{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 95.60229445506693, "eval_steps": 500, "global_step": 150000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.006373486297004461, "grad_norm": 9.50260066986084, "learning_rate": 6e-08, "loss": 0.8141, "step": 10 }, { "epoch": 0.012746972594008922, "grad_norm": 11.740574836730957, "learning_rate": 1.2666666666666666e-07, "loss": 0.8284, "step": 20 }, { "epoch": 0.019120458891013385, "grad_norm": 16.99808692932129, "learning_rate": 1.9333333333333334e-07, "loss": 0.8016, "step": 30 }, { "epoch": 0.025493945188017845, "grad_norm": 11.063021659851074, "learning_rate": 2.6e-07, "loss": 0.8329, "step": 40 }, { "epoch": 0.03186743148502231, "grad_norm": 18.429140090942383, "learning_rate": 3.266666666666667e-07, "loss": 0.9546, "step": 50 }, { "epoch": 0.03824091778202677, "grad_norm": 10.585908889770508, "learning_rate": 3.933333333333333e-07, "loss": 0.8555, "step": 60 }, { "epoch": 0.04461440407903123, "grad_norm": 13.518843650817871, "learning_rate": 4.6e-07, "loss": 0.8202, "step": 70 }, { "epoch": 0.05098789037603569, "grad_norm": 8.653311729431152, "learning_rate": 5.266666666666667e-07, "loss": 0.6907, "step": 80 }, { "epoch": 0.05736137667304015, "grad_norm": 9.918243408203125, "learning_rate": 5.933333333333333e-07, "loss": 0.5854, "step": 90 }, { "epoch": 0.06373486297004462, "grad_norm": 6.905433654785156, "learning_rate": 6.6e-07, "loss": 0.5492, "step": 100 }, { "epoch": 0.07010834926704908, "grad_norm": 3.717879056930542, "learning_rate": 7.266666666666668e-07, "loss": 0.4975, "step": 110 }, { "epoch": 0.07648183556405354, "grad_norm": 4.2783684730529785, "learning_rate": 7.933333333333334e-07, "loss": 0.3872, "step": 120 }, { "epoch": 0.082855321861058, "grad_norm": 2.390763998031616, "learning_rate": 8.6e-07, "loss": 0.2653, "step": 130 }, { "epoch": 0.08922880815806246, "grad_norm": 2.8818657398223877, "learning_rate": 9.266666666666667e-07, "loss": 0.2991, "step": 140 }, { "epoch": 0.09560229445506692, "grad_norm": 2.502659797668457, "learning_rate": 9.933333333333335e-07, "loss": 0.2696, "step": 150 }, { "epoch": 0.10197578075207138, "grad_norm": 1.6304689645767212, "learning_rate": 1.06e-06, "loss": 0.2705, "step": 160 }, { "epoch": 0.10834926704907584, "grad_norm": 1.7764955759048462, "learning_rate": 1.1266666666666667e-06, "loss": 0.1726, "step": 170 }, { "epoch": 0.1147227533460803, "grad_norm": 2.821688652038574, "learning_rate": 1.1933333333333335e-06, "loss": 0.2019, "step": 180 }, { "epoch": 0.12109623964308477, "grad_norm": 2.787238836288452, "learning_rate": 1.26e-06, "loss": 0.2024, "step": 190 }, { "epoch": 0.12746972594008923, "grad_norm": 2.0788207054138184, "learning_rate": 1.3266666666666667e-06, "loss": 0.2077, "step": 200 }, { "epoch": 0.1338432122370937, "grad_norm": 2.0954220294952393, "learning_rate": 1.3933333333333335e-06, "loss": 0.1764, "step": 210 }, { "epoch": 0.14021669853409816, "grad_norm": 1.7847673892974854, "learning_rate": 1.46e-06, "loss": 0.1948, "step": 220 }, { "epoch": 0.14659018483110262, "grad_norm": 1.7281008958816528, "learning_rate": 1.5266666666666667e-06, "loss": 0.1823, "step": 230 }, { "epoch": 0.15296367112810708, "grad_norm": 2.379450798034668, "learning_rate": 1.5933333333333335e-06, "loss": 0.1772, "step": 240 }, { "epoch": 0.15933715742511154, "grad_norm": 2.32367205619812, "learning_rate": 1.6600000000000002e-06, "loss": 0.1728, "step": 250 }, { "epoch": 0.165710643722116, "grad_norm": 2.5996227264404297, "learning_rate": 1.7266666666666667e-06, "loss": 0.172, "step": 260 }, { "epoch": 0.17208413001912046, "grad_norm": 2.1081321239471436, "learning_rate": 1.7933333333333332e-06, "loss": 0.1574, "step": 270 }, { "epoch": 0.17845761631612492, "grad_norm": 3.374861478805542, "learning_rate": 1.86e-06, "loss": 0.1547, "step": 280 }, { "epoch": 0.18483110261312938, "grad_norm": 2.300109624862671, "learning_rate": 1.9266666666666667e-06, "loss": 0.1748, "step": 290 }, { "epoch": 0.19120458891013384, "grad_norm": 2.2415449619293213, "learning_rate": 1.9933333333333334e-06, "loss": 0.1773, "step": 300 }, { "epoch": 0.1975780752071383, "grad_norm": 2.5365710258483887, "learning_rate": 2.06e-06, "loss": 0.1671, "step": 310 }, { "epoch": 0.20395156150414276, "grad_norm": 2.430039167404175, "learning_rate": 2.126666666666667e-06, "loss": 0.1341, "step": 320 }, { "epoch": 0.21032504780114722, "grad_norm": 2.3952574729919434, "learning_rate": 2.1933333333333332e-06, "loss": 0.1666, "step": 330 }, { "epoch": 0.21669853409815168, "grad_norm": 2.762911796569824, "learning_rate": 2.26e-06, "loss": 0.1726, "step": 340 }, { "epoch": 0.22307202039515614, "grad_norm": 2.1954283714294434, "learning_rate": 2.326666666666667e-06, "loss": 0.1511, "step": 350 }, { "epoch": 0.2294455066921606, "grad_norm": 2.528956890106201, "learning_rate": 2.3933333333333334e-06, "loss": 0.151, "step": 360 }, { "epoch": 0.23581899298916506, "grad_norm": 2.1332144737243652, "learning_rate": 2.46e-06, "loss": 0.158, "step": 370 }, { "epoch": 0.24219247928616955, "grad_norm": 1.6394823789596558, "learning_rate": 2.526666666666667e-06, "loss": 0.1562, "step": 380 }, { "epoch": 0.248565965583174, "grad_norm": 1.5594468116760254, "learning_rate": 2.593333333333333e-06, "loss": 0.1625, "step": 390 }, { "epoch": 0.25493945188017847, "grad_norm": 2.135986089706421, "learning_rate": 2.66e-06, "loss": 0.1656, "step": 400 }, { "epoch": 0.26131293817718293, "grad_norm": 1.1881332397460938, "learning_rate": 2.726666666666667e-06, "loss": 0.1632, "step": 410 }, { "epoch": 0.2676864244741874, "grad_norm": 2.092585802078247, "learning_rate": 2.7933333333333334e-06, "loss": 0.1415, "step": 420 }, { "epoch": 0.27405991077119185, "grad_norm": 1.8596680164337158, "learning_rate": 2.86e-06, "loss": 0.1776, "step": 430 }, { "epoch": 0.2804333970681963, "grad_norm": 2.1945323944091797, "learning_rate": 2.926666666666667e-06, "loss": 0.1672, "step": 440 }, { "epoch": 0.28680688336520077, "grad_norm": 0.947393536567688, "learning_rate": 2.9933333333333336e-06, "loss": 0.1229, "step": 450 }, { "epoch": 0.29318036966220523, "grad_norm": 3.158277750015259, "learning_rate": 3.06e-06, "loss": 0.138, "step": 460 }, { "epoch": 0.2995538559592097, "grad_norm": 1.4556961059570312, "learning_rate": 3.1266666666666667e-06, "loss": 0.1444, "step": 470 }, { "epoch": 0.30592734225621415, "grad_norm": 1.6013380289077759, "learning_rate": 3.1933333333333334e-06, "loss": 0.1618, "step": 480 }, { "epoch": 0.3123008285532186, "grad_norm": 1.6442179679870605, "learning_rate": 3.2599999999999997e-06, "loss": 0.1677, "step": 490 }, { "epoch": 0.3186743148502231, "grad_norm": 3.0498273372650146, "learning_rate": 3.326666666666667e-06, "loss": 0.1414, "step": 500 }, { "epoch": 0.32504780114722753, "grad_norm": 1.5951189994812012, "learning_rate": 3.3933333333333336e-06, "loss": 0.1558, "step": 510 }, { "epoch": 0.331421287444232, "grad_norm": 2.0183534622192383, "learning_rate": 3.46e-06, "loss": 0.1464, "step": 520 }, { "epoch": 0.33779477374123645, "grad_norm": 2.1769559383392334, "learning_rate": 3.526666666666667e-06, "loss": 0.1649, "step": 530 }, { "epoch": 0.3441682600382409, "grad_norm": 2.2509684562683105, "learning_rate": 3.5933333333333334e-06, "loss": 0.1371, "step": 540 }, { "epoch": 0.3505417463352454, "grad_norm": 1.3416584730148315, "learning_rate": 3.66e-06, "loss": 0.1477, "step": 550 }, { "epoch": 0.35691523263224983, "grad_norm": 1.9828554391860962, "learning_rate": 3.7266666666666664e-06, "loss": 0.1613, "step": 560 }, { "epoch": 0.3632887189292543, "grad_norm": 1.1496295928955078, "learning_rate": 3.7933333333333336e-06, "loss": 0.1274, "step": 570 }, { "epoch": 0.36966220522625876, "grad_norm": 1.9544023275375366, "learning_rate": 3.86e-06, "loss": 0.1302, "step": 580 }, { "epoch": 0.3760356915232632, "grad_norm": 1.5555181503295898, "learning_rate": 3.926666666666667e-06, "loss": 0.1251, "step": 590 }, { "epoch": 0.3824091778202677, "grad_norm": 1.9086092710494995, "learning_rate": 3.993333333333334e-06, "loss": 0.1565, "step": 600 }, { "epoch": 0.38878266411727214, "grad_norm": 2.619568109512329, "learning_rate": 4.06e-06, "loss": 0.1337, "step": 610 }, { "epoch": 0.3951561504142766, "grad_norm": 1.56656813621521, "learning_rate": 4.126666666666667e-06, "loss": 0.1366, "step": 620 }, { "epoch": 0.40152963671128106, "grad_norm": 2.150005340576172, "learning_rate": 4.1933333333333336e-06, "loss": 0.1543, "step": 630 }, { "epoch": 0.4079031230082855, "grad_norm": 1.9915505647659302, "learning_rate": 4.26e-06, "loss": 0.1369, "step": 640 }, { "epoch": 0.41427660930529, "grad_norm": 1.766042709350586, "learning_rate": 4.326666666666667e-06, "loss": 0.1255, "step": 650 }, { "epoch": 0.42065009560229444, "grad_norm": 2.2898106575012207, "learning_rate": 4.393333333333333e-06, "loss": 0.1621, "step": 660 }, { "epoch": 0.4270235818992989, "grad_norm": 2.1378886699676514, "learning_rate": 4.4600000000000005e-06, "loss": 0.1411, "step": 670 }, { "epoch": 0.43339706819630336, "grad_norm": 3.2848191261291504, "learning_rate": 4.526666666666667e-06, "loss": 0.1481, "step": 680 }, { "epoch": 0.4397705544933078, "grad_norm": 1.6556960344314575, "learning_rate": 4.593333333333333e-06, "loss": 0.1452, "step": 690 }, { "epoch": 0.4461440407903123, "grad_norm": 2.0749261379241943, "learning_rate": 4.66e-06, "loss": 0.1507, "step": 700 }, { "epoch": 0.45251752708731674, "grad_norm": 2.629366397857666, "learning_rate": 4.726666666666667e-06, "loss": 0.1446, "step": 710 }, { "epoch": 0.4588910133843212, "grad_norm": 1.869321584701538, "learning_rate": 4.793333333333334e-06, "loss": 0.1481, "step": 720 }, { "epoch": 0.46526449968132566, "grad_norm": 1.557539939880371, "learning_rate": 4.86e-06, "loss": 0.1342, "step": 730 }, { "epoch": 0.4716379859783301, "grad_norm": 1.2169694900512695, "learning_rate": 4.926666666666667e-06, "loss": 0.113, "step": 740 }, { "epoch": 0.4780114722753346, "grad_norm": 1.5026923418045044, "learning_rate": 4.9933333333333335e-06, "loss": 0.1396, "step": 750 }, { "epoch": 0.4843849585723391, "grad_norm": 1.6789538860321045, "learning_rate": 5.06e-06, "loss": 0.1273, "step": 760 }, { "epoch": 0.49075844486934356, "grad_norm": 1.173251986503601, "learning_rate": 5.126666666666667e-06, "loss": 0.1433, "step": 770 }, { "epoch": 0.497131931166348, "grad_norm": 2.4430761337280273, "learning_rate": 5.193333333333333e-06, "loss": 0.1631, "step": 780 }, { "epoch": 0.5035054174633524, "grad_norm": 1.9218428134918213, "learning_rate": 5.2600000000000005e-06, "loss": 0.1375, "step": 790 }, { "epoch": 0.5098789037603569, "grad_norm": 1.734273910522461, "learning_rate": 5.326666666666667e-06, "loss": 0.1482, "step": 800 }, { "epoch": 0.5162523900573613, "grad_norm": 1.686924695968628, "learning_rate": 5.393333333333334e-06, "loss": 0.1378, "step": 810 }, { "epoch": 0.5226258763543659, "grad_norm": 1.936724066734314, "learning_rate": 5.46e-06, "loss": 0.1444, "step": 820 }, { "epoch": 0.5289993626513703, "grad_norm": 1.8305740356445312, "learning_rate": 5.5266666666666666e-06, "loss": 0.156, "step": 830 }, { "epoch": 0.5353728489483748, "grad_norm": 1.2125890254974365, "learning_rate": 5.593333333333334e-06, "loss": 0.1254, "step": 840 }, { "epoch": 0.5417463352453792, "grad_norm": 1.9790780544281006, "learning_rate": 5.66e-06, "loss": 0.1397, "step": 850 }, { "epoch": 0.5481198215423837, "grad_norm": 1.9570019245147705, "learning_rate": 5.726666666666667e-06, "loss": 0.1421, "step": 860 }, { "epoch": 0.5544933078393881, "grad_norm": 1.190658450126648, "learning_rate": 5.793333333333334e-06, "loss": 0.1328, "step": 870 }, { "epoch": 0.5608667941363926, "grad_norm": 2.010284423828125, "learning_rate": 5.86e-06, "loss": 0.1241, "step": 880 }, { "epoch": 0.567240280433397, "grad_norm": 1.7606861591339111, "learning_rate": 5.926666666666667e-06, "loss": 0.132, "step": 890 }, { "epoch": 0.5736137667304015, "grad_norm": 0.8547699451446533, "learning_rate": 5.993333333333333e-06, "loss": 0.1231, "step": 900 }, { "epoch": 0.579987253027406, "grad_norm": 1.2310320138931274, "learning_rate": 6.0600000000000004e-06, "loss": 0.1247, "step": 910 }, { "epoch": 0.5863607393244105, "grad_norm": 1.368887186050415, "learning_rate": 6.126666666666667e-06, "loss": 0.1196, "step": 920 }, { "epoch": 0.5927342256214149, "grad_norm": 1.5329285860061646, "learning_rate": 6.193333333333334e-06, "loss": 0.1284, "step": 930 }, { "epoch": 0.5991077119184194, "grad_norm": 1.4317708015441895, "learning_rate": 6.26e-06, "loss": 0.1322, "step": 940 }, { "epoch": 0.6054811982154238, "grad_norm": 1.296454668045044, "learning_rate": 6.3266666666666665e-06, "loss": 0.1133, "step": 950 }, { "epoch": 0.6118546845124283, "grad_norm": 1.8704478740692139, "learning_rate": 6.393333333333333e-06, "loss": 0.131, "step": 960 }, { "epoch": 0.6182281708094327, "grad_norm": 2.585780382156372, "learning_rate": 6.460000000000001e-06, "loss": 0.1445, "step": 970 }, { "epoch": 0.6246016571064372, "grad_norm": 1.1671991348266602, "learning_rate": 6.526666666666667e-06, "loss": 0.1135, "step": 980 }, { "epoch": 0.6309751434034416, "grad_norm": 1.8688873052597046, "learning_rate": 6.5933333333333335e-06, "loss": 0.1204, "step": 990 }, { "epoch": 0.6373486297004461, "grad_norm": 2.5886378288269043, "learning_rate": 6.660000000000001e-06, "loss": 0.135, "step": 1000 }, { "epoch": 0.6437221159974506, "grad_norm": 2.5049116611480713, "learning_rate": 6.726666666666667e-06, "loss": 0.1513, "step": 1010 }, { "epoch": 0.6500956022944551, "grad_norm": 1.8935885429382324, "learning_rate": 6.793333333333333e-06, "loss": 0.1421, "step": 1020 }, { "epoch": 0.6564690885914596, "grad_norm": 1.5030502080917358, "learning_rate": 6.8599999999999995e-06, "loss": 0.11, "step": 1030 }, { "epoch": 0.662842574888464, "grad_norm": 1.3139643669128418, "learning_rate": 6.9266666666666675e-06, "loss": 0.1206, "step": 1040 }, { "epoch": 0.6692160611854685, "grad_norm": 1.3345052003860474, "learning_rate": 6.993333333333334e-06, "loss": 0.1156, "step": 1050 }, { "epoch": 0.6755895474824729, "grad_norm": 1.3748908042907715, "learning_rate": 7.06e-06, "loss": 0.1339, "step": 1060 }, { "epoch": 0.6819630337794774, "grad_norm": 1.4622594118118286, "learning_rate": 7.126666666666667e-06, "loss": 0.1259, "step": 1070 }, { "epoch": 0.6883365200764818, "grad_norm": 1.5618497133255005, "learning_rate": 7.193333333333334e-06, "loss": 0.1482, "step": 1080 }, { "epoch": 0.6947100063734863, "grad_norm": 1.174689769744873, "learning_rate": 7.26e-06, "loss": 0.1198, "step": 1090 }, { "epoch": 0.7010834926704907, "grad_norm": 1.6729302406311035, "learning_rate": 7.326666666666666e-06, "loss": 0.1109, "step": 1100 }, { "epoch": 0.7074569789674953, "grad_norm": 1.5087976455688477, "learning_rate": 7.393333333333334e-06, "loss": 0.1275, "step": 1110 }, { "epoch": 0.7138304652644997, "grad_norm": 1.3268859386444092, "learning_rate": 7.4600000000000006e-06, "loss": 0.1097, "step": 1120 }, { "epoch": 0.7202039515615042, "grad_norm": 1.0452150106430054, "learning_rate": 7.526666666666667e-06, "loss": 0.1268, "step": 1130 }, { "epoch": 0.7265774378585086, "grad_norm": 1.128790020942688, "learning_rate": 7.593333333333334e-06, "loss": 0.148, "step": 1140 }, { "epoch": 0.7329509241555131, "grad_norm": 2.143430233001709, "learning_rate": 7.660000000000001e-06, "loss": 0.1335, "step": 1150 }, { "epoch": 0.7393244104525175, "grad_norm": 2.210578203201294, "learning_rate": 7.726666666666667e-06, "loss": 0.1194, "step": 1160 }, { "epoch": 0.745697896749522, "grad_norm": 1.5607821941375732, "learning_rate": 7.793333333333334e-06, "loss": 0.1258, "step": 1170 }, { "epoch": 0.7520713830465264, "grad_norm": 1.99208402633667, "learning_rate": 7.860000000000001e-06, "loss": 0.1021, "step": 1180 }, { "epoch": 0.758444869343531, "grad_norm": 1.4818758964538574, "learning_rate": 7.926666666666666e-06, "loss": 0.1228, "step": 1190 }, { "epoch": 0.7648183556405354, "grad_norm": 2.065819501876831, "learning_rate": 7.993333333333334e-06, "loss": 0.1289, "step": 1200 }, { "epoch": 0.7711918419375399, "grad_norm": 2.015576124191284, "learning_rate": 8.06e-06, "loss": 0.137, "step": 1210 }, { "epoch": 0.7775653282345443, "grad_norm": 1.9455817937850952, "learning_rate": 8.126666666666668e-06, "loss": 0.1242, "step": 1220 }, { "epoch": 0.7839388145315488, "grad_norm": 1.1709002256393433, "learning_rate": 8.193333333333333e-06, "loss": 0.1233, "step": 1230 }, { "epoch": 0.7903123008285532, "grad_norm": 1.6992369890213013, "learning_rate": 8.26e-06, "loss": 0.1131, "step": 1240 }, { "epoch": 0.7966857871255577, "grad_norm": 1.7543067932128906, "learning_rate": 8.326666666666668e-06, "loss": 0.1021, "step": 1250 }, { "epoch": 0.8030592734225621, "grad_norm": 1.9983257055282593, "learning_rate": 8.393333333333333e-06, "loss": 0.1187, "step": 1260 }, { "epoch": 0.8094327597195666, "grad_norm": 1.5104361772537231, "learning_rate": 8.46e-06, "loss": 0.1122, "step": 1270 }, { "epoch": 0.815806246016571, "grad_norm": 1.4873898029327393, "learning_rate": 8.526666666666667e-06, "loss": 0.1176, "step": 1280 }, { "epoch": 0.8221797323135756, "grad_norm": 2.229330062866211, "learning_rate": 8.593333333333335e-06, "loss": 0.1139, "step": 1290 }, { "epoch": 0.82855321861058, "grad_norm": 1.1514588594436646, "learning_rate": 8.66e-06, "loss": 0.127, "step": 1300 }, { "epoch": 0.8349267049075845, "grad_norm": 1.0396504402160645, "learning_rate": 8.726666666666667e-06, "loss": 0.1243, "step": 1310 }, { "epoch": 0.8413001912045889, "grad_norm": 1.17670738697052, "learning_rate": 8.793333333333334e-06, "loss": 0.1207, "step": 1320 }, { "epoch": 0.8476736775015934, "grad_norm": 1.0504316091537476, "learning_rate": 8.86e-06, "loss": 0.1064, "step": 1330 }, { "epoch": 0.8540471637985978, "grad_norm": 1.0945193767547607, "learning_rate": 8.926666666666667e-06, "loss": 0.1051, "step": 1340 }, { "epoch": 0.8604206500956023, "grad_norm": 0.6964119672775269, "learning_rate": 8.993333333333334e-06, "loss": 0.1262, "step": 1350 }, { "epoch": 0.8667941363926067, "grad_norm": 2.0203912258148193, "learning_rate": 9.06e-06, "loss": 0.1269, "step": 1360 }, { "epoch": 0.8731676226896112, "grad_norm": 2.3909099102020264, "learning_rate": 9.126666666666667e-06, "loss": 0.1114, "step": 1370 }, { "epoch": 0.8795411089866156, "grad_norm": 1.132116436958313, "learning_rate": 9.193333333333334e-06, "loss": 0.1202, "step": 1380 }, { "epoch": 0.8859145952836202, "grad_norm": 1.6137852668762207, "learning_rate": 9.260000000000001e-06, "loss": 0.1088, "step": 1390 }, { "epoch": 0.8922880815806246, "grad_norm": 2.0166521072387695, "learning_rate": 9.326666666666667e-06, "loss": 0.1014, "step": 1400 }, { "epoch": 0.8986615678776291, "grad_norm": 1.089432954788208, "learning_rate": 9.393333333333334e-06, "loss": 0.1186, "step": 1410 }, { "epoch": 0.9050350541746335, "grad_norm": 1.5007840394973755, "learning_rate": 9.460000000000001e-06, "loss": 0.1165, "step": 1420 }, { "epoch": 0.911408540471638, "grad_norm": 1.334079384803772, "learning_rate": 9.526666666666666e-06, "loss": 0.0929, "step": 1430 }, { "epoch": 0.9177820267686424, "grad_norm": 2.3709166049957275, "learning_rate": 9.593333333333334e-06, "loss": 0.1198, "step": 1440 }, { "epoch": 0.9241555130656469, "grad_norm": 1.677902102470398, "learning_rate": 9.66e-06, "loss": 0.1239, "step": 1450 }, { "epoch": 0.9305289993626513, "grad_norm": 1.6223987340927124, "learning_rate": 9.726666666666668e-06, "loss": 0.1049, "step": 1460 }, { "epoch": 0.9369024856596558, "grad_norm": 1.6598254442214966, "learning_rate": 9.793333333333333e-06, "loss": 0.1128, "step": 1470 }, { "epoch": 0.9432759719566602, "grad_norm": 2.6494898796081543, "learning_rate": 9.86e-06, "loss": 0.1157, "step": 1480 }, { "epoch": 0.9496494582536648, "grad_norm": 1.5539672374725342, "learning_rate": 9.926666666666668e-06, "loss": 0.1048, "step": 1490 }, { "epoch": 0.9560229445506692, "grad_norm": 2.537384271621704, "learning_rate": 9.993333333333333e-06, "loss": 0.1218, "step": 1500 }, { "epoch": 0.9623964308476737, "grad_norm": 1.366798996925354, "learning_rate": 1.006e-05, "loss": 0.1126, "step": 1510 }, { "epoch": 0.9687699171446782, "grad_norm": 1.446845293045044, "learning_rate": 1.0126666666666667e-05, "loss": 0.1201, "step": 1520 }, { "epoch": 0.9751434034416826, "grad_norm": 1.1611850261688232, "learning_rate": 1.0193333333333335e-05, "loss": 0.1075, "step": 1530 }, { "epoch": 0.9815168897386871, "grad_norm": 1.2011315822601318, "learning_rate": 1.026e-05, "loss": 0.1019, "step": 1540 }, { "epoch": 0.9878903760356915, "grad_norm": 2.2322311401367188, "learning_rate": 1.0326666666666667e-05, "loss": 0.1226, "step": 1550 }, { "epoch": 0.994263862332696, "grad_norm": 0.8722662329673767, "learning_rate": 1.0393333333333334e-05, "loss": 0.113, "step": 1560 }, { "epoch": 1.0006373486297004, "grad_norm": 1.3424410820007324, "learning_rate": 1.046e-05, "loss": 0.0845, "step": 1570 }, { "epoch": 1.0070108349267048, "grad_norm": 1.2302227020263672, "learning_rate": 1.0526666666666667e-05, "loss": 0.0988, "step": 1580 }, { "epoch": 1.0133843212237095, "grad_norm": 1.23185396194458, "learning_rate": 1.0593333333333334e-05, "loss": 0.1074, "step": 1590 }, { "epoch": 1.0197578075207139, "grad_norm": 1.995958685874939, "learning_rate": 1.0660000000000001e-05, "loss": 0.1182, "step": 1600 }, { "epoch": 1.0261312938177183, "grad_norm": 0.5706260800361633, "learning_rate": 1.0726666666666667e-05, "loss": 0.1015, "step": 1610 }, { "epoch": 1.0325047801147227, "grad_norm": 1.0292627811431885, "learning_rate": 1.0793333333333334e-05, "loss": 0.1, "step": 1620 }, { "epoch": 1.0388782664117273, "grad_norm": 1.7296640872955322, "learning_rate": 1.0860000000000001e-05, "loss": 0.1326, "step": 1630 }, { "epoch": 1.0452517527087317, "grad_norm": 1.43869948387146, "learning_rate": 1.0926666666666667e-05, "loss": 0.1292, "step": 1640 }, { "epoch": 1.0516252390057361, "grad_norm": 1.0628490447998047, "learning_rate": 1.0993333333333334e-05, "loss": 0.1034, "step": 1650 }, { "epoch": 1.0579987253027405, "grad_norm": 1.5121562480926514, "learning_rate": 1.106e-05, "loss": 0.1076, "step": 1660 }, { "epoch": 1.0643722115997452, "grad_norm": 1.4332911968231201, "learning_rate": 1.1126666666666668e-05, "loss": 0.0942, "step": 1670 }, { "epoch": 1.0707456978967496, "grad_norm": 1.0733493566513062, "learning_rate": 1.1193333333333333e-05, "loss": 0.103, "step": 1680 }, { "epoch": 1.077119184193754, "grad_norm": 1.2125245332717896, "learning_rate": 1.126e-05, "loss": 0.1027, "step": 1690 }, { "epoch": 1.0834926704907584, "grad_norm": 1.374131202697754, "learning_rate": 1.1326666666666668e-05, "loss": 0.1063, "step": 1700 }, { "epoch": 1.089866156787763, "grad_norm": 1.331972360610962, "learning_rate": 1.1393333333333333e-05, "loss": 0.1032, "step": 1710 }, { "epoch": 1.0962396430847674, "grad_norm": 1.8142362833023071, "learning_rate": 1.146e-05, "loss": 0.1147, "step": 1720 }, { "epoch": 1.1026131293817718, "grad_norm": 1.474809169769287, "learning_rate": 1.1526666666666668e-05, "loss": 0.1012, "step": 1730 }, { "epoch": 1.1089866156787762, "grad_norm": 1.0845487117767334, "learning_rate": 1.1593333333333333e-05, "loss": 0.1036, "step": 1740 }, { "epoch": 1.1153601019757808, "grad_norm": 1.7612701654434204, "learning_rate": 1.166e-05, "loss": 0.1063, "step": 1750 }, { "epoch": 1.1217335882727852, "grad_norm": 0.7480101585388184, "learning_rate": 1.1726666666666667e-05, "loss": 0.1016, "step": 1760 }, { "epoch": 1.1281070745697896, "grad_norm": 0.9497768878936768, "learning_rate": 1.1793333333333334e-05, "loss": 0.0917, "step": 1770 }, { "epoch": 1.134480560866794, "grad_norm": 1.1517006158828735, "learning_rate": 1.186e-05, "loss": 0.1166, "step": 1780 }, { "epoch": 1.1408540471637987, "grad_norm": 1.1664561033248901, "learning_rate": 1.1926666666666667e-05, "loss": 0.0917, "step": 1790 }, { "epoch": 1.147227533460803, "grad_norm": 2.021282434463501, "learning_rate": 1.1993333333333334e-05, "loss": 0.1115, "step": 1800 }, { "epoch": 1.1536010197578075, "grad_norm": 1.5622092485427856, "learning_rate": 1.206e-05, "loss": 0.1136, "step": 1810 }, { "epoch": 1.159974506054812, "grad_norm": 1.4180028438568115, "learning_rate": 1.2126666666666667e-05, "loss": 0.1024, "step": 1820 }, { "epoch": 1.1663479923518165, "grad_norm": 2.3993589878082275, "learning_rate": 1.2193333333333334e-05, "loss": 0.1037, "step": 1830 }, { "epoch": 1.172721478648821, "grad_norm": 1.6957080364227295, "learning_rate": 1.2260000000000001e-05, "loss": 0.0915, "step": 1840 }, { "epoch": 1.1790949649458253, "grad_norm": 1.1840903759002686, "learning_rate": 1.2326666666666667e-05, "loss": 0.1119, "step": 1850 }, { "epoch": 1.1854684512428297, "grad_norm": 1.1787384748458862, "learning_rate": 1.2393333333333334e-05, "loss": 0.1034, "step": 1860 }, { "epoch": 1.1918419375398344, "grad_norm": 1.8163771629333496, "learning_rate": 1.2460000000000001e-05, "loss": 0.1029, "step": 1870 }, { "epoch": 1.1982154238368388, "grad_norm": 1.1066467761993408, "learning_rate": 1.2526666666666666e-05, "loss": 0.086, "step": 1880 }, { "epoch": 1.2045889101338432, "grad_norm": 1.2255226373672485, "learning_rate": 1.2593333333333335e-05, "loss": 0.1021, "step": 1890 }, { "epoch": 1.2109623964308476, "grad_norm": 1.2998727560043335, "learning_rate": 1.2659999999999999e-05, "loss": 0.0968, "step": 1900 }, { "epoch": 1.2173358827278522, "grad_norm": 1.1388373374938965, "learning_rate": 1.2726666666666668e-05, "loss": 0.1145, "step": 1910 }, { "epoch": 1.2237093690248566, "grad_norm": 2.0356638431549072, "learning_rate": 1.2793333333333335e-05, "loss": 0.1188, "step": 1920 }, { "epoch": 1.230082855321861, "grad_norm": 1.5201404094696045, "learning_rate": 1.286e-05, "loss": 0.0939, "step": 1930 }, { "epoch": 1.2364563416188656, "grad_norm": 2.302424907684326, "learning_rate": 1.2926666666666668e-05, "loss": 0.1095, "step": 1940 }, { "epoch": 1.24282982791587, "grad_norm": 1.1126807928085327, "learning_rate": 1.2993333333333335e-05, "loss": 0.0864, "step": 1950 }, { "epoch": 1.2492033142128744, "grad_norm": 1.4584306478500366, "learning_rate": 1.306e-05, "loss": 0.0968, "step": 1960 }, { "epoch": 1.2555768005098789, "grad_norm": 1.5292973518371582, "learning_rate": 1.3126666666666667e-05, "loss": 0.0891, "step": 1970 }, { "epoch": 1.2619502868068833, "grad_norm": 1.7257049083709717, "learning_rate": 1.3193333333333335e-05, "loss": 0.1062, "step": 1980 }, { "epoch": 1.2683237731038879, "grad_norm": 0.7163188457489014, "learning_rate": 1.326e-05, "loss": 0.1001, "step": 1990 }, { "epoch": 1.2746972594008923, "grad_norm": 0.9911763072013855, "learning_rate": 1.3326666666666667e-05, "loss": 0.1065, "step": 2000 }, { "epoch": 1.2810707456978967, "grad_norm": 1.2415435314178467, "learning_rate": 1.3393333333333333e-05, "loss": 0.104, "step": 2010 }, { "epoch": 1.2874442319949013, "grad_norm": 1.0172119140625, "learning_rate": 1.346e-05, "loss": 0.0911, "step": 2020 }, { "epoch": 1.2938177182919057, "grad_norm": 1.5576133728027344, "learning_rate": 1.3526666666666669e-05, "loss": 0.0859, "step": 2030 }, { "epoch": 1.3001912045889101, "grad_norm": 1.4438010454177856, "learning_rate": 1.3593333333333332e-05, "loss": 0.1007, "step": 2040 }, { "epoch": 1.3065646908859145, "grad_norm": 1.0720605850219727, "learning_rate": 1.3660000000000001e-05, "loss": 0.0864, "step": 2050 }, { "epoch": 1.312938177182919, "grad_norm": 1.5696890354156494, "learning_rate": 1.3726666666666669e-05, "loss": 0.0873, "step": 2060 }, { "epoch": 1.3193116634799236, "grad_norm": 1.2315691709518433, "learning_rate": 1.3793333333333332e-05, "loss": 0.1094, "step": 2070 }, { "epoch": 1.325685149776928, "grad_norm": 1.1183645725250244, "learning_rate": 1.3860000000000001e-05, "loss": 0.0857, "step": 2080 }, { "epoch": 1.3320586360739324, "grad_norm": 1.6838148832321167, "learning_rate": 1.3926666666666668e-05, "loss": 0.1003, "step": 2090 }, { "epoch": 1.338432122370937, "grad_norm": 1.1183300018310547, "learning_rate": 1.3993333333333334e-05, "loss": 0.0831, "step": 2100 }, { "epoch": 1.3448056086679414, "grad_norm": 1.8366953134536743, "learning_rate": 1.4060000000000001e-05, "loss": 0.111, "step": 2110 }, { "epoch": 1.3511790949649458, "grad_norm": 1.9242589473724365, "learning_rate": 1.4126666666666668e-05, "loss": 0.1073, "step": 2120 }, { "epoch": 1.3575525812619502, "grad_norm": 1.864620327949524, "learning_rate": 1.4193333333333334e-05, "loss": 0.0901, "step": 2130 }, { "epoch": 1.3639260675589546, "grad_norm": 1.115146279335022, "learning_rate": 1.426e-05, "loss": 0.0825, "step": 2140 }, { "epoch": 1.3702995538559593, "grad_norm": 1.6277066469192505, "learning_rate": 1.4326666666666666e-05, "loss": 0.1161, "step": 2150 }, { "epoch": 1.3766730401529637, "grad_norm": 1.6365883350372314, "learning_rate": 1.4393333333333333e-05, "loss": 0.0848, "step": 2160 }, { "epoch": 1.383046526449968, "grad_norm": 1.2011946439743042, "learning_rate": 1.4460000000000002e-05, "loss": 0.0996, "step": 2170 }, { "epoch": 1.3894200127469727, "grad_norm": 1.8417644500732422, "learning_rate": 1.4526666666666666e-05, "loss": 0.1142, "step": 2180 }, { "epoch": 1.395793499043977, "grad_norm": 1.2948949337005615, "learning_rate": 1.4593333333333333e-05, "loss": 0.1095, "step": 2190 }, { "epoch": 1.4021669853409815, "grad_norm": 1.2903964519500732, "learning_rate": 1.4660000000000002e-05, "loss": 0.0864, "step": 2200 }, { "epoch": 1.408540471637986, "grad_norm": 1.643641710281372, "learning_rate": 1.4726666666666666e-05, "loss": 0.094, "step": 2210 }, { "epoch": 1.4149139579349903, "grad_norm": 1.1676900386810303, "learning_rate": 1.4793333333333335e-05, "loss": 0.0972, "step": 2220 }, { "epoch": 1.421287444231995, "grad_norm": 0.9590888023376465, "learning_rate": 1.4860000000000002e-05, "loss": 0.0826, "step": 2230 }, { "epoch": 1.4276609305289993, "grad_norm": 1.325158953666687, "learning_rate": 1.4926666666666667e-05, "loss": 0.1018, "step": 2240 }, { "epoch": 1.4340344168260037, "grad_norm": 0.6889841556549072, "learning_rate": 1.4993333333333334e-05, "loss": 0.0875, "step": 2250 }, { "epoch": 1.4404079031230084, "grad_norm": 1.7010300159454346, "learning_rate": 1.5060000000000001e-05, "loss": 0.088, "step": 2260 }, { "epoch": 1.4467813894200128, "grad_norm": 0.9164239168167114, "learning_rate": 1.5126666666666667e-05, "loss": 0.0886, "step": 2270 }, { "epoch": 1.4531548757170172, "grad_norm": 1.0260887145996094, "learning_rate": 1.5193333333333334e-05, "loss": 0.0828, "step": 2280 }, { "epoch": 1.4595283620140216, "grad_norm": 1.2509478330612183, "learning_rate": 1.5260000000000003e-05, "loss": 0.1153, "step": 2290 }, { "epoch": 1.4659018483110262, "grad_norm": 1.040419578552246, "learning_rate": 1.5326666666666667e-05, "loss": 0.0992, "step": 2300 }, { "epoch": 1.4722753346080306, "grad_norm": 1.8140629529953003, "learning_rate": 1.5393333333333334e-05, "loss": 0.0923, "step": 2310 }, { "epoch": 1.478648820905035, "grad_norm": 1.4857326745986938, "learning_rate": 1.546e-05, "loss": 0.0777, "step": 2320 }, { "epoch": 1.4850223072020396, "grad_norm": 1.146477460861206, "learning_rate": 1.5526666666666668e-05, "loss": 0.1113, "step": 2330 }, { "epoch": 1.491395793499044, "grad_norm": 1.3774482011795044, "learning_rate": 1.5593333333333335e-05, "loss": 0.0964, "step": 2340 }, { "epoch": 1.4977692797960485, "grad_norm": 0.9920369982719421, "learning_rate": 1.566e-05, "loss": 0.0965, "step": 2350 }, { "epoch": 1.5041427660930529, "grad_norm": 0.6378938555717468, "learning_rate": 1.5726666666666666e-05, "loss": 0.0765, "step": 2360 }, { "epoch": 1.5105162523900573, "grad_norm": 0.9574321508407593, "learning_rate": 1.5793333333333333e-05, "loss": 0.1047, "step": 2370 }, { "epoch": 1.5168897386870617, "grad_norm": 1.6412683725357056, "learning_rate": 1.586e-05, "loss": 0.083, "step": 2380 }, { "epoch": 1.5232632249840663, "grad_norm": 1.1442257165908813, "learning_rate": 1.5926666666666668e-05, "loss": 0.1049, "step": 2390 }, { "epoch": 1.5296367112810707, "grad_norm": 1.5450184345245361, "learning_rate": 1.5993333333333335e-05, "loss": 0.0973, "step": 2400 }, { "epoch": 1.5360101975780753, "grad_norm": 1.226181149482727, "learning_rate": 1.606e-05, "loss": 0.093, "step": 2410 }, { "epoch": 1.5423836838750797, "grad_norm": 0.7279766798019409, "learning_rate": 1.612666666666667e-05, "loss": 0.1147, "step": 2420 }, { "epoch": 1.5487571701720841, "grad_norm": 0.9621545076370239, "learning_rate": 1.6193333333333336e-05, "loss": 0.0895, "step": 2430 }, { "epoch": 1.5551306564690885, "grad_norm": 2.1044633388519287, "learning_rate": 1.626e-05, "loss": 0.1017, "step": 2440 }, { "epoch": 1.561504142766093, "grad_norm": 0.9241976737976074, "learning_rate": 1.6326666666666667e-05, "loss": 0.0952, "step": 2450 }, { "epoch": 1.5678776290630974, "grad_norm": 1.1970354318618774, "learning_rate": 1.639333333333333e-05, "loss": 0.0748, "step": 2460 }, { "epoch": 1.574251115360102, "grad_norm": 0.6390184760093689, "learning_rate": 1.646e-05, "loss": 0.0833, "step": 2470 }, { "epoch": 1.5806246016571064, "grad_norm": 1.1987965106964111, "learning_rate": 1.652666666666667e-05, "loss": 0.0919, "step": 2480 }, { "epoch": 1.586998087954111, "grad_norm": 1.0885887145996094, "learning_rate": 1.6593333333333333e-05, "loss": 0.0812, "step": 2490 }, { "epoch": 1.5933715742511154, "grad_norm": 0.7863695025444031, "learning_rate": 1.666e-05, "loss": 0.0861, "step": 2500 }, { "epoch": 1.5997450605481198, "grad_norm": 1.272106409072876, "learning_rate": 1.6726666666666667e-05, "loss": 0.0728, "step": 2510 }, { "epoch": 1.6061185468451242, "grad_norm": 0.9821702837944031, "learning_rate": 1.6793333333333334e-05, "loss": 0.0795, "step": 2520 }, { "epoch": 1.6124920331421286, "grad_norm": 0.9293336868286133, "learning_rate": 1.686e-05, "loss": 0.0723, "step": 2530 }, { "epoch": 1.6188655194391333, "grad_norm": 2.0479531288146973, "learning_rate": 1.692666666666667e-05, "loss": 0.0938, "step": 2540 }, { "epoch": 1.6252390057361377, "grad_norm": 1.4463709592819214, "learning_rate": 1.6993333333333332e-05, "loss": 0.0931, "step": 2550 }, { "epoch": 1.6316124920331423, "grad_norm": 1.106081485748291, "learning_rate": 1.706e-05, "loss": 0.084, "step": 2560 }, { "epoch": 1.6379859783301467, "grad_norm": 1.3168553113937378, "learning_rate": 1.712666666666667e-05, "loss": 0.0861, "step": 2570 }, { "epoch": 1.644359464627151, "grad_norm": 1.2882040739059448, "learning_rate": 1.7193333333333334e-05, "loss": 0.0885, "step": 2580 }, { "epoch": 1.6507329509241555, "grad_norm": 1.623707890510559, "learning_rate": 1.726e-05, "loss": 0.087, "step": 2590 }, { "epoch": 1.65710643722116, "grad_norm": 1.4060157537460327, "learning_rate": 1.7326666666666668e-05, "loss": 0.1084, "step": 2600 }, { "epoch": 1.6634799235181643, "grad_norm": 1.3172701597213745, "learning_rate": 1.7393333333333335e-05, "loss": 0.0881, "step": 2610 }, { "epoch": 1.669853409815169, "grad_norm": 1.099621295928955, "learning_rate": 1.7460000000000002e-05, "loss": 0.072, "step": 2620 }, { "epoch": 1.6762268961121733, "grad_norm": 1.0552417039871216, "learning_rate": 1.7526666666666666e-05, "loss": 0.1015, "step": 2630 }, { "epoch": 1.682600382409178, "grad_norm": 1.4513311386108398, "learning_rate": 1.7593333333333333e-05, "loss": 0.0931, "step": 2640 }, { "epoch": 1.6889738687061824, "grad_norm": 0.7885988354682922, "learning_rate": 1.766e-05, "loss": 0.0897, "step": 2650 }, { "epoch": 1.6953473550031868, "grad_norm": 0.7710033059120178, "learning_rate": 1.7726666666666667e-05, "loss": 0.0821, "step": 2660 }, { "epoch": 1.7017208413001912, "grad_norm": 1.3100814819335938, "learning_rate": 1.7793333333333335e-05, "loss": 0.1115, "step": 2670 }, { "epoch": 1.7080943275971956, "grad_norm": 1.3566712141036987, "learning_rate": 1.7860000000000002e-05, "loss": 0.1096, "step": 2680 }, { "epoch": 1.7144678138942, "grad_norm": 0.9596314430236816, "learning_rate": 1.7926666666666666e-05, "loss": 0.0943, "step": 2690 }, { "epoch": 1.7208413001912046, "grad_norm": 1.5328532457351685, "learning_rate": 1.7993333333333333e-05, "loss": 0.0847, "step": 2700 }, { "epoch": 1.727214786488209, "grad_norm": 1.2858617305755615, "learning_rate": 1.8060000000000003e-05, "loss": 0.098, "step": 2710 }, { "epoch": 1.7335882727852137, "grad_norm": 1.620842456817627, "learning_rate": 1.8126666666666667e-05, "loss": 0.1128, "step": 2720 }, { "epoch": 1.739961759082218, "grad_norm": 1.5014934539794922, "learning_rate": 1.8193333333333334e-05, "loss": 0.0933, "step": 2730 }, { "epoch": 1.7463352453792225, "grad_norm": 1.610737919807434, "learning_rate": 1.826e-05, "loss": 0.0838, "step": 2740 }, { "epoch": 1.7527087316762269, "grad_norm": 1.0811647176742554, "learning_rate": 1.832666666666667e-05, "loss": 0.0785, "step": 2750 }, { "epoch": 1.7590822179732313, "grad_norm": 1.580681562423706, "learning_rate": 1.8393333333333336e-05, "loss": 0.0839, "step": 2760 }, { "epoch": 1.7654557042702357, "grad_norm": 1.1532455682754517, "learning_rate": 1.846e-05, "loss": 0.0958, "step": 2770 }, { "epoch": 1.7718291905672403, "grad_norm": 1.2616368532180786, "learning_rate": 1.8526666666666667e-05, "loss": 0.0991, "step": 2780 }, { "epoch": 1.7782026768642447, "grad_norm": 0.7279543876647949, "learning_rate": 1.8593333333333334e-05, "loss": 0.0807, "step": 2790 }, { "epoch": 1.7845761631612493, "grad_norm": 1.0883718729019165, "learning_rate": 1.866e-05, "loss": 0.0908, "step": 2800 }, { "epoch": 1.7909496494582537, "grad_norm": 0.8580917716026306, "learning_rate": 1.8726666666666668e-05, "loss": 0.099, "step": 2810 }, { "epoch": 1.7973231357552581, "grad_norm": 1.086074948310852, "learning_rate": 1.8793333333333335e-05, "loss": 0.1012, "step": 2820 }, { "epoch": 1.8036966220522626, "grad_norm": 0.8041730523109436, "learning_rate": 1.886e-05, "loss": 0.0895, "step": 2830 }, { "epoch": 1.810070108349267, "grad_norm": 0.9425352811813354, "learning_rate": 1.8926666666666666e-05, "loss": 0.0876, "step": 2840 }, { "epoch": 1.8164435946462714, "grad_norm": 0.9173586964607239, "learning_rate": 1.8993333333333337e-05, "loss": 0.0871, "step": 2850 }, { "epoch": 1.822817080943276, "grad_norm": 1.0493013858795166, "learning_rate": 1.906e-05, "loss": 0.0835, "step": 2860 }, { "epoch": 1.8291905672402804, "grad_norm": 1.195335030555725, "learning_rate": 1.9126666666666668e-05, "loss": 0.0834, "step": 2870 }, { "epoch": 1.835564053537285, "grad_norm": 1.6238062381744385, "learning_rate": 1.9193333333333335e-05, "loss": 0.0706, "step": 2880 }, { "epoch": 1.8419375398342894, "grad_norm": 1.0624531507492065, "learning_rate": 1.9260000000000002e-05, "loss": 0.0985, "step": 2890 }, { "epoch": 1.8483110261312938, "grad_norm": 1.3885860443115234, "learning_rate": 1.932666666666667e-05, "loss": 0.0773, "step": 2900 }, { "epoch": 1.8546845124282982, "grad_norm": 1.0686782598495483, "learning_rate": 1.9393333333333336e-05, "loss": 0.0889, "step": 2910 }, { "epoch": 1.8610579987253026, "grad_norm": 0.9558495879173279, "learning_rate": 1.946e-05, "loss": 0.0956, "step": 2920 }, { "epoch": 1.867431485022307, "grad_norm": 1.1304773092269897, "learning_rate": 1.9526666666666667e-05, "loss": 0.0631, "step": 2930 }, { "epoch": 1.8738049713193117, "grad_norm": 1.3271628618240356, "learning_rate": 1.9593333333333334e-05, "loss": 0.0731, "step": 2940 }, { "epoch": 1.880178457616316, "grad_norm": 0.8929280638694763, "learning_rate": 1.966e-05, "loss": 0.0736, "step": 2950 }, { "epoch": 1.8865519439133207, "grad_norm": 0.6617772579193115, "learning_rate": 1.972666666666667e-05, "loss": 0.0883, "step": 2960 }, { "epoch": 1.892925430210325, "grad_norm": 1.1443796157836914, "learning_rate": 1.9793333333333332e-05, "loss": 0.0795, "step": 2970 }, { "epoch": 1.8992989165073295, "grad_norm": 1.21043860912323, "learning_rate": 1.986e-05, "loss": 0.0801, "step": 2980 }, { "epoch": 1.905672402804334, "grad_norm": 1.102117896080017, "learning_rate": 1.992666666666667e-05, "loss": 0.0873, "step": 2990 }, { "epoch": 1.9120458891013383, "grad_norm": 1.3731114864349365, "learning_rate": 1.9993333333333334e-05, "loss": 0.0952, "step": 3000 }, { "epoch": 1.9184193753983427, "grad_norm": 1.1329739093780518, "learning_rate": 2.006e-05, "loss": 0.0706, "step": 3010 }, { "epoch": 1.9247928616953474, "grad_norm": 0.5612623691558838, "learning_rate": 2.0126666666666668e-05, "loss": 0.0698, "step": 3020 }, { "epoch": 1.9311663479923518, "grad_norm": 1.3422327041625977, "learning_rate": 2.0193333333333332e-05, "loss": 0.0916, "step": 3030 }, { "epoch": 1.9375398342893564, "grad_norm": 1.1616122722625732, "learning_rate": 2.0260000000000003e-05, "loss": 0.0706, "step": 3040 }, { "epoch": 1.9439133205863608, "grad_norm": 0.9410479068756104, "learning_rate": 2.032666666666667e-05, "loss": 0.0767, "step": 3050 }, { "epoch": 1.9502868068833652, "grad_norm": 1.0274105072021484, "learning_rate": 2.0393333333333333e-05, "loss": 0.0841, "step": 3060 }, { "epoch": 1.9566602931803696, "grad_norm": 0.9492813348770142, "learning_rate": 2.046e-05, "loss": 0.0653, "step": 3070 }, { "epoch": 1.963033779477374, "grad_norm": 1.442031741142273, "learning_rate": 2.0526666666666668e-05, "loss": 0.0876, "step": 3080 }, { "epoch": 1.9694072657743786, "grad_norm": 1.257333755493164, "learning_rate": 2.0593333333333335e-05, "loss": 0.1017, "step": 3090 }, { "epoch": 1.975780752071383, "grad_norm": 1.355202317237854, "learning_rate": 2.0660000000000002e-05, "loss": 0.0953, "step": 3100 }, { "epoch": 1.9821542383683877, "grad_norm": 1.1883312463760376, "learning_rate": 2.0726666666666666e-05, "loss": 0.0733, "step": 3110 }, { "epoch": 1.988527724665392, "grad_norm": 0.946861207485199, "learning_rate": 2.0793333333333333e-05, "loss": 0.0594, "step": 3120 }, { "epoch": 1.9949012109623965, "grad_norm": 1.412591814994812, "learning_rate": 2.086e-05, "loss": 0.0817, "step": 3130 }, { "epoch": 2.001274697259401, "grad_norm": 1.3307621479034424, "learning_rate": 2.0926666666666667e-05, "loss": 0.0782, "step": 3140 }, { "epoch": 2.0076481835564053, "grad_norm": 1.5508772134780884, "learning_rate": 2.0993333333333334e-05, "loss": 0.0769, "step": 3150 }, { "epoch": 2.0140216698534097, "grad_norm": 1.8503458499908447, "learning_rate": 2.106e-05, "loss": 0.0853, "step": 3160 }, { "epoch": 2.020395156150414, "grad_norm": 1.7811330556869507, "learning_rate": 2.1126666666666665e-05, "loss": 0.0721, "step": 3170 }, { "epoch": 2.026768642447419, "grad_norm": 1.1020867824554443, "learning_rate": 2.1193333333333336e-05, "loss": 0.0847, "step": 3180 }, { "epoch": 2.0331421287444233, "grad_norm": 1.0944749116897583, "learning_rate": 2.1260000000000003e-05, "loss": 0.0853, "step": 3190 }, { "epoch": 2.0395156150414278, "grad_norm": 1.286003828048706, "learning_rate": 2.1326666666666667e-05, "loss": 0.0749, "step": 3200 }, { "epoch": 2.045889101338432, "grad_norm": 0.9382231831550598, "learning_rate": 2.1393333333333334e-05, "loss": 0.0794, "step": 3210 }, { "epoch": 2.0522625876354366, "grad_norm": 1.6390314102172852, "learning_rate": 2.146e-05, "loss": 0.1117, "step": 3220 }, { "epoch": 2.058636073932441, "grad_norm": 0.8031259775161743, "learning_rate": 2.152666666666667e-05, "loss": 0.0954, "step": 3230 }, { "epoch": 2.0650095602294454, "grad_norm": 1.2209105491638184, "learning_rate": 2.1593333333333336e-05, "loss": 0.0822, "step": 3240 }, { "epoch": 2.07138304652645, "grad_norm": 1.52094566822052, "learning_rate": 2.166e-05, "loss": 0.0915, "step": 3250 }, { "epoch": 2.0777565328234546, "grad_norm": 1.176224708557129, "learning_rate": 2.1726666666666666e-05, "loss": 0.0861, "step": 3260 }, { "epoch": 2.084130019120459, "grad_norm": 0.6373295783996582, "learning_rate": 2.1793333333333334e-05, "loss": 0.0731, "step": 3270 }, { "epoch": 2.0905035054174634, "grad_norm": 1.152594804763794, "learning_rate": 2.186e-05, "loss": 0.0755, "step": 3280 }, { "epoch": 2.096876991714468, "grad_norm": 1.1399019956588745, "learning_rate": 2.1926666666666668e-05, "loss": 0.084, "step": 3290 }, { "epoch": 2.1032504780114722, "grad_norm": 1.0591745376586914, "learning_rate": 2.1993333333333335e-05, "loss": 0.0769, "step": 3300 }, { "epoch": 2.1096239643084767, "grad_norm": 0.8013867139816284, "learning_rate": 2.206e-05, "loss": 0.0816, "step": 3310 }, { "epoch": 2.115997450605481, "grad_norm": 1.0244139432907104, "learning_rate": 2.212666666666667e-05, "loss": 0.0906, "step": 3320 }, { "epoch": 2.1223709369024855, "grad_norm": 1.5283808708190918, "learning_rate": 2.2193333333333337e-05, "loss": 0.0755, "step": 3330 }, { "epoch": 2.1287444231994903, "grad_norm": 1.2056852579116821, "learning_rate": 2.226e-05, "loss": 0.0719, "step": 3340 }, { "epoch": 2.1351179094964947, "grad_norm": 0.6835068464279175, "learning_rate": 2.2326666666666667e-05, "loss": 0.079, "step": 3350 }, { "epoch": 2.141491395793499, "grad_norm": 0.9059255719184875, "learning_rate": 2.2393333333333335e-05, "loss": 0.0686, "step": 3360 }, { "epoch": 2.1478648820905035, "grad_norm": 1.4286351203918457, "learning_rate": 2.2460000000000002e-05, "loss": 0.0956, "step": 3370 }, { "epoch": 2.154238368387508, "grad_norm": 1.0044485330581665, "learning_rate": 2.252666666666667e-05, "loss": 0.1064, "step": 3380 }, { "epoch": 2.1606118546845123, "grad_norm": 1.326513409614563, "learning_rate": 2.2593333333333336e-05, "loss": 0.076, "step": 3390 }, { "epoch": 2.1669853409815167, "grad_norm": 1.187397837638855, "learning_rate": 2.266e-05, "loss": 0.0863, "step": 3400 }, { "epoch": 2.173358827278521, "grad_norm": 1.5882831811904907, "learning_rate": 2.2726666666666667e-05, "loss": 0.0848, "step": 3410 }, { "epoch": 2.179732313575526, "grad_norm": 0.9371598362922668, "learning_rate": 2.2793333333333334e-05, "loss": 0.0665, "step": 3420 }, { "epoch": 2.1861057998725304, "grad_norm": 0.8871974349021912, "learning_rate": 2.286e-05, "loss": 0.0729, "step": 3430 }, { "epoch": 2.192479286169535, "grad_norm": 1.0369257926940918, "learning_rate": 2.292666666666667e-05, "loss": 0.0834, "step": 3440 }, { "epoch": 2.198852772466539, "grad_norm": 1.0512542724609375, "learning_rate": 2.2993333333333332e-05, "loss": 0.0655, "step": 3450 }, { "epoch": 2.2052262587635436, "grad_norm": 1.6292001008987427, "learning_rate": 2.306e-05, "loss": 0.0735, "step": 3460 }, { "epoch": 2.211599745060548, "grad_norm": 1.6295171976089478, "learning_rate": 2.312666666666667e-05, "loss": 0.0862, "step": 3470 }, { "epoch": 2.2179732313575524, "grad_norm": 1.0787677764892578, "learning_rate": 2.3193333333333334e-05, "loss": 0.0896, "step": 3480 }, { "epoch": 2.224346717654557, "grad_norm": 1.106917142868042, "learning_rate": 2.326e-05, "loss": 0.0891, "step": 3490 }, { "epoch": 2.2307202039515617, "grad_norm": 0.6044514775276184, "learning_rate": 2.3326666666666668e-05, "loss": 0.0665, "step": 3500 }, { "epoch": 2.237093690248566, "grad_norm": 1.183518886566162, "learning_rate": 2.3393333333333335e-05, "loss": 0.0681, "step": 3510 }, { "epoch": 2.2434671765455705, "grad_norm": 1.143831729888916, "learning_rate": 2.3460000000000002e-05, "loss": 0.0831, "step": 3520 }, { "epoch": 2.249840662842575, "grad_norm": 1.1012294292449951, "learning_rate": 2.352666666666667e-05, "loss": 0.079, "step": 3530 }, { "epoch": 2.2562141491395793, "grad_norm": 0.9378037452697754, "learning_rate": 2.3593333333333333e-05, "loss": 0.0695, "step": 3540 }, { "epoch": 2.2625876354365837, "grad_norm": 0.7672236561775208, "learning_rate": 2.366e-05, "loss": 0.0677, "step": 3550 }, { "epoch": 2.268961121733588, "grad_norm": 0.7897099256515503, "learning_rate": 2.3726666666666668e-05, "loss": 0.0631, "step": 3560 }, { "epoch": 2.275334608030593, "grad_norm": 1.0121351480484009, "learning_rate": 2.3793333333333335e-05, "loss": 0.0847, "step": 3570 }, { "epoch": 2.2817080943275974, "grad_norm": 1.0353792905807495, "learning_rate": 2.3860000000000002e-05, "loss": 0.0957, "step": 3580 }, { "epoch": 2.2880815806246018, "grad_norm": 0.9366838335990906, "learning_rate": 2.3926666666666666e-05, "loss": 0.089, "step": 3590 }, { "epoch": 2.294455066921606, "grad_norm": 1.0918177366256714, "learning_rate": 2.3993333333333333e-05, "loss": 0.0732, "step": 3600 }, { "epoch": 2.3008285532186106, "grad_norm": 0.9088542461395264, "learning_rate": 2.4060000000000003e-05, "loss": 0.0696, "step": 3610 }, { "epoch": 2.307202039515615, "grad_norm": 0.7565183043479919, "learning_rate": 2.4126666666666667e-05, "loss": 0.0745, "step": 3620 }, { "epoch": 2.3135755258126194, "grad_norm": 0.739345133304596, "learning_rate": 2.4193333333333334e-05, "loss": 0.0619, "step": 3630 }, { "epoch": 2.319949012109624, "grad_norm": 0.9682939648628235, "learning_rate": 2.426e-05, "loss": 0.0828, "step": 3640 }, { "epoch": 2.326322498406628, "grad_norm": 1.3012282848358154, "learning_rate": 2.432666666666667e-05, "loss": 0.0697, "step": 3650 }, { "epoch": 2.332695984703633, "grad_norm": 1.3370519876480103, "learning_rate": 2.4393333333333336e-05, "loss": 0.0693, "step": 3660 }, { "epoch": 2.3390694710006374, "grad_norm": 1.290480136871338, "learning_rate": 2.4460000000000003e-05, "loss": 0.0806, "step": 3670 }, { "epoch": 2.345442957297642, "grad_norm": 1.7388986349105835, "learning_rate": 2.4526666666666667e-05, "loss": 0.0871, "step": 3680 }, { "epoch": 2.3518164435946463, "grad_norm": 1.2476669549942017, "learning_rate": 2.4593333333333334e-05, "loss": 0.0777, "step": 3690 }, { "epoch": 2.3581899298916507, "grad_norm": 1.3203452825546265, "learning_rate": 2.466e-05, "loss": 0.0826, "step": 3700 }, { "epoch": 2.364563416188655, "grad_norm": 0.8643007278442383, "learning_rate": 2.4726666666666668e-05, "loss": 0.0802, "step": 3710 }, { "epoch": 2.3709369024856595, "grad_norm": 1.2604588270187378, "learning_rate": 2.4793333333333335e-05, "loss": 0.0928, "step": 3720 }, { "epoch": 2.3773103887826643, "grad_norm": 1.1747511625289917, "learning_rate": 2.486e-05, "loss": 0.074, "step": 3730 }, { "epoch": 2.3836838750796687, "grad_norm": 1.140848994255066, "learning_rate": 2.4926666666666666e-05, "loss": 0.0805, "step": 3740 }, { "epoch": 2.390057361376673, "grad_norm": 0.9357328414916992, "learning_rate": 2.4993333333333337e-05, "loss": 0.0816, "step": 3750 }, { "epoch": 2.3964308476736775, "grad_norm": 0.9001865983009338, "learning_rate": 2.506e-05, "loss": 0.0637, "step": 3760 }, { "epoch": 2.402804333970682, "grad_norm": 0.8584229946136475, "learning_rate": 2.512666666666667e-05, "loss": 0.0862, "step": 3770 }, { "epoch": 2.4091778202676863, "grad_norm": 1.199390172958374, "learning_rate": 2.5193333333333335e-05, "loss": 0.089, "step": 3780 }, { "epoch": 2.4155513065646907, "grad_norm": 1.5466145277023315, "learning_rate": 2.526e-05, "loss": 0.0731, "step": 3790 }, { "epoch": 2.421924792861695, "grad_norm": 0.89247727394104, "learning_rate": 2.5326666666666666e-05, "loss": 0.0951, "step": 3800 }, { "epoch": 2.4282982791587, "grad_norm": 0.83308345079422, "learning_rate": 2.5393333333333336e-05, "loss": 0.0605, "step": 3810 }, { "epoch": 2.4346717654557044, "grad_norm": 1.4955074787139893, "learning_rate": 2.546e-05, "loss": 0.0738, "step": 3820 }, { "epoch": 2.441045251752709, "grad_norm": 0.7073909640312195, "learning_rate": 2.5526666666666664e-05, "loss": 0.0712, "step": 3830 }, { "epoch": 2.447418738049713, "grad_norm": 1.0291156768798828, "learning_rate": 2.5593333333333334e-05, "loss": 0.0624, "step": 3840 }, { "epoch": 2.4537922243467176, "grad_norm": 1.3625471591949463, "learning_rate": 2.566e-05, "loss": 0.1009, "step": 3850 }, { "epoch": 2.460165710643722, "grad_norm": 1.9747023582458496, "learning_rate": 2.5726666666666665e-05, "loss": 0.0915, "step": 3860 }, { "epoch": 2.4665391969407264, "grad_norm": 1.3161057233810425, "learning_rate": 2.5793333333333336e-05, "loss": 0.0742, "step": 3870 }, { "epoch": 2.4729126832377313, "grad_norm": 0.5482892394065857, "learning_rate": 2.586e-05, "loss": 0.0676, "step": 3880 }, { "epoch": 2.4792861695347357, "grad_norm": 0.6921859383583069, "learning_rate": 2.5926666666666667e-05, "loss": 0.0746, "step": 3890 }, { "epoch": 2.48565965583174, "grad_norm": 1.1666264533996582, "learning_rate": 2.5993333333333337e-05, "loss": 0.0915, "step": 3900 }, { "epoch": 2.4920331421287445, "grad_norm": 0.7556964159011841, "learning_rate": 2.606e-05, "loss": 0.0715, "step": 3910 }, { "epoch": 2.498406628425749, "grad_norm": 0.768217921257019, "learning_rate": 2.6126666666666665e-05, "loss": 0.0741, "step": 3920 }, { "epoch": 2.5047801147227533, "grad_norm": 0.9767629504203796, "learning_rate": 2.6193333333333336e-05, "loss": 0.0756, "step": 3930 }, { "epoch": 2.5111536010197577, "grad_norm": 1.4233574867248535, "learning_rate": 2.6260000000000003e-05, "loss": 0.0894, "step": 3940 }, { "epoch": 2.517527087316762, "grad_norm": 1.6541444063186646, "learning_rate": 2.6326666666666666e-05, "loss": 0.0844, "step": 3950 }, { "epoch": 2.5239005736137665, "grad_norm": 1.1772117614746094, "learning_rate": 2.6393333333333337e-05, "loss": 0.0785, "step": 3960 }, { "epoch": 2.530274059910771, "grad_norm": 1.4708929061889648, "learning_rate": 2.646e-05, "loss": 0.0936, "step": 3970 }, { "epoch": 2.5366475462077758, "grad_norm": 0.8775594830513, "learning_rate": 2.6526666666666668e-05, "loss": 0.0816, "step": 3980 }, { "epoch": 2.54302103250478, "grad_norm": 0.8465993404388428, "learning_rate": 2.6593333333333335e-05, "loss": 0.071, "step": 3990 }, { "epoch": 2.5493945188017846, "grad_norm": 0.7804232239723206, "learning_rate": 2.6660000000000002e-05, "loss": 0.066, "step": 4000 }, { "epoch": 2.555768005098789, "grad_norm": 0.7311035990715027, "learning_rate": 2.6726666666666666e-05, "loss": 0.0809, "step": 4010 }, { "epoch": 2.5621414913957934, "grad_norm": 1.1837273836135864, "learning_rate": 2.6793333333333337e-05, "loss": 0.0825, "step": 4020 }, { "epoch": 2.568514977692798, "grad_norm": 2.203735589981079, "learning_rate": 2.686e-05, "loss": 0.0705, "step": 4030 }, { "epoch": 2.5748884639898026, "grad_norm": 0.9898331165313721, "learning_rate": 2.6926666666666667e-05, "loss": 0.0634, "step": 4040 }, { "epoch": 2.581261950286807, "grad_norm": 1.2964098453521729, "learning_rate": 2.6993333333333338e-05, "loss": 0.0735, "step": 4050 }, { "epoch": 2.5876354365838115, "grad_norm": 1.2006776332855225, "learning_rate": 2.7060000000000002e-05, "loss": 0.0716, "step": 4060 }, { "epoch": 2.594008922880816, "grad_norm": 1.2477272748947144, "learning_rate": 2.7126666666666666e-05, "loss": 0.0559, "step": 4070 }, { "epoch": 2.6003824091778203, "grad_norm": 0.6044740080833435, "learning_rate": 2.7193333333333336e-05, "loss": 0.0813, "step": 4080 }, { "epoch": 2.6067558954748247, "grad_norm": 0.7347907423973083, "learning_rate": 2.7260000000000003e-05, "loss": 0.0765, "step": 4090 }, { "epoch": 2.613129381771829, "grad_norm": 0.7471100091934204, "learning_rate": 2.7326666666666667e-05, "loss": 0.06, "step": 4100 }, { "epoch": 2.6195028680688335, "grad_norm": 1.0894196033477783, "learning_rate": 2.739333333333333e-05, "loss": 0.0755, "step": 4110 }, { "epoch": 2.625876354365838, "grad_norm": 0.8457331657409668, "learning_rate": 2.746e-05, "loss": 0.0706, "step": 4120 }, { "epoch": 2.6322498406628427, "grad_norm": 2.0056066513061523, "learning_rate": 2.752666666666667e-05, "loss": 0.0814, "step": 4130 }, { "epoch": 2.638623326959847, "grad_norm": 1.1211191415786743, "learning_rate": 2.7593333333333332e-05, "loss": 0.0741, "step": 4140 }, { "epoch": 2.6449968132568515, "grad_norm": 1.2720351219177246, "learning_rate": 2.7660000000000003e-05, "loss": 0.0698, "step": 4150 }, { "epoch": 2.651370299553856, "grad_norm": 1.2304461002349854, "learning_rate": 2.7726666666666667e-05, "loss": 0.0737, "step": 4160 }, { "epoch": 2.6577437858508604, "grad_norm": 1.1421581506729126, "learning_rate": 2.7793333333333334e-05, "loss": 0.0642, "step": 4170 }, { "epoch": 2.6641172721478648, "grad_norm": 1.1228699684143066, "learning_rate": 2.7860000000000004e-05, "loss": 0.0668, "step": 4180 }, { "epoch": 2.6704907584448696, "grad_norm": 0.8477457165718079, "learning_rate": 2.7926666666666668e-05, "loss": 0.0578, "step": 4190 }, { "epoch": 2.676864244741874, "grad_norm": 0.6787257790565491, "learning_rate": 2.7993333333333332e-05, "loss": 0.0798, "step": 4200 }, { "epoch": 2.6832377310388784, "grad_norm": 0.6480321884155273, "learning_rate": 2.8060000000000002e-05, "loss": 0.0713, "step": 4210 }, { "epoch": 2.689611217335883, "grad_norm": 0.8902840614318848, "learning_rate": 2.8126666666666666e-05, "loss": 0.0874, "step": 4220 }, { "epoch": 2.6959847036328872, "grad_norm": 0.9456228017807007, "learning_rate": 2.8193333333333333e-05, "loss": 0.0844, "step": 4230 }, { "epoch": 2.7023581899298916, "grad_norm": 0.6799092888832092, "learning_rate": 2.8260000000000004e-05, "loss": 0.0867, "step": 4240 }, { "epoch": 2.708731676226896, "grad_norm": 0.6477775573730469, "learning_rate": 2.8326666666666668e-05, "loss": 0.0819, "step": 4250 }, { "epoch": 2.7151051625239004, "grad_norm": 0.9661011099815369, "learning_rate": 2.839333333333333e-05, "loss": 0.0833, "step": 4260 }, { "epoch": 2.721478648820905, "grad_norm": 0.9668269753456116, "learning_rate": 2.8460000000000002e-05, "loss": 0.0801, "step": 4270 }, { "epoch": 2.7278521351179092, "grad_norm": 1.0115479230880737, "learning_rate": 2.852666666666667e-05, "loss": 0.0655, "step": 4280 }, { "epoch": 2.734225621414914, "grad_norm": 1.2372345924377441, "learning_rate": 2.8593333333333333e-05, "loss": 0.0733, "step": 4290 }, { "epoch": 2.7405991077119185, "grad_norm": 1.1563453674316406, "learning_rate": 2.8660000000000003e-05, "loss": 0.0682, "step": 4300 }, { "epoch": 2.746972594008923, "grad_norm": 1.0354273319244385, "learning_rate": 2.8726666666666667e-05, "loss": 0.0683, "step": 4310 }, { "epoch": 2.7533460803059273, "grad_norm": 0.8020623922348022, "learning_rate": 2.8793333333333334e-05, "loss": 0.0845, "step": 4320 }, { "epoch": 2.7597195666029317, "grad_norm": 0.9466817378997803, "learning_rate": 2.8860000000000005e-05, "loss": 0.082, "step": 4330 }, { "epoch": 2.766093052899936, "grad_norm": 1.0749906301498413, "learning_rate": 2.892666666666667e-05, "loss": 0.0688, "step": 4340 }, { "epoch": 2.772466539196941, "grad_norm": 0.8603000044822693, "learning_rate": 2.8993333333333332e-05, "loss": 0.0611, "step": 4350 }, { "epoch": 2.7788400254939454, "grad_norm": 1.47003972530365, "learning_rate": 2.9060000000000003e-05, "loss": 0.0677, "step": 4360 }, { "epoch": 2.78521351179095, "grad_norm": 0.6875565052032471, "learning_rate": 2.912666666666667e-05, "loss": 0.0693, "step": 4370 }, { "epoch": 2.791586998087954, "grad_norm": 1.0347020626068115, "learning_rate": 2.9193333333333334e-05, "loss": 0.0718, "step": 4380 }, { "epoch": 2.7979604843849586, "grad_norm": 0.49027177691459656, "learning_rate": 2.9260000000000004e-05, "loss": 0.0617, "step": 4390 }, { "epoch": 2.804333970681963, "grad_norm": 0.6546693444252014, "learning_rate": 2.9326666666666668e-05, "loss": 0.0777, "step": 4400 }, { "epoch": 2.8107074569789674, "grad_norm": 0.9329594373703003, "learning_rate": 2.9393333333333335e-05, "loss": 0.0737, "step": 4410 }, { "epoch": 2.817080943275972, "grad_norm": 1.2554725408554077, "learning_rate": 2.946e-05, "loss": 0.0938, "step": 4420 }, { "epoch": 2.823454429572976, "grad_norm": 0.8347150087356567, "learning_rate": 2.952666666666667e-05, "loss": 0.0726, "step": 4430 }, { "epoch": 2.8298279158699806, "grad_norm": 1.3713853359222412, "learning_rate": 2.9593333333333333e-05, "loss": 0.0747, "step": 4440 }, { "epoch": 2.8362014021669855, "grad_norm": 1.1022520065307617, "learning_rate": 2.9659999999999997e-05, "loss": 0.0737, "step": 4450 }, { "epoch": 2.84257488846399, "grad_norm": 0.5521169900894165, "learning_rate": 2.9726666666666668e-05, "loss": 0.0592, "step": 4460 }, { "epoch": 2.8489483747609943, "grad_norm": 1.3383432626724243, "learning_rate": 2.9793333333333335e-05, "loss": 0.0631, "step": 4470 }, { "epoch": 2.8553218610579987, "grad_norm": 0.5947378873825073, "learning_rate": 2.986e-05, "loss": 0.0632, "step": 4480 }, { "epoch": 2.861695347355003, "grad_norm": 0.8243908882141113, "learning_rate": 2.992666666666667e-05, "loss": 0.0661, "step": 4490 }, { "epoch": 2.8680688336520075, "grad_norm": 0.5694260597229004, "learning_rate": 2.9993333333333333e-05, "loss": 0.0749, "step": 4500 }, { "epoch": 2.8744423199490123, "grad_norm": 1.345126748085022, "learning_rate": 3.006e-05, "loss": 0.1045, "step": 4510 }, { "epoch": 2.8808158062460167, "grad_norm": 0.8360680937767029, "learning_rate": 3.012666666666667e-05, "loss": 0.0756, "step": 4520 }, { "epoch": 2.887189292543021, "grad_norm": 0.7815035581588745, "learning_rate": 3.0193333333333335e-05, "loss": 0.07, "step": 4530 }, { "epoch": 2.8935627788400256, "grad_norm": 0.9285995960235596, "learning_rate": 3.0259999999999998e-05, "loss": 0.0913, "step": 4540 }, { "epoch": 2.89993626513703, "grad_norm": 1.153631567955017, "learning_rate": 3.032666666666667e-05, "loss": 0.083, "step": 4550 }, { "epoch": 2.9063097514340344, "grad_norm": 0.7436297535896301, "learning_rate": 3.0393333333333336e-05, "loss": 0.0617, "step": 4560 }, { "epoch": 2.9126832377310388, "grad_norm": 0.9512943029403687, "learning_rate": 3.046e-05, "loss": 0.0872, "step": 4570 }, { "epoch": 2.919056724028043, "grad_norm": 1.808171033859253, "learning_rate": 3.052666666666667e-05, "loss": 0.0686, "step": 4580 }, { "epoch": 2.9254302103250476, "grad_norm": 0.7790931463241577, "learning_rate": 3.059333333333334e-05, "loss": 0.0871, "step": 4590 }, { "epoch": 2.9318036966220524, "grad_norm": 0.6268696188926697, "learning_rate": 3.066e-05, "loss": 0.0682, "step": 4600 }, { "epoch": 2.938177182919057, "grad_norm": 0.8465157747268677, "learning_rate": 3.072666666666667e-05, "loss": 0.0772, "step": 4610 }, { "epoch": 2.9445506692160612, "grad_norm": 0.7893638014793396, "learning_rate": 3.0793333333333336e-05, "loss": 0.0576, "step": 4620 }, { "epoch": 2.9509241555130656, "grad_norm": 1.395041823387146, "learning_rate": 3.086e-05, "loss": 0.0766, "step": 4630 }, { "epoch": 2.95729764181007, "grad_norm": 0.8314698338508606, "learning_rate": 3.092666666666667e-05, "loss": 0.0791, "step": 4640 }, { "epoch": 2.9636711281070744, "grad_norm": 0.7972301244735718, "learning_rate": 3.0993333333333334e-05, "loss": 0.0679, "step": 4650 }, { "epoch": 2.9700446144040793, "grad_norm": 1.237327218055725, "learning_rate": 3.106e-05, "loss": 0.093, "step": 4660 }, { "epoch": 2.9764181007010837, "grad_norm": 0.9242610931396484, "learning_rate": 3.112666666666667e-05, "loss": 0.066, "step": 4670 }, { "epoch": 2.982791586998088, "grad_norm": 1.5865044593811035, "learning_rate": 3.119333333333334e-05, "loss": 0.0794, "step": 4680 }, { "epoch": 2.9891650732950925, "grad_norm": 1.0787328481674194, "learning_rate": 3.126e-05, "loss": 0.0823, "step": 4690 }, { "epoch": 2.995538559592097, "grad_norm": 1.1212153434753418, "learning_rate": 3.132666666666667e-05, "loss": 0.0738, "step": 4700 }, { "epoch": 3.0019120458891013, "grad_norm": 0.6594878435134888, "learning_rate": 3.1393333333333337e-05, "loss": 0.0717, "step": 4710 }, { "epoch": 3.0082855321861057, "grad_norm": 1.0209071636199951, "learning_rate": 3.146e-05, "loss": 0.0613, "step": 4720 }, { "epoch": 3.01465901848311, "grad_norm": 1.0822089910507202, "learning_rate": 3.1526666666666664e-05, "loss": 0.083, "step": 4730 }, { "epoch": 3.0210325047801145, "grad_norm": 0.9222205281257629, "learning_rate": 3.1593333333333335e-05, "loss": 0.0692, "step": 4740 }, { "epoch": 3.0274059910771194, "grad_norm": 0.7965155243873596, "learning_rate": 3.166e-05, "loss": 0.0638, "step": 4750 }, { "epoch": 3.033779477374124, "grad_norm": 0.8105457425117493, "learning_rate": 3.172666666666667e-05, "loss": 0.0752, "step": 4760 }, { "epoch": 3.040152963671128, "grad_norm": 0.623116135597229, "learning_rate": 3.179333333333333e-05, "loss": 0.0722, "step": 4770 }, { "epoch": 3.0465264499681326, "grad_norm": 0.9211954474449158, "learning_rate": 3.186e-05, "loss": 0.0743, "step": 4780 }, { "epoch": 3.052899936265137, "grad_norm": 1.0942955017089844, "learning_rate": 3.192666666666667e-05, "loss": 0.0619, "step": 4790 }, { "epoch": 3.0592734225621414, "grad_norm": 1.1815531253814697, "learning_rate": 3.199333333333334e-05, "loss": 0.0725, "step": 4800 }, { "epoch": 3.065646908859146, "grad_norm": 1.0571568012237549, "learning_rate": 3.206e-05, "loss": 0.0707, "step": 4810 }, { "epoch": 3.07202039515615, "grad_norm": 1.062965750694275, "learning_rate": 3.2126666666666665e-05, "loss": 0.0796, "step": 4820 }, { "epoch": 3.078393881453155, "grad_norm": 1.058998942375183, "learning_rate": 3.2193333333333336e-05, "loss": 0.0742, "step": 4830 }, { "epoch": 3.0847673677501595, "grad_norm": 0.47686275839805603, "learning_rate": 3.226e-05, "loss": 0.0592, "step": 4840 }, { "epoch": 3.091140854047164, "grad_norm": 0.9305811524391174, "learning_rate": 3.232666666666666e-05, "loss": 0.062, "step": 4850 }, { "epoch": 3.0975143403441683, "grad_norm": 0.6117062568664551, "learning_rate": 3.2393333333333334e-05, "loss": 0.0625, "step": 4860 }, { "epoch": 3.1038878266411727, "grad_norm": 0.8363755941390991, "learning_rate": 3.2460000000000004e-05, "loss": 0.054, "step": 4870 }, { "epoch": 3.110261312938177, "grad_norm": 2.193182945251465, "learning_rate": 3.252666666666667e-05, "loss": 0.0827, "step": 4880 }, { "epoch": 3.1166347992351815, "grad_norm": 0.8609194755554199, "learning_rate": 3.259333333333334e-05, "loss": 0.0881, "step": 4890 }, { "epoch": 3.123008285532186, "grad_norm": 0.7422274947166443, "learning_rate": 3.266e-05, "loss": 0.0608, "step": 4900 }, { "epoch": 3.1293817718291908, "grad_norm": 1.7015436887741089, "learning_rate": 3.2726666666666666e-05, "loss": 0.07, "step": 4910 }, { "epoch": 3.135755258126195, "grad_norm": 0.9272896647453308, "learning_rate": 3.279333333333334e-05, "loss": 0.0746, "step": 4920 }, { "epoch": 3.1421287444231996, "grad_norm": 1.6632956266403198, "learning_rate": 3.286e-05, "loss": 0.0769, "step": 4930 }, { "epoch": 3.148502230720204, "grad_norm": 0.9707281589508057, "learning_rate": 3.2926666666666664e-05, "loss": 0.0664, "step": 4940 }, { "epoch": 3.1548757170172084, "grad_norm": 0.7668961882591248, "learning_rate": 3.2993333333333335e-05, "loss": 0.066, "step": 4950 }, { "epoch": 3.1612492033142128, "grad_norm": 1.1414968967437744, "learning_rate": 3.3060000000000005e-05, "loss": 0.0668, "step": 4960 }, { "epoch": 3.167622689611217, "grad_norm": 0.6386382579803467, "learning_rate": 3.312666666666667e-05, "loss": 0.0731, "step": 4970 }, { "epoch": 3.173996175908222, "grad_norm": 0.937657356262207, "learning_rate": 3.319333333333334e-05, "loss": 0.0612, "step": 4980 }, { "epoch": 3.1803696622052264, "grad_norm": 1.0628135204315186, "learning_rate": 3.3260000000000003e-05, "loss": 0.0671, "step": 4990 }, { "epoch": 3.186743148502231, "grad_norm": 1.070125937461853, "learning_rate": 3.332666666666667e-05, "loss": 0.0723, "step": 5000 }, { "epoch": 3.1931166347992352, "grad_norm": 1.0119048357009888, "learning_rate": 3.339333333333334e-05, "loss": 0.0728, "step": 5010 }, { "epoch": 3.1994901210962396, "grad_norm": 0.9777017831802368, "learning_rate": 3.346e-05, "loss": 0.0652, "step": 5020 }, { "epoch": 3.205863607393244, "grad_norm": 1.3063814640045166, "learning_rate": 3.3526666666666665e-05, "loss": 0.0728, "step": 5030 }, { "epoch": 3.2122370936902485, "grad_norm": 0.6333829164505005, "learning_rate": 3.359333333333333e-05, "loss": 0.0548, "step": 5040 }, { "epoch": 3.218610579987253, "grad_norm": 0.7940689325332642, "learning_rate": 3.366e-05, "loss": 0.0869, "step": 5050 }, { "epoch": 3.2249840662842573, "grad_norm": 0.7025421261787415, "learning_rate": 3.372666666666667e-05, "loss": 0.053, "step": 5060 }, { "epoch": 3.231357552581262, "grad_norm": 1.060881495475769, "learning_rate": 3.3793333333333334e-05, "loss": 0.0776, "step": 5070 }, { "epoch": 3.2377310388782665, "grad_norm": 0.6675065159797668, "learning_rate": 3.3860000000000004e-05, "loss": 0.0808, "step": 5080 }, { "epoch": 3.244104525175271, "grad_norm": 0.788928747177124, "learning_rate": 3.392666666666667e-05, "loss": 0.0651, "step": 5090 }, { "epoch": 3.2504780114722753, "grad_norm": 0.9996426105499268, "learning_rate": 3.399333333333333e-05, "loss": 0.0649, "step": 5100 }, { "epoch": 3.2568514977692797, "grad_norm": 0.7572866082191467, "learning_rate": 3.406e-05, "loss": 0.0704, "step": 5110 }, { "epoch": 3.263224984066284, "grad_norm": 1.0348575115203857, "learning_rate": 3.4126666666666666e-05, "loss": 0.0815, "step": 5120 }, { "epoch": 3.2695984703632885, "grad_norm": 1.1521161794662476, "learning_rate": 3.419333333333333e-05, "loss": 0.075, "step": 5130 }, { "epoch": 3.2759719566602934, "grad_norm": 0.6772708892822266, "learning_rate": 3.426e-05, "loss": 0.0643, "step": 5140 }, { "epoch": 3.282345442957298, "grad_norm": 0.8704012632369995, "learning_rate": 3.432666666666667e-05, "loss": 0.0706, "step": 5150 }, { "epoch": 3.288718929254302, "grad_norm": 0.8287102580070496, "learning_rate": 3.4393333333333335e-05, "loss": 0.0844, "step": 5160 }, { "epoch": 3.2950924155513066, "grad_norm": 0.8892394304275513, "learning_rate": 3.4460000000000005e-05, "loss": 0.0694, "step": 5170 }, { "epoch": 3.301465901848311, "grad_norm": 1.0021601915359497, "learning_rate": 3.452666666666667e-05, "loss": 0.0617, "step": 5180 }, { "epoch": 3.3078393881453154, "grad_norm": 1.0118314027786255, "learning_rate": 3.459333333333333e-05, "loss": 0.0706, "step": 5190 }, { "epoch": 3.31421287444232, "grad_norm": 0.93133544921875, "learning_rate": 3.4660000000000004e-05, "loss": 0.0634, "step": 5200 }, { "epoch": 3.3205863607393242, "grad_norm": 0.9466410279273987, "learning_rate": 3.472666666666667e-05, "loss": 0.068, "step": 5210 }, { "epoch": 3.3269598470363286, "grad_norm": 0.8412697911262512, "learning_rate": 3.479333333333333e-05, "loss": 0.0636, "step": 5220 }, { "epoch": 3.3333333333333335, "grad_norm": 0.7940225005149841, "learning_rate": 3.486e-05, "loss": 0.066, "step": 5230 }, { "epoch": 3.339706819630338, "grad_norm": 0.9359369874000549, "learning_rate": 3.4926666666666665e-05, "loss": 0.0719, "step": 5240 }, { "epoch": 3.3460803059273423, "grad_norm": 1.3645020723342896, "learning_rate": 3.4993333333333336e-05, "loss": 0.0707, "step": 5250 }, { "epoch": 3.3524537922243467, "grad_norm": 1.2952816486358643, "learning_rate": 3.5060000000000007e-05, "loss": 0.0685, "step": 5260 }, { "epoch": 3.358827278521351, "grad_norm": 0.7421273589134216, "learning_rate": 3.512666666666667e-05, "loss": 0.0744, "step": 5270 }, { "epoch": 3.3652007648183555, "grad_norm": 0.7207891941070557, "learning_rate": 3.5193333333333334e-05, "loss": 0.0781, "step": 5280 }, { "epoch": 3.37157425111536, "grad_norm": 0.9531093239784241, "learning_rate": 3.5260000000000005e-05, "loss": 0.0676, "step": 5290 }, { "epoch": 3.3779477374123648, "grad_norm": 0.5877613425254822, "learning_rate": 3.532666666666667e-05, "loss": 0.0574, "step": 5300 }, { "epoch": 3.384321223709369, "grad_norm": 0.6459888815879822, "learning_rate": 3.539333333333333e-05, "loss": 0.0681, "step": 5310 }, { "epoch": 3.3906947100063736, "grad_norm": 1.011889100074768, "learning_rate": 3.546e-05, "loss": 0.0724, "step": 5320 }, { "epoch": 3.397068196303378, "grad_norm": 0.5205362439155579, "learning_rate": 3.5526666666666666e-05, "loss": 0.0797, "step": 5330 }, { "epoch": 3.4034416826003824, "grad_norm": 0.8124321103096008, "learning_rate": 3.559333333333334e-05, "loss": 0.0727, "step": 5340 }, { "epoch": 3.409815168897387, "grad_norm": 0.7917166352272034, "learning_rate": 3.566e-05, "loss": 0.0701, "step": 5350 }, { "epoch": 3.416188655194391, "grad_norm": 0.8654747009277344, "learning_rate": 3.572666666666667e-05, "loss": 0.0672, "step": 5360 }, { "epoch": 3.4225621414913956, "grad_norm": 1.0261168479919434, "learning_rate": 3.5793333333333335e-05, "loss": 0.0593, "step": 5370 }, { "epoch": 3.4289356277884004, "grad_norm": 0.7219008207321167, "learning_rate": 3.586e-05, "loss": 0.068, "step": 5380 }, { "epoch": 3.435309114085405, "grad_norm": 0.8611841201782227, "learning_rate": 3.592666666666667e-05, "loss": 0.0607, "step": 5390 }, { "epoch": 3.4416826003824093, "grad_norm": 0.7135807275772095, "learning_rate": 3.599333333333333e-05, "loss": 0.0698, "step": 5400 }, { "epoch": 3.4480560866794137, "grad_norm": 0.9602314829826355, "learning_rate": 3.606e-05, "loss": 0.0678, "step": 5410 }, { "epoch": 3.454429572976418, "grad_norm": 1.1248433589935303, "learning_rate": 3.612666666666667e-05, "loss": 0.0614, "step": 5420 }, { "epoch": 3.4608030592734225, "grad_norm": 0.6453667879104614, "learning_rate": 3.619333333333333e-05, "loss": 0.0599, "step": 5430 }, { "epoch": 3.467176545570427, "grad_norm": 1.391933798789978, "learning_rate": 3.626e-05, "loss": 0.0909, "step": 5440 }, { "epoch": 3.4735500318674317, "grad_norm": 0.8513909578323364, "learning_rate": 3.632666666666667e-05, "loss": 0.0529, "step": 5450 }, { "epoch": 3.479923518164436, "grad_norm": 1.0372978448867798, "learning_rate": 3.6393333333333336e-05, "loss": 0.0663, "step": 5460 }, { "epoch": 3.4862970044614405, "grad_norm": 0.6336553692817688, "learning_rate": 3.646e-05, "loss": 0.0559, "step": 5470 }, { "epoch": 3.492670490758445, "grad_norm": 0.7353127598762512, "learning_rate": 3.652666666666667e-05, "loss": 0.0592, "step": 5480 }, { "epoch": 3.4990439770554493, "grad_norm": 0.48831209540367126, "learning_rate": 3.6593333333333334e-05, "loss": 0.0602, "step": 5490 }, { "epoch": 3.5054174633524537, "grad_norm": 0.9713586568832397, "learning_rate": 3.666e-05, "loss": 0.0717, "step": 5500 }, { "epoch": 3.511790949649458, "grad_norm": 0.9127499461174011, "learning_rate": 3.672666666666667e-05, "loss": 0.0747, "step": 5510 }, { "epoch": 3.5181644359464626, "grad_norm": 0.7769306302070618, "learning_rate": 3.679333333333333e-05, "loss": 0.0631, "step": 5520 }, { "epoch": 3.524537922243467, "grad_norm": 0.7864208817481995, "learning_rate": 3.686e-05, "loss": 0.059, "step": 5530 }, { "epoch": 3.5309114085404714, "grad_norm": 0.9494785666465759, "learning_rate": 3.6926666666666673e-05, "loss": 0.0561, "step": 5540 }, { "epoch": 3.537284894837476, "grad_norm": 0.6086204648017883, "learning_rate": 3.699333333333334e-05, "loss": 0.0484, "step": 5550 }, { "epoch": 3.5436583811344806, "grad_norm": 0.9108319282531738, "learning_rate": 3.706e-05, "loss": 0.0739, "step": 5560 }, { "epoch": 3.550031867431485, "grad_norm": 0.8275277018547058, "learning_rate": 3.712666666666667e-05, "loss": 0.0649, "step": 5570 }, { "epoch": 3.5564053537284894, "grad_norm": 0.8741169571876526, "learning_rate": 3.7193333333333335e-05, "loss": 0.0711, "step": 5580 }, { "epoch": 3.562778840025494, "grad_norm": 0.5867934226989746, "learning_rate": 3.726e-05, "loss": 0.0666, "step": 5590 }, { "epoch": 3.5691523263224987, "grad_norm": 0.9893175363540649, "learning_rate": 3.732666666666667e-05, "loss": 0.0707, "step": 5600 }, { "epoch": 3.575525812619503, "grad_norm": 1.4422320127487183, "learning_rate": 3.739333333333333e-05, "loss": 0.0654, "step": 5610 }, { "epoch": 3.5818992989165075, "grad_norm": 1.280185580253601, "learning_rate": 3.7460000000000004e-05, "loss": 0.0674, "step": 5620 }, { "epoch": 3.588272785213512, "grad_norm": 1.0105539560317993, "learning_rate": 3.7526666666666674e-05, "loss": 0.0645, "step": 5630 }, { "epoch": 3.5946462715105163, "grad_norm": 0.8940385580062866, "learning_rate": 3.759333333333334e-05, "loss": 0.0635, "step": 5640 }, { "epoch": 3.6010197578075207, "grad_norm": 1.0427013635635376, "learning_rate": 3.766e-05, "loss": 0.0773, "step": 5650 }, { "epoch": 3.607393244104525, "grad_norm": 0.7193406224250793, "learning_rate": 3.7726666666666666e-05, "loss": 0.0583, "step": 5660 }, { "epoch": 3.6137667304015295, "grad_norm": 0.7543081641197205, "learning_rate": 3.7793333333333336e-05, "loss": 0.0707, "step": 5670 }, { "epoch": 3.620140216698534, "grad_norm": 0.8943052291870117, "learning_rate": 3.786e-05, "loss": 0.0679, "step": 5680 }, { "epoch": 3.6265137029955383, "grad_norm": 0.6760386228561401, "learning_rate": 3.7926666666666664e-05, "loss": 0.0647, "step": 5690 }, { "epoch": 3.632887189292543, "grad_norm": 0.9791440367698669, "learning_rate": 3.7993333333333334e-05, "loss": 0.0727, "step": 5700 }, { "epoch": 3.6392606755895476, "grad_norm": 0.9366337060928345, "learning_rate": 3.806e-05, "loss": 0.08, "step": 5710 }, { "epoch": 3.645634161886552, "grad_norm": 0.9262896180152893, "learning_rate": 3.812666666666667e-05, "loss": 0.072, "step": 5720 }, { "epoch": 3.6520076481835564, "grad_norm": 1.2537277936935425, "learning_rate": 3.819333333333334e-05, "loss": 0.0691, "step": 5730 }, { "epoch": 3.658381134480561, "grad_norm": 1.117384672164917, "learning_rate": 3.826e-05, "loss": 0.0754, "step": 5740 }, { "epoch": 3.664754620777565, "grad_norm": 0.6348002552986145, "learning_rate": 3.832666666666667e-05, "loss": 0.0603, "step": 5750 }, { "epoch": 3.67112810707457, "grad_norm": 0.7391278743743896, "learning_rate": 3.839333333333334e-05, "loss": 0.0656, "step": 5760 }, { "epoch": 3.6775015933715745, "grad_norm": 0.7247368097305298, "learning_rate": 3.846e-05, "loss": 0.064, "step": 5770 }, { "epoch": 3.683875079668579, "grad_norm": 0.6577849388122559, "learning_rate": 3.8526666666666665e-05, "loss": 0.0842, "step": 5780 }, { "epoch": 3.6902485659655833, "grad_norm": 0.8825833797454834, "learning_rate": 3.8593333333333335e-05, "loss": 0.06, "step": 5790 }, { "epoch": 3.6966220522625877, "grad_norm": 0.5579012632369995, "learning_rate": 3.866e-05, "loss": 0.0787, "step": 5800 }, { "epoch": 3.702995538559592, "grad_norm": 0.6694490909576416, "learning_rate": 3.872666666666667e-05, "loss": 0.0594, "step": 5810 }, { "epoch": 3.7093690248565965, "grad_norm": 0.8608837127685547, "learning_rate": 3.879333333333334e-05, "loss": 0.0632, "step": 5820 }, { "epoch": 3.715742511153601, "grad_norm": 0.8467159271240234, "learning_rate": 3.8860000000000004e-05, "loss": 0.0635, "step": 5830 }, { "epoch": 3.7221159974506053, "grad_norm": 0.8490839004516602, "learning_rate": 3.892666666666667e-05, "loss": 0.058, "step": 5840 }, { "epoch": 3.7284894837476097, "grad_norm": 1.2588324546813965, "learning_rate": 3.899333333333334e-05, "loss": 0.0562, "step": 5850 }, { "epoch": 3.7348629700446145, "grad_norm": 0.9957313537597656, "learning_rate": 3.906e-05, "loss": 0.0723, "step": 5860 }, { "epoch": 3.741236456341619, "grad_norm": 1.0650949478149414, "learning_rate": 3.9126666666666666e-05, "loss": 0.0862, "step": 5870 }, { "epoch": 3.7476099426386233, "grad_norm": 0.5892736911773682, "learning_rate": 3.9193333333333336e-05, "loss": 0.0807, "step": 5880 }, { "epoch": 3.7539834289356278, "grad_norm": 0.6370896697044373, "learning_rate": 3.926e-05, "loss": 0.0636, "step": 5890 }, { "epoch": 3.760356915232632, "grad_norm": 0.5198182463645935, "learning_rate": 3.9326666666666664e-05, "loss": 0.0796, "step": 5900 }, { "epoch": 3.7667304015296366, "grad_norm": 0.8571606278419495, "learning_rate": 3.9393333333333335e-05, "loss": 0.0782, "step": 5910 }, { "epoch": 3.7731038878266414, "grad_norm": 0.7116801738739014, "learning_rate": 3.9460000000000005e-05, "loss": 0.052, "step": 5920 }, { "epoch": 3.779477374123646, "grad_norm": 0.6442444920539856, "learning_rate": 3.952666666666667e-05, "loss": 0.0588, "step": 5930 }, { "epoch": 3.78585086042065, "grad_norm": 1.0575484037399292, "learning_rate": 3.959333333333334e-05, "loss": 0.0758, "step": 5940 }, { "epoch": 3.7922243467176546, "grad_norm": 0.8098411560058594, "learning_rate": 3.966e-05, "loss": 0.0496, "step": 5950 }, { "epoch": 3.798597833014659, "grad_norm": 0.7580195069313049, "learning_rate": 3.972666666666667e-05, "loss": 0.0574, "step": 5960 }, { "epoch": 3.8049713193116634, "grad_norm": 0.9306786060333252, "learning_rate": 3.979333333333333e-05, "loss": 0.0608, "step": 5970 }, { "epoch": 3.811344805608668, "grad_norm": 1.4234120845794678, "learning_rate": 3.986e-05, "loss": 0.0893, "step": 5980 }, { "epoch": 3.8177182919056722, "grad_norm": 0.7272818684577942, "learning_rate": 3.9926666666666665e-05, "loss": 0.0803, "step": 5990 }, { "epoch": 3.8240917782026767, "grad_norm": 0.7398123145103455, "learning_rate": 3.9993333333333336e-05, "loss": 0.0733, "step": 6000 }, { "epoch": 3.830465264499681, "grad_norm": 0.8781506419181824, "learning_rate": 4.0060000000000006e-05, "loss": 0.0715, "step": 6010 }, { "epoch": 3.836838750796686, "grad_norm": 0.8230180144309998, "learning_rate": 4.012666666666667e-05, "loss": 0.0831, "step": 6020 }, { "epoch": 3.8432122370936903, "grad_norm": 1.5217406749725342, "learning_rate": 4.0193333333333334e-05, "loss": 0.0818, "step": 6030 }, { "epoch": 3.8495857233906947, "grad_norm": 1.135979175567627, "learning_rate": 4.0260000000000004e-05, "loss": 0.0771, "step": 6040 }, { "epoch": 3.855959209687699, "grad_norm": 1.0574519634246826, "learning_rate": 4.032666666666667e-05, "loss": 0.0725, "step": 6050 }, { "epoch": 3.8623326959847035, "grad_norm": 1.0695313215255737, "learning_rate": 4.039333333333333e-05, "loss": 0.0555, "step": 6060 }, { "epoch": 3.868706182281708, "grad_norm": 1.1515299081802368, "learning_rate": 4.046e-05, "loss": 0.0753, "step": 6070 }, { "epoch": 3.8750796685787128, "grad_norm": 0.9582716822624207, "learning_rate": 4.0526666666666666e-05, "loss": 0.0961, "step": 6080 }, { "epoch": 3.881453154875717, "grad_norm": 0.7768780589103699, "learning_rate": 4.0593333333333337e-05, "loss": 0.0693, "step": 6090 }, { "epoch": 3.8878266411727216, "grad_norm": 0.7570106983184814, "learning_rate": 4.066e-05, "loss": 0.0826, "step": 6100 }, { "epoch": 3.894200127469726, "grad_norm": 0.531088650226593, "learning_rate": 4.072666666666667e-05, "loss": 0.0521, "step": 6110 }, { "epoch": 3.9005736137667304, "grad_norm": 0.9833720922470093, "learning_rate": 4.0793333333333335e-05, "loss": 0.0692, "step": 6120 }, { "epoch": 3.906947100063735, "grad_norm": 0.6525686383247375, "learning_rate": 4.0860000000000005e-05, "loss": 0.0628, "step": 6130 }, { "epoch": 3.913320586360739, "grad_norm": 1.07793128490448, "learning_rate": 4.092666666666667e-05, "loss": 0.0693, "step": 6140 }, { "epoch": 3.9196940726577436, "grad_norm": 0.8308563828468323, "learning_rate": 4.099333333333333e-05, "loss": 0.0807, "step": 6150 }, { "epoch": 3.926067558954748, "grad_norm": 0.7288112640380859, "learning_rate": 4.106e-05, "loss": 0.0723, "step": 6160 }, { "epoch": 3.932441045251753, "grad_norm": 0.8844555616378784, "learning_rate": 4.112666666666667e-05, "loss": 0.0737, "step": 6170 }, { "epoch": 3.9388145315487573, "grad_norm": 0.6023966073989868, "learning_rate": 4.119333333333333e-05, "loss": 0.0572, "step": 6180 }, { "epoch": 3.9451880178457617, "grad_norm": 0.6090594530105591, "learning_rate": 4.126e-05, "loss": 0.0747, "step": 6190 }, { "epoch": 3.951561504142766, "grad_norm": 0.8953942656517029, "learning_rate": 4.132666666666667e-05, "loss": 0.0598, "step": 6200 }, { "epoch": 3.9579349904397705, "grad_norm": 0.8100000023841858, "learning_rate": 4.1393333333333336e-05, "loss": 0.0667, "step": 6210 }, { "epoch": 3.964308476736775, "grad_norm": 0.7682720422744751, "learning_rate": 4.1460000000000006e-05, "loss": 0.067, "step": 6220 }, { "epoch": 3.9706819630337797, "grad_norm": 1.0075936317443848, "learning_rate": 4.152666666666667e-05, "loss": 0.0737, "step": 6230 }, { "epoch": 3.977055449330784, "grad_norm": 0.7655991911888123, "learning_rate": 4.1593333333333334e-05, "loss": 0.0686, "step": 6240 }, { "epoch": 3.9834289356277885, "grad_norm": 1.4795335531234741, "learning_rate": 4.1660000000000004e-05, "loss": 0.0937, "step": 6250 }, { "epoch": 3.989802421924793, "grad_norm": 0.5925967693328857, "learning_rate": 4.172666666666667e-05, "loss": 0.0531, "step": 6260 }, { "epoch": 3.9961759082217974, "grad_norm": 1.1187046766281128, "learning_rate": 4.179333333333333e-05, "loss": 0.0732, "step": 6270 }, { "epoch": 4.002549394518802, "grad_norm": 0.9150164127349854, "learning_rate": 4.186e-05, "loss": 0.0666, "step": 6280 }, { "epoch": 4.008922880815806, "grad_norm": 0.49149367213249207, "learning_rate": 4.192666666666667e-05, "loss": 0.0548, "step": 6290 }, { "epoch": 4.015296367112811, "grad_norm": 0.6512823104858398, "learning_rate": 4.199333333333334e-05, "loss": 0.0641, "step": 6300 }, { "epoch": 4.021669853409815, "grad_norm": 0.7342813014984131, "learning_rate": 4.206e-05, "loss": 0.0582, "step": 6310 }, { "epoch": 4.028043339706819, "grad_norm": 0.8499735593795776, "learning_rate": 4.212666666666667e-05, "loss": 0.0668, "step": 6320 }, { "epoch": 4.034416826003824, "grad_norm": 1.5677454471588135, "learning_rate": 4.2193333333333335e-05, "loss": 0.0775, "step": 6330 }, { "epoch": 4.040790312300828, "grad_norm": 0.5655584931373596, "learning_rate": 4.226e-05, "loss": 0.0737, "step": 6340 }, { "epoch": 4.047163798597833, "grad_norm": 0.5354357957839966, "learning_rate": 4.232666666666667e-05, "loss": 0.0576, "step": 6350 }, { "epoch": 4.053537284894838, "grad_norm": 0.6646163463592529, "learning_rate": 4.239333333333333e-05, "loss": 0.0713, "step": 6360 }, { "epoch": 4.059910771191842, "grad_norm": 0.8326684236526489, "learning_rate": 4.246e-05, "loss": 0.0512, "step": 6370 }, { "epoch": 4.066284257488847, "grad_norm": 0.6551247239112854, "learning_rate": 4.252666666666667e-05, "loss": 0.0547, "step": 6380 }, { "epoch": 4.072657743785851, "grad_norm": 0.8713095784187317, "learning_rate": 4.259333333333334e-05, "loss": 0.0754, "step": 6390 }, { "epoch": 4.0790312300828555, "grad_norm": 0.8737226724624634, "learning_rate": 4.266e-05, "loss": 0.0649, "step": 6400 }, { "epoch": 4.08540471637986, "grad_norm": 1.3375325202941895, "learning_rate": 4.272666666666667e-05, "loss": 0.0572, "step": 6410 }, { "epoch": 4.091778202676864, "grad_norm": 1.0645341873168945, "learning_rate": 4.2793333333333336e-05, "loss": 0.0579, "step": 6420 }, { "epoch": 4.098151688973869, "grad_norm": 0.6480242013931274, "learning_rate": 4.286e-05, "loss": 0.0657, "step": 6430 }, { "epoch": 4.104525175270873, "grad_norm": 0.8946130275726318, "learning_rate": 4.292666666666667e-05, "loss": 0.0584, "step": 6440 }, { "epoch": 4.1108986615678775, "grad_norm": 0.82254558801651, "learning_rate": 4.2993333333333334e-05, "loss": 0.0621, "step": 6450 }, { "epoch": 4.117272147864882, "grad_norm": 0.8301029205322266, "learning_rate": 4.306e-05, "loss": 0.0519, "step": 6460 }, { "epoch": 4.123645634161886, "grad_norm": 0.9191573262214661, "learning_rate": 4.312666666666667e-05, "loss": 0.0648, "step": 6470 }, { "epoch": 4.130019120458891, "grad_norm": 0.6837590336799622, "learning_rate": 4.319333333333334e-05, "loss": 0.0578, "step": 6480 }, { "epoch": 4.136392606755895, "grad_norm": 0.8105356097221375, "learning_rate": 4.326e-05, "loss": 0.0776, "step": 6490 }, { "epoch": 4.1427660930529, "grad_norm": 0.8152103424072266, "learning_rate": 4.332666666666667e-05, "loss": 0.0684, "step": 6500 }, { "epoch": 4.149139579349905, "grad_norm": 0.9396456480026245, "learning_rate": 4.339333333333334e-05, "loss": 0.0523, "step": 6510 }, { "epoch": 4.155513065646909, "grad_norm": 0.8193104863166809, "learning_rate": 4.346e-05, "loss": 0.0574, "step": 6520 }, { "epoch": 4.161886551943914, "grad_norm": 0.7371982336044312, "learning_rate": 4.352666666666667e-05, "loss": 0.0645, "step": 6530 }, { "epoch": 4.168260038240918, "grad_norm": 1.027175784111023, "learning_rate": 4.3593333333333335e-05, "loss": 0.0729, "step": 6540 }, { "epoch": 4.1746335245379225, "grad_norm": 0.6634669303894043, "learning_rate": 4.366e-05, "loss": 0.0523, "step": 6550 }, { "epoch": 4.181007010834927, "grad_norm": 0.7883057594299316, "learning_rate": 4.372666666666667e-05, "loss": 0.0818, "step": 6560 }, { "epoch": 4.187380497131931, "grad_norm": 1.2026309967041016, "learning_rate": 4.379333333333333e-05, "loss": 0.0756, "step": 6570 }, { "epoch": 4.193753983428936, "grad_norm": 1.1520119905471802, "learning_rate": 4.3860000000000004e-05, "loss": 0.0725, "step": 6580 }, { "epoch": 4.20012746972594, "grad_norm": 1.096225380897522, "learning_rate": 4.3926666666666674e-05, "loss": 0.0737, "step": 6590 }, { "epoch": 4.2065009560229445, "grad_norm": 1.2484513521194458, "learning_rate": 4.399333333333334e-05, "loss": 0.0759, "step": 6600 }, { "epoch": 4.212874442319949, "grad_norm": 0.33926501870155334, "learning_rate": 4.406e-05, "loss": 0.0621, "step": 6610 }, { "epoch": 4.219247928616953, "grad_norm": 0.8289417624473572, "learning_rate": 4.4126666666666665e-05, "loss": 0.0551, "step": 6620 }, { "epoch": 4.225621414913958, "grad_norm": 1.2392948865890503, "learning_rate": 4.4193333333333336e-05, "loss": 0.067, "step": 6630 }, { "epoch": 4.231994901210962, "grad_norm": 0.6331199407577515, "learning_rate": 4.426e-05, "loss": 0.0573, "step": 6640 }, { "epoch": 4.2383683875079665, "grad_norm": 0.7887037396430969, "learning_rate": 4.4326666666666664e-05, "loss": 0.0767, "step": 6650 }, { "epoch": 4.244741873804971, "grad_norm": 0.6712101101875305, "learning_rate": 4.4393333333333334e-05, "loss": 0.0501, "step": 6660 }, { "epoch": 4.251115360101975, "grad_norm": 0.5873392820358276, "learning_rate": 4.4460000000000005e-05, "loss": 0.0629, "step": 6670 }, { "epoch": 4.257488846398981, "grad_norm": 0.8280028104782104, "learning_rate": 4.452666666666667e-05, "loss": 0.0693, "step": 6680 }, { "epoch": 4.263862332695985, "grad_norm": 1.1430957317352295, "learning_rate": 4.459333333333334e-05, "loss": 0.055, "step": 6690 }, { "epoch": 4.270235818992989, "grad_norm": 0.5776487588882446, "learning_rate": 4.466e-05, "loss": 0.0741, "step": 6700 }, { "epoch": 4.276609305289994, "grad_norm": 0.657620906829834, "learning_rate": 4.4726666666666666e-05, "loss": 0.0602, "step": 6710 }, { "epoch": 4.282982791586998, "grad_norm": 0.7546238899230957, "learning_rate": 4.479333333333334e-05, "loss": 0.0772, "step": 6720 }, { "epoch": 4.289356277884003, "grad_norm": 0.7995972633361816, "learning_rate": 4.486e-05, "loss": 0.078, "step": 6730 }, { "epoch": 4.295729764181007, "grad_norm": 0.6277379989624023, "learning_rate": 4.4926666666666665e-05, "loss": 0.08, "step": 6740 }, { "epoch": 4.3021032504780115, "grad_norm": 0.7915437817573547, "learning_rate": 4.4993333333333335e-05, "loss": 0.0641, "step": 6750 }, { "epoch": 4.308476736775016, "grad_norm": 0.7979368567466736, "learning_rate": 4.506e-05, "loss": 0.0663, "step": 6760 }, { "epoch": 4.31485022307202, "grad_norm": 0.43743348121643066, "learning_rate": 4.512666666666667e-05, "loss": 0.0508, "step": 6770 }, { "epoch": 4.321223709369025, "grad_norm": 0.5177498459815979, "learning_rate": 4.519333333333334e-05, "loss": 0.0656, "step": 6780 }, { "epoch": 4.327597195666029, "grad_norm": 0.6387514472007751, "learning_rate": 4.5260000000000004e-05, "loss": 0.0755, "step": 6790 }, { "epoch": 4.3339706819630335, "grad_norm": 0.8713686466217041, "learning_rate": 4.532666666666667e-05, "loss": 0.0658, "step": 6800 }, { "epoch": 4.340344168260038, "grad_norm": 0.4039125442504883, "learning_rate": 4.539333333333334e-05, "loss": 0.0559, "step": 6810 }, { "epoch": 4.346717654557042, "grad_norm": 0.5320805907249451, "learning_rate": 4.546e-05, "loss": 0.0692, "step": 6820 }, { "epoch": 4.353091140854048, "grad_norm": 1.0053350925445557, "learning_rate": 4.5526666666666666e-05, "loss": 0.0738, "step": 6830 }, { "epoch": 4.359464627151052, "grad_norm": 0.8067516088485718, "learning_rate": 4.5593333333333336e-05, "loss": 0.0754, "step": 6840 }, { "epoch": 4.365838113448056, "grad_norm": 0.8567308783531189, "learning_rate": 4.566e-05, "loss": 0.0709, "step": 6850 }, { "epoch": 4.372211599745061, "grad_norm": 0.8834486603736877, "learning_rate": 4.572666666666667e-05, "loss": 0.054, "step": 6860 }, { "epoch": 4.378585086042065, "grad_norm": 0.8575840592384338, "learning_rate": 4.579333333333334e-05, "loss": 0.0828, "step": 6870 }, { "epoch": 4.38495857233907, "grad_norm": 0.5521349310874939, "learning_rate": 4.5860000000000005e-05, "loss": 0.0721, "step": 6880 }, { "epoch": 4.391332058636074, "grad_norm": 0.5701805949211121, "learning_rate": 4.592666666666667e-05, "loss": 0.0562, "step": 6890 }, { "epoch": 4.397705544933078, "grad_norm": 0.5939077138900757, "learning_rate": 4.599333333333334e-05, "loss": 0.0599, "step": 6900 }, { "epoch": 4.404079031230083, "grad_norm": 0.5022218823432922, "learning_rate": 4.606e-05, "loss": 0.0574, "step": 6910 }, { "epoch": 4.410452517527087, "grad_norm": 0.9328749775886536, "learning_rate": 4.612666666666667e-05, "loss": 0.0684, "step": 6920 }, { "epoch": 4.416826003824092, "grad_norm": 0.8469638824462891, "learning_rate": 4.619333333333333e-05, "loss": 0.0641, "step": 6930 }, { "epoch": 4.423199490121096, "grad_norm": 0.7458369731903076, "learning_rate": 4.626e-05, "loss": 0.057, "step": 6940 }, { "epoch": 4.4295729764181, "grad_norm": 0.7582130432128906, "learning_rate": 4.632666666666667e-05, "loss": 0.0622, "step": 6950 }, { "epoch": 4.435946462715105, "grad_norm": 0.7022218108177185, "learning_rate": 4.6393333333333335e-05, "loss": 0.0607, "step": 6960 }, { "epoch": 4.442319949012109, "grad_norm": 1.1240204572677612, "learning_rate": 4.6460000000000006e-05, "loss": 0.0668, "step": 6970 }, { "epoch": 4.448693435309114, "grad_norm": 0.8612990379333496, "learning_rate": 4.652666666666667e-05, "loss": 0.0726, "step": 6980 }, { "epoch": 4.455066921606119, "grad_norm": 0.8591046929359436, "learning_rate": 4.659333333333333e-05, "loss": 0.0598, "step": 6990 }, { "epoch": 4.461440407903123, "grad_norm": 0.9302867650985718, "learning_rate": 4.6660000000000004e-05, "loss": 0.0778, "step": 7000 }, { "epoch": 4.467813894200128, "grad_norm": 0.7860726714134216, "learning_rate": 4.672666666666667e-05, "loss": 0.0745, "step": 7010 }, { "epoch": 4.474187380497132, "grad_norm": 0.8013750910758972, "learning_rate": 4.679333333333333e-05, "loss": 0.0594, "step": 7020 }, { "epoch": 4.480560866794137, "grad_norm": 1.153072476387024, "learning_rate": 4.686e-05, "loss": 0.0818, "step": 7030 }, { "epoch": 4.486934353091141, "grad_norm": 0.8857479691505432, "learning_rate": 4.6926666666666666e-05, "loss": 0.0751, "step": 7040 }, { "epoch": 4.493307839388145, "grad_norm": 0.6867552399635315, "learning_rate": 4.6993333333333336e-05, "loss": 0.0498, "step": 7050 }, { "epoch": 4.49968132568515, "grad_norm": 0.696652889251709, "learning_rate": 4.706000000000001e-05, "loss": 0.0578, "step": 7060 }, { "epoch": 4.506054811982154, "grad_norm": 0.5428764820098877, "learning_rate": 4.712666666666667e-05, "loss": 0.0595, "step": 7070 }, { "epoch": 4.512428298279159, "grad_norm": 1.3017147779464722, "learning_rate": 4.7193333333333334e-05, "loss": 0.0929, "step": 7080 }, { "epoch": 4.518801784576163, "grad_norm": 0.595328688621521, "learning_rate": 4.7260000000000005e-05, "loss": 0.0652, "step": 7090 }, { "epoch": 4.525175270873167, "grad_norm": 0.9018744826316833, "learning_rate": 4.732666666666667e-05, "loss": 0.0642, "step": 7100 }, { "epoch": 4.531548757170172, "grad_norm": 0.7857619524002075, "learning_rate": 4.739333333333333e-05, "loss": 0.0836, "step": 7110 }, { "epoch": 4.537922243467176, "grad_norm": 0.6209820508956909, "learning_rate": 4.746e-05, "loss": 0.0481, "step": 7120 }, { "epoch": 4.5442957297641815, "grad_norm": 0.9665376543998718, "learning_rate": 4.752666666666667e-05, "loss": 0.0557, "step": 7130 }, { "epoch": 4.550669216061186, "grad_norm": 1.2759356498718262, "learning_rate": 4.759333333333334e-05, "loss": 0.0737, "step": 7140 }, { "epoch": 4.55704270235819, "grad_norm": 0.674159824848175, "learning_rate": 4.766000000000001e-05, "loss": 0.0615, "step": 7150 }, { "epoch": 4.563416188655195, "grad_norm": 1.0083224773406982, "learning_rate": 4.772666666666667e-05, "loss": 0.0712, "step": 7160 }, { "epoch": 4.569789674952199, "grad_norm": 0.5094786882400513, "learning_rate": 4.7793333333333335e-05, "loss": 0.0684, "step": 7170 }, { "epoch": 4.5761631612492035, "grad_norm": 0.8377692103385925, "learning_rate": 4.7860000000000006e-05, "loss": 0.0599, "step": 7180 }, { "epoch": 4.582536647546208, "grad_norm": 0.43251752853393555, "learning_rate": 4.792666666666667e-05, "loss": 0.0707, "step": 7190 }, { "epoch": 4.588910133843212, "grad_norm": 1.0879417657852173, "learning_rate": 4.7993333333333333e-05, "loss": 0.0525, "step": 7200 }, { "epoch": 4.595283620140217, "grad_norm": 0.9131572842597961, "learning_rate": 4.8060000000000004e-05, "loss": 0.0667, "step": 7210 }, { "epoch": 4.601657106437221, "grad_norm": 0.4205023944377899, "learning_rate": 4.812666666666667e-05, "loss": 0.0656, "step": 7220 }, { "epoch": 4.6080305927342256, "grad_norm": 0.4257108271121979, "learning_rate": 4.819333333333333e-05, "loss": 0.0538, "step": 7230 }, { "epoch": 4.61440407903123, "grad_norm": 0.9521909952163696, "learning_rate": 4.826e-05, "loss": 0.0673, "step": 7240 }, { "epoch": 4.620777565328234, "grad_norm": 0.8961038589477539, "learning_rate": 4.832666666666667e-05, "loss": 0.0717, "step": 7250 }, { "epoch": 4.627151051625239, "grad_norm": 0.6734869480133057, "learning_rate": 4.8393333333333336e-05, "loss": 0.0622, "step": 7260 }, { "epoch": 4.633524537922243, "grad_norm": 0.5155494213104248, "learning_rate": 4.846e-05, "loss": 0.0624, "step": 7270 }, { "epoch": 4.639898024219248, "grad_norm": 0.8553544282913208, "learning_rate": 4.852666666666667e-05, "loss": 0.059, "step": 7280 }, { "epoch": 4.646271510516252, "grad_norm": 1.181668758392334, "learning_rate": 4.8593333333333335e-05, "loss": 0.0696, "step": 7290 }, { "epoch": 4.652644996813256, "grad_norm": 0.7553808093070984, "learning_rate": 4.866e-05, "loss": 0.0633, "step": 7300 }, { "epoch": 4.659018483110262, "grad_norm": 0.834030270576477, "learning_rate": 4.872666666666667e-05, "loss": 0.0611, "step": 7310 }, { "epoch": 4.665391969407266, "grad_norm": 0.5208401679992676, "learning_rate": 4.879333333333333e-05, "loss": 0.0529, "step": 7320 }, { "epoch": 4.6717654557042705, "grad_norm": 0.8630608320236206, "learning_rate": 4.886e-05, "loss": 0.0572, "step": 7330 }, { "epoch": 4.678138942001275, "grad_norm": 1.0148797035217285, "learning_rate": 4.8926666666666674e-05, "loss": 0.0787, "step": 7340 }, { "epoch": 4.684512428298279, "grad_norm": 0.5508243441581726, "learning_rate": 4.899333333333334e-05, "loss": 0.0485, "step": 7350 }, { "epoch": 4.690885914595284, "grad_norm": 0.783965528011322, "learning_rate": 4.906e-05, "loss": 0.0709, "step": 7360 }, { "epoch": 4.697259400892288, "grad_norm": 0.7949073910713196, "learning_rate": 4.912666666666667e-05, "loss": 0.0656, "step": 7370 }, { "epoch": 4.7036328871892925, "grad_norm": 0.8483895659446716, "learning_rate": 4.9193333333333336e-05, "loss": 0.0712, "step": 7380 }, { "epoch": 4.710006373486297, "grad_norm": 0.504065752029419, "learning_rate": 4.926e-05, "loss": 0.064, "step": 7390 }, { "epoch": 4.716379859783301, "grad_norm": 0.6554313898086548, "learning_rate": 4.932666666666667e-05, "loss": 0.0585, "step": 7400 }, { "epoch": 4.722753346080306, "grad_norm": 0.6560725569725037, "learning_rate": 4.9393333333333334e-05, "loss": 0.068, "step": 7410 }, { "epoch": 4.72912683237731, "grad_norm": 0.7313477396965027, "learning_rate": 4.946e-05, "loss": 0.066, "step": 7420 }, { "epoch": 4.7355003186743145, "grad_norm": 0.8558791279792786, "learning_rate": 4.952666666666667e-05, "loss": 0.0774, "step": 7430 }, { "epoch": 4.741873804971319, "grad_norm": 0.6880666017532349, "learning_rate": 4.959333333333334e-05, "loss": 0.0684, "step": 7440 }, { "epoch": 4.748247291268324, "grad_norm": 0.9492466449737549, "learning_rate": 4.966e-05, "loss": 0.067, "step": 7450 }, { "epoch": 4.754620777565329, "grad_norm": 0.786095142364502, "learning_rate": 4.972666666666667e-05, "loss": 0.0584, "step": 7460 }, { "epoch": 4.760994263862333, "grad_norm": 0.6420959234237671, "learning_rate": 4.9793333333333337e-05, "loss": 0.072, "step": 7470 }, { "epoch": 4.7673677501593374, "grad_norm": 0.598442792892456, "learning_rate": 4.986e-05, "loss": 0.0655, "step": 7480 }, { "epoch": 4.773741236456342, "grad_norm": 1.136210560798645, "learning_rate": 4.992666666666667e-05, "loss": 0.0667, "step": 7490 }, { "epoch": 4.780114722753346, "grad_norm": 1.0655301809310913, "learning_rate": 4.9993333333333335e-05, "loss": 0.0606, "step": 7500 }, { "epoch": 4.786488209050351, "grad_norm": 0.7716424465179443, "learning_rate": 5.0060000000000005e-05, "loss": 0.0501, "step": 7510 }, { "epoch": 4.792861695347355, "grad_norm": 0.7212883234024048, "learning_rate": 5.012666666666667e-05, "loss": 0.0641, "step": 7520 }, { "epoch": 4.7992351816443595, "grad_norm": 0.8329992294311523, "learning_rate": 5.019333333333334e-05, "loss": 0.0429, "step": 7530 }, { "epoch": 4.805608667941364, "grad_norm": 0.5419008731842041, "learning_rate": 5.026000000000001e-05, "loss": 0.0558, "step": 7540 }, { "epoch": 4.811982154238368, "grad_norm": 0.4046176075935364, "learning_rate": 5.032666666666667e-05, "loss": 0.0522, "step": 7550 }, { "epoch": 4.818355640535373, "grad_norm": 0.7823494076728821, "learning_rate": 5.039333333333334e-05, "loss": 0.0629, "step": 7560 }, { "epoch": 4.824729126832377, "grad_norm": 1.1260772943496704, "learning_rate": 5.046000000000001e-05, "loss": 0.0743, "step": 7570 }, { "epoch": 4.8311026131293815, "grad_norm": 1.0845251083374023, "learning_rate": 5.0526666666666665e-05, "loss": 0.0669, "step": 7580 }, { "epoch": 4.837476099426386, "grad_norm": 0.9632096886634827, "learning_rate": 5.0593333333333336e-05, "loss": 0.0655, "step": 7590 }, { "epoch": 4.84384958572339, "grad_norm": 0.7473251223564148, "learning_rate": 5.0660000000000006e-05, "loss": 0.0737, "step": 7600 }, { "epoch": 4.850223072020395, "grad_norm": 0.8181489109992981, "learning_rate": 5.072666666666667e-05, "loss": 0.0695, "step": 7610 }, { "epoch": 4.8565965583174, "grad_norm": 0.8925595879554749, "learning_rate": 5.079333333333334e-05, "loss": 0.0621, "step": 7620 }, { "epoch": 4.862970044614404, "grad_norm": 0.9254628419876099, "learning_rate": 5.0860000000000004e-05, "loss": 0.0559, "step": 7630 }, { "epoch": 4.869343530911409, "grad_norm": 0.7745023965835571, "learning_rate": 5.092666666666667e-05, "loss": 0.0587, "step": 7640 }, { "epoch": 4.875717017208413, "grad_norm": 1.0639039278030396, "learning_rate": 5.099333333333334e-05, "loss": 0.0738, "step": 7650 }, { "epoch": 4.882090503505418, "grad_norm": 0.8726522922515869, "learning_rate": 5.106000000000001e-05, "loss": 0.0648, "step": 7660 }, { "epoch": 4.888463989802422, "grad_norm": 1.01661217212677, "learning_rate": 5.1126666666666666e-05, "loss": 0.0696, "step": 7670 }, { "epoch": 4.894837476099426, "grad_norm": 0.8254091739654541, "learning_rate": 5.119333333333334e-05, "loss": 0.0679, "step": 7680 }, { "epoch": 4.901210962396431, "grad_norm": 1.0373421907424927, "learning_rate": 5.1259999999999994e-05, "loss": 0.0675, "step": 7690 }, { "epoch": 4.907584448693435, "grad_norm": 1.0218459367752075, "learning_rate": 5.1326666666666664e-05, "loss": 0.0655, "step": 7700 }, { "epoch": 4.91395793499044, "grad_norm": 1.1121819019317627, "learning_rate": 5.1393333333333335e-05, "loss": 0.0672, "step": 7710 }, { "epoch": 4.920331421287444, "grad_norm": 0.6847512125968933, "learning_rate": 5.146e-05, "loss": 0.0812, "step": 7720 }, { "epoch": 4.9267049075844485, "grad_norm": 0.5663353204727173, "learning_rate": 5.152666666666667e-05, "loss": 0.0619, "step": 7730 }, { "epoch": 4.933078393881453, "grad_norm": 0.7508721947669983, "learning_rate": 5.159333333333334e-05, "loss": 0.0695, "step": 7740 }, { "epoch": 4.939451880178457, "grad_norm": 0.8792316317558289, "learning_rate": 5.166e-05, "loss": 0.0621, "step": 7750 }, { "epoch": 4.945825366475463, "grad_norm": 1.2483041286468506, "learning_rate": 5.172666666666667e-05, "loss": 0.0749, "step": 7760 }, { "epoch": 4.952198852772467, "grad_norm": 1.3614460229873657, "learning_rate": 5.179333333333334e-05, "loss": 0.0638, "step": 7770 }, { "epoch": 4.958572339069471, "grad_norm": 0.8524526953697205, "learning_rate": 5.1859999999999995e-05, "loss": 0.0772, "step": 7780 }, { "epoch": 4.964945825366476, "grad_norm": 0.7259225845336914, "learning_rate": 5.1926666666666665e-05, "loss": 0.0716, "step": 7790 }, { "epoch": 4.97131931166348, "grad_norm": 0.46326547861099243, "learning_rate": 5.1993333333333336e-05, "loss": 0.0617, "step": 7800 }, { "epoch": 4.977692797960485, "grad_norm": 0.9748802185058594, "learning_rate": 5.206e-05, "loss": 0.0578, "step": 7810 }, { "epoch": 4.984066284257489, "grad_norm": 0.8784742951393127, "learning_rate": 5.212666666666667e-05, "loss": 0.0532, "step": 7820 }, { "epoch": 4.990439770554493, "grad_norm": 0.6700788140296936, "learning_rate": 5.219333333333334e-05, "loss": 0.0597, "step": 7830 }, { "epoch": 4.996813256851498, "grad_norm": 0.5486948490142822, "learning_rate": 5.226e-05, "loss": 0.0622, "step": 7840 }, { "epoch": 5.003186743148502, "grad_norm": 0.5421954989433289, "learning_rate": 5.232666666666667e-05, "loss": 0.0669, "step": 7850 }, { "epoch": 5.009560229445507, "grad_norm": 0.5749859809875488, "learning_rate": 5.239333333333334e-05, "loss": 0.0637, "step": 7860 }, { "epoch": 5.015933715742511, "grad_norm": 0.7803623676300049, "learning_rate": 5.2459999999999996e-05, "loss": 0.0662, "step": 7870 }, { "epoch": 5.022307202039515, "grad_norm": 1.003782868385315, "learning_rate": 5.2526666666666666e-05, "loss": 0.0667, "step": 7880 }, { "epoch": 5.02868068833652, "grad_norm": 1.059926986694336, "learning_rate": 5.259333333333334e-05, "loss": 0.0552, "step": 7890 }, { "epoch": 5.035054174633524, "grad_norm": 0.7520424127578735, "learning_rate": 5.266e-05, "loss": 0.064, "step": 7900 }, { "epoch": 5.041427660930529, "grad_norm": 0.5341896414756775, "learning_rate": 5.272666666666667e-05, "loss": 0.0636, "step": 7910 }, { "epoch": 5.047801147227533, "grad_norm": 0.7045947313308716, "learning_rate": 5.279333333333334e-05, "loss": 0.0758, "step": 7920 }, { "epoch": 5.054174633524538, "grad_norm": 0.7141209840774536, "learning_rate": 5.286e-05, "loss": 0.053, "step": 7930 }, { "epoch": 5.060548119821543, "grad_norm": 0.7655815482139587, "learning_rate": 5.292666666666667e-05, "loss": 0.0647, "step": 7940 }, { "epoch": 5.066921606118547, "grad_norm": 0.6709862947463989, "learning_rate": 5.299333333333334e-05, "loss": 0.0603, "step": 7950 }, { "epoch": 5.0732950924155515, "grad_norm": 1.0763871669769287, "learning_rate": 5.306e-05, "loss": 0.0716, "step": 7960 }, { "epoch": 5.079668578712556, "grad_norm": 0.6360602974891663, "learning_rate": 5.312666666666667e-05, "loss": 0.0578, "step": 7970 }, { "epoch": 5.08604206500956, "grad_norm": 0.6845023036003113, "learning_rate": 5.319333333333334e-05, "loss": 0.0599, "step": 7980 }, { "epoch": 5.092415551306565, "grad_norm": 1.253475546836853, "learning_rate": 5.326e-05, "loss": 0.0686, "step": 7990 }, { "epoch": 5.098789037603569, "grad_norm": 0.881453812122345, "learning_rate": 5.332666666666667e-05, "loss": 0.0545, "step": 8000 }, { "epoch": 5.105162523900574, "grad_norm": 1.1186039447784424, "learning_rate": 5.339333333333334e-05, "loss": 0.063, "step": 8010 }, { "epoch": 5.111536010197578, "grad_norm": 0.542425811290741, "learning_rate": 5.346e-05, "loss": 0.0566, "step": 8020 }, { "epoch": 5.117909496494582, "grad_norm": 0.6032022833824158, "learning_rate": 5.352666666666667e-05, "loss": 0.0604, "step": 8030 }, { "epoch": 5.124282982791587, "grad_norm": 0.7036418318748474, "learning_rate": 5.359333333333334e-05, "loss": 0.0585, "step": 8040 }, { "epoch": 5.130656469088591, "grad_norm": 0.5876562595367432, "learning_rate": 5.366e-05, "loss": 0.0699, "step": 8050 }, { "epoch": 5.137029955385596, "grad_norm": 1.1779491901397705, "learning_rate": 5.372666666666667e-05, "loss": 0.0626, "step": 8060 }, { "epoch": 5.1434034416826, "grad_norm": 0.5704241991043091, "learning_rate": 5.379333333333334e-05, "loss": 0.0604, "step": 8070 }, { "epoch": 5.149776927979605, "grad_norm": 0.7771600484848022, "learning_rate": 5.386e-05, "loss": 0.0565, "step": 8080 }, { "epoch": 5.15615041427661, "grad_norm": 0.694642186164856, "learning_rate": 5.3926666666666666e-05, "loss": 0.0733, "step": 8090 }, { "epoch": 5.162523900573614, "grad_norm": 0.9646157026290894, "learning_rate": 5.399333333333334e-05, "loss": 0.0704, "step": 8100 }, { "epoch": 5.1688973868706185, "grad_norm": 0.5102450847625732, "learning_rate": 5.406e-05, "loss": 0.0633, "step": 8110 }, { "epoch": 5.175270873167623, "grad_norm": 0.8537219762802124, "learning_rate": 5.412666666666667e-05, "loss": 0.0601, "step": 8120 }, { "epoch": 5.181644359464627, "grad_norm": 0.8907761573791504, "learning_rate": 5.419333333333334e-05, "loss": 0.0616, "step": 8130 }, { "epoch": 5.188017845761632, "grad_norm": 0.9172653555870056, "learning_rate": 5.426e-05, "loss": 0.087, "step": 8140 }, { "epoch": 5.194391332058636, "grad_norm": 0.6922829747200012, "learning_rate": 5.432666666666667e-05, "loss": 0.0627, "step": 8150 }, { "epoch": 5.2007648183556405, "grad_norm": 0.6184099912643433, "learning_rate": 5.439333333333334e-05, "loss": 0.0682, "step": 8160 }, { "epoch": 5.207138304652645, "grad_norm": 0.3626927137374878, "learning_rate": 5.446e-05, "loss": 0.0694, "step": 8170 }, { "epoch": 5.213511790949649, "grad_norm": 0.7159704566001892, "learning_rate": 5.452666666666667e-05, "loss": 0.0532, "step": 8180 }, { "epoch": 5.219885277246654, "grad_norm": 0.42213982343673706, "learning_rate": 5.459333333333334e-05, "loss": 0.0588, "step": 8190 }, { "epoch": 5.226258763543658, "grad_norm": 1.0273957252502441, "learning_rate": 5.466e-05, "loss": 0.0654, "step": 8200 }, { "epoch": 5.2326322498406626, "grad_norm": 0.8868750929832458, "learning_rate": 5.472666666666667e-05, "loss": 0.0466, "step": 8210 }, { "epoch": 5.239005736137667, "grad_norm": 0.42625030875205994, "learning_rate": 5.479333333333334e-05, "loss": 0.0525, "step": 8220 }, { "epoch": 5.245379222434671, "grad_norm": 0.40689733624458313, "learning_rate": 5.486e-05, "loss": 0.05, "step": 8230 }, { "epoch": 5.251752708731676, "grad_norm": 0.838309645652771, "learning_rate": 5.492666666666667e-05, "loss": 0.0655, "step": 8240 }, { "epoch": 5.258126195028681, "grad_norm": 0.6672176122665405, "learning_rate": 5.499333333333334e-05, "loss": 0.0616, "step": 8250 }, { "epoch": 5.2644996813256855, "grad_norm": 0.6883916854858398, "learning_rate": 5.506e-05, "loss": 0.044, "step": 8260 }, { "epoch": 5.27087316762269, "grad_norm": 0.7790850400924683, "learning_rate": 5.512666666666667e-05, "loss": 0.0619, "step": 8270 }, { "epoch": 5.277246653919694, "grad_norm": 0.6095466613769531, "learning_rate": 5.519333333333334e-05, "loss": 0.0685, "step": 8280 }, { "epoch": 5.283620140216699, "grad_norm": 1.0407490730285645, "learning_rate": 5.526e-05, "loss": 0.0696, "step": 8290 }, { "epoch": 5.289993626513703, "grad_norm": 0.9175785183906555, "learning_rate": 5.5326666666666673e-05, "loss": 0.0649, "step": 8300 }, { "epoch": 5.2963671128107075, "grad_norm": 0.8231756687164307, "learning_rate": 5.5393333333333344e-05, "loss": 0.0678, "step": 8310 }, { "epoch": 5.302740599107712, "grad_norm": 0.7577474117279053, "learning_rate": 5.546e-05, "loss": 0.0609, "step": 8320 }, { "epoch": 5.309114085404716, "grad_norm": 0.8225215077400208, "learning_rate": 5.552666666666667e-05, "loss": 0.0566, "step": 8330 }, { "epoch": 5.315487571701721, "grad_norm": 0.7148070335388184, "learning_rate": 5.559333333333333e-05, "loss": 0.0538, "step": 8340 }, { "epoch": 5.321861057998725, "grad_norm": 0.5415747761726379, "learning_rate": 5.566e-05, "loss": 0.0656, "step": 8350 }, { "epoch": 5.3282345442957295, "grad_norm": 0.5351488590240479, "learning_rate": 5.572666666666667e-05, "loss": 0.0517, "step": 8360 }, { "epoch": 5.334608030592734, "grad_norm": 0.988207995891571, "learning_rate": 5.579333333333333e-05, "loss": 0.0588, "step": 8370 }, { "epoch": 5.340981516889738, "grad_norm": 0.5303449630737305, "learning_rate": 5.5860000000000004e-05, "loss": 0.0578, "step": 8380 }, { "epoch": 5.347355003186744, "grad_norm": 0.3682520091533661, "learning_rate": 5.5926666666666674e-05, "loss": 0.0539, "step": 8390 }, { "epoch": 5.353728489483748, "grad_norm": 0.9374811053276062, "learning_rate": 5.599333333333333e-05, "loss": 0.0565, "step": 8400 }, { "epoch": 5.360101975780752, "grad_norm": 0.8362928032875061, "learning_rate": 5.606e-05, "loss": 0.0605, "step": 8410 }, { "epoch": 5.366475462077757, "grad_norm": 0.6892371773719788, "learning_rate": 5.612666666666667e-05, "loss": 0.0514, "step": 8420 }, { "epoch": 5.372848948374761, "grad_norm": 0.6773945689201355, "learning_rate": 5.619333333333333e-05, "loss": 0.0512, "step": 8430 }, { "epoch": 5.379222434671766, "grad_norm": 0.5313262939453125, "learning_rate": 5.626e-05, "loss": 0.0533, "step": 8440 }, { "epoch": 5.38559592096877, "grad_norm": 0.554133951663971, "learning_rate": 5.632666666666667e-05, "loss": 0.0545, "step": 8450 }, { "epoch": 5.3919694072657744, "grad_norm": 0.7247320413589478, "learning_rate": 5.6393333333333334e-05, "loss": 0.0521, "step": 8460 }, { "epoch": 5.398342893562779, "grad_norm": 0.561715304851532, "learning_rate": 5.6460000000000005e-05, "loss": 0.0433, "step": 8470 }, { "epoch": 5.404716379859783, "grad_norm": 0.6583260297775269, "learning_rate": 5.6526666666666675e-05, "loss": 0.0461, "step": 8480 }, { "epoch": 5.411089866156788, "grad_norm": 1.1962138414382935, "learning_rate": 5.659333333333333e-05, "loss": 0.0592, "step": 8490 }, { "epoch": 5.417463352453792, "grad_norm": 0.8981515765190125, "learning_rate": 5.666e-05, "loss": 0.084, "step": 8500 }, { "epoch": 5.4238368387507965, "grad_norm": 0.5341464877128601, "learning_rate": 5.6726666666666674e-05, "loss": 0.0573, "step": 8510 }, { "epoch": 5.430210325047801, "grad_norm": 0.5179601311683655, "learning_rate": 5.679333333333333e-05, "loss": 0.0512, "step": 8520 }, { "epoch": 5.436583811344805, "grad_norm": 0.8599264025688171, "learning_rate": 5.686e-05, "loss": 0.0575, "step": 8530 }, { "epoch": 5.44295729764181, "grad_norm": 0.9956772923469543, "learning_rate": 5.692666666666667e-05, "loss": 0.0611, "step": 8540 }, { "epoch": 5.449330783938814, "grad_norm": 0.5431164503097534, "learning_rate": 5.699333333333333e-05, "loss": 0.0617, "step": 8550 }, { "epoch": 5.455704270235819, "grad_norm": 0.3895617723464966, "learning_rate": 5.706e-05, "loss": 0.0623, "step": 8560 }, { "epoch": 5.462077756532824, "grad_norm": 0.5155817270278931, "learning_rate": 5.712666666666667e-05, "loss": 0.077, "step": 8570 }, { "epoch": 5.468451242829828, "grad_norm": 0.7464935779571533, "learning_rate": 5.7193333333333333e-05, "loss": 0.0585, "step": 8580 }, { "epoch": 5.474824729126833, "grad_norm": 0.47496673464775085, "learning_rate": 5.7260000000000004e-05, "loss": 0.0725, "step": 8590 }, { "epoch": 5.481198215423837, "grad_norm": 0.48573318123817444, "learning_rate": 5.7326666666666675e-05, "loss": 0.0423, "step": 8600 }, { "epoch": 5.487571701720841, "grad_norm": 0.47488465905189514, "learning_rate": 5.739333333333333e-05, "loss": 0.0457, "step": 8610 }, { "epoch": 5.493945188017846, "grad_norm": 0.698132336139679, "learning_rate": 5.746e-05, "loss": 0.0593, "step": 8620 }, { "epoch": 5.50031867431485, "grad_norm": 0.6548746824264526, "learning_rate": 5.752666666666667e-05, "loss": 0.073, "step": 8630 }, { "epoch": 5.506692160611855, "grad_norm": 0.8008704781532288, "learning_rate": 5.759333333333333e-05, "loss": 0.0519, "step": 8640 }, { "epoch": 5.513065646908859, "grad_norm": 0.5915107131004333, "learning_rate": 5.766e-05, "loss": 0.069, "step": 8650 }, { "epoch": 5.519439133205863, "grad_norm": 0.5799915790557861, "learning_rate": 5.772666666666667e-05, "loss": 0.0645, "step": 8660 }, { "epoch": 5.525812619502868, "grad_norm": 1.300042748451233, "learning_rate": 5.7793333333333335e-05, "loss": 0.0705, "step": 8670 }, { "epoch": 5.532186105799872, "grad_norm": 0.764645516872406, "learning_rate": 5.7860000000000005e-05, "loss": 0.0476, "step": 8680 }, { "epoch": 5.538559592096877, "grad_norm": 0.7687389850616455, "learning_rate": 5.7926666666666676e-05, "loss": 0.0611, "step": 8690 }, { "epoch": 5.544933078393882, "grad_norm": 1.612897276878357, "learning_rate": 5.799333333333333e-05, "loss": 0.0652, "step": 8700 }, { "epoch": 5.551306564690886, "grad_norm": 0.5542522668838501, "learning_rate": 5.806e-05, "loss": 0.0563, "step": 8710 }, { "epoch": 5.557680050987891, "grad_norm": 0.8147121667861938, "learning_rate": 5.8126666666666674e-05, "loss": 0.0678, "step": 8720 }, { "epoch": 5.564053537284895, "grad_norm": 0.9129625558853149, "learning_rate": 5.819333333333333e-05, "loss": 0.0637, "step": 8730 }, { "epoch": 5.5704270235819, "grad_norm": 0.6277623176574707, "learning_rate": 5.826e-05, "loss": 0.0552, "step": 8740 }, { "epoch": 5.576800509878904, "grad_norm": 0.5057451128959656, "learning_rate": 5.832666666666667e-05, "loss": 0.0566, "step": 8750 }, { "epoch": 5.583173996175908, "grad_norm": 0.5900144577026367, "learning_rate": 5.8393333333333336e-05, "loss": 0.0577, "step": 8760 }, { "epoch": 5.589547482472913, "grad_norm": 0.6781315803527832, "learning_rate": 5.8460000000000006e-05, "loss": 0.0549, "step": 8770 }, { "epoch": 5.595920968769917, "grad_norm": 0.7219472527503967, "learning_rate": 5.852666666666668e-05, "loss": 0.063, "step": 8780 }, { "epoch": 5.602294455066922, "grad_norm": 0.7543862462043762, "learning_rate": 5.8593333333333334e-05, "loss": 0.0578, "step": 8790 }, { "epoch": 5.608667941363926, "grad_norm": 0.8206136226654053, "learning_rate": 5.8660000000000004e-05, "loss": 0.0491, "step": 8800 }, { "epoch": 5.61504142766093, "grad_norm": 0.9566856622695923, "learning_rate": 5.8726666666666675e-05, "loss": 0.0557, "step": 8810 }, { "epoch": 5.621414913957935, "grad_norm": 0.9145086407661438, "learning_rate": 5.879333333333333e-05, "loss": 0.0543, "step": 8820 }, { "epoch": 5.627788400254939, "grad_norm": 0.48449772596359253, "learning_rate": 5.886e-05, "loss": 0.0548, "step": 8830 }, { "epoch": 5.634161886551944, "grad_norm": 0.7416876554489136, "learning_rate": 5.892666666666667e-05, "loss": 0.0565, "step": 8840 }, { "epoch": 5.640535372848948, "grad_norm": 0.5322074890136719, "learning_rate": 5.8993333333333337e-05, "loss": 0.0528, "step": 8850 }, { "epoch": 5.646908859145952, "grad_norm": 0.7602063417434692, "learning_rate": 5.906000000000001e-05, "loss": 0.0569, "step": 8860 }, { "epoch": 5.653282345442957, "grad_norm": 0.6393805146217346, "learning_rate": 5.912666666666668e-05, "loss": 0.0476, "step": 8870 }, { "epoch": 5.659655831739962, "grad_norm": 0.8366895914077759, "learning_rate": 5.9193333333333335e-05, "loss": 0.0758, "step": 8880 }, { "epoch": 5.6660293180369665, "grad_norm": 0.6291630268096924, "learning_rate": 5.9260000000000005e-05, "loss": 0.0673, "step": 8890 }, { "epoch": 5.672402804333971, "grad_norm": 0.8264247179031372, "learning_rate": 5.9326666666666676e-05, "loss": 0.0578, "step": 8900 }, { "epoch": 5.678776290630975, "grad_norm": 0.8180743455886841, "learning_rate": 5.939333333333333e-05, "loss": 0.0611, "step": 8910 }, { "epoch": 5.68514977692798, "grad_norm": 1.2748998403549194, "learning_rate": 5.946e-05, "loss": 0.0734, "step": 8920 }, { "epoch": 5.691523263224984, "grad_norm": 0.7235389351844788, "learning_rate": 5.9526666666666674e-05, "loss": 0.0628, "step": 8930 }, { "epoch": 5.6978967495219885, "grad_norm": 0.7432854175567627, "learning_rate": 5.959333333333334e-05, "loss": 0.0727, "step": 8940 }, { "epoch": 5.704270235818993, "grad_norm": 0.45918038487434387, "learning_rate": 5.966000000000001e-05, "loss": 0.0519, "step": 8950 }, { "epoch": 5.710643722115997, "grad_norm": 0.797531008720398, "learning_rate": 5.9726666666666665e-05, "loss": 0.0549, "step": 8960 }, { "epoch": 5.717017208413002, "grad_norm": 0.8184925317764282, "learning_rate": 5.9793333333333336e-05, "loss": 0.0537, "step": 8970 }, { "epoch": 5.723390694710006, "grad_norm": 0.7258854508399963, "learning_rate": 5.9860000000000006e-05, "loss": 0.0672, "step": 8980 }, { "epoch": 5.729764181007011, "grad_norm": 0.8069808483123779, "learning_rate": 5.992666666666666e-05, "loss": 0.0845, "step": 8990 }, { "epoch": 5.736137667304015, "grad_norm": 0.6261904239654541, "learning_rate": 5.9993333333333334e-05, "loss": 0.066, "step": 9000 }, { "epoch": 5.742511153601019, "grad_norm": 0.8198938369750977, "learning_rate": 6.0060000000000004e-05, "loss": 0.0539, "step": 9010 }, { "epoch": 5.748884639898025, "grad_norm": 0.5557858347892761, "learning_rate": 6.012666666666666e-05, "loss": 0.0566, "step": 9020 }, { "epoch": 5.755258126195029, "grad_norm": 0.5786111354827881, "learning_rate": 6.019333333333333e-05, "loss": 0.0729, "step": 9030 }, { "epoch": 5.7616316124920335, "grad_norm": 0.4758230149745941, "learning_rate": 6.026e-05, "loss": 0.0581, "step": 9040 }, { "epoch": 5.768005098789038, "grad_norm": 0.7065956592559814, "learning_rate": 6.0326666666666666e-05, "loss": 0.0617, "step": 9050 }, { "epoch": 5.774378585086042, "grad_norm": 0.8448079228401184, "learning_rate": 6.039333333333334e-05, "loss": 0.0676, "step": 9060 }, { "epoch": 5.780752071383047, "grad_norm": 0.6785176992416382, "learning_rate": 6.046000000000001e-05, "loss": 0.0585, "step": 9070 }, { "epoch": 5.787125557680051, "grad_norm": 1.0336339473724365, "learning_rate": 6.0526666666666664e-05, "loss": 0.0655, "step": 9080 }, { "epoch": 5.7934990439770555, "grad_norm": 0.620617151260376, "learning_rate": 6.0593333333333335e-05, "loss": 0.0644, "step": 9090 }, { "epoch": 5.79987253027406, "grad_norm": 0.9712967872619629, "learning_rate": 6.0660000000000005e-05, "loss": 0.0697, "step": 9100 }, { "epoch": 5.806246016571064, "grad_norm": 0.8093079328536987, "learning_rate": 6.072666666666666e-05, "loss": 0.0604, "step": 9110 }, { "epoch": 5.812619502868069, "grad_norm": 0.714555025100708, "learning_rate": 6.079333333333333e-05, "loss": 0.0614, "step": 9120 }, { "epoch": 5.818992989165073, "grad_norm": 0.6867609620094299, "learning_rate": 6.0860000000000003e-05, "loss": 0.0707, "step": 9130 }, { "epoch": 5.8253664754620775, "grad_norm": 0.6431583762168884, "learning_rate": 6.092666666666667e-05, "loss": 0.0573, "step": 9140 }, { "epoch": 5.831739961759082, "grad_norm": 0.5615859031677246, "learning_rate": 6.099333333333334e-05, "loss": 0.07, "step": 9150 }, { "epoch": 5.838113448056086, "grad_norm": 0.4224008619785309, "learning_rate": 6.106e-05, "loss": 0.0514, "step": 9160 }, { "epoch": 5.844486934353091, "grad_norm": 0.769130289554596, "learning_rate": 6.112666666666667e-05, "loss": 0.0467, "step": 9170 }, { "epoch": 5.850860420650095, "grad_norm": 1.0572727918624878, "learning_rate": 6.119333333333333e-05, "loss": 0.0726, "step": 9180 }, { "epoch": 5.8572339069471, "grad_norm": 0.9408935308456421, "learning_rate": 6.126e-05, "loss": 0.0555, "step": 9190 }, { "epoch": 5.863607393244105, "grad_norm": 0.7971311211585999, "learning_rate": 6.132666666666667e-05, "loss": 0.0618, "step": 9200 }, { "epoch": 5.869980879541109, "grad_norm": 0.6989256739616394, "learning_rate": 6.139333333333333e-05, "loss": 0.0646, "step": 9210 }, { "epoch": 5.876354365838114, "grad_norm": 0.7526812553405762, "learning_rate": 6.146000000000001e-05, "loss": 0.0731, "step": 9220 }, { "epoch": 5.882727852135118, "grad_norm": 0.9330829381942749, "learning_rate": 6.152666666666666e-05, "loss": 0.0659, "step": 9230 }, { "epoch": 5.8891013384321225, "grad_norm": 0.5467667579650879, "learning_rate": 6.159333333333334e-05, "loss": 0.0655, "step": 9240 }, { "epoch": 5.895474824729127, "grad_norm": 0.5973659157752991, "learning_rate": 6.166e-05, "loss": 0.0488, "step": 9250 }, { "epoch": 5.901848311026131, "grad_norm": 0.4567822813987732, "learning_rate": 6.172666666666667e-05, "loss": 0.0544, "step": 9260 }, { "epoch": 5.908221797323136, "grad_norm": 0.6740135550498962, "learning_rate": 6.179333333333333e-05, "loss": 0.056, "step": 9270 }, { "epoch": 5.91459528362014, "grad_norm": 0.6389235854148865, "learning_rate": 6.186000000000001e-05, "loss": 0.0704, "step": 9280 }, { "epoch": 5.9209687699171445, "grad_norm": 0.5664303302764893, "learning_rate": 6.192666666666667e-05, "loss": 0.0544, "step": 9290 }, { "epoch": 5.927342256214149, "grad_norm": 0.39746248722076416, "learning_rate": 6.199333333333333e-05, "loss": 0.0693, "step": 9300 }, { "epoch": 5.933715742511153, "grad_norm": 0.6555776596069336, "learning_rate": 6.206000000000001e-05, "loss": 0.0724, "step": 9310 }, { "epoch": 5.940089228808158, "grad_norm": 1.0497114658355713, "learning_rate": 6.212666666666666e-05, "loss": 0.067, "step": 9320 }, { "epoch": 5.946462715105163, "grad_norm": 0.6835218667984009, "learning_rate": 6.219333333333334e-05, "loss": 0.0704, "step": 9330 }, { "epoch": 5.952836201402167, "grad_norm": 0.6713403463363647, "learning_rate": 6.226e-05, "loss": 0.0713, "step": 9340 }, { "epoch": 5.959209687699172, "grad_norm": 0.7689964175224304, "learning_rate": 6.232666666666667e-05, "loss": 0.0566, "step": 9350 }, { "epoch": 5.965583173996176, "grad_norm": 0.7639724016189575, "learning_rate": 6.239333333333333e-05, "loss": 0.0516, "step": 9360 }, { "epoch": 5.971956660293181, "grad_norm": 0.4493684470653534, "learning_rate": 6.246000000000001e-05, "loss": 0.0686, "step": 9370 }, { "epoch": 5.978330146590185, "grad_norm": 0.5575623512268066, "learning_rate": 6.252666666666667e-05, "loss": 0.0507, "step": 9380 }, { "epoch": 5.984703632887189, "grad_norm": 0.4197111427783966, "learning_rate": 6.259333333333334e-05, "loss": 0.0624, "step": 9390 }, { "epoch": 5.991077119184194, "grad_norm": 0.8092382550239563, "learning_rate": 6.266000000000001e-05, "loss": 0.0648, "step": 9400 }, { "epoch": 5.997450605481198, "grad_norm": 0.5399866104125977, "learning_rate": 6.272666666666666e-05, "loss": 0.0617, "step": 9410 }, { "epoch": 6.003824091778203, "grad_norm": 0.5049969553947449, "learning_rate": 6.279333333333334e-05, "loss": 0.0478, "step": 9420 }, { "epoch": 6.010197578075207, "grad_norm": 0.5460518002510071, "learning_rate": 6.286e-05, "loss": 0.0553, "step": 9430 }, { "epoch": 6.0165710643722115, "grad_norm": 0.7110340595245361, "learning_rate": 6.292666666666667e-05, "loss": 0.0535, "step": 9440 }, { "epoch": 6.022944550669216, "grad_norm": 0.6920067071914673, "learning_rate": 6.299333333333333e-05, "loss": 0.0651, "step": 9450 }, { "epoch": 6.02931803696622, "grad_norm": 0.466105192899704, "learning_rate": 6.306000000000001e-05, "loss": 0.0488, "step": 9460 }, { "epoch": 6.035691523263225, "grad_norm": 0.4926980435848236, "learning_rate": 6.312666666666667e-05, "loss": 0.0718, "step": 9470 }, { "epoch": 6.042065009560229, "grad_norm": 0.6662371158599854, "learning_rate": 6.319333333333334e-05, "loss": 0.0612, "step": 9480 }, { "epoch": 6.0484384958572335, "grad_norm": 0.9834148287773132, "learning_rate": 6.326000000000001e-05, "loss": 0.0528, "step": 9490 }, { "epoch": 6.054811982154239, "grad_norm": 0.7669191360473633, "learning_rate": 6.332666666666666e-05, "loss": 0.054, "step": 9500 }, { "epoch": 6.061185468451243, "grad_norm": 0.9060104489326477, "learning_rate": 6.339333333333334e-05, "loss": 0.0551, "step": 9510 }, { "epoch": 6.067558954748248, "grad_norm": 0.5581040382385254, "learning_rate": 6.346e-05, "loss": 0.0454, "step": 9520 }, { "epoch": 6.073932441045252, "grad_norm": 0.6774866580963135, "learning_rate": 6.352666666666667e-05, "loss": 0.0656, "step": 9530 }, { "epoch": 6.080305927342256, "grad_norm": 0.6943061351776123, "learning_rate": 6.359333333333333e-05, "loss": 0.0607, "step": 9540 }, { "epoch": 6.086679413639261, "grad_norm": 0.7664393782615662, "learning_rate": 6.366000000000001e-05, "loss": 0.0647, "step": 9550 }, { "epoch": 6.093052899936265, "grad_norm": 0.6050956845283508, "learning_rate": 6.372666666666667e-05, "loss": 0.0576, "step": 9560 }, { "epoch": 6.09942638623327, "grad_norm": 0.8026940226554871, "learning_rate": 6.379333333333334e-05, "loss": 0.0586, "step": 9570 }, { "epoch": 6.105799872530274, "grad_norm": 0.7366113662719727, "learning_rate": 6.386e-05, "loss": 0.0788, "step": 9580 }, { "epoch": 6.112173358827278, "grad_norm": 0.6704623103141785, "learning_rate": 6.392666666666667e-05, "loss": 0.045, "step": 9590 }, { "epoch": 6.118546845124283, "grad_norm": 0.7459873557090759, "learning_rate": 6.399333333333334e-05, "loss": 0.0701, "step": 9600 }, { "epoch": 6.124920331421287, "grad_norm": 0.5031256675720215, "learning_rate": 6.405999999999999e-05, "loss": 0.0514, "step": 9610 }, { "epoch": 6.131293817718292, "grad_norm": 0.8962956070899963, "learning_rate": 6.412666666666667e-05, "loss": 0.0709, "step": 9620 }, { "epoch": 6.137667304015296, "grad_norm": 0.5765659213066101, "learning_rate": 6.419333333333333e-05, "loss": 0.0525, "step": 9630 }, { "epoch": 6.1440407903123, "grad_norm": 0.7239110469818115, "learning_rate": 6.426e-05, "loss": 0.0506, "step": 9640 }, { "epoch": 6.150414276609306, "grad_norm": 0.7703732848167419, "learning_rate": 6.432666666666666e-05, "loss": 0.0564, "step": 9650 }, { "epoch": 6.15678776290631, "grad_norm": 0.6895337104797363, "learning_rate": 6.439333333333334e-05, "loss": 0.0554, "step": 9660 }, { "epoch": 6.1631612492033145, "grad_norm": 0.8975412249565125, "learning_rate": 6.446e-05, "loss": 0.071, "step": 9670 }, { "epoch": 6.169534735500319, "grad_norm": 0.7354782223701477, "learning_rate": 6.452666666666667e-05, "loss": 0.0735, "step": 9680 }, { "epoch": 6.175908221797323, "grad_norm": 0.6420743465423584, "learning_rate": 6.459333333333334e-05, "loss": 0.0668, "step": 9690 }, { "epoch": 6.182281708094328, "grad_norm": 0.9225689768791199, "learning_rate": 6.466e-05, "loss": 0.0569, "step": 9700 }, { "epoch": 6.188655194391332, "grad_norm": 1.1190143823623657, "learning_rate": 6.472666666666667e-05, "loss": 0.0593, "step": 9710 }, { "epoch": 6.195028680688337, "grad_norm": 0.5006576180458069, "learning_rate": 6.479333333333334e-05, "loss": 0.0521, "step": 9720 }, { "epoch": 6.201402166985341, "grad_norm": 0.9682351350784302, "learning_rate": 6.486e-05, "loss": 0.0696, "step": 9730 }, { "epoch": 6.207775653282345, "grad_norm": 0.7219080328941345, "learning_rate": 6.492666666666666e-05, "loss": 0.0609, "step": 9740 }, { "epoch": 6.21414913957935, "grad_norm": 0.5833954811096191, "learning_rate": 6.499333333333334e-05, "loss": 0.0587, "step": 9750 }, { "epoch": 6.220522625876354, "grad_norm": 0.43249595165252686, "learning_rate": 6.506e-05, "loss": 0.0561, "step": 9760 }, { "epoch": 6.226896112173359, "grad_norm": 0.47439971566200256, "learning_rate": 6.512666666666667e-05, "loss": 0.0531, "step": 9770 }, { "epoch": 6.233269598470363, "grad_norm": 0.43036776781082153, "learning_rate": 6.519333333333334e-05, "loss": 0.0667, "step": 9780 }, { "epoch": 6.239643084767367, "grad_norm": 0.6738905906677246, "learning_rate": 6.526e-05, "loss": 0.0516, "step": 9790 }, { "epoch": 6.246016571064372, "grad_norm": 0.6552261114120483, "learning_rate": 6.532666666666667e-05, "loss": 0.0528, "step": 9800 }, { "epoch": 6.252390057361376, "grad_norm": 0.7906575202941895, "learning_rate": 6.539333333333334e-05, "loss": 0.053, "step": 9810 }, { "epoch": 6.2587635436583815, "grad_norm": 0.4689423441886902, "learning_rate": 6.546e-05, "loss": 0.0632, "step": 9820 }, { "epoch": 6.265137029955386, "grad_norm": 0.41271916031837463, "learning_rate": 6.552666666666666e-05, "loss": 0.0568, "step": 9830 }, { "epoch": 6.27151051625239, "grad_norm": 0.7616137862205505, "learning_rate": 6.559333333333334e-05, "loss": 0.0648, "step": 9840 }, { "epoch": 6.277884002549395, "grad_norm": 0.6170486807823181, "learning_rate": 6.566e-05, "loss": 0.0626, "step": 9850 }, { "epoch": 6.284257488846399, "grad_norm": 0.894328236579895, "learning_rate": 6.572666666666667e-05, "loss": 0.0673, "step": 9860 }, { "epoch": 6.2906309751434035, "grad_norm": 0.743172824382782, "learning_rate": 6.579333333333335e-05, "loss": 0.0527, "step": 9870 }, { "epoch": 6.297004461440408, "grad_norm": 0.8911871314048767, "learning_rate": 6.586e-05, "loss": 0.0708, "step": 9880 }, { "epoch": 6.303377947737412, "grad_norm": 0.7575047016143799, "learning_rate": 6.592666666666667e-05, "loss": 0.0528, "step": 9890 }, { "epoch": 6.309751434034417, "grad_norm": 0.6481773853302002, "learning_rate": 6.599333333333334e-05, "loss": 0.0582, "step": 9900 }, { "epoch": 6.316124920331421, "grad_norm": 0.6251875162124634, "learning_rate": 6.606e-05, "loss": 0.051, "step": 9910 }, { "epoch": 6.3224984066284255, "grad_norm": 1.0031306743621826, "learning_rate": 6.612666666666666e-05, "loss": 0.0596, "step": 9920 }, { "epoch": 6.32887189292543, "grad_norm": 0.934403121471405, "learning_rate": 6.619333333333334e-05, "loss": 0.0658, "step": 9930 }, { "epoch": 6.335245379222434, "grad_norm": 0.749822199344635, "learning_rate": 6.626e-05, "loss": 0.0563, "step": 9940 }, { "epoch": 6.341618865519439, "grad_norm": 0.7393651008605957, "learning_rate": 6.632666666666667e-05, "loss": 0.0547, "step": 9950 }, { "epoch": 6.347992351816444, "grad_norm": 0.6630147695541382, "learning_rate": 6.639333333333335e-05, "loss": 0.0574, "step": 9960 }, { "epoch": 6.3543658381134485, "grad_norm": 1.2484126091003418, "learning_rate": 6.646e-05, "loss": 0.0624, "step": 9970 }, { "epoch": 6.360739324410453, "grad_norm": 0.4653473496437073, "learning_rate": 6.652666666666667e-05, "loss": 0.0511, "step": 9980 }, { "epoch": 6.367112810707457, "grad_norm": 0.6409396529197693, "learning_rate": 6.659333333333334e-05, "loss": 0.0613, "step": 9990 }, { "epoch": 6.373486297004462, "grad_norm": 0.7591714262962341, "learning_rate": 6.666e-05, "loss": 0.059, "step": 10000 }, { "epoch": 6.379859783301466, "grad_norm": 0.5877882838249207, "learning_rate": 6.672666666666667e-05, "loss": 0.0732, "step": 10010 }, { "epoch": 6.3862332695984705, "grad_norm": 0.9828495979309082, "learning_rate": 6.679333333333334e-05, "loss": 0.0619, "step": 10020 }, { "epoch": 6.392606755895475, "grad_norm": 0.6354754567146301, "learning_rate": 6.686000000000001e-05, "loss": 0.0457, "step": 10030 }, { "epoch": 6.398980242192479, "grad_norm": 0.8597481846809387, "learning_rate": 6.692666666666667e-05, "loss": 0.0615, "step": 10040 }, { "epoch": 6.405353728489484, "grad_norm": 0.5619955062866211, "learning_rate": 6.699333333333333e-05, "loss": 0.0764, "step": 10050 }, { "epoch": 6.411727214786488, "grad_norm": 0.5440130829811096, "learning_rate": 6.706e-05, "loss": 0.0566, "step": 10060 }, { "epoch": 6.4181007010834925, "grad_norm": 0.3928253948688507, "learning_rate": 6.712666666666668e-05, "loss": 0.0571, "step": 10070 }, { "epoch": 6.424474187380497, "grad_norm": 0.9080679416656494, "learning_rate": 6.719333333333334e-05, "loss": 0.062, "step": 10080 }, { "epoch": 6.430847673677501, "grad_norm": 0.6312495470046997, "learning_rate": 6.726e-05, "loss": 0.0624, "step": 10090 }, { "epoch": 6.437221159974506, "grad_norm": 0.6919388771057129, "learning_rate": 6.732666666666667e-05, "loss": 0.0683, "step": 10100 }, { "epoch": 6.44359464627151, "grad_norm": 0.48411840200424194, "learning_rate": 6.739333333333334e-05, "loss": 0.0503, "step": 10110 }, { "epoch": 6.4499681325685145, "grad_norm": 0.8454859256744385, "learning_rate": 6.746e-05, "loss": 0.053, "step": 10120 }, { "epoch": 6.45634161886552, "grad_norm": 0.71438068151474, "learning_rate": 6.752666666666667e-05, "loss": 0.0549, "step": 10130 }, { "epoch": 6.462715105162524, "grad_norm": 0.8116707801818848, "learning_rate": 6.759333333333334e-05, "loss": 0.0712, "step": 10140 }, { "epoch": 6.469088591459529, "grad_norm": 0.5926610231399536, "learning_rate": 6.766e-05, "loss": 0.0688, "step": 10150 }, { "epoch": 6.475462077756533, "grad_norm": 0.6376394629478455, "learning_rate": 6.772666666666668e-05, "loss": 0.0586, "step": 10160 }, { "epoch": 6.4818355640535374, "grad_norm": 0.7720407247543335, "learning_rate": 6.779333333333334e-05, "loss": 0.0567, "step": 10170 }, { "epoch": 6.488209050350542, "grad_norm": 0.5263776183128357, "learning_rate": 6.786e-05, "loss": 0.0665, "step": 10180 }, { "epoch": 6.494582536647546, "grad_norm": 0.902793288230896, "learning_rate": 6.792666666666667e-05, "loss": 0.0677, "step": 10190 }, { "epoch": 6.500956022944551, "grad_norm": 0.6537356376647949, "learning_rate": 6.799333333333333e-05, "loss": 0.0629, "step": 10200 }, { "epoch": 6.507329509241555, "grad_norm": 0.9066295623779297, "learning_rate": 6.806e-05, "loss": 0.0605, "step": 10210 }, { "epoch": 6.5137029955385595, "grad_norm": 0.9434707164764404, "learning_rate": 6.812666666666667e-05, "loss": 0.0546, "step": 10220 }, { "epoch": 6.520076481835564, "grad_norm": 1.070278525352478, "learning_rate": 6.819333333333334e-05, "loss": 0.0595, "step": 10230 }, { "epoch": 6.526449968132568, "grad_norm": 0.5275779366493225, "learning_rate": 6.826e-05, "loss": 0.0563, "step": 10240 }, { "epoch": 6.532823454429573, "grad_norm": 0.4790928363800049, "learning_rate": 6.832666666666668e-05, "loss": 0.0577, "step": 10250 }, { "epoch": 6.539196940726577, "grad_norm": 0.6474365592002869, "learning_rate": 6.839333333333333e-05, "loss": 0.0521, "step": 10260 }, { "epoch": 6.545570427023582, "grad_norm": 0.6602675318717957, "learning_rate": 6.846e-05, "loss": 0.0606, "step": 10270 }, { "epoch": 6.551943913320587, "grad_norm": 1.0174161195755005, "learning_rate": 6.852666666666667e-05, "loss": 0.0693, "step": 10280 }, { "epoch": 6.558317399617591, "grad_norm": 0.5428268909454346, "learning_rate": 6.859333333333333e-05, "loss": 0.0638, "step": 10290 }, { "epoch": 6.564690885914596, "grad_norm": 0.6684935688972473, "learning_rate": 6.866e-05, "loss": 0.0545, "step": 10300 }, { "epoch": 6.5710643722116, "grad_norm": 0.4289809465408325, "learning_rate": 6.872666666666667e-05, "loss": 0.0492, "step": 10310 }, { "epoch": 6.577437858508604, "grad_norm": 0.619526743888855, "learning_rate": 6.879333333333334e-05, "loss": 0.0594, "step": 10320 }, { "epoch": 6.583811344805609, "grad_norm": 0.6994621753692627, "learning_rate": 6.886e-05, "loss": 0.053, "step": 10330 }, { "epoch": 6.590184831102613, "grad_norm": 0.7008401155471802, "learning_rate": 6.892666666666668e-05, "loss": 0.0593, "step": 10340 }, { "epoch": 6.596558317399618, "grad_norm": 0.8790857195854187, "learning_rate": 6.899333333333333e-05, "loss": 0.067, "step": 10350 }, { "epoch": 6.602931803696622, "grad_norm": 0.628646194934845, "learning_rate": 6.906e-05, "loss": 0.0629, "step": 10360 }, { "epoch": 6.609305289993626, "grad_norm": 0.5086628794670105, "learning_rate": 6.912666666666667e-05, "loss": 0.0522, "step": 10370 }, { "epoch": 6.615678776290631, "grad_norm": 0.5419976115226746, "learning_rate": 6.919333333333333e-05, "loss": 0.0593, "step": 10380 }, { "epoch": 6.622052262587635, "grad_norm": 0.4934231638908386, "learning_rate": 6.926e-05, "loss": 0.0443, "step": 10390 }, { "epoch": 6.62842574888464, "grad_norm": 0.552006185054779, "learning_rate": 6.932666666666667e-05, "loss": 0.0623, "step": 10400 }, { "epoch": 6.634799235181644, "grad_norm": 0.6392134428024292, "learning_rate": 6.939333333333334e-05, "loss": 0.0586, "step": 10410 }, { "epoch": 6.6411727214786485, "grad_norm": 0.8226262331008911, "learning_rate": 6.946e-05, "loss": 0.064, "step": 10420 }, { "epoch": 6.647546207775653, "grad_norm": 0.552422285079956, "learning_rate": 6.952666666666668e-05, "loss": 0.0427, "step": 10430 }, { "epoch": 6.653919694072657, "grad_norm": 0.590422511100769, "learning_rate": 6.959333333333333e-05, "loss": 0.0497, "step": 10440 }, { "epoch": 6.660293180369663, "grad_norm": 0.6564462184906006, "learning_rate": 6.966000000000001e-05, "loss": 0.0635, "step": 10450 }, { "epoch": 6.666666666666667, "grad_norm": 0.5087324380874634, "learning_rate": 6.972666666666667e-05, "loss": 0.0546, "step": 10460 }, { "epoch": 6.673040152963671, "grad_norm": 0.5284962058067322, "learning_rate": 6.979333333333333e-05, "loss": 0.0554, "step": 10470 }, { "epoch": 6.679413639260676, "grad_norm": 0.7105791568756104, "learning_rate": 6.986e-05, "loss": 0.0459, "step": 10480 }, { "epoch": 6.68578712555768, "grad_norm": 0.5709689259529114, "learning_rate": 6.992666666666668e-05, "loss": 0.0508, "step": 10490 }, { "epoch": 6.692160611854685, "grad_norm": 0.7488101720809937, "learning_rate": 6.999333333333334e-05, "loss": 0.0471, "step": 10500 }, { "epoch": 6.698534098151689, "grad_norm": 0.7387686371803284, "learning_rate": 7.006e-05, "loss": 0.0576, "step": 10510 }, { "epoch": 6.704907584448693, "grad_norm": 0.6705060005187988, "learning_rate": 7.012666666666667e-05, "loss": 0.0488, "step": 10520 }, { "epoch": 6.711281070745698, "grad_norm": 0.6967387795448303, "learning_rate": 7.019333333333333e-05, "loss": 0.0614, "step": 10530 }, { "epoch": 6.717654557042702, "grad_norm": 0.4333428144454956, "learning_rate": 7.026000000000001e-05, "loss": 0.0483, "step": 10540 }, { "epoch": 6.724028043339707, "grad_norm": 0.4872308373451233, "learning_rate": 7.032666666666667e-05, "loss": 0.0518, "step": 10550 }, { "epoch": 6.730401529636711, "grad_norm": 0.5130177736282349, "learning_rate": 7.039333333333334e-05, "loss": 0.0553, "step": 10560 }, { "epoch": 6.736775015933715, "grad_norm": 0.421485960483551, "learning_rate": 7.046e-05, "loss": 0.0494, "step": 10570 }, { "epoch": 6.74314850223072, "grad_norm": 0.8604413866996765, "learning_rate": 7.052666666666668e-05, "loss": 0.0643, "step": 10580 }, { "epoch": 6.749521988527725, "grad_norm": 0.7307182550430298, "learning_rate": 7.059333333333333e-05, "loss": 0.0555, "step": 10590 }, { "epoch": 6.7558954748247295, "grad_norm": 0.7127339839935303, "learning_rate": 7.066e-05, "loss": 0.0744, "step": 10600 }, { "epoch": 6.762268961121734, "grad_norm": 1.1702367067337036, "learning_rate": 7.072666666666667e-05, "loss": 0.0697, "step": 10610 }, { "epoch": 6.768642447418738, "grad_norm": 1.2847914695739746, "learning_rate": 7.079333333333333e-05, "loss": 0.0651, "step": 10620 }, { "epoch": 6.775015933715743, "grad_norm": 0.593812108039856, "learning_rate": 7.086000000000001e-05, "loss": 0.0742, "step": 10630 }, { "epoch": 6.781389420012747, "grad_norm": 0.706718385219574, "learning_rate": 7.092666666666667e-05, "loss": 0.0703, "step": 10640 }, { "epoch": 6.7877629063097515, "grad_norm": 0.6615965962409973, "learning_rate": 7.099333333333334e-05, "loss": 0.0654, "step": 10650 }, { "epoch": 6.794136392606756, "grad_norm": 0.6300081610679626, "learning_rate": 7.106e-05, "loss": 0.0664, "step": 10660 }, { "epoch": 6.80050987890376, "grad_norm": 0.547766387462616, "learning_rate": 7.112666666666668e-05, "loss": 0.0666, "step": 10670 }, { "epoch": 6.806883365200765, "grad_norm": 0.6945487260818481, "learning_rate": 7.119333333333333e-05, "loss": 0.057, "step": 10680 }, { "epoch": 6.813256851497769, "grad_norm": 1.0761439800262451, "learning_rate": 7.126e-05, "loss": 0.0545, "step": 10690 }, { "epoch": 6.819630337794774, "grad_norm": 0.5777052044868469, "learning_rate": 7.132666666666667e-05, "loss": 0.0634, "step": 10700 }, { "epoch": 6.826003824091778, "grad_norm": 0.4692111909389496, "learning_rate": 7.139333333333333e-05, "loss": 0.068, "step": 10710 }, { "epoch": 6.832377310388782, "grad_norm": 0.5378316044807434, "learning_rate": 7.146000000000001e-05, "loss": 0.0661, "step": 10720 }, { "epoch": 6.838750796685787, "grad_norm": 0.7132922410964966, "learning_rate": 7.152666666666667e-05, "loss": 0.0506, "step": 10730 }, { "epoch": 6.845124282982791, "grad_norm": 0.5381823182106018, "learning_rate": 7.159333333333334e-05, "loss": 0.0509, "step": 10740 }, { "epoch": 6.851497769279796, "grad_norm": 0.5761995315551758, "learning_rate": 7.166e-05, "loss": 0.0566, "step": 10750 }, { "epoch": 6.857871255576801, "grad_norm": 0.781377911567688, "learning_rate": 7.172666666666668e-05, "loss": 0.0657, "step": 10760 }, { "epoch": 6.864244741873805, "grad_norm": 0.6733283996582031, "learning_rate": 7.179333333333333e-05, "loss": 0.0577, "step": 10770 }, { "epoch": 6.87061822817081, "grad_norm": 0.5615326762199402, "learning_rate": 7.186e-05, "loss": 0.0652, "step": 10780 }, { "epoch": 6.876991714467814, "grad_norm": 0.6769412159919739, "learning_rate": 7.192666666666667e-05, "loss": 0.0573, "step": 10790 }, { "epoch": 6.8833652007648185, "grad_norm": 0.7781015038490295, "learning_rate": 7.199333333333333e-05, "loss": 0.0753, "step": 10800 }, { "epoch": 6.889738687061823, "grad_norm": 0.7718249559402466, "learning_rate": 7.206000000000001e-05, "loss": 0.052, "step": 10810 }, { "epoch": 6.896112173358827, "grad_norm": 0.4445921778678894, "learning_rate": 7.212666666666666e-05, "loss": 0.0563, "step": 10820 }, { "epoch": 6.902485659655832, "grad_norm": 0.5905616283416748, "learning_rate": 7.219333333333334e-05, "loss": 0.0609, "step": 10830 }, { "epoch": 6.908859145952836, "grad_norm": 0.9250876903533936, "learning_rate": 7.226e-05, "loss": 0.0698, "step": 10840 }, { "epoch": 6.9152326322498405, "grad_norm": 0.6766232252120972, "learning_rate": 7.232666666666667e-05, "loss": 0.0522, "step": 10850 }, { "epoch": 6.921606118546845, "grad_norm": 0.8802348375320435, "learning_rate": 7.239333333333333e-05, "loss": 0.0596, "step": 10860 }, { "epoch": 6.927979604843849, "grad_norm": 0.48499956727027893, "learning_rate": 7.246000000000001e-05, "loss": 0.0647, "step": 10870 }, { "epoch": 6.934353091140854, "grad_norm": 0.6914782524108887, "learning_rate": 7.252666666666667e-05, "loss": 0.0642, "step": 10880 }, { "epoch": 6.940726577437858, "grad_norm": 1.1734741926193237, "learning_rate": 7.259333333333334e-05, "loss": 0.0668, "step": 10890 }, { "epoch": 6.947100063734863, "grad_norm": 0.9609723687171936, "learning_rate": 7.266000000000001e-05, "loss": 0.0726, "step": 10900 }, { "epoch": 6.953473550031868, "grad_norm": 0.6944030523300171, "learning_rate": 7.272666666666666e-05, "loss": 0.057, "step": 10910 }, { "epoch": 6.959847036328872, "grad_norm": 0.8145813941955566, "learning_rate": 7.279333333333334e-05, "loss": 0.0574, "step": 10920 }, { "epoch": 6.966220522625877, "grad_norm": 0.518361508846283, "learning_rate": 7.286e-05, "loss": 0.0637, "step": 10930 }, { "epoch": 6.972594008922881, "grad_norm": 0.7921425104141235, "learning_rate": 7.292666666666667e-05, "loss": 0.0622, "step": 10940 }, { "epoch": 6.9789674952198855, "grad_norm": 0.44566255807876587, "learning_rate": 7.299333333333333e-05, "loss": 0.0716, "step": 10950 }, { "epoch": 6.98534098151689, "grad_norm": 0.7134578824043274, "learning_rate": 7.306000000000001e-05, "loss": 0.0636, "step": 10960 }, { "epoch": 6.991714467813894, "grad_norm": 0.7634859681129456, "learning_rate": 7.312666666666666e-05, "loss": 0.0587, "step": 10970 }, { "epoch": 6.998087954110899, "grad_norm": 0.620833694934845, "learning_rate": 7.319333333333334e-05, "loss": 0.0572, "step": 10980 }, { "epoch": 7.004461440407903, "grad_norm": 0.8117725253105164, "learning_rate": 7.326e-05, "loss": 0.0549, "step": 10990 }, { "epoch": 7.0108349267049075, "grad_norm": 0.9135079383850098, "learning_rate": 7.332666666666666e-05, "loss": 0.0686, "step": 11000 }, { "epoch": 7.017208413001912, "grad_norm": 0.4884015917778015, "learning_rate": 7.339333333333334e-05, "loss": 0.057, "step": 11010 }, { "epoch": 7.023581899298916, "grad_norm": 1.0910640954971313, "learning_rate": 7.346e-05, "loss": 0.0727, "step": 11020 }, { "epoch": 7.029955385595921, "grad_norm": 0.8113101720809937, "learning_rate": 7.352666666666667e-05, "loss": 0.0687, "step": 11030 }, { "epoch": 7.036328871892925, "grad_norm": 0.9775698184967041, "learning_rate": 7.359333333333333e-05, "loss": 0.0618, "step": 11040 }, { "epoch": 7.0427023581899295, "grad_norm": 0.9888875484466553, "learning_rate": 7.366000000000001e-05, "loss": 0.0568, "step": 11050 }, { "epoch": 7.049075844486934, "grad_norm": 0.5883205533027649, "learning_rate": 7.372666666666666e-05, "loss": 0.0589, "step": 11060 }, { "epoch": 7.055449330783939, "grad_norm": 0.42911091446876526, "learning_rate": 7.379333333333334e-05, "loss": 0.06, "step": 11070 }, { "epoch": 7.061822817080944, "grad_norm": 0.4612443745136261, "learning_rate": 7.386e-05, "loss": 0.0504, "step": 11080 }, { "epoch": 7.068196303377948, "grad_norm": 0.8134729266166687, "learning_rate": 7.392666666666666e-05, "loss": 0.0673, "step": 11090 }, { "epoch": 7.074569789674952, "grad_norm": 0.68392014503479, "learning_rate": 7.399333333333334e-05, "loss": 0.0466, "step": 11100 }, { "epoch": 7.080943275971957, "grad_norm": 0.5725587606430054, "learning_rate": 7.406e-05, "loss": 0.0655, "step": 11110 }, { "epoch": 7.087316762268961, "grad_norm": 0.8250055313110352, "learning_rate": 7.412666666666667e-05, "loss": 0.0615, "step": 11120 }, { "epoch": 7.093690248565966, "grad_norm": 0.6771624684333801, "learning_rate": 7.419333333333333e-05, "loss": 0.0619, "step": 11130 }, { "epoch": 7.10006373486297, "grad_norm": 0.5988875031471252, "learning_rate": 7.426000000000001e-05, "loss": 0.0585, "step": 11140 }, { "epoch": 7.1064372211599744, "grad_norm": 0.4609868824481964, "learning_rate": 7.432666666666666e-05, "loss": 0.0344, "step": 11150 }, { "epoch": 7.112810707456979, "grad_norm": 0.5410454869270325, "learning_rate": 7.439333333333334e-05, "loss": 0.0512, "step": 11160 }, { "epoch": 7.119184193753983, "grad_norm": 0.7448944449424744, "learning_rate": 7.446e-05, "loss": 0.0532, "step": 11170 }, { "epoch": 7.125557680050988, "grad_norm": 0.7174357175827026, "learning_rate": 7.452666666666667e-05, "loss": 0.0531, "step": 11180 }, { "epoch": 7.131931166347992, "grad_norm": 0.43492981791496277, "learning_rate": 7.459333333333334e-05, "loss": 0.0562, "step": 11190 }, { "epoch": 7.1383046526449965, "grad_norm": 0.4520013928413391, "learning_rate": 7.466000000000001e-05, "loss": 0.0584, "step": 11200 }, { "epoch": 7.144678138942001, "grad_norm": 0.6327972412109375, "learning_rate": 7.472666666666667e-05, "loss": 0.0542, "step": 11210 }, { "epoch": 7.151051625239006, "grad_norm": 0.8993403911590576, "learning_rate": 7.479333333333333e-05, "loss": 0.0687, "step": 11220 }, { "epoch": 7.157425111536011, "grad_norm": 0.8367977738380432, "learning_rate": 7.486000000000001e-05, "loss": 0.0896, "step": 11230 }, { "epoch": 7.163798597833015, "grad_norm": 0.4823082685470581, "learning_rate": 7.492666666666666e-05, "loss": 0.0556, "step": 11240 }, { "epoch": 7.170172084130019, "grad_norm": 0.7253256440162659, "learning_rate": 7.499333333333334e-05, "loss": 0.0449, "step": 11250 }, { "epoch": 7.176545570427024, "grad_norm": 0.42491480708122253, "learning_rate": 7.506e-05, "loss": 0.0462, "step": 11260 }, { "epoch": 7.182919056724028, "grad_norm": 0.3749558627605438, "learning_rate": 7.512666666666667e-05, "loss": 0.0577, "step": 11270 }, { "epoch": 7.189292543021033, "grad_norm": 1.0233200788497925, "learning_rate": 7.519333333333334e-05, "loss": 0.0658, "step": 11280 }, { "epoch": 7.195666029318037, "grad_norm": 0.6975396275520325, "learning_rate": 7.526000000000001e-05, "loss": 0.0684, "step": 11290 }, { "epoch": 7.202039515615041, "grad_norm": 0.6800739169120789, "learning_rate": 7.532666666666667e-05, "loss": 0.0527, "step": 11300 }, { "epoch": 7.208413001912046, "grad_norm": 0.6258073449134827, "learning_rate": 7.539333333333334e-05, "loss": 0.0577, "step": 11310 }, { "epoch": 7.21478648820905, "grad_norm": 0.6886470317840576, "learning_rate": 7.546000000000001e-05, "loss": 0.0655, "step": 11320 }, { "epoch": 7.221159974506055, "grad_norm": 0.8964318037033081, "learning_rate": 7.552666666666666e-05, "loss": 0.0394, "step": 11330 }, { "epoch": 7.227533460803059, "grad_norm": 0.6888375282287598, "learning_rate": 7.559333333333334e-05, "loss": 0.0471, "step": 11340 }, { "epoch": 7.233906947100063, "grad_norm": 0.5329063534736633, "learning_rate": 7.566e-05, "loss": 0.0699, "step": 11350 }, { "epoch": 7.240280433397068, "grad_norm": 0.638798177242279, "learning_rate": 7.572666666666667e-05, "loss": 0.0492, "step": 11360 }, { "epoch": 7.246653919694072, "grad_norm": 0.6056517958641052, "learning_rate": 7.579333333333335e-05, "loss": 0.053, "step": 11370 }, { "epoch": 7.253027405991077, "grad_norm": 0.5125308632850647, "learning_rate": 7.586000000000001e-05, "loss": 0.0481, "step": 11380 }, { "epoch": 7.259400892288082, "grad_norm": 0.5904332399368286, "learning_rate": 7.592666666666667e-05, "loss": 0.0647, "step": 11390 }, { "epoch": 7.265774378585086, "grad_norm": 0.8789222240447998, "learning_rate": 7.599333333333334e-05, "loss": 0.0627, "step": 11400 }, { "epoch": 7.272147864882091, "grad_norm": 0.4663279950618744, "learning_rate": 7.606000000000001e-05, "loss": 0.0491, "step": 11410 }, { "epoch": 7.278521351179095, "grad_norm": 0.4175785779953003, "learning_rate": 7.612666666666666e-05, "loss": 0.0694, "step": 11420 }, { "epoch": 7.2848948374761, "grad_norm": 0.6583784222602844, "learning_rate": 7.619333333333334e-05, "loss": 0.0385, "step": 11430 }, { "epoch": 7.291268323773104, "grad_norm": 0.49252942204475403, "learning_rate": 7.625999999999999e-05, "loss": 0.056, "step": 11440 }, { "epoch": 7.297641810070108, "grad_norm": 0.6419488191604614, "learning_rate": 7.632666666666667e-05, "loss": 0.0538, "step": 11450 }, { "epoch": 7.304015296367113, "grad_norm": 0.4881172478199005, "learning_rate": 7.639333333333333e-05, "loss": 0.0673, "step": 11460 }, { "epoch": 7.310388782664117, "grad_norm": 0.6967611908912659, "learning_rate": 7.646e-05, "loss": 0.0476, "step": 11470 }, { "epoch": 7.316762268961122, "grad_norm": 0.9382917881011963, "learning_rate": 7.652666666666667e-05, "loss": 0.0533, "step": 11480 }, { "epoch": 7.323135755258126, "grad_norm": 0.5202028155326843, "learning_rate": 7.659333333333334e-05, "loss": 0.0597, "step": 11490 }, { "epoch": 7.32950924155513, "grad_norm": 0.6725322604179382, "learning_rate": 7.666e-05, "loss": 0.0612, "step": 11500 }, { "epoch": 7.335882727852135, "grad_norm": 0.9001889228820801, "learning_rate": 7.672666666666667e-05, "loss": 0.0528, "step": 11510 }, { "epoch": 7.342256214149139, "grad_norm": 1.0062668323516846, "learning_rate": 7.679333333333334e-05, "loss": 0.0594, "step": 11520 }, { "epoch": 7.3486297004461445, "grad_norm": 0.3967103660106659, "learning_rate": 7.685999999999999e-05, "loss": 0.0456, "step": 11530 }, { "epoch": 7.355003186743149, "grad_norm": 0.8786386847496033, "learning_rate": 7.692666666666667e-05, "loss": 0.0513, "step": 11540 }, { "epoch": 7.361376673040153, "grad_norm": 0.5655531883239746, "learning_rate": 7.699333333333333e-05, "loss": 0.0648, "step": 11550 }, { "epoch": 7.367750159337158, "grad_norm": 0.5367156863212585, "learning_rate": 7.706e-05, "loss": 0.0525, "step": 11560 }, { "epoch": 7.374123645634162, "grad_norm": 0.49696165323257446, "learning_rate": 7.712666666666667e-05, "loss": 0.0468, "step": 11570 }, { "epoch": 7.3804971319311665, "grad_norm": 0.6176717281341553, "learning_rate": 7.719333333333334e-05, "loss": 0.0599, "step": 11580 }, { "epoch": 7.386870618228171, "grad_norm": 0.5491601228713989, "learning_rate": 7.726e-05, "loss": 0.06, "step": 11590 }, { "epoch": 7.393244104525175, "grad_norm": 0.5185981392860413, "learning_rate": 7.732666666666667e-05, "loss": 0.0436, "step": 11600 }, { "epoch": 7.39961759082218, "grad_norm": 0.7929900884628296, "learning_rate": 7.739333333333334e-05, "loss": 0.0691, "step": 11610 }, { "epoch": 7.405991077119184, "grad_norm": 0.8049429655075073, "learning_rate": 7.746e-05, "loss": 0.0616, "step": 11620 }, { "epoch": 7.4123645634161885, "grad_norm": 1.1048583984375, "learning_rate": 7.752666666666667e-05, "loss": 0.0815, "step": 11630 }, { "epoch": 7.418738049713193, "grad_norm": 0.6788186430931091, "learning_rate": 7.759333333333333e-05, "loss": 0.0635, "step": 11640 }, { "epoch": 7.425111536010197, "grad_norm": 0.6070594787597656, "learning_rate": 7.766e-05, "loss": 0.0576, "step": 11650 }, { "epoch": 7.431485022307202, "grad_norm": 0.6177842617034912, "learning_rate": 7.772666666666668e-05, "loss": 0.063, "step": 11660 }, { "epoch": 7.437858508604206, "grad_norm": 0.502066969871521, "learning_rate": 7.779333333333334e-05, "loss": 0.0389, "step": 11670 }, { "epoch": 7.444231994901211, "grad_norm": 0.552257239818573, "learning_rate": 7.786e-05, "loss": 0.056, "step": 11680 }, { "epoch": 7.450605481198215, "grad_norm": 0.593451738357544, "learning_rate": 7.792666666666667e-05, "loss": 0.0525, "step": 11690 }, { "epoch": 7.45697896749522, "grad_norm": 0.437322199344635, "learning_rate": 7.799333333333334e-05, "loss": 0.0647, "step": 11700 }, { "epoch": 7.463352453792225, "grad_norm": 0.7175058722496033, "learning_rate": 7.806e-05, "loss": 0.0544, "step": 11710 }, { "epoch": 7.469725940089229, "grad_norm": 0.3645232617855072, "learning_rate": 7.812666666666667e-05, "loss": 0.0575, "step": 11720 }, { "epoch": 7.4760994263862335, "grad_norm": 0.9268646836280823, "learning_rate": 7.819333333333334e-05, "loss": 0.0551, "step": 11730 }, { "epoch": 7.482472912683238, "grad_norm": 0.8866416811943054, "learning_rate": 7.826e-05, "loss": 0.0542, "step": 11740 }, { "epoch": 7.488846398980242, "grad_norm": 0.7859415411949158, "learning_rate": 7.832666666666668e-05, "loss": 0.0612, "step": 11750 }, { "epoch": 7.495219885277247, "grad_norm": 0.6096539497375488, "learning_rate": 7.839333333333334e-05, "loss": 0.0526, "step": 11760 }, { "epoch": 7.501593371574251, "grad_norm": 0.6642547845840454, "learning_rate": 7.846e-05, "loss": 0.0473, "step": 11770 }, { "epoch": 7.5079668578712555, "grad_norm": 0.580796480178833, "learning_rate": 7.852666666666667e-05, "loss": 0.0554, "step": 11780 }, { "epoch": 7.51434034416826, "grad_norm": 0.5509164929389954, "learning_rate": 7.859333333333335e-05, "loss": 0.064, "step": 11790 }, { "epoch": 7.520713830465264, "grad_norm": 0.5541414618492126, "learning_rate": 7.866e-05, "loss": 0.0575, "step": 11800 }, { "epoch": 7.527087316762269, "grad_norm": 0.4667292535305023, "learning_rate": 7.872666666666667e-05, "loss": 0.0751, "step": 11810 }, { "epoch": 7.533460803059273, "grad_norm": 0.7596257328987122, "learning_rate": 7.879333333333334e-05, "loss": 0.0572, "step": 11820 }, { "epoch": 7.5398342893562775, "grad_norm": 0.8888062834739685, "learning_rate": 7.886e-05, "loss": 0.0741, "step": 11830 }, { "epoch": 7.546207775653283, "grad_norm": 1.5570415258407593, "learning_rate": 7.892666666666666e-05, "loss": 0.068, "step": 11840 }, { "epoch": 7.552581261950287, "grad_norm": 0.6390975117683411, "learning_rate": 7.899333333333334e-05, "loss": 0.0627, "step": 11850 }, { "epoch": 7.558954748247292, "grad_norm": 0.5417156219482422, "learning_rate": 7.906e-05, "loss": 0.0684, "step": 11860 }, { "epoch": 7.565328234544296, "grad_norm": 0.7917418479919434, "learning_rate": 7.912666666666667e-05, "loss": 0.0521, "step": 11870 }, { "epoch": 7.5717017208413, "grad_norm": 0.9175390601158142, "learning_rate": 7.919333333333335e-05, "loss": 0.094, "step": 11880 }, { "epoch": 7.578075207138305, "grad_norm": 0.6375300288200378, "learning_rate": 7.926e-05, "loss": 0.0583, "step": 11890 }, { "epoch": 7.584448693435309, "grad_norm": 0.7368698120117188, "learning_rate": 7.932666666666667e-05, "loss": 0.0567, "step": 11900 }, { "epoch": 7.590822179732314, "grad_norm": 0.42117226123809814, "learning_rate": 7.939333333333334e-05, "loss": 0.0657, "step": 11910 }, { "epoch": 7.597195666029318, "grad_norm": 0.7985228896141052, "learning_rate": 7.946e-05, "loss": 0.0551, "step": 11920 }, { "epoch": 7.6035691523263225, "grad_norm": 0.5908305048942566, "learning_rate": 7.952666666666667e-05, "loss": 0.0581, "step": 11930 }, { "epoch": 7.609942638623327, "grad_norm": 0.6016819477081299, "learning_rate": 7.959333333333334e-05, "loss": 0.0512, "step": 11940 }, { "epoch": 7.616316124920331, "grad_norm": 0.6850925087928772, "learning_rate": 7.966e-05, "loss": 0.0517, "step": 11950 }, { "epoch": 7.622689611217336, "grad_norm": 0.7231594324111938, "learning_rate": 7.972666666666667e-05, "loss": 0.0501, "step": 11960 }, { "epoch": 7.62906309751434, "grad_norm": 0.5239086747169495, "learning_rate": 7.979333333333335e-05, "loss": 0.0461, "step": 11970 }, { "epoch": 7.6354365838113445, "grad_norm": 0.6435546875, "learning_rate": 7.986e-05, "loss": 0.061, "step": 11980 }, { "epoch": 7.641810070108349, "grad_norm": 0.5770428776741028, "learning_rate": 7.992666666666668e-05, "loss": 0.0643, "step": 11990 }, { "epoch": 7.648183556405353, "grad_norm": 0.7032604813575745, "learning_rate": 7.999333333333334e-05, "loss": 0.0626, "step": 12000 }, { "epoch": 7.654557042702358, "grad_norm": 0.7884879112243652, "learning_rate": 8.006e-05, "loss": 0.0692, "step": 12010 }, { "epoch": 7.660930528999363, "grad_norm": 0.61533522605896, "learning_rate": 8.012666666666667e-05, "loss": 0.0492, "step": 12020 }, { "epoch": 7.667304015296367, "grad_norm": 0.41875240206718445, "learning_rate": 8.019333333333334e-05, "loss": 0.0464, "step": 12030 }, { "epoch": 7.673677501593372, "grad_norm": 0.27094653248786926, "learning_rate": 8.026000000000001e-05, "loss": 0.0438, "step": 12040 }, { "epoch": 7.680050987890376, "grad_norm": 0.724066436290741, "learning_rate": 8.032666666666667e-05, "loss": 0.0619, "step": 12050 }, { "epoch": 7.686424474187381, "grad_norm": 0.5979686975479126, "learning_rate": 8.039333333333335e-05, "loss": 0.0578, "step": 12060 }, { "epoch": 7.692797960484385, "grad_norm": 0.8723482489585876, "learning_rate": 8.046e-05, "loss": 0.0808, "step": 12070 }, { "epoch": 7.699171446781389, "grad_norm": 0.8083339929580688, "learning_rate": 8.052666666666668e-05, "loss": 0.0446, "step": 12080 }, { "epoch": 7.705544933078394, "grad_norm": 0.6310730576515198, "learning_rate": 8.059333333333333e-05, "loss": 0.0489, "step": 12090 }, { "epoch": 7.711918419375398, "grad_norm": 0.7891052961349487, "learning_rate": 8.066e-05, "loss": 0.0607, "step": 12100 }, { "epoch": 7.718291905672403, "grad_norm": 0.44173556566238403, "learning_rate": 8.072666666666667e-05, "loss": 0.0583, "step": 12110 }, { "epoch": 7.724665391969407, "grad_norm": 0.7001399397850037, "learning_rate": 8.079333333333333e-05, "loss": 0.0444, "step": 12120 }, { "epoch": 7.7310388782664115, "grad_norm": 0.7543017268180847, "learning_rate": 8.086000000000001e-05, "loss": 0.0674, "step": 12130 }, { "epoch": 7.737412364563416, "grad_norm": 0.5297566056251526, "learning_rate": 8.092666666666667e-05, "loss": 0.0475, "step": 12140 }, { "epoch": 7.743785850860421, "grad_norm": 0.4382438361644745, "learning_rate": 8.099333333333334e-05, "loss": 0.0507, "step": 12150 }, { "epoch": 7.7501593371574256, "grad_norm": 0.7441959977149963, "learning_rate": 8.106e-05, "loss": 0.0505, "step": 12160 }, { "epoch": 7.75653282345443, "grad_norm": 0.7295777201652527, "learning_rate": 8.112666666666668e-05, "loss": 0.0585, "step": 12170 }, { "epoch": 7.762906309751434, "grad_norm": 0.7317323088645935, "learning_rate": 8.119333333333333e-05, "loss": 0.0499, "step": 12180 }, { "epoch": 7.769279796048439, "grad_norm": 0.5636889934539795, "learning_rate": 8.126e-05, "loss": 0.0497, "step": 12190 }, { "epoch": 7.775653282345443, "grad_norm": 0.7215498685836792, "learning_rate": 8.132666666666667e-05, "loss": 0.073, "step": 12200 }, { "epoch": 7.782026768642448, "grad_norm": 0.8393809199333191, "learning_rate": 8.139333333333333e-05, "loss": 0.072, "step": 12210 }, { "epoch": 7.788400254939452, "grad_norm": 0.8304045796394348, "learning_rate": 8.146000000000001e-05, "loss": 0.0584, "step": 12220 }, { "epoch": 7.794773741236456, "grad_norm": 0.7566322684288025, "learning_rate": 8.152666666666667e-05, "loss": 0.0581, "step": 12230 }, { "epoch": 7.801147227533461, "grad_norm": 0.3760037124156952, "learning_rate": 8.159333333333334e-05, "loss": 0.0516, "step": 12240 }, { "epoch": 7.807520713830465, "grad_norm": 0.5697731971740723, "learning_rate": 8.166e-05, "loss": 0.0728, "step": 12250 }, { "epoch": 7.81389420012747, "grad_norm": 0.5997262001037598, "learning_rate": 8.172666666666668e-05, "loss": 0.0607, "step": 12260 }, { "epoch": 7.820267686424474, "grad_norm": 0.48271444439888, "learning_rate": 8.179333333333333e-05, "loss": 0.0587, "step": 12270 }, { "epoch": 7.826641172721478, "grad_norm": 0.4417996406555176, "learning_rate": 8.186e-05, "loss": 0.0498, "step": 12280 }, { "epoch": 7.833014659018483, "grad_norm": 0.34783264994621277, "learning_rate": 8.192666666666667e-05, "loss": 0.0531, "step": 12290 }, { "epoch": 7.839388145315487, "grad_norm": 0.7350356578826904, "learning_rate": 8.199333333333333e-05, "loss": 0.0543, "step": 12300 }, { "epoch": 7.845761631612492, "grad_norm": 0.878852367401123, "learning_rate": 8.206e-05, "loss": 0.0563, "step": 12310 }, { "epoch": 7.852135117909496, "grad_norm": 0.7118002772331238, "learning_rate": 8.212666666666667e-05, "loss": 0.06, "step": 12320 }, { "epoch": 7.858508604206501, "grad_norm": 0.4555775225162506, "learning_rate": 8.219333333333334e-05, "loss": 0.0556, "step": 12330 }, { "epoch": 7.864882090503506, "grad_norm": 0.711097002029419, "learning_rate": 8.226e-05, "loss": 0.0652, "step": 12340 }, { "epoch": 7.87125557680051, "grad_norm": 0.7447531223297119, "learning_rate": 8.232666666666668e-05, "loss": 0.0703, "step": 12350 }, { "epoch": 7.8776290630975145, "grad_norm": 0.48015764355659485, "learning_rate": 8.239333333333333e-05, "loss": 0.0651, "step": 12360 }, { "epoch": 7.884002549394519, "grad_norm": 0.8711662888526917, "learning_rate": 8.246e-05, "loss": 0.0475, "step": 12370 }, { "epoch": 7.890376035691523, "grad_norm": 1.0463588237762451, "learning_rate": 8.252666666666667e-05, "loss": 0.0508, "step": 12380 }, { "epoch": 7.896749521988528, "grad_norm": 0.67466801404953, "learning_rate": 8.259333333333333e-05, "loss": 0.0695, "step": 12390 }, { "epoch": 7.903123008285532, "grad_norm": 0.824320375919342, "learning_rate": 8.266e-05, "loss": 0.066, "step": 12400 }, { "epoch": 7.909496494582537, "grad_norm": 0.7048618197441101, "learning_rate": 8.272666666666668e-05, "loss": 0.0647, "step": 12410 }, { "epoch": 7.915869980879541, "grad_norm": 0.5582943558692932, "learning_rate": 8.279333333333334e-05, "loss": 0.0588, "step": 12420 }, { "epoch": 7.922243467176545, "grad_norm": 0.5646951794624329, "learning_rate": 8.286e-05, "loss": 0.0628, "step": 12430 }, { "epoch": 7.92861695347355, "grad_norm": 0.48201754689216614, "learning_rate": 8.292666666666668e-05, "loss": 0.0726, "step": 12440 }, { "epoch": 7.934990439770554, "grad_norm": 0.45752134919166565, "learning_rate": 8.299333333333333e-05, "loss": 0.0551, "step": 12450 }, { "epoch": 7.941363926067559, "grad_norm": 0.5571115612983704, "learning_rate": 8.306000000000001e-05, "loss": 0.0554, "step": 12460 }, { "epoch": 7.947737412364564, "grad_norm": 0.6592827439308167, "learning_rate": 8.312666666666667e-05, "loss": 0.0715, "step": 12470 }, { "epoch": 7.954110898661568, "grad_norm": 1.0221611261367798, "learning_rate": 8.319333333333334e-05, "loss": 0.0474, "step": 12480 }, { "epoch": 7.960484384958573, "grad_norm": 0.4267323911190033, "learning_rate": 8.326e-05, "loss": 0.0526, "step": 12490 }, { "epoch": 7.966857871255577, "grad_norm": 1.247265100479126, "learning_rate": 8.332666666666668e-05, "loss": 0.0757, "step": 12500 }, { "epoch": 7.9732313575525815, "grad_norm": 0.6217453479766846, "learning_rate": 8.339333333333334e-05, "loss": 0.0644, "step": 12510 }, { "epoch": 7.979604843849586, "grad_norm": 0.7512181401252747, "learning_rate": 8.346e-05, "loss": 0.0827, "step": 12520 }, { "epoch": 7.98597833014659, "grad_norm": 0.606593906879425, "learning_rate": 8.352666666666668e-05, "loss": 0.0625, "step": 12530 }, { "epoch": 7.992351816443595, "grad_norm": 0.5596364140510559, "learning_rate": 8.359333333333333e-05, "loss": 0.0633, "step": 12540 }, { "epoch": 7.998725302740599, "grad_norm": 0.6581247448921204, "learning_rate": 8.366000000000001e-05, "loss": 0.0799, "step": 12550 }, { "epoch": 8.005098789037604, "grad_norm": 0.7492228746414185, "learning_rate": 8.372666666666667e-05, "loss": 0.0564, "step": 12560 }, { "epoch": 8.011472275334608, "grad_norm": 0.8005379438400269, "learning_rate": 8.379333333333334e-05, "loss": 0.0585, "step": 12570 }, { "epoch": 8.017845761631612, "grad_norm": 0.7363327145576477, "learning_rate": 8.386e-05, "loss": 0.0621, "step": 12580 }, { "epoch": 8.024219247928617, "grad_norm": 0.5561438798904419, "learning_rate": 8.392666666666668e-05, "loss": 0.0647, "step": 12590 }, { "epoch": 8.030592734225621, "grad_norm": 1.0865379571914673, "learning_rate": 8.399333333333334e-05, "loss": 0.0646, "step": 12600 }, { "epoch": 8.036966220522626, "grad_norm": 0.8478971719741821, "learning_rate": 8.406e-05, "loss": 0.0562, "step": 12610 }, { "epoch": 8.04333970681963, "grad_norm": 0.874611496925354, "learning_rate": 8.412666666666668e-05, "loss": 0.0602, "step": 12620 }, { "epoch": 8.049713193116634, "grad_norm": 0.5931339263916016, "learning_rate": 8.419333333333333e-05, "loss": 0.0406, "step": 12630 }, { "epoch": 8.056086679413639, "grad_norm": 0.917545735836029, "learning_rate": 8.426000000000001e-05, "loss": 0.0618, "step": 12640 }, { "epoch": 8.062460165710643, "grad_norm": 0.6005682945251465, "learning_rate": 8.432666666666667e-05, "loss": 0.0652, "step": 12650 }, { "epoch": 8.068833652007648, "grad_norm": 0.6540262699127197, "learning_rate": 8.439333333333334e-05, "loss": 0.0546, "step": 12660 }, { "epoch": 8.075207138304652, "grad_norm": 0.8592332601547241, "learning_rate": 8.446e-05, "loss": 0.0631, "step": 12670 }, { "epoch": 8.081580624601656, "grad_norm": 0.7033776640892029, "learning_rate": 8.452666666666668e-05, "loss": 0.0495, "step": 12680 }, { "epoch": 8.08795411089866, "grad_norm": 0.6539597511291504, "learning_rate": 8.459333333333334e-05, "loss": 0.0661, "step": 12690 }, { "epoch": 8.094327597195665, "grad_norm": 0.7465230822563171, "learning_rate": 8.466e-05, "loss": 0.057, "step": 12700 }, { "epoch": 8.10070108349267, "grad_norm": 0.9498315453529358, "learning_rate": 8.472666666666667e-05, "loss": 0.0645, "step": 12710 }, { "epoch": 8.107074569789676, "grad_norm": 0.562738299369812, "learning_rate": 8.479333333333333e-05, "loss": 0.0557, "step": 12720 }, { "epoch": 8.11344805608668, "grad_norm": 0.6365389227867126, "learning_rate": 8.486000000000001e-05, "loss": 0.0452, "step": 12730 }, { "epoch": 8.119821542383685, "grad_norm": 0.7993060946464539, "learning_rate": 8.492666666666666e-05, "loss": 0.0633, "step": 12740 }, { "epoch": 8.126195028680689, "grad_norm": 0.7765477895736694, "learning_rate": 8.499333333333334e-05, "loss": 0.0632, "step": 12750 }, { "epoch": 8.132568514977693, "grad_norm": 0.8996695280075073, "learning_rate": 8.506e-05, "loss": 0.0668, "step": 12760 }, { "epoch": 8.138942001274698, "grad_norm": 0.6463532447814941, "learning_rate": 8.512666666666667e-05, "loss": 0.0758, "step": 12770 }, { "epoch": 8.145315487571702, "grad_norm": 0.7349516153335571, "learning_rate": 8.519333333333333e-05, "loss": 0.0669, "step": 12780 }, { "epoch": 8.151688973868707, "grad_norm": 0.6427165865898132, "learning_rate": 8.526000000000001e-05, "loss": 0.0617, "step": 12790 }, { "epoch": 8.158062460165711, "grad_norm": 0.742243230342865, "learning_rate": 8.532666666666667e-05, "loss": 0.0618, "step": 12800 }, { "epoch": 8.164435946462715, "grad_norm": 0.5856829285621643, "learning_rate": 8.539333333333333e-05, "loss": 0.0564, "step": 12810 }, { "epoch": 8.17080943275972, "grad_norm": 0.8810951709747314, "learning_rate": 8.546000000000001e-05, "loss": 0.0706, "step": 12820 }, { "epoch": 8.177182919056724, "grad_norm": 0.6175669431686401, "learning_rate": 8.552666666666666e-05, "loss": 0.0576, "step": 12830 }, { "epoch": 8.183556405353729, "grad_norm": 0.6240518093109131, "learning_rate": 8.559333333333334e-05, "loss": 0.0535, "step": 12840 }, { "epoch": 8.189929891650733, "grad_norm": 0.924400806427002, "learning_rate": 8.566e-05, "loss": 0.0549, "step": 12850 }, { "epoch": 8.196303377947737, "grad_norm": 0.7381389737129211, "learning_rate": 8.572666666666667e-05, "loss": 0.0648, "step": 12860 }, { "epoch": 8.202676864244742, "grad_norm": 0.6493406295776367, "learning_rate": 8.579333333333333e-05, "loss": 0.0478, "step": 12870 }, { "epoch": 8.209050350541746, "grad_norm": 0.49466875195503235, "learning_rate": 8.586000000000001e-05, "loss": 0.0479, "step": 12880 }, { "epoch": 8.21542383683875, "grad_norm": 0.5850510001182556, "learning_rate": 8.592666666666667e-05, "loss": 0.06, "step": 12890 }, { "epoch": 8.221797323135755, "grad_norm": 0.7878484725952148, "learning_rate": 8.599333333333334e-05, "loss": 0.074, "step": 12900 }, { "epoch": 8.22817080943276, "grad_norm": 0.8049789071083069, "learning_rate": 8.606000000000001e-05, "loss": 0.0603, "step": 12910 }, { "epoch": 8.234544295729764, "grad_norm": 0.7117138504981995, "learning_rate": 8.612666666666666e-05, "loss": 0.0422, "step": 12920 }, { "epoch": 8.240917782026768, "grad_norm": 1.0727053880691528, "learning_rate": 8.619333333333334e-05, "loss": 0.0529, "step": 12930 }, { "epoch": 8.247291268323773, "grad_norm": 0.576468825340271, "learning_rate": 8.626e-05, "loss": 0.0501, "step": 12940 }, { "epoch": 8.253664754620777, "grad_norm": 0.6107454299926758, "learning_rate": 8.632666666666667e-05, "loss": 0.0636, "step": 12950 }, { "epoch": 8.260038240917781, "grad_norm": 0.7168861627578735, "learning_rate": 8.639333333333333e-05, "loss": 0.0449, "step": 12960 }, { "epoch": 8.266411727214786, "grad_norm": 0.5779502391815186, "learning_rate": 8.646000000000001e-05, "loss": 0.0498, "step": 12970 }, { "epoch": 8.27278521351179, "grad_norm": 0.42147284746170044, "learning_rate": 8.652666666666667e-05, "loss": 0.0487, "step": 12980 }, { "epoch": 8.279158699808795, "grad_norm": 0.435723215341568, "learning_rate": 8.659333333333334e-05, "loss": 0.0493, "step": 12990 }, { "epoch": 8.2855321861058, "grad_norm": 0.8571350574493408, "learning_rate": 8.666000000000001e-05, "loss": 0.0562, "step": 13000 }, { "epoch": 8.291905672402804, "grad_norm": 0.9729090929031372, "learning_rate": 8.672666666666666e-05, "loss": 0.0666, "step": 13010 }, { "epoch": 8.29827915869981, "grad_norm": 0.5585812926292419, "learning_rate": 8.679333333333334e-05, "loss": 0.0569, "step": 13020 }, { "epoch": 8.304652644996814, "grad_norm": 0.9320511817932129, "learning_rate": 8.686e-05, "loss": 0.0748, "step": 13030 }, { "epoch": 8.311026131293819, "grad_norm": 0.5809386968612671, "learning_rate": 8.692666666666667e-05, "loss": 0.0504, "step": 13040 }, { "epoch": 8.317399617590823, "grad_norm": 0.6346476078033447, "learning_rate": 8.699333333333333e-05, "loss": 0.0596, "step": 13050 }, { "epoch": 8.323773103887827, "grad_norm": 0.6254171133041382, "learning_rate": 8.706000000000001e-05, "loss": 0.0483, "step": 13060 }, { "epoch": 8.330146590184832, "grad_norm": 0.6562458872795105, "learning_rate": 8.712666666666667e-05, "loss": 0.0645, "step": 13070 }, { "epoch": 8.336520076481836, "grad_norm": 0.6064593195915222, "learning_rate": 8.719333333333334e-05, "loss": 0.0661, "step": 13080 }, { "epoch": 8.34289356277884, "grad_norm": 1.0466936826705933, "learning_rate": 8.726000000000001e-05, "loss": 0.064, "step": 13090 }, { "epoch": 8.349267049075845, "grad_norm": 0.708739161491394, "learning_rate": 8.732666666666667e-05, "loss": 0.0827, "step": 13100 }, { "epoch": 8.35564053537285, "grad_norm": 0.4305594861507416, "learning_rate": 8.739333333333334e-05, "loss": 0.0573, "step": 13110 }, { "epoch": 8.362014021669854, "grad_norm": 0.581961989402771, "learning_rate": 8.746e-05, "loss": 0.0661, "step": 13120 }, { "epoch": 8.368387507966858, "grad_norm": 0.5669571161270142, "learning_rate": 8.752666666666667e-05, "loss": 0.0654, "step": 13130 }, { "epoch": 8.374760994263863, "grad_norm": 0.7762645483016968, "learning_rate": 8.759333333333333e-05, "loss": 0.0656, "step": 13140 }, { "epoch": 8.381134480560867, "grad_norm": 0.5275634527206421, "learning_rate": 8.766000000000001e-05, "loss": 0.0713, "step": 13150 }, { "epoch": 8.387507966857871, "grad_norm": 0.5054892897605896, "learning_rate": 8.772666666666666e-05, "loss": 0.0544, "step": 13160 }, { "epoch": 8.393881453154876, "grad_norm": 0.7584949731826782, "learning_rate": 8.779333333333334e-05, "loss": 0.0508, "step": 13170 }, { "epoch": 8.40025493945188, "grad_norm": 0.8117373585700989, "learning_rate": 8.786e-05, "loss": 0.065, "step": 13180 }, { "epoch": 8.406628425748885, "grad_norm": 0.5433506965637207, "learning_rate": 8.792666666666667e-05, "loss": 0.0513, "step": 13190 }, { "epoch": 8.413001912045889, "grad_norm": 0.6876084804534912, "learning_rate": 8.799333333333334e-05, "loss": 0.0525, "step": 13200 }, { "epoch": 8.419375398342893, "grad_norm": 0.40120765566825867, "learning_rate": 8.806000000000001e-05, "loss": 0.0678, "step": 13210 }, { "epoch": 8.425748884639898, "grad_norm": 0.7087835669517517, "learning_rate": 8.812666666666667e-05, "loss": 0.0603, "step": 13220 }, { "epoch": 8.432122370936902, "grad_norm": 0.6004530787467957, "learning_rate": 8.819333333333333e-05, "loss": 0.0559, "step": 13230 }, { "epoch": 8.438495857233907, "grad_norm": 0.4934057295322418, "learning_rate": 8.826000000000001e-05, "loss": 0.0649, "step": 13240 }, { "epoch": 8.444869343530911, "grad_norm": 0.9917402863502502, "learning_rate": 8.832666666666666e-05, "loss": 0.0628, "step": 13250 }, { "epoch": 8.451242829827915, "grad_norm": 0.7195178866386414, "learning_rate": 8.839333333333334e-05, "loss": 0.0511, "step": 13260 }, { "epoch": 8.45761631612492, "grad_norm": 0.8430331349372864, "learning_rate": 8.846e-05, "loss": 0.0555, "step": 13270 }, { "epoch": 8.463989802421924, "grad_norm": 0.42990657687187195, "learning_rate": 8.852666666666667e-05, "loss": 0.0472, "step": 13280 }, { "epoch": 8.470363288718929, "grad_norm": 0.6589018106460571, "learning_rate": 8.859333333333334e-05, "loss": 0.0627, "step": 13290 }, { "epoch": 8.476736775015933, "grad_norm": 0.5364127159118652, "learning_rate": 8.866000000000001e-05, "loss": 0.0499, "step": 13300 }, { "epoch": 8.483110261312937, "grad_norm": 0.8255322575569153, "learning_rate": 8.872666666666667e-05, "loss": 0.0555, "step": 13310 }, { "epoch": 8.489483747609942, "grad_norm": 0.7342777252197266, "learning_rate": 8.879333333333334e-05, "loss": 0.062, "step": 13320 }, { "epoch": 8.495857233906946, "grad_norm": 0.6865614652633667, "learning_rate": 8.886e-05, "loss": 0.0653, "step": 13330 }, { "epoch": 8.50223072020395, "grad_norm": 0.886336624622345, "learning_rate": 8.892666666666666e-05, "loss": 0.0651, "step": 13340 }, { "epoch": 8.508604206500957, "grad_norm": 0.8968199491500854, "learning_rate": 8.899333333333334e-05, "loss": 0.0542, "step": 13350 }, { "epoch": 8.514977692797961, "grad_norm": 0.6704058647155762, "learning_rate": 8.906e-05, "loss": 0.0604, "step": 13360 }, { "epoch": 8.521351179094966, "grad_norm": 0.47855129837989807, "learning_rate": 8.912666666666667e-05, "loss": 0.0649, "step": 13370 }, { "epoch": 8.52772466539197, "grad_norm": 0.5004100799560547, "learning_rate": 8.919333333333335e-05, "loss": 0.0473, "step": 13380 }, { "epoch": 8.534098151688974, "grad_norm": 0.5227582454681396, "learning_rate": 8.926e-05, "loss": 0.0782, "step": 13390 }, { "epoch": 8.540471637985979, "grad_norm": 0.6243178248405457, "learning_rate": 8.932666666666667e-05, "loss": 0.0713, "step": 13400 }, { "epoch": 8.546845124282983, "grad_norm": 0.6021842360496521, "learning_rate": 8.939333333333334e-05, "loss": 0.0601, "step": 13410 }, { "epoch": 8.553218610579988, "grad_norm": 0.8270822167396545, "learning_rate": 8.946e-05, "loss": 0.0643, "step": 13420 }, { "epoch": 8.559592096876992, "grad_norm": 0.7009769678115845, "learning_rate": 8.952666666666666e-05, "loss": 0.0477, "step": 13430 }, { "epoch": 8.565965583173996, "grad_norm": 0.7166087031364441, "learning_rate": 8.959333333333334e-05, "loss": 0.0633, "step": 13440 }, { "epoch": 8.572339069471, "grad_norm": 0.7389613389968872, "learning_rate": 8.966e-05, "loss": 0.0631, "step": 13450 }, { "epoch": 8.578712555768005, "grad_norm": 0.6592992544174194, "learning_rate": 8.972666666666667e-05, "loss": 0.077, "step": 13460 }, { "epoch": 8.58508604206501, "grad_norm": 0.5244159698486328, "learning_rate": 8.979333333333335e-05, "loss": 0.0703, "step": 13470 }, { "epoch": 8.591459528362014, "grad_norm": 0.8833013772964478, "learning_rate": 8.986e-05, "loss": 0.0618, "step": 13480 }, { "epoch": 8.597833014659019, "grad_norm": 0.38170257210731506, "learning_rate": 8.992666666666667e-05, "loss": 0.0694, "step": 13490 }, { "epoch": 8.604206500956023, "grad_norm": 0.8855859041213989, "learning_rate": 8.999333333333334e-05, "loss": 0.0506, "step": 13500 }, { "epoch": 8.610579987253027, "grad_norm": 0.6504349112510681, "learning_rate": 9.006e-05, "loss": 0.0692, "step": 13510 }, { "epoch": 8.616953473550032, "grad_norm": 0.6342761516571045, "learning_rate": 9.012666666666667e-05, "loss": 0.083, "step": 13520 }, { "epoch": 8.623326959847036, "grad_norm": 0.7263931632041931, "learning_rate": 9.019333333333334e-05, "loss": 0.0715, "step": 13530 }, { "epoch": 8.62970044614404, "grad_norm": 0.625174880027771, "learning_rate": 9.026e-05, "loss": 0.0668, "step": 13540 }, { "epoch": 8.636073932441045, "grad_norm": 0.6043142080307007, "learning_rate": 9.032666666666667e-05, "loss": 0.0574, "step": 13550 }, { "epoch": 8.64244741873805, "grad_norm": 1.0910696983337402, "learning_rate": 9.039333333333335e-05, "loss": 0.0701, "step": 13560 }, { "epoch": 8.648820905035054, "grad_norm": 0.7509660720825195, "learning_rate": 9.046e-05, "loss": 0.0426, "step": 13570 }, { "epoch": 8.655194391332058, "grad_norm": 0.7571632862091064, "learning_rate": 9.052666666666668e-05, "loss": 0.0551, "step": 13580 }, { "epoch": 8.661567877629063, "grad_norm": 0.9736087918281555, "learning_rate": 9.059333333333334e-05, "loss": 0.0666, "step": 13590 }, { "epoch": 8.667941363926067, "grad_norm": 0.7005413770675659, "learning_rate": 9.066e-05, "loss": 0.0648, "step": 13600 }, { "epoch": 8.674314850223071, "grad_norm": 0.64781254529953, "learning_rate": 9.072666666666667e-05, "loss": 0.0552, "step": 13610 }, { "epoch": 8.680688336520076, "grad_norm": 0.6549044847488403, "learning_rate": 9.079333333333334e-05, "loss": 0.0625, "step": 13620 }, { "epoch": 8.68706182281708, "grad_norm": 0.8877256512641907, "learning_rate": 9.086e-05, "loss": 0.0557, "step": 13630 }, { "epoch": 8.693435309114085, "grad_norm": 0.7637243270874023, "learning_rate": 9.092666666666667e-05, "loss": 0.0504, "step": 13640 }, { "epoch": 8.69980879541109, "grad_norm": 0.7398011088371277, "learning_rate": 9.099333333333334e-05, "loss": 0.0626, "step": 13650 }, { "epoch": 8.706182281708095, "grad_norm": 0.4594363272190094, "learning_rate": 9.106e-05, "loss": 0.0644, "step": 13660 }, { "epoch": 8.7125557680051, "grad_norm": 0.5682175755500793, "learning_rate": 9.112666666666668e-05, "loss": 0.077, "step": 13670 }, { "epoch": 8.718929254302104, "grad_norm": 0.4889727830886841, "learning_rate": 9.119333333333334e-05, "loss": 0.0686, "step": 13680 }, { "epoch": 8.725302740599108, "grad_norm": 0.7815690040588379, "learning_rate": 9.126e-05, "loss": 0.0561, "step": 13690 }, { "epoch": 8.731676226896113, "grad_norm": 0.756198525428772, "learning_rate": 9.132666666666667e-05, "loss": 0.0658, "step": 13700 }, { "epoch": 8.738049713193117, "grad_norm": 0.4668671190738678, "learning_rate": 9.139333333333334e-05, "loss": 0.0552, "step": 13710 }, { "epoch": 8.744423199490122, "grad_norm": 0.7657446265220642, "learning_rate": 9.146e-05, "loss": 0.056, "step": 13720 }, { "epoch": 8.750796685787126, "grad_norm": 0.5875611305236816, "learning_rate": 9.152666666666667e-05, "loss": 0.049, "step": 13730 }, { "epoch": 8.75717017208413, "grad_norm": 0.5725999474525452, "learning_rate": 9.159333333333334e-05, "loss": 0.0641, "step": 13740 }, { "epoch": 8.763543658381135, "grad_norm": 0.5676485300064087, "learning_rate": 9.166e-05, "loss": 0.0556, "step": 13750 }, { "epoch": 8.76991714467814, "grad_norm": 0.6588584780693054, "learning_rate": 9.172666666666668e-05, "loss": 0.0559, "step": 13760 }, { "epoch": 8.776290630975144, "grad_norm": 0.7619849443435669, "learning_rate": 9.179333333333334e-05, "loss": 0.0781, "step": 13770 }, { "epoch": 8.782664117272148, "grad_norm": 0.5273615717887878, "learning_rate": 9.186e-05, "loss": 0.0504, "step": 13780 }, { "epoch": 8.789037603569152, "grad_norm": 0.6352676153182983, "learning_rate": 9.192666666666667e-05, "loss": 0.0545, "step": 13790 }, { "epoch": 8.795411089866157, "grad_norm": 0.5762425661087036, "learning_rate": 9.199333333333335e-05, "loss": 0.0541, "step": 13800 }, { "epoch": 8.801784576163161, "grad_norm": 0.6435329914093018, "learning_rate": 9.206e-05, "loss": 0.0551, "step": 13810 }, { "epoch": 8.808158062460166, "grad_norm": 0.6709913015365601, "learning_rate": 9.212666666666667e-05, "loss": 0.0658, "step": 13820 }, { "epoch": 8.81453154875717, "grad_norm": 0.794775664806366, "learning_rate": 9.219333333333334e-05, "loss": 0.0687, "step": 13830 }, { "epoch": 8.820905035054174, "grad_norm": 0.5512056946754456, "learning_rate": 9.226e-05, "loss": 0.0581, "step": 13840 }, { "epoch": 8.827278521351179, "grad_norm": 0.4298240840435028, "learning_rate": 9.232666666666668e-05, "loss": 0.0494, "step": 13850 }, { "epoch": 8.833652007648183, "grad_norm": 0.425416499376297, "learning_rate": 9.239333333333334e-05, "loss": 0.065, "step": 13860 }, { "epoch": 8.840025493945188, "grad_norm": 0.5146231651306152, "learning_rate": 9.246e-05, "loss": 0.0577, "step": 13870 }, { "epoch": 8.846398980242192, "grad_norm": 0.5774735808372498, "learning_rate": 9.252666666666667e-05, "loss": 0.0407, "step": 13880 }, { "epoch": 8.852772466539196, "grad_norm": 0.580212414264679, "learning_rate": 9.259333333333335e-05, "loss": 0.0366, "step": 13890 }, { "epoch": 8.8591459528362, "grad_norm": 0.46737149357795715, "learning_rate": 9.266e-05, "loss": 0.0604, "step": 13900 }, { "epoch": 8.865519439133205, "grad_norm": 0.6332247853279114, "learning_rate": 9.272666666666667e-05, "loss": 0.0602, "step": 13910 }, { "epoch": 8.87189292543021, "grad_norm": 0.38513123989105225, "learning_rate": 9.279333333333334e-05, "loss": 0.0618, "step": 13920 }, { "epoch": 8.878266411727214, "grad_norm": 0.7670353055000305, "learning_rate": 9.286e-05, "loss": 0.0665, "step": 13930 }, { "epoch": 8.884639898024218, "grad_norm": 0.5132635831832886, "learning_rate": 9.292666666666668e-05, "loss": 0.058, "step": 13940 }, { "epoch": 8.891013384321223, "grad_norm": 0.5709401369094849, "learning_rate": 9.299333333333333e-05, "loss": 0.0545, "step": 13950 }, { "epoch": 8.897386870618227, "grad_norm": 0.469900518655777, "learning_rate": 9.306000000000001e-05, "loss": 0.0579, "step": 13960 }, { "epoch": 8.903760356915232, "grad_norm": 0.3554374873638153, "learning_rate": 9.312666666666667e-05, "loss": 0.076, "step": 13970 }, { "epoch": 8.910133843212238, "grad_norm": 0.5653266310691833, "learning_rate": 9.319333333333333e-05, "loss": 0.0668, "step": 13980 }, { "epoch": 8.916507329509242, "grad_norm": 0.9526334404945374, "learning_rate": 9.326e-05, "loss": 0.0744, "step": 13990 }, { "epoch": 8.922880815806247, "grad_norm": 0.5525512099266052, "learning_rate": 9.332666666666668e-05, "loss": 0.0585, "step": 14000 }, { "epoch": 8.929254302103251, "grad_norm": 0.5350542068481445, "learning_rate": 9.339333333333334e-05, "loss": 0.0504, "step": 14010 }, { "epoch": 8.935627788400256, "grad_norm": 0.6447840929031372, "learning_rate": 9.346e-05, "loss": 0.069, "step": 14020 }, { "epoch": 8.94200127469726, "grad_norm": 0.6802664995193481, "learning_rate": 9.352666666666667e-05, "loss": 0.0651, "step": 14030 }, { "epoch": 8.948374760994264, "grad_norm": 0.7058776617050171, "learning_rate": 9.359333333333333e-05, "loss": 0.0524, "step": 14040 }, { "epoch": 8.954748247291269, "grad_norm": 0.9878630638122559, "learning_rate": 9.366000000000001e-05, "loss": 0.052, "step": 14050 }, { "epoch": 8.961121733588273, "grad_norm": 0.7848569750785828, "learning_rate": 9.372666666666667e-05, "loss": 0.0616, "step": 14060 }, { "epoch": 8.967495219885278, "grad_norm": 0.5310623049736023, "learning_rate": 9.379333333333334e-05, "loss": 0.0505, "step": 14070 }, { "epoch": 8.973868706182282, "grad_norm": 0.4780806303024292, "learning_rate": 9.386e-05, "loss": 0.0493, "step": 14080 }, { "epoch": 8.980242192479286, "grad_norm": 0.5557371377944946, "learning_rate": 9.392666666666668e-05, "loss": 0.0494, "step": 14090 }, { "epoch": 8.98661567877629, "grad_norm": 0.9081065058708191, "learning_rate": 9.399333333333333e-05, "loss": 0.059, "step": 14100 }, { "epoch": 8.992989165073295, "grad_norm": 1.0077979564666748, "learning_rate": 9.406e-05, "loss": 0.0605, "step": 14110 }, { "epoch": 8.9993626513703, "grad_norm": 0.7335388660430908, "learning_rate": 9.412666666666667e-05, "loss": 0.0495, "step": 14120 }, { "epoch": 9.005736137667304, "grad_norm": 0.7435501217842102, "learning_rate": 9.419333333333333e-05, "loss": 0.055, "step": 14130 }, { "epoch": 9.012109623964308, "grad_norm": 0.6691561937332153, "learning_rate": 9.426000000000001e-05, "loss": 0.072, "step": 14140 }, { "epoch": 9.018483110261313, "grad_norm": 0.4774843156337738, "learning_rate": 9.432666666666667e-05, "loss": 0.041, "step": 14150 }, { "epoch": 9.024856596558317, "grad_norm": 0.8308414220809937, "learning_rate": 9.439333333333334e-05, "loss": 0.0728, "step": 14160 }, { "epoch": 9.031230082855322, "grad_norm": 0.8462230563163757, "learning_rate": 9.446e-05, "loss": 0.0631, "step": 14170 }, { "epoch": 9.037603569152326, "grad_norm": 0.435079962015152, "learning_rate": 9.452666666666668e-05, "loss": 0.0539, "step": 14180 }, { "epoch": 9.04397705544933, "grad_norm": 0.46461808681488037, "learning_rate": 9.459333333333333e-05, "loss": 0.0646, "step": 14190 }, { "epoch": 9.050350541746335, "grad_norm": 0.6792469620704651, "learning_rate": 9.466e-05, "loss": 0.0594, "step": 14200 }, { "epoch": 9.05672402804334, "grad_norm": 0.6812877058982849, "learning_rate": 9.472666666666667e-05, "loss": 0.0655, "step": 14210 }, { "epoch": 9.063097514340344, "grad_norm": 0.5176808834075928, "learning_rate": 9.479333333333333e-05, "loss": 0.0489, "step": 14220 }, { "epoch": 9.069471000637348, "grad_norm": 0.5170048475265503, "learning_rate": 9.486000000000001e-05, "loss": 0.0583, "step": 14230 }, { "epoch": 9.075844486934352, "grad_norm": 0.5931413769721985, "learning_rate": 9.492666666666667e-05, "loss": 0.0473, "step": 14240 }, { "epoch": 9.082217973231357, "grad_norm": 0.7961833477020264, "learning_rate": 9.499333333333334e-05, "loss": 0.0671, "step": 14250 }, { "epoch": 9.088591459528361, "grad_norm": 0.5670689940452576, "learning_rate": 9.506e-05, "loss": 0.0483, "step": 14260 }, { "epoch": 9.094964945825366, "grad_norm": 0.598980724811554, "learning_rate": 9.512666666666668e-05, "loss": 0.0544, "step": 14270 }, { "epoch": 9.101338432122372, "grad_norm": 0.6023126244544983, "learning_rate": 9.519333333333333e-05, "loss": 0.0638, "step": 14280 }, { "epoch": 9.107711918419376, "grad_norm": 0.6292657256126404, "learning_rate": 9.526e-05, "loss": 0.0595, "step": 14290 }, { "epoch": 9.11408540471638, "grad_norm": 0.715130627155304, "learning_rate": 9.532666666666667e-05, "loss": 0.054, "step": 14300 }, { "epoch": 9.120458891013385, "grad_norm": 0.6806856989860535, "learning_rate": 9.539333333333333e-05, "loss": 0.0517, "step": 14310 }, { "epoch": 9.12683237731039, "grad_norm": 0.505490243434906, "learning_rate": 9.546000000000001e-05, "loss": 0.071, "step": 14320 }, { "epoch": 9.133205863607394, "grad_norm": 0.6481647491455078, "learning_rate": 9.552666666666667e-05, "loss": 0.0511, "step": 14330 }, { "epoch": 9.139579349904398, "grad_norm": 0.41041502356529236, "learning_rate": 9.559333333333334e-05, "loss": 0.0504, "step": 14340 }, { "epoch": 9.145952836201403, "grad_norm": 0.4806886315345764, "learning_rate": 9.566e-05, "loss": 0.0588, "step": 14350 }, { "epoch": 9.152326322498407, "grad_norm": 1.0263683795928955, "learning_rate": 9.572666666666668e-05, "loss": 0.0579, "step": 14360 }, { "epoch": 9.158699808795411, "grad_norm": 0.3896993398666382, "learning_rate": 9.579333333333333e-05, "loss": 0.0475, "step": 14370 }, { "epoch": 9.165073295092416, "grad_norm": 0.6258089542388916, "learning_rate": 9.586000000000001e-05, "loss": 0.0552, "step": 14380 }, { "epoch": 9.17144678138942, "grad_norm": 0.675748884677887, "learning_rate": 9.592666666666667e-05, "loss": 0.0624, "step": 14390 }, { "epoch": 9.177820267686425, "grad_norm": 0.8424323201179504, "learning_rate": 9.599333333333333e-05, "loss": 0.0698, "step": 14400 }, { "epoch": 9.184193753983429, "grad_norm": 0.703313946723938, "learning_rate": 9.606000000000001e-05, "loss": 0.0445, "step": 14410 }, { "epoch": 9.190567240280433, "grad_norm": 0.6200817227363586, "learning_rate": 9.612666666666668e-05, "loss": 0.0578, "step": 14420 }, { "epoch": 9.196940726577438, "grad_norm": 0.3376152217388153, "learning_rate": 9.619333333333334e-05, "loss": 0.055, "step": 14430 }, { "epoch": 9.203314212874442, "grad_norm": 0.4722602665424347, "learning_rate": 9.626e-05, "loss": 0.0646, "step": 14440 }, { "epoch": 9.209687699171447, "grad_norm": 0.7795982360839844, "learning_rate": 9.632666666666668e-05, "loss": 0.0671, "step": 14450 }, { "epoch": 9.216061185468451, "grad_norm": 0.8860408663749695, "learning_rate": 9.639333333333333e-05, "loss": 0.057, "step": 14460 }, { "epoch": 9.222434671765456, "grad_norm": 0.9807704091072083, "learning_rate": 9.646000000000001e-05, "loss": 0.0587, "step": 14470 }, { "epoch": 9.22880815806246, "grad_norm": 0.4594630300998688, "learning_rate": 9.652666666666667e-05, "loss": 0.0416, "step": 14480 }, { "epoch": 9.235181644359464, "grad_norm": 0.6166505813598633, "learning_rate": 9.659333333333334e-05, "loss": 0.0478, "step": 14490 }, { "epoch": 9.241555130656469, "grad_norm": 0.45989012718200684, "learning_rate": 9.666e-05, "loss": 0.0597, "step": 14500 }, { "epoch": 9.247928616953473, "grad_norm": 0.7244271636009216, "learning_rate": 9.672666666666668e-05, "loss": 0.0591, "step": 14510 }, { "epoch": 9.254302103250478, "grad_norm": 0.635843813419342, "learning_rate": 9.679333333333334e-05, "loss": 0.0509, "step": 14520 }, { "epoch": 9.260675589547482, "grad_norm": 0.6153807640075684, "learning_rate": 9.686e-05, "loss": 0.0752, "step": 14530 }, { "epoch": 9.267049075844486, "grad_norm": 0.6309207081794739, "learning_rate": 9.692666666666668e-05, "loss": 0.0523, "step": 14540 }, { "epoch": 9.27342256214149, "grad_norm": 0.6113809943199158, "learning_rate": 9.699333333333333e-05, "loss": 0.0533, "step": 14550 }, { "epoch": 9.279796048438495, "grad_norm": 0.7630850672721863, "learning_rate": 9.706000000000001e-05, "loss": 0.0497, "step": 14560 }, { "epoch": 9.2861695347355, "grad_norm": 0.7670928835868835, "learning_rate": 9.712666666666666e-05, "loss": 0.0447, "step": 14570 }, { "epoch": 9.292543021032504, "grad_norm": 0.5984125137329102, "learning_rate": 9.719333333333334e-05, "loss": 0.0516, "step": 14580 }, { "epoch": 9.298916507329508, "grad_norm": 0.5009859800338745, "learning_rate": 9.726e-05, "loss": 0.0647, "step": 14590 }, { "epoch": 9.305289993626515, "grad_norm": 1.0185586214065552, "learning_rate": 9.732666666666666e-05, "loss": 0.0597, "step": 14600 }, { "epoch": 9.311663479923519, "grad_norm": 0.6494755744934082, "learning_rate": 9.739333333333334e-05, "loss": 0.0504, "step": 14610 }, { "epoch": 9.318036966220523, "grad_norm": 0.7207589149475098, "learning_rate": 9.746e-05, "loss": 0.0532, "step": 14620 }, { "epoch": 9.324410452517528, "grad_norm": 0.5885887145996094, "learning_rate": 9.752666666666667e-05, "loss": 0.0533, "step": 14630 }, { "epoch": 9.330783938814532, "grad_norm": 0.9548124670982361, "learning_rate": 9.759333333333333e-05, "loss": 0.0541, "step": 14640 }, { "epoch": 9.337157425111537, "grad_norm": 0.6898685097694397, "learning_rate": 9.766000000000001e-05, "loss": 0.048, "step": 14650 }, { "epoch": 9.343530911408541, "grad_norm": 0.8245808482170105, "learning_rate": 9.772666666666666e-05, "loss": 0.0583, "step": 14660 }, { "epoch": 9.349904397705545, "grad_norm": 0.6053099632263184, "learning_rate": 9.779333333333334e-05, "loss": 0.066, "step": 14670 }, { "epoch": 9.35627788400255, "grad_norm": 0.6548680067062378, "learning_rate": 9.786e-05, "loss": 0.0485, "step": 14680 }, { "epoch": 9.362651370299554, "grad_norm": 0.6803520917892456, "learning_rate": 9.792666666666667e-05, "loss": 0.0548, "step": 14690 }, { "epoch": 9.369024856596559, "grad_norm": 0.6984466314315796, "learning_rate": 9.799333333333334e-05, "loss": 0.0462, "step": 14700 }, { "epoch": 9.375398342893563, "grad_norm": 0.5420809984207153, "learning_rate": 9.806e-05, "loss": 0.0531, "step": 14710 }, { "epoch": 9.381771829190567, "grad_norm": 0.70583575963974, "learning_rate": 9.812666666666667e-05, "loss": 0.0716, "step": 14720 }, { "epoch": 9.388145315487572, "grad_norm": 0.9024677872657776, "learning_rate": 9.819333333333333e-05, "loss": 0.0508, "step": 14730 }, { "epoch": 9.394518801784576, "grad_norm": 0.9504936337471008, "learning_rate": 9.826000000000001e-05, "loss": 0.0914, "step": 14740 }, { "epoch": 9.40089228808158, "grad_norm": 0.5353940725326538, "learning_rate": 9.832666666666666e-05, "loss": 0.0551, "step": 14750 }, { "epoch": 9.407265774378585, "grad_norm": 0.9119231700897217, "learning_rate": 9.839333333333334e-05, "loss": 0.0744, "step": 14760 }, { "epoch": 9.41363926067559, "grad_norm": 0.8097683787345886, "learning_rate": 9.846e-05, "loss": 0.0662, "step": 14770 }, { "epoch": 9.420012746972594, "grad_norm": 0.7556730508804321, "learning_rate": 9.852666666666667e-05, "loss": 0.0588, "step": 14780 }, { "epoch": 9.426386233269598, "grad_norm": 0.5082266330718994, "learning_rate": 9.859333333333334e-05, "loss": 0.0523, "step": 14790 }, { "epoch": 9.432759719566603, "grad_norm": 0.4727637767791748, "learning_rate": 9.866000000000001e-05, "loss": 0.0565, "step": 14800 }, { "epoch": 9.439133205863607, "grad_norm": 0.7173406481742859, "learning_rate": 9.872666666666667e-05, "loss": 0.0618, "step": 14810 }, { "epoch": 9.445506692160611, "grad_norm": 0.6705032587051392, "learning_rate": 9.879333333333333e-05, "loss": 0.0496, "step": 14820 }, { "epoch": 9.451880178457616, "grad_norm": 0.5275325179100037, "learning_rate": 9.886000000000001e-05, "loss": 0.0737, "step": 14830 }, { "epoch": 9.45825366475462, "grad_norm": 0.8255791068077087, "learning_rate": 9.892666666666666e-05, "loss": 0.0759, "step": 14840 }, { "epoch": 9.464627151051625, "grad_norm": 0.8190925717353821, "learning_rate": 9.899333333333334e-05, "loss": 0.0459, "step": 14850 }, { "epoch": 9.471000637348629, "grad_norm": 0.967989444732666, "learning_rate": 9.906e-05, "loss": 0.0828, "step": 14860 }, { "epoch": 9.477374123645633, "grad_norm": 0.6458868384361267, "learning_rate": 9.912666666666667e-05, "loss": 0.0643, "step": 14870 }, { "epoch": 9.483747609942638, "grad_norm": 0.577491283416748, "learning_rate": 9.919333333333334e-05, "loss": 0.056, "step": 14880 }, { "epoch": 9.490121096239642, "grad_norm": 0.48064032196998596, "learning_rate": 9.926000000000001e-05, "loss": 0.0672, "step": 14890 }, { "epoch": 9.496494582536647, "grad_norm": 0.7018259763717651, "learning_rate": 9.932666666666667e-05, "loss": 0.0638, "step": 14900 }, { "epoch": 9.502868068833653, "grad_norm": 0.5664310455322266, "learning_rate": 9.939333333333334e-05, "loss": 0.0641, "step": 14910 }, { "epoch": 9.509241555130657, "grad_norm": 0.786356508731842, "learning_rate": 9.946000000000001e-05, "loss": 0.0559, "step": 14920 }, { "epoch": 9.515615041427662, "grad_norm": 0.5695542097091675, "learning_rate": 9.952666666666666e-05, "loss": 0.0567, "step": 14930 }, { "epoch": 9.521988527724666, "grad_norm": 0.7495202422142029, "learning_rate": 9.959333333333334e-05, "loss": 0.0528, "step": 14940 }, { "epoch": 9.52836201402167, "grad_norm": 0.7040093541145325, "learning_rate": 9.966e-05, "loss": 0.0584, "step": 14950 }, { "epoch": 9.534735500318675, "grad_norm": 0.6281925439834595, "learning_rate": 9.972666666666667e-05, "loss": 0.0637, "step": 14960 }, { "epoch": 9.54110898661568, "grad_norm": 0.5572054386138916, "learning_rate": 9.979333333333333e-05, "loss": 0.0604, "step": 14970 }, { "epoch": 9.547482472912684, "grad_norm": 0.7755315899848938, "learning_rate": 9.986000000000001e-05, "loss": 0.0568, "step": 14980 }, { "epoch": 9.553855959209688, "grad_norm": 0.4848751723766327, "learning_rate": 9.992666666666667e-05, "loss": 0.0624, "step": 14990 }, { "epoch": 9.560229445506693, "grad_norm": 0.6210675835609436, "learning_rate": 9.999333333333334e-05, "loss": 0.0678, "step": 15000 }, { "epoch": 9.566602931803697, "grad_norm": 0.42524453997612, "learning_rate": 9.999999975394338e-05, "loss": 0.0494, "step": 15010 }, { "epoch": 9.572976418100701, "grad_norm": 0.36582857370376587, "learning_rate": 9.999999890337731e-05, "loss": 0.0519, "step": 15020 }, { "epoch": 9.579349904397706, "grad_norm": 0.47146666049957275, "learning_rate": 9.999999744526401e-05, "loss": 0.0613, "step": 15030 }, { "epoch": 9.58572339069471, "grad_norm": 0.5648419857025146, "learning_rate": 9.999999537960356e-05, "loss": 0.0738, "step": 15040 }, { "epoch": 9.592096876991715, "grad_norm": 0.6596403121948242, "learning_rate": 9.999999270639596e-05, "loss": 0.0512, "step": 15050 }, { "epoch": 9.598470363288719, "grad_norm": 0.6999940276145935, "learning_rate": 9.999998942564122e-05, "loss": 0.0812, "step": 15060 }, { "epoch": 9.604843849585723, "grad_norm": 1.080168604850769, "learning_rate": 9.99999855373394e-05, "loss": 0.0596, "step": 15070 }, { "epoch": 9.611217335882728, "grad_norm": 0.8175768256187439, "learning_rate": 9.999998104149056e-05, "loss": 0.0555, "step": 15080 }, { "epoch": 9.617590822179732, "grad_norm": 0.5036365985870361, "learning_rate": 9.999997593809473e-05, "loss": 0.0606, "step": 15090 }, { "epoch": 9.623964308476737, "grad_norm": 0.7283940315246582, "learning_rate": 9.999997022715201e-05, "loss": 0.0679, "step": 15100 }, { "epoch": 9.630337794773741, "grad_norm": 0.4878939390182495, "learning_rate": 9.999996390866243e-05, "loss": 0.0527, "step": 15110 }, { "epoch": 9.636711281070745, "grad_norm": 0.5333446860313416, "learning_rate": 9.999995698262608e-05, "loss": 0.0503, "step": 15120 }, { "epoch": 9.64308476736775, "grad_norm": 0.5796135067939758, "learning_rate": 9.999994944904306e-05, "loss": 0.0565, "step": 15130 }, { "epoch": 9.649458253664754, "grad_norm": 0.6852944493293762, "learning_rate": 9.999994130791341e-05, "loss": 0.0422, "step": 15140 }, { "epoch": 9.655831739961759, "grad_norm": 0.8792035579681396, "learning_rate": 9.99999325592373e-05, "loss": 0.0655, "step": 15150 }, { "epoch": 9.662205226258763, "grad_norm": 0.5844212770462036, "learning_rate": 9.999992320301478e-05, "loss": 0.0566, "step": 15160 }, { "epoch": 9.668578712555767, "grad_norm": 0.5131065845489502, "learning_rate": 9.9999913239246e-05, "loss": 0.0602, "step": 15170 }, { "epoch": 9.674952198852772, "grad_norm": 0.8779581785202026, "learning_rate": 9.999990266793106e-05, "loss": 0.0692, "step": 15180 }, { "epoch": 9.681325685149776, "grad_norm": 0.4830127954483032, "learning_rate": 9.99998914890701e-05, "loss": 0.0568, "step": 15190 }, { "epoch": 9.68769917144678, "grad_norm": 0.6114006042480469, "learning_rate": 9.999987970266323e-05, "loss": 0.0726, "step": 15200 }, { "epoch": 9.694072657743785, "grad_norm": 0.8563700914382935, "learning_rate": 9.999986730871064e-05, "loss": 0.0685, "step": 15210 }, { "epoch": 9.70044614404079, "grad_norm": 0.7654949426651001, "learning_rate": 9.999985430721243e-05, "loss": 0.051, "step": 15220 }, { "epoch": 9.706819630337796, "grad_norm": 0.8609695434570312, "learning_rate": 9.999984069816881e-05, "loss": 0.0626, "step": 15230 }, { "epoch": 9.7131931166348, "grad_norm": 0.7848291397094727, "learning_rate": 9.99998264815799e-05, "loss": 0.0457, "step": 15240 }, { "epoch": 9.719566602931804, "grad_norm": 0.631909966468811, "learning_rate": 9.99998116574459e-05, "loss": 0.0645, "step": 15250 }, { "epoch": 9.725940089228809, "grad_norm": 0.6485551595687866, "learning_rate": 9.999979622576698e-05, "loss": 0.0753, "step": 15260 }, { "epoch": 9.732313575525813, "grad_norm": 1.1122710704803467, "learning_rate": 9.999978018654333e-05, "loss": 0.0663, "step": 15270 }, { "epoch": 9.738687061822818, "grad_norm": 0.516740620136261, "learning_rate": 9.999976353977511e-05, "loss": 0.053, "step": 15280 }, { "epoch": 9.745060548119822, "grad_norm": 0.7186670899391174, "learning_rate": 9.99997462854626e-05, "loss": 0.0623, "step": 15290 }, { "epoch": 9.751434034416826, "grad_norm": 0.48222893476486206, "learning_rate": 9.999972842360594e-05, "loss": 0.0624, "step": 15300 }, { "epoch": 9.75780752071383, "grad_norm": 0.6196928024291992, "learning_rate": 9.999970995420537e-05, "loss": 0.0857, "step": 15310 }, { "epoch": 9.764181007010835, "grad_norm": 0.7647284865379333, "learning_rate": 9.999969087726114e-05, "loss": 0.0486, "step": 15320 }, { "epoch": 9.77055449330784, "grad_norm": 0.629514217376709, "learning_rate": 9.999967119277344e-05, "loss": 0.0475, "step": 15330 }, { "epoch": 9.776927979604844, "grad_norm": 0.4425369203090668, "learning_rate": 9.999965090074254e-05, "loss": 0.0557, "step": 15340 }, { "epoch": 9.783301465901848, "grad_norm": 0.7280276417732239, "learning_rate": 9.999963000116865e-05, "loss": 0.0549, "step": 15350 }, { "epoch": 9.789674952198853, "grad_norm": 0.8114652037620544, "learning_rate": 9.999960849405208e-05, "loss": 0.0519, "step": 15360 }, { "epoch": 9.796048438495857, "grad_norm": 0.7324187159538269, "learning_rate": 9.999958637939304e-05, "loss": 0.0611, "step": 15370 }, { "epoch": 9.802421924792862, "grad_norm": 0.5522525310516357, "learning_rate": 9.999956365719182e-05, "loss": 0.0573, "step": 15380 }, { "epoch": 9.808795411089866, "grad_norm": 0.664623498916626, "learning_rate": 9.99995403274487e-05, "loss": 0.04, "step": 15390 }, { "epoch": 9.81516889738687, "grad_norm": 0.696927547454834, "learning_rate": 9.999951639016395e-05, "loss": 0.0543, "step": 15400 }, { "epoch": 9.821542383683875, "grad_norm": 1.258500099182129, "learning_rate": 9.999949184533788e-05, "loss": 0.0614, "step": 15410 }, { "epoch": 9.82791586998088, "grad_norm": 0.6368541121482849, "learning_rate": 9.999946669297076e-05, "loss": 0.057, "step": 15420 }, { "epoch": 9.834289356277884, "grad_norm": 0.6270894408226013, "learning_rate": 9.999944093306293e-05, "loss": 0.0686, "step": 15430 }, { "epoch": 9.840662842574888, "grad_norm": 0.5678898692131042, "learning_rate": 9.999941456561465e-05, "loss": 0.0553, "step": 15440 }, { "epoch": 9.847036328871893, "grad_norm": 0.5162479281425476, "learning_rate": 9.999938759062633e-05, "loss": 0.0637, "step": 15450 }, { "epoch": 9.853409815168897, "grad_norm": 1.0005308389663696, "learning_rate": 9.999936000809819e-05, "loss": 0.0607, "step": 15460 }, { "epoch": 9.859783301465901, "grad_norm": 0.5103743672370911, "learning_rate": 9.999933181803066e-05, "loss": 0.0464, "step": 15470 }, { "epoch": 9.866156787762906, "grad_norm": 0.70570969581604, "learning_rate": 9.999930302042402e-05, "loss": 0.0477, "step": 15480 }, { "epoch": 9.87253027405991, "grad_norm": 0.3010031282901764, "learning_rate": 9.999927361527866e-05, "loss": 0.0578, "step": 15490 }, { "epoch": 9.878903760356915, "grad_norm": 0.5186561346054077, "learning_rate": 9.99992436025949e-05, "loss": 0.0531, "step": 15500 }, { "epoch": 9.885277246653919, "grad_norm": 0.5804335474967957, "learning_rate": 9.999921298237311e-05, "loss": 0.0633, "step": 15510 }, { "epoch": 9.891650732950923, "grad_norm": 0.824968159198761, "learning_rate": 9.999918175461371e-05, "loss": 0.0533, "step": 15520 }, { "epoch": 9.89802421924793, "grad_norm": 0.6402810215950012, "learning_rate": 9.999914991931702e-05, "loss": 0.0539, "step": 15530 }, { "epoch": 9.904397705544934, "grad_norm": 1.041672706604004, "learning_rate": 9.999911747648346e-05, "loss": 0.0579, "step": 15540 }, { "epoch": 9.910771191841938, "grad_norm": 0.5121114253997803, "learning_rate": 9.999908442611341e-05, "loss": 0.0424, "step": 15550 }, { "epoch": 9.917144678138943, "grad_norm": 0.3592798411846161, "learning_rate": 9.999905076820728e-05, "loss": 0.0579, "step": 15560 }, { "epoch": 9.923518164435947, "grad_norm": 0.6497433185577393, "learning_rate": 9.999901650276549e-05, "loss": 0.0585, "step": 15570 }, { "epoch": 9.929891650732952, "grad_norm": 0.6613918542861938, "learning_rate": 9.999898162978843e-05, "loss": 0.0824, "step": 15580 }, { "epoch": 9.936265137029956, "grad_norm": 0.5419838428497314, "learning_rate": 9.999894614927653e-05, "loss": 0.0525, "step": 15590 }, { "epoch": 9.94263862332696, "grad_norm": 0.41045868396759033, "learning_rate": 9.999891006123023e-05, "loss": 0.0507, "step": 15600 }, { "epoch": 9.949012109623965, "grad_norm": 0.6143380403518677, "learning_rate": 9.999887336564998e-05, "loss": 0.0612, "step": 15610 }, { "epoch": 9.95538559592097, "grad_norm": 0.7102112174034119, "learning_rate": 9.99988360625362e-05, "loss": 0.0799, "step": 15620 }, { "epoch": 9.961759082217974, "grad_norm": 0.5740867853164673, "learning_rate": 9.999879815188936e-05, "loss": 0.0605, "step": 15630 }, { "epoch": 9.968132568514978, "grad_norm": 0.4061245024204254, "learning_rate": 9.999875963370991e-05, "loss": 0.0647, "step": 15640 }, { "epoch": 9.974506054811982, "grad_norm": 0.7223836183547974, "learning_rate": 9.999872050799833e-05, "loss": 0.0654, "step": 15650 }, { "epoch": 9.980879541108987, "grad_norm": 0.5730621218681335, "learning_rate": 9.999868077475509e-05, "loss": 0.0509, "step": 15660 }, { "epoch": 9.987253027405991, "grad_norm": 0.5930061936378479, "learning_rate": 9.999864043398066e-05, "loss": 0.0638, "step": 15670 }, { "epoch": 9.993626513702996, "grad_norm": 0.6156326532363892, "learning_rate": 9.999859948567557e-05, "loss": 0.0514, "step": 15680 }, { "epoch": 10.0, "grad_norm": 0.7921385765075684, "learning_rate": 9.999855792984025e-05, "loss": 0.072, "step": 15690 }, { "epoch": 10.006373486297004, "grad_norm": 0.5710336565971375, "learning_rate": 9.999851576647528e-05, "loss": 0.0582, "step": 15700 }, { "epoch": 10.012746972594009, "grad_norm": 0.6952888369560242, "learning_rate": 9.999847299558111e-05, "loss": 0.0716, "step": 15710 }, { "epoch": 10.019120458891013, "grad_norm": 0.6671890616416931, "learning_rate": 9.999842961715831e-05, "loss": 0.059, "step": 15720 }, { "epoch": 10.025493945188018, "grad_norm": 0.5703734755516052, "learning_rate": 9.999838563120737e-05, "loss": 0.0665, "step": 15730 }, { "epoch": 10.031867431485022, "grad_norm": 0.7158628702163696, "learning_rate": 9.999834103772883e-05, "loss": 0.0625, "step": 15740 }, { "epoch": 10.038240917782026, "grad_norm": 0.7780296206474304, "learning_rate": 9.999829583672327e-05, "loss": 0.0554, "step": 15750 }, { "epoch": 10.04461440407903, "grad_norm": 0.7620775103569031, "learning_rate": 9.999825002819119e-05, "loss": 0.0526, "step": 15760 }, { "epoch": 10.050987890376035, "grad_norm": 0.7956362962722778, "learning_rate": 9.999820361213316e-05, "loss": 0.0603, "step": 15770 }, { "epoch": 10.05736137667304, "grad_norm": 0.5731815099716187, "learning_rate": 9.999815658854976e-05, "loss": 0.0446, "step": 15780 }, { "epoch": 10.063734862970044, "grad_norm": 0.9944173693656921, "learning_rate": 9.999810895744155e-05, "loss": 0.0616, "step": 15790 }, { "epoch": 10.070108349267048, "grad_norm": 0.4661408066749573, "learning_rate": 9.99980607188091e-05, "loss": 0.054, "step": 15800 }, { "epoch": 10.076481835564053, "grad_norm": 0.6796442866325378, "learning_rate": 9.999801187265301e-05, "loss": 0.0579, "step": 15810 }, { "epoch": 10.082855321861057, "grad_norm": 0.448863685131073, "learning_rate": 9.999796241897387e-05, "loss": 0.0556, "step": 15820 }, { "epoch": 10.089228808158062, "grad_norm": 0.644807755947113, "learning_rate": 9.99979123577723e-05, "loss": 0.0562, "step": 15830 }, { "epoch": 10.095602294455066, "grad_norm": 0.6758536100387573, "learning_rate": 9.999786168904886e-05, "loss": 0.049, "step": 15840 }, { "epoch": 10.101975780752072, "grad_norm": 0.6082631945610046, "learning_rate": 9.999781041280422e-05, "loss": 0.0472, "step": 15850 }, { "epoch": 10.108349267049077, "grad_norm": 1.315745234489441, "learning_rate": 9.999775852903896e-05, "loss": 0.0771, "step": 15860 }, { "epoch": 10.114722753346081, "grad_norm": 0.5523437857627869, "learning_rate": 9.999770603775373e-05, "loss": 0.0686, "step": 15870 }, { "epoch": 10.121096239643085, "grad_norm": 0.6934453845024109, "learning_rate": 9.999765293894918e-05, "loss": 0.0705, "step": 15880 }, { "epoch": 10.12746972594009, "grad_norm": 0.7601489424705505, "learning_rate": 9.999759923262593e-05, "loss": 0.0586, "step": 15890 }, { "epoch": 10.133843212237094, "grad_norm": 0.5398861765861511, "learning_rate": 9.999754491878464e-05, "loss": 0.0511, "step": 15900 }, { "epoch": 10.140216698534099, "grad_norm": 0.5415521264076233, "learning_rate": 9.999748999742597e-05, "loss": 0.0471, "step": 15910 }, { "epoch": 10.146590184831103, "grad_norm": 0.7853138446807861, "learning_rate": 9.999743446855061e-05, "loss": 0.0622, "step": 15920 }, { "epoch": 10.152963671128107, "grad_norm": 0.6591348052024841, "learning_rate": 9.99973783321592e-05, "loss": 0.064, "step": 15930 }, { "epoch": 10.159337157425112, "grad_norm": 0.7076617479324341, "learning_rate": 9.999732158825246e-05, "loss": 0.0599, "step": 15940 }, { "epoch": 10.165710643722116, "grad_norm": 0.6937175989151001, "learning_rate": 9.999726423683104e-05, "loss": 0.0727, "step": 15950 }, { "epoch": 10.17208413001912, "grad_norm": 0.5288292169570923, "learning_rate": 9.999720627789566e-05, "loss": 0.0526, "step": 15960 }, { "epoch": 10.178457616316125, "grad_norm": 0.6662324666976929, "learning_rate": 9.999714771144701e-05, "loss": 0.0488, "step": 15970 }, { "epoch": 10.18483110261313, "grad_norm": 0.3469081521034241, "learning_rate": 9.99970885374858e-05, "loss": 0.0599, "step": 15980 }, { "epoch": 10.191204588910134, "grad_norm": 0.5644394159317017, "learning_rate": 9.999702875601277e-05, "loss": 0.0518, "step": 15990 }, { "epoch": 10.197578075207138, "grad_norm": 0.38171422481536865, "learning_rate": 9.999696836702865e-05, "loss": 0.062, "step": 16000 }, { "epoch": 10.203951561504143, "grad_norm": 0.6088692545890808, "learning_rate": 9.999690737053416e-05, "loss": 0.0637, "step": 16010 }, { "epoch": 10.210325047801147, "grad_norm": 0.8294176459312439, "learning_rate": 9.999684576653001e-05, "loss": 0.0608, "step": 16020 }, { "epoch": 10.216698534098152, "grad_norm": 0.5365344285964966, "learning_rate": 9.999678355501701e-05, "loss": 0.0564, "step": 16030 }, { "epoch": 10.223072020395156, "grad_norm": 0.4190749526023865, "learning_rate": 9.999672073599586e-05, "loss": 0.049, "step": 16040 }, { "epoch": 10.22944550669216, "grad_norm": 0.7462962865829468, "learning_rate": 9.999665730946736e-05, "loss": 0.0517, "step": 16050 }, { "epoch": 10.235818992989165, "grad_norm": 0.47202739119529724, "learning_rate": 9.999659327543228e-05, "loss": 0.0506, "step": 16060 }, { "epoch": 10.24219247928617, "grad_norm": 0.598147988319397, "learning_rate": 9.999652863389138e-05, "loss": 0.0497, "step": 16070 }, { "epoch": 10.248565965583174, "grad_norm": 0.941378116607666, "learning_rate": 9.999646338484545e-05, "loss": 0.0555, "step": 16080 }, { "epoch": 10.254939451880178, "grad_norm": 0.6677988767623901, "learning_rate": 9.999639752829528e-05, "loss": 0.0526, "step": 16090 }, { "epoch": 10.261312938177182, "grad_norm": 0.6185131669044495, "learning_rate": 9.999633106424169e-05, "loss": 0.0551, "step": 16100 }, { "epoch": 10.267686424474187, "grad_norm": 0.5114786028862, "learning_rate": 9.999626399268546e-05, "loss": 0.0612, "step": 16110 }, { "epoch": 10.274059910771191, "grad_norm": 0.600007176399231, "learning_rate": 9.999619631362742e-05, "loss": 0.0668, "step": 16120 }, { "epoch": 10.280433397068196, "grad_norm": 0.6373008489608765, "learning_rate": 9.99961280270684e-05, "loss": 0.068, "step": 16130 }, { "epoch": 10.2868068833652, "grad_norm": 0.49713751673698425, "learning_rate": 9.99960591330092e-05, "loss": 0.0608, "step": 16140 }, { "epoch": 10.293180369662204, "grad_norm": 0.8552665114402771, "learning_rate": 9.999598963145069e-05, "loss": 0.0654, "step": 16150 }, { "epoch": 10.29955385595921, "grad_norm": 0.7540515065193176, "learning_rate": 9.999591952239371e-05, "loss": 0.0679, "step": 16160 }, { "epoch": 10.305927342256215, "grad_norm": 1.0121618509292603, "learning_rate": 9.99958488058391e-05, "loss": 0.0522, "step": 16170 }, { "epoch": 10.31230082855322, "grad_norm": 0.602270245552063, "learning_rate": 9.999577748178773e-05, "loss": 0.0565, "step": 16180 }, { "epoch": 10.318674314850224, "grad_norm": 0.7276262044906616, "learning_rate": 9.999570555024043e-05, "loss": 0.0559, "step": 16190 }, { "epoch": 10.325047801147228, "grad_norm": 0.6497871279716492, "learning_rate": 9.999563301119811e-05, "loss": 0.0641, "step": 16200 }, { "epoch": 10.331421287444233, "grad_norm": 0.5377486348152161, "learning_rate": 9.999555986466166e-05, "loss": 0.0569, "step": 16210 }, { "epoch": 10.337794773741237, "grad_norm": 0.6051101684570312, "learning_rate": 9.999548611063197e-05, "loss": 0.0524, "step": 16220 }, { "epoch": 10.344168260038241, "grad_norm": 0.8002655506134033, "learning_rate": 9.99954117491099e-05, "loss": 0.0597, "step": 16230 }, { "epoch": 10.350541746335246, "grad_norm": 0.507246732711792, "learning_rate": 9.999533678009639e-05, "loss": 0.0581, "step": 16240 }, { "epoch": 10.35691523263225, "grad_norm": 0.48768991231918335, "learning_rate": 9.99952612035923e-05, "loss": 0.0688, "step": 16250 }, { "epoch": 10.363288718929255, "grad_norm": 0.795545220375061, "learning_rate": 9.999518501959861e-05, "loss": 0.0643, "step": 16260 }, { "epoch": 10.369662205226259, "grad_norm": 0.6196532845497131, "learning_rate": 9.999510822811621e-05, "loss": 0.0582, "step": 16270 }, { "epoch": 10.376035691523263, "grad_norm": 0.6797712445259094, "learning_rate": 9.999503082914605e-05, "loss": 0.0608, "step": 16280 }, { "epoch": 10.382409177820268, "grad_norm": 0.9830362200737, "learning_rate": 9.999495282268905e-05, "loss": 0.065, "step": 16290 }, { "epoch": 10.388782664117272, "grad_norm": 0.676116943359375, "learning_rate": 9.999487420874618e-05, "loss": 0.0498, "step": 16300 }, { "epoch": 10.395156150414277, "grad_norm": 0.795210599899292, "learning_rate": 9.999479498731839e-05, "loss": 0.0505, "step": 16310 }, { "epoch": 10.401529636711281, "grad_norm": 0.7064853310585022, "learning_rate": 9.999471515840662e-05, "loss": 0.0558, "step": 16320 }, { "epoch": 10.407903123008285, "grad_norm": 0.5844698548316956, "learning_rate": 9.999463472201187e-05, "loss": 0.0658, "step": 16330 }, { "epoch": 10.41427660930529, "grad_norm": 0.5698778033256531, "learning_rate": 9.999455367813511e-05, "loss": 0.0532, "step": 16340 }, { "epoch": 10.420650095602294, "grad_norm": 0.6656918525695801, "learning_rate": 9.999447202677732e-05, "loss": 0.0536, "step": 16350 }, { "epoch": 10.427023581899299, "grad_norm": 0.5382378101348877, "learning_rate": 9.999438976793947e-05, "loss": 0.0574, "step": 16360 }, { "epoch": 10.433397068196303, "grad_norm": 0.8292346000671387, "learning_rate": 9.999430690162261e-05, "loss": 0.0612, "step": 16370 }, { "epoch": 10.439770554493307, "grad_norm": 0.7137431502342224, "learning_rate": 9.99942234278277e-05, "loss": 0.0666, "step": 16380 }, { "epoch": 10.446144040790312, "grad_norm": 0.7976674437522888, "learning_rate": 9.999413934655578e-05, "loss": 0.0576, "step": 16390 }, { "epoch": 10.452517527087316, "grad_norm": 0.7755173444747925, "learning_rate": 9.999405465780787e-05, "loss": 0.0648, "step": 16400 }, { "epoch": 10.45889101338432, "grad_norm": 0.519244372844696, "learning_rate": 9.999396936158498e-05, "loss": 0.0504, "step": 16410 }, { "epoch": 10.465264499681325, "grad_norm": 0.46889597177505493, "learning_rate": 9.999388345788819e-05, "loss": 0.0606, "step": 16420 }, { "epoch": 10.47163798597833, "grad_norm": 0.5492410659790039, "learning_rate": 9.999379694671847e-05, "loss": 0.0579, "step": 16430 }, { "epoch": 10.478011472275334, "grad_norm": 0.5425655245780945, "learning_rate": 9.999370982807695e-05, "loss": 0.0572, "step": 16440 }, { "epoch": 10.484384958572338, "grad_norm": 0.5655606985092163, "learning_rate": 9.999362210196463e-05, "loss": 0.0519, "step": 16450 }, { "epoch": 10.490758444869343, "grad_norm": 0.475955992937088, "learning_rate": 9.99935337683826e-05, "loss": 0.0534, "step": 16460 }, { "epoch": 10.497131931166347, "grad_norm": 0.462833970785141, "learning_rate": 9.999344482733197e-05, "loss": 0.0565, "step": 16470 }, { "epoch": 10.503505417463352, "grad_norm": 0.5556862354278564, "learning_rate": 9.999335527881375e-05, "loss": 0.049, "step": 16480 }, { "epoch": 10.509878903760358, "grad_norm": 0.8238747715950012, "learning_rate": 9.999326512282908e-05, "loss": 0.0707, "step": 16490 }, { "epoch": 10.516252390057362, "grad_norm": 0.9749704599380493, "learning_rate": 9.9993174359379e-05, "loss": 0.0549, "step": 16500 }, { "epoch": 10.522625876354367, "grad_norm": 0.5116440653800964, "learning_rate": 9.999308298846468e-05, "loss": 0.0558, "step": 16510 }, { "epoch": 10.528999362651371, "grad_norm": 0.5135734677314758, "learning_rate": 9.999299101008721e-05, "loss": 0.0598, "step": 16520 }, { "epoch": 10.535372848948375, "grad_norm": 0.5870323777198792, "learning_rate": 9.999289842424768e-05, "loss": 0.0581, "step": 16530 }, { "epoch": 10.54174633524538, "grad_norm": 0.6875941157341003, "learning_rate": 9.999280523094724e-05, "loss": 0.0499, "step": 16540 }, { "epoch": 10.548119821542384, "grad_norm": 0.4197682738304138, "learning_rate": 9.999271143018702e-05, "loss": 0.0662, "step": 16550 }, { "epoch": 10.554493307839389, "grad_norm": 0.840517520904541, "learning_rate": 9.999261702196813e-05, "loss": 0.0674, "step": 16560 }, { "epoch": 10.560866794136393, "grad_norm": 0.6064279675483704, "learning_rate": 9.999252200629176e-05, "loss": 0.0608, "step": 16570 }, { "epoch": 10.567240280433397, "grad_norm": 0.6202387809753418, "learning_rate": 9.999242638315906e-05, "loss": 0.0605, "step": 16580 }, { "epoch": 10.573613766730402, "grad_norm": 0.674312174320221, "learning_rate": 9.999233015257116e-05, "loss": 0.0517, "step": 16590 }, { "epoch": 10.579987253027406, "grad_norm": 0.3645886778831482, "learning_rate": 9.999223331452926e-05, "loss": 0.0444, "step": 16600 }, { "epoch": 10.58636073932441, "grad_norm": 1.1831674575805664, "learning_rate": 9.999213586903452e-05, "loss": 0.0578, "step": 16610 }, { "epoch": 10.592734225621415, "grad_norm": 0.5315924882888794, "learning_rate": 9.999203781608813e-05, "loss": 0.0432, "step": 16620 }, { "epoch": 10.59910771191842, "grad_norm": 0.5224555134773254, "learning_rate": 9.999193915569127e-05, "loss": 0.0422, "step": 16630 }, { "epoch": 10.605481198215424, "grad_norm": 0.3949601650238037, "learning_rate": 9.999183988784516e-05, "loss": 0.0509, "step": 16640 }, { "epoch": 10.611854684512428, "grad_norm": 0.6192469596862793, "learning_rate": 9.999174001255099e-05, "loss": 0.0613, "step": 16650 }, { "epoch": 10.618228170809433, "grad_norm": 0.557607114315033, "learning_rate": 9.999163952981e-05, "loss": 0.0606, "step": 16660 }, { "epoch": 10.624601657106437, "grad_norm": 0.9131451845169067, "learning_rate": 9.999153843962336e-05, "loss": 0.0428, "step": 16670 }, { "epoch": 10.630975143403441, "grad_norm": 0.3558861315250397, "learning_rate": 9.999143674199234e-05, "loss": 0.0611, "step": 16680 }, { "epoch": 10.637348629700446, "grad_norm": 0.7525888681411743, "learning_rate": 9.999133443691817e-05, "loss": 0.0635, "step": 16690 }, { "epoch": 10.64372211599745, "grad_norm": 0.6017816066741943, "learning_rate": 9.999123152440208e-05, "loss": 0.0566, "step": 16700 }, { "epoch": 10.650095602294455, "grad_norm": 0.5555442571640015, "learning_rate": 9.999112800444533e-05, "loss": 0.057, "step": 16710 }, { "epoch": 10.656469088591459, "grad_norm": 0.8302058577537537, "learning_rate": 9.999102387704917e-05, "loss": 0.059, "step": 16720 }, { "epoch": 10.662842574888463, "grad_norm": 0.7159201502799988, "learning_rate": 9.999091914221486e-05, "loss": 0.0647, "step": 16730 }, { "epoch": 10.669216061185468, "grad_norm": 0.5368846654891968, "learning_rate": 9.99908137999437e-05, "loss": 0.0516, "step": 16740 }, { "epoch": 10.675589547482472, "grad_norm": 0.4882710576057434, "learning_rate": 9.999070785023696e-05, "loss": 0.0496, "step": 16750 }, { "epoch": 10.681963033779477, "grad_norm": 0.8025274276733398, "learning_rate": 9.99906012930959e-05, "loss": 0.0626, "step": 16760 }, { "epoch": 10.688336520076481, "grad_norm": 0.7932184338569641, "learning_rate": 9.999049412852186e-05, "loss": 0.0549, "step": 16770 }, { "epoch": 10.694710006373487, "grad_norm": 0.7863204479217529, "learning_rate": 9.99903863565161e-05, "loss": 0.0638, "step": 16780 }, { "epoch": 10.701083492670492, "grad_norm": 0.6114450693130493, "learning_rate": 9.999027797707994e-05, "loss": 0.059, "step": 16790 }, { "epoch": 10.707456978967496, "grad_norm": 0.35417380928993225, "learning_rate": 9.999016899021472e-05, "loss": 0.0677, "step": 16800 }, { "epoch": 10.7138304652645, "grad_norm": 0.6127309799194336, "learning_rate": 9.999005939592174e-05, "loss": 0.0523, "step": 16810 }, { "epoch": 10.720203951561505, "grad_norm": 0.7677340507507324, "learning_rate": 9.998994919420234e-05, "loss": 0.0572, "step": 16820 }, { "epoch": 10.72657743785851, "grad_norm": 0.41578641533851624, "learning_rate": 9.998983838505786e-05, "loss": 0.0533, "step": 16830 }, { "epoch": 10.732950924155514, "grad_norm": 0.7479081153869629, "learning_rate": 9.998972696848965e-05, "loss": 0.0488, "step": 16840 }, { "epoch": 10.739324410452518, "grad_norm": 0.5703463554382324, "learning_rate": 9.998961494449906e-05, "loss": 0.0424, "step": 16850 }, { "epoch": 10.745697896749522, "grad_norm": 0.7256620526313782, "learning_rate": 9.998950231308744e-05, "loss": 0.0473, "step": 16860 }, { "epoch": 10.752071383046527, "grad_norm": 0.38508841395378113, "learning_rate": 9.998938907425616e-05, "loss": 0.0587, "step": 16870 }, { "epoch": 10.758444869343531, "grad_norm": 0.39906227588653564, "learning_rate": 9.998927522800663e-05, "loss": 0.0418, "step": 16880 }, { "epoch": 10.764818355640536, "grad_norm": 0.42261654138565063, "learning_rate": 9.998916077434019e-05, "loss": 0.0505, "step": 16890 }, { "epoch": 10.77119184193754, "grad_norm": 0.8031429648399353, "learning_rate": 9.998904571325827e-05, "loss": 0.0528, "step": 16900 }, { "epoch": 10.777565328234544, "grad_norm": 0.29207149147987366, "learning_rate": 9.998893004476224e-05, "loss": 0.0494, "step": 16910 }, { "epoch": 10.783938814531549, "grad_norm": 0.6371474862098694, "learning_rate": 9.99888137688535e-05, "loss": 0.0546, "step": 16920 }, { "epoch": 10.790312300828553, "grad_norm": 0.7001533508300781, "learning_rate": 9.998869688553348e-05, "loss": 0.052, "step": 16930 }, { "epoch": 10.796685787125558, "grad_norm": 0.5313620567321777, "learning_rate": 9.998857939480361e-05, "loss": 0.0553, "step": 16940 }, { "epoch": 10.803059273422562, "grad_norm": 0.470103919506073, "learning_rate": 9.998846129666529e-05, "loss": 0.0477, "step": 16950 }, { "epoch": 10.809432759719567, "grad_norm": 0.8891826868057251, "learning_rate": 9.998834259111996e-05, "loss": 0.0509, "step": 16960 }, { "epoch": 10.815806246016571, "grad_norm": 0.6524693369865417, "learning_rate": 9.998822327816908e-05, "loss": 0.0565, "step": 16970 }, { "epoch": 10.822179732313575, "grad_norm": 0.7200140357017517, "learning_rate": 9.99881033578141e-05, "loss": 0.0557, "step": 16980 }, { "epoch": 10.82855321861058, "grad_norm": 0.6414451003074646, "learning_rate": 9.998798283005647e-05, "loss": 0.061, "step": 16990 }, { "epoch": 10.834926704907584, "grad_norm": 0.398269385099411, "learning_rate": 9.998786169489762e-05, "loss": 0.0512, "step": 17000 }, { "epoch": 10.841300191204589, "grad_norm": 0.6985471844673157, "learning_rate": 9.998773995233908e-05, "loss": 0.0615, "step": 17010 }, { "epoch": 10.847673677501593, "grad_norm": 0.5928990840911865, "learning_rate": 9.99876176023823e-05, "loss": 0.0631, "step": 17020 }, { "epoch": 10.854047163798597, "grad_norm": 0.5422001481056213, "learning_rate": 9.998749464502877e-05, "loss": 0.0546, "step": 17030 }, { "epoch": 10.860420650095602, "grad_norm": 0.5168808102607727, "learning_rate": 9.998737108027998e-05, "loss": 0.0441, "step": 17040 }, { "epoch": 10.866794136392606, "grad_norm": 0.4506094753742218, "learning_rate": 9.998724690813744e-05, "loss": 0.0393, "step": 17050 }, { "epoch": 10.87316762268961, "grad_norm": 0.7235383987426758, "learning_rate": 9.998712212860264e-05, "loss": 0.0453, "step": 17060 }, { "epoch": 10.879541108986615, "grad_norm": 0.7417569756507874, "learning_rate": 9.998699674167712e-05, "loss": 0.0526, "step": 17070 }, { "epoch": 10.88591459528362, "grad_norm": 0.7921003103256226, "learning_rate": 9.998687074736241e-05, "loss": 0.0502, "step": 17080 }, { "epoch": 10.892288081580624, "grad_norm": 0.58517986536026, "learning_rate": 9.998674414566001e-05, "loss": 0.0582, "step": 17090 }, { "epoch": 10.898661567877628, "grad_norm": 0.48357248306274414, "learning_rate": 9.998661693657148e-05, "loss": 0.0438, "step": 17100 }, { "epoch": 10.905035054174633, "grad_norm": 0.5111726522445679, "learning_rate": 9.998648912009835e-05, "loss": 0.0503, "step": 17110 }, { "epoch": 10.911408540471639, "grad_norm": 0.3606181740760803, "learning_rate": 9.998636069624219e-05, "loss": 0.0579, "step": 17120 }, { "epoch": 10.917782026768643, "grad_norm": 0.8223302960395813, "learning_rate": 9.998623166500454e-05, "loss": 0.0471, "step": 17130 }, { "epoch": 10.924155513065648, "grad_norm": 0.47486552596092224, "learning_rate": 9.998610202638699e-05, "loss": 0.0386, "step": 17140 }, { "epoch": 10.930528999362652, "grad_norm": 0.7364180684089661, "learning_rate": 9.998597178039111e-05, "loss": 0.0638, "step": 17150 }, { "epoch": 10.936902485659656, "grad_norm": 0.6910980939865112, "learning_rate": 9.998584092701848e-05, "loss": 0.0577, "step": 17160 }, { "epoch": 10.94327597195666, "grad_norm": 0.40659859776496887, "learning_rate": 9.998570946627067e-05, "loss": 0.0568, "step": 17170 }, { "epoch": 10.949649458253665, "grad_norm": 0.4734037220478058, "learning_rate": 9.998557739814931e-05, "loss": 0.0587, "step": 17180 }, { "epoch": 10.95602294455067, "grad_norm": 0.626246452331543, "learning_rate": 9.998544472265598e-05, "loss": 0.0556, "step": 17190 }, { "epoch": 10.962396430847674, "grad_norm": 1.0859874486923218, "learning_rate": 9.998531143979232e-05, "loss": 0.0494, "step": 17200 }, { "epoch": 10.968769917144678, "grad_norm": 0.5683014392852783, "learning_rate": 9.998517754955992e-05, "loss": 0.0611, "step": 17210 }, { "epoch": 10.975143403441683, "grad_norm": 0.6202378273010254, "learning_rate": 9.998504305196042e-05, "loss": 0.0566, "step": 17220 }, { "epoch": 10.981516889738687, "grad_norm": 0.5997171998023987, "learning_rate": 9.998490794699546e-05, "loss": 0.0432, "step": 17230 }, { "epoch": 10.987890376035692, "grad_norm": 0.5708463191986084, "learning_rate": 9.998477223466667e-05, "loss": 0.045, "step": 17240 }, { "epoch": 10.994263862332696, "grad_norm": 0.5977245569229126, "learning_rate": 9.998463591497569e-05, "loss": 0.0844, "step": 17250 }, { "epoch": 11.0006373486297, "grad_norm": 0.7333068251609802, "learning_rate": 9.998449898792422e-05, "loss": 0.0603, "step": 17260 }, { "epoch": 11.007010834926705, "grad_norm": 0.584259569644928, "learning_rate": 9.998436145351386e-05, "loss": 0.0456, "step": 17270 }, { "epoch": 11.01338432122371, "grad_norm": 0.43404531478881836, "learning_rate": 9.998422331174634e-05, "loss": 0.0522, "step": 17280 }, { "epoch": 11.019757807520714, "grad_norm": 1.031726360321045, "learning_rate": 9.99840845626233e-05, "loss": 0.0598, "step": 17290 }, { "epoch": 11.026131293817718, "grad_norm": 0.798851728439331, "learning_rate": 9.998394520614645e-05, "loss": 0.052, "step": 17300 }, { "epoch": 11.032504780114722, "grad_norm": 0.46099844574928284, "learning_rate": 9.998380524231747e-05, "loss": 0.0574, "step": 17310 }, { "epoch": 11.038878266411727, "grad_norm": 0.6638126373291016, "learning_rate": 9.998366467113807e-05, "loss": 0.0344, "step": 17320 }, { "epoch": 11.045251752708731, "grad_norm": 0.6141223907470703, "learning_rate": 9.998352349260993e-05, "loss": 0.0466, "step": 17330 }, { "epoch": 11.051625239005736, "grad_norm": 0.7332136034965515, "learning_rate": 9.99833817067348e-05, "loss": 0.0467, "step": 17340 }, { "epoch": 11.05799872530274, "grad_norm": 0.565891683101654, "learning_rate": 9.998323931351441e-05, "loss": 0.0442, "step": 17350 }, { "epoch": 11.064372211599744, "grad_norm": 0.6422147154808044, "learning_rate": 9.998309631295045e-05, "loss": 0.0495, "step": 17360 }, { "epoch": 11.070745697896749, "grad_norm": 0.6818075180053711, "learning_rate": 9.998295270504468e-05, "loss": 0.0532, "step": 17370 }, { "epoch": 11.077119184193753, "grad_norm": 0.7112347483634949, "learning_rate": 9.998280848979884e-05, "loss": 0.0665, "step": 17380 }, { "epoch": 11.083492670490758, "grad_norm": 0.8333026766777039, "learning_rate": 9.99826636672147e-05, "loss": 0.0528, "step": 17390 }, { "epoch": 11.089866156787762, "grad_norm": 0.6337835192680359, "learning_rate": 9.998251823729399e-05, "loss": 0.0444, "step": 17400 }, { "epoch": 11.096239643084767, "grad_norm": 0.5257356762886047, "learning_rate": 9.998237220003848e-05, "loss": 0.0559, "step": 17410 }, { "epoch": 11.102613129381773, "grad_norm": 0.6054905652999878, "learning_rate": 9.998222555544999e-05, "loss": 0.0477, "step": 17420 }, { "epoch": 11.108986615678777, "grad_norm": 0.47715023159980774, "learning_rate": 9.998207830353024e-05, "loss": 0.0562, "step": 17430 }, { "epoch": 11.115360101975782, "grad_norm": 1.4385261535644531, "learning_rate": 9.998193044428106e-05, "loss": 0.0393, "step": 17440 }, { "epoch": 11.121733588272786, "grad_norm": 0.53253173828125, "learning_rate": 9.998178197770422e-05, "loss": 0.0644, "step": 17450 }, { "epoch": 11.12810707456979, "grad_norm": 0.7377299070358276, "learning_rate": 9.998163290380156e-05, "loss": 0.0491, "step": 17460 }, { "epoch": 11.134480560866795, "grad_norm": 0.31071048974990845, "learning_rate": 9.998148322257486e-05, "loss": 0.0494, "step": 17470 }, { "epoch": 11.1408540471638, "grad_norm": 0.428337424993515, "learning_rate": 9.998133293402595e-05, "loss": 0.0396, "step": 17480 }, { "epoch": 11.147227533460804, "grad_norm": 0.6270567178726196, "learning_rate": 9.998118203815665e-05, "loss": 0.0583, "step": 17490 }, { "epoch": 11.153601019757808, "grad_norm": 0.36213913559913635, "learning_rate": 9.998103053496882e-05, "loss": 0.043, "step": 17500 }, { "epoch": 11.159974506054812, "grad_norm": 0.4144083857536316, "learning_rate": 9.998087842446426e-05, "loss": 0.042, "step": 17510 }, { "epoch": 11.166347992351817, "grad_norm": 0.45778176188468933, "learning_rate": 9.998072570664483e-05, "loss": 0.0467, "step": 17520 }, { "epoch": 11.172721478648821, "grad_norm": 0.4047306776046753, "learning_rate": 9.99805723815124e-05, "loss": 0.0574, "step": 17530 }, { "epoch": 11.179094964945826, "grad_norm": 0.958371102809906, "learning_rate": 9.998041844906883e-05, "loss": 0.0617, "step": 17540 }, { "epoch": 11.18546845124283, "grad_norm": 0.7824062705039978, "learning_rate": 9.998026390931598e-05, "loss": 0.0434, "step": 17550 }, { "epoch": 11.191841937539834, "grad_norm": 0.48062580823898315, "learning_rate": 9.998010876225574e-05, "loss": 0.0562, "step": 17560 }, { "epoch": 11.198215423836839, "grad_norm": 0.6278024315834045, "learning_rate": 9.997995300789e-05, "loss": 0.0632, "step": 17570 }, { "epoch": 11.204588910133843, "grad_norm": 0.7428111433982849, "learning_rate": 9.997979664622062e-05, "loss": 0.0563, "step": 17580 }, { "epoch": 11.210962396430848, "grad_norm": 0.9911885857582092, "learning_rate": 9.997963967724952e-05, "loss": 0.0588, "step": 17590 }, { "epoch": 11.217335882727852, "grad_norm": 0.631261944770813, "learning_rate": 9.997948210097863e-05, "loss": 0.0539, "step": 17600 }, { "epoch": 11.223709369024856, "grad_norm": 0.8290361166000366, "learning_rate": 9.997932391740983e-05, "loss": 0.0761, "step": 17610 }, { "epoch": 11.23008285532186, "grad_norm": 0.7124612331390381, "learning_rate": 9.997916512654507e-05, "loss": 0.0587, "step": 17620 }, { "epoch": 11.236456341618865, "grad_norm": 0.30079036951065063, "learning_rate": 9.997900572838626e-05, "loss": 0.0404, "step": 17630 }, { "epoch": 11.24282982791587, "grad_norm": 0.7114712595939636, "learning_rate": 9.997884572293534e-05, "loss": 0.06, "step": 17640 }, { "epoch": 11.249203314212874, "grad_norm": 0.6289411783218384, "learning_rate": 9.997868511019425e-05, "loss": 0.0498, "step": 17650 }, { "epoch": 11.255576800509878, "grad_norm": 1.10546875, "learning_rate": 9.997852389016496e-05, "loss": 0.0443, "step": 17660 }, { "epoch": 11.261950286806883, "grad_norm": 0.8365715742111206, "learning_rate": 9.997836206284941e-05, "loss": 0.0646, "step": 17670 }, { "epoch": 11.268323773103887, "grad_norm": 0.629873514175415, "learning_rate": 9.997819962824957e-05, "loss": 0.061, "step": 17680 }, { "epoch": 11.274697259400892, "grad_norm": 0.8870893120765686, "learning_rate": 9.997803658636742e-05, "loss": 0.0548, "step": 17690 }, { "epoch": 11.281070745697896, "grad_norm": 0.4535408318042755, "learning_rate": 9.997787293720495e-05, "loss": 0.0551, "step": 17700 }, { "epoch": 11.2874442319949, "grad_norm": 1.0586836338043213, "learning_rate": 9.997770868076413e-05, "loss": 0.0565, "step": 17710 }, { "epoch": 11.293817718291905, "grad_norm": 0.4891342222690582, "learning_rate": 9.997754381704696e-05, "loss": 0.0463, "step": 17720 }, { "epoch": 11.30019120458891, "grad_norm": 0.3861699402332306, "learning_rate": 9.997737834605545e-05, "loss": 0.0494, "step": 17730 }, { "epoch": 11.306564690885915, "grad_norm": 0.7491887211799622, "learning_rate": 9.99772122677916e-05, "loss": 0.0532, "step": 17740 }, { "epoch": 11.31293817718292, "grad_norm": 0.3956933319568634, "learning_rate": 9.997704558225743e-05, "loss": 0.0579, "step": 17750 }, { "epoch": 11.319311663479924, "grad_norm": 0.5724512934684753, "learning_rate": 9.997687828945499e-05, "loss": 0.0657, "step": 17760 }, { "epoch": 11.325685149776929, "grad_norm": 0.6473221182823181, "learning_rate": 9.997671038938628e-05, "loss": 0.0658, "step": 17770 }, { "epoch": 11.332058636073933, "grad_norm": 0.7001258134841919, "learning_rate": 9.997654188205334e-05, "loss": 0.0496, "step": 17780 }, { "epoch": 11.338432122370937, "grad_norm": 0.560114860534668, "learning_rate": 9.997637276745826e-05, "loss": 0.0615, "step": 17790 }, { "epoch": 11.344805608667942, "grad_norm": 0.6450788378715515, "learning_rate": 9.997620304560304e-05, "loss": 0.0521, "step": 17800 }, { "epoch": 11.351179094964946, "grad_norm": 0.4742850959300995, "learning_rate": 9.997603271648979e-05, "loss": 0.0454, "step": 17810 }, { "epoch": 11.35755258126195, "grad_norm": 1.1459397077560425, "learning_rate": 9.997586178012052e-05, "loss": 0.0524, "step": 17820 }, { "epoch": 11.363926067558955, "grad_norm": 0.4577571749687195, "learning_rate": 9.997569023649737e-05, "loss": 0.0503, "step": 17830 }, { "epoch": 11.37029955385596, "grad_norm": 0.6457903981208801, "learning_rate": 9.997551808562241e-05, "loss": 0.0497, "step": 17840 }, { "epoch": 11.376673040152964, "grad_norm": 0.9518571496009827, "learning_rate": 9.997534532749771e-05, "loss": 0.058, "step": 17850 }, { "epoch": 11.383046526449968, "grad_norm": 0.5197674632072449, "learning_rate": 9.997517196212538e-05, "loss": 0.0513, "step": 17860 }, { "epoch": 11.389420012746973, "grad_norm": 0.6322747468948364, "learning_rate": 9.997499798950751e-05, "loss": 0.0563, "step": 17870 }, { "epoch": 11.395793499043977, "grad_norm": 0.6855553984642029, "learning_rate": 9.997482340964625e-05, "loss": 0.0491, "step": 17880 }, { "epoch": 11.402166985340981, "grad_norm": 0.37808260321617126, "learning_rate": 9.99746482225437e-05, "loss": 0.0497, "step": 17890 }, { "epoch": 11.408540471637986, "grad_norm": 0.7514583468437195, "learning_rate": 9.997447242820199e-05, "loss": 0.0503, "step": 17900 }, { "epoch": 11.41491395793499, "grad_norm": 0.5832131505012512, "learning_rate": 9.997429602662325e-05, "loss": 0.0438, "step": 17910 }, { "epoch": 11.421287444231995, "grad_norm": 0.5786134004592896, "learning_rate": 9.997411901780963e-05, "loss": 0.0539, "step": 17920 }, { "epoch": 11.427660930529, "grad_norm": 0.474710613489151, "learning_rate": 9.997394140176328e-05, "loss": 0.0454, "step": 17930 }, { "epoch": 11.434034416826004, "grad_norm": 0.8023557662963867, "learning_rate": 9.997376317848636e-05, "loss": 0.0593, "step": 17940 }, { "epoch": 11.440407903123008, "grad_norm": 1.0486323833465576, "learning_rate": 9.997358434798103e-05, "loss": 0.0554, "step": 17950 }, { "epoch": 11.446781389420012, "grad_norm": 0.5281891822814941, "learning_rate": 9.997340491024948e-05, "loss": 0.0503, "step": 17960 }, { "epoch": 11.453154875717017, "grad_norm": 0.6257161498069763, "learning_rate": 9.997322486529387e-05, "loss": 0.0533, "step": 17970 }, { "epoch": 11.459528362014021, "grad_norm": 0.7398996949195862, "learning_rate": 9.997304421311639e-05, "loss": 0.0537, "step": 17980 }, { "epoch": 11.465901848311026, "grad_norm": 0.47160854935646057, "learning_rate": 9.997286295371926e-05, "loss": 0.05, "step": 17990 }, { "epoch": 11.47227533460803, "grad_norm": 0.5948900580406189, "learning_rate": 9.997268108710464e-05, "loss": 0.0575, "step": 18000 }, { "epoch": 11.478648820905034, "grad_norm": 0.5790464878082275, "learning_rate": 9.997249861327478e-05, "loss": 0.044, "step": 18010 }, { "epoch": 11.485022307202039, "grad_norm": 0.5215743184089661, "learning_rate": 9.997231553223186e-05, "loss": 0.0634, "step": 18020 }, { "epoch": 11.491395793499043, "grad_norm": 0.7115923762321472, "learning_rate": 9.997213184397814e-05, "loss": 0.0559, "step": 18030 }, { "epoch": 11.49776927979605, "grad_norm": 0.6699000000953674, "learning_rate": 9.997194754851584e-05, "loss": 0.0547, "step": 18040 }, { "epoch": 11.504142766093054, "grad_norm": 0.7512786388397217, "learning_rate": 9.997176264584717e-05, "loss": 0.0627, "step": 18050 }, { "epoch": 11.510516252390058, "grad_norm": 0.5977685451507568, "learning_rate": 9.997157713597444e-05, "loss": 0.0505, "step": 18060 }, { "epoch": 11.516889738687063, "grad_norm": 1.0020976066589355, "learning_rate": 9.997139101889986e-05, "loss": 0.0602, "step": 18070 }, { "epoch": 11.523263224984067, "grad_norm": 0.6246644258499146, "learning_rate": 9.997120429462567e-05, "loss": 0.0536, "step": 18080 }, { "epoch": 11.529636711281071, "grad_norm": 0.7035977840423584, "learning_rate": 9.997101696315418e-05, "loss": 0.0597, "step": 18090 }, { "epoch": 11.536010197578076, "grad_norm": 0.7818805575370789, "learning_rate": 9.997082902448767e-05, "loss": 0.0601, "step": 18100 }, { "epoch": 11.54238368387508, "grad_norm": 0.9022552967071533, "learning_rate": 9.997064047862839e-05, "loss": 0.0569, "step": 18110 }, { "epoch": 11.548757170172085, "grad_norm": 0.6126748919487, "learning_rate": 9.997045132557866e-05, "loss": 0.056, "step": 18120 }, { "epoch": 11.555130656469089, "grad_norm": 0.6667356491088867, "learning_rate": 9.997026156534075e-05, "loss": 0.0488, "step": 18130 }, { "epoch": 11.561504142766093, "grad_norm": 0.4176846444606781, "learning_rate": 9.9970071197917e-05, "loss": 0.045, "step": 18140 }, { "epoch": 11.567877629063098, "grad_norm": 0.6282650232315063, "learning_rate": 9.996988022330969e-05, "loss": 0.0547, "step": 18150 }, { "epoch": 11.574251115360102, "grad_norm": 0.46769824624061584, "learning_rate": 9.996968864152116e-05, "loss": 0.0562, "step": 18160 }, { "epoch": 11.580624601657107, "grad_norm": 0.6001009941101074, "learning_rate": 9.996949645255374e-05, "loss": 0.053, "step": 18170 }, { "epoch": 11.586998087954111, "grad_norm": 0.4135037362575531, "learning_rate": 9.996930365640975e-05, "loss": 0.0403, "step": 18180 }, { "epoch": 11.593371574251115, "grad_norm": 0.563538134098053, "learning_rate": 9.996911025309155e-05, "loss": 0.0516, "step": 18190 }, { "epoch": 11.59974506054812, "grad_norm": 0.6317570209503174, "learning_rate": 9.996891624260147e-05, "loss": 0.0551, "step": 18200 }, { "epoch": 11.606118546845124, "grad_norm": 0.591675341129303, "learning_rate": 9.996872162494189e-05, "loss": 0.0524, "step": 18210 }, { "epoch": 11.612492033142129, "grad_norm": 0.8648858070373535, "learning_rate": 9.996852640011516e-05, "loss": 0.0417, "step": 18220 }, { "epoch": 11.618865519439133, "grad_norm": 0.63719242811203, "learning_rate": 9.996833056812367e-05, "loss": 0.0472, "step": 18230 }, { "epoch": 11.625239005736137, "grad_norm": 0.6004825234413147, "learning_rate": 9.996813412896977e-05, "loss": 0.0563, "step": 18240 }, { "epoch": 11.631612492033142, "grad_norm": 1.0001980066299438, "learning_rate": 9.996793708265586e-05, "loss": 0.0636, "step": 18250 }, { "epoch": 11.637985978330146, "grad_norm": 1.067087173461914, "learning_rate": 9.996773942918435e-05, "loss": 0.0608, "step": 18260 }, { "epoch": 11.64435946462715, "grad_norm": 0.847445547580719, "learning_rate": 9.996754116855763e-05, "loss": 0.0565, "step": 18270 }, { "epoch": 11.650732950924155, "grad_norm": 0.3594801425933838, "learning_rate": 9.99673423007781e-05, "loss": 0.0517, "step": 18280 }, { "epoch": 11.65710643722116, "grad_norm": 0.6227310299873352, "learning_rate": 9.996714282584818e-05, "loss": 0.0522, "step": 18290 }, { "epoch": 11.663479923518164, "grad_norm": 0.4727412760257721, "learning_rate": 9.99669427437703e-05, "loss": 0.0482, "step": 18300 }, { "epoch": 11.669853409815168, "grad_norm": 0.9138240814208984, "learning_rate": 9.996674205454688e-05, "loss": 0.0668, "step": 18310 }, { "epoch": 11.676226896112173, "grad_norm": 0.6075490117073059, "learning_rate": 9.996654075818039e-05, "loss": 0.0509, "step": 18320 }, { "epoch": 11.682600382409177, "grad_norm": 0.7808625102043152, "learning_rate": 9.996633885467324e-05, "loss": 0.0592, "step": 18330 }, { "epoch": 11.688973868706181, "grad_norm": 0.7400175929069519, "learning_rate": 9.996613634402789e-05, "loss": 0.0629, "step": 18340 }, { "epoch": 11.695347355003186, "grad_norm": 0.755127489566803, "learning_rate": 9.996593322624682e-05, "loss": 0.058, "step": 18350 }, { "epoch": 11.70172084130019, "grad_norm": 0.44330671429634094, "learning_rate": 9.996572950133248e-05, "loss": 0.0517, "step": 18360 }, { "epoch": 11.708094327597196, "grad_norm": 0.5726228356361389, "learning_rate": 9.996552516928735e-05, "loss": 0.0571, "step": 18370 }, { "epoch": 11.7144678138942, "grad_norm": 0.591275691986084, "learning_rate": 9.996532023011392e-05, "loss": 0.054, "step": 18380 }, { "epoch": 11.720841300191205, "grad_norm": 1.0331820249557495, "learning_rate": 9.996511468381467e-05, "loss": 0.0497, "step": 18390 }, { "epoch": 11.72721478648821, "grad_norm": 0.9469217658042908, "learning_rate": 9.996490853039211e-05, "loss": 0.0519, "step": 18400 }, { "epoch": 11.733588272785214, "grad_norm": 0.5066688060760498, "learning_rate": 9.996470176984873e-05, "loss": 0.0558, "step": 18410 }, { "epoch": 11.739961759082219, "grad_norm": 0.586126446723938, "learning_rate": 9.996449440218704e-05, "loss": 0.0772, "step": 18420 }, { "epoch": 11.746335245379223, "grad_norm": 0.5571174621582031, "learning_rate": 9.996428642740958e-05, "loss": 0.055, "step": 18430 }, { "epoch": 11.752708731676227, "grad_norm": 0.6182481050491333, "learning_rate": 9.996407784551887e-05, "loss": 0.0509, "step": 18440 }, { "epoch": 11.759082217973232, "grad_norm": 0.4466564357280731, "learning_rate": 9.996386865651744e-05, "loss": 0.0388, "step": 18450 }, { "epoch": 11.765455704270236, "grad_norm": 0.4253198206424713, "learning_rate": 9.996365886040782e-05, "loss": 0.0555, "step": 18460 }, { "epoch": 11.77182919056724, "grad_norm": 0.7634193897247314, "learning_rate": 9.996344845719257e-05, "loss": 0.0592, "step": 18470 }, { "epoch": 11.778202676864245, "grad_norm": 0.7888400554656982, "learning_rate": 9.996323744687426e-05, "loss": 0.061, "step": 18480 }, { "epoch": 11.78457616316125, "grad_norm": 0.29007822275161743, "learning_rate": 9.996302582945543e-05, "loss": 0.0378, "step": 18490 }, { "epoch": 11.790949649458254, "grad_norm": 0.3867831230163574, "learning_rate": 9.996281360493866e-05, "loss": 0.0545, "step": 18500 }, { "epoch": 11.797323135755258, "grad_norm": 0.8648092150688171, "learning_rate": 9.996260077332656e-05, "loss": 0.0539, "step": 18510 }, { "epoch": 11.803696622052263, "grad_norm": 0.5546586513519287, "learning_rate": 9.996238733462167e-05, "loss": 0.045, "step": 18520 }, { "epoch": 11.810070108349267, "grad_norm": 0.6702616810798645, "learning_rate": 9.99621732888266e-05, "loss": 0.0542, "step": 18530 }, { "epoch": 11.816443594646271, "grad_norm": 0.5179563760757446, "learning_rate": 9.996195863594396e-05, "loss": 0.0494, "step": 18540 }, { "epoch": 11.822817080943276, "grad_norm": 0.5806443095207214, "learning_rate": 9.996174337597632e-05, "loss": 0.0441, "step": 18550 }, { "epoch": 11.82919056724028, "grad_norm": 0.6398141384124756, "learning_rate": 9.996152750892635e-05, "loss": 0.0592, "step": 18560 }, { "epoch": 11.835564053537285, "grad_norm": 0.4812298119068146, "learning_rate": 9.996131103479667e-05, "loss": 0.0609, "step": 18570 }, { "epoch": 11.841937539834289, "grad_norm": 0.4223327338695526, "learning_rate": 9.996109395358986e-05, "loss": 0.0545, "step": 18580 }, { "epoch": 11.848311026131293, "grad_norm": 0.585468590259552, "learning_rate": 9.99608762653086e-05, "loss": 0.0539, "step": 18590 }, { "epoch": 11.854684512428298, "grad_norm": 0.5610795617103577, "learning_rate": 9.996065796995552e-05, "loss": 0.0601, "step": 18600 }, { "epoch": 11.861057998725302, "grad_norm": 1.0543278455734253, "learning_rate": 9.996043906753328e-05, "loss": 0.0565, "step": 18610 }, { "epoch": 11.867431485022307, "grad_norm": 0.8825480341911316, "learning_rate": 9.996021955804453e-05, "loss": 0.046, "step": 18620 }, { "epoch": 11.873804971319311, "grad_norm": 0.6645362973213196, "learning_rate": 9.995999944149192e-05, "loss": 0.0795, "step": 18630 }, { "epoch": 11.880178457616315, "grad_norm": 0.6627642512321472, "learning_rate": 9.995977871787818e-05, "loss": 0.0591, "step": 18640 }, { "epoch": 11.88655194391332, "grad_norm": 0.5781305432319641, "learning_rate": 9.995955738720595e-05, "loss": 0.0549, "step": 18650 }, { "epoch": 11.892925430210324, "grad_norm": 0.28119561076164246, "learning_rate": 9.995933544947792e-05, "loss": 0.0444, "step": 18660 }, { "epoch": 11.89929891650733, "grad_norm": 0.49158886075019836, "learning_rate": 9.995911290469679e-05, "loss": 0.0555, "step": 18670 }, { "epoch": 11.905672402804335, "grad_norm": 1.0449167490005493, "learning_rate": 9.995888975286529e-05, "loss": 0.0584, "step": 18680 }, { "epoch": 11.91204588910134, "grad_norm": 0.4692186117172241, "learning_rate": 9.99586659939861e-05, "loss": 0.0498, "step": 18690 }, { "epoch": 11.918419375398344, "grad_norm": 0.6601914167404175, "learning_rate": 9.995844162806195e-05, "loss": 0.0345, "step": 18700 }, { "epoch": 11.924792861695348, "grad_norm": 0.7166926264762878, "learning_rate": 9.995821665509558e-05, "loss": 0.0573, "step": 18710 }, { "epoch": 11.931166347992352, "grad_norm": 0.507392942905426, "learning_rate": 9.995799107508968e-05, "loss": 0.0398, "step": 18720 }, { "epoch": 11.937539834289357, "grad_norm": 0.569351315498352, "learning_rate": 9.995776488804702e-05, "loss": 0.0726, "step": 18730 }, { "epoch": 11.943913320586361, "grad_norm": 0.5389679074287415, "learning_rate": 9.995753809397037e-05, "loss": 0.0484, "step": 18740 }, { "epoch": 11.950286806883366, "grad_norm": 0.7490147948265076, "learning_rate": 9.995731069286246e-05, "loss": 0.0383, "step": 18750 }, { "epoch": 11.95666029318037, "grad_norm": 0.38610783219337463, "learning_rate": 9.995708268472606e-05, "loss": 0.0375, "step": 18760 }, { "epoch": 11.963033779477374, "grad_norm": 1.1472092866897583, "learning_rate": 9.995685406956392e-05, "loss": 0.0632, "step": 18770 }, { "epoch": 11.969407265774379, "grad_norm": 1.152759313583374, "learning_rate": 9.995662484737885e-05, "loss": 0.0506, "step": 18780 }, { "epoch": 11.975780752071383, "grad_norm": 0.6192075610160828, "learning_rate": 9.995639501817362e-05, "loss": 0.059, "step": 18790 }, { "epoch": 11.982154238368388, "grad_norm": 0.3730803430080414, "learning_rate": 9.995616458195103e-05, "loss": 0.0435, "step": 18800 }, { "epoch": 11.988527724665392, "grad_norm": 0.4230964481830597, "learning_rate": 9.995593353871386e-05, "loss": 0.0552, "step": 18810 }, { "epoch": 11.994901210962396, "grad_norm": 0.28632041811943054, "learning_rate": 9.995570188846495e-05, "loss": 0.0394, "step": 18820 }, { "epoch": 12.0012746972594, "grad_norm": 0.6016038060188293, "learning_rate": 9.995546963120708e-05, "loss": 0.0531, "step": 18830 }, { "epoch": 12.007648183556405, "grad_norm": 1.32315993309021, "learning_rate": 9.995523676694309e-05, "loss": 0.0474, "step": 18840 }, { "epoch": 12.01402166985341, "grad_norm": 0.4533834755420685, "learning_rate": 9.995500329567583e-05, "loss": 0.0464, "step": 18850 }, { "epoch": 12.020395156150414, "grad_norm": 0.4621172249317169, "learning_rate": 9.995476921740807e-05, "loss": 0.0544, "step": 18860 }, { "epoch": 12.026768642447419, "grad_norm": 0.5771495699882507, "learning_rate": 9.995453453214274e-05, "loss": 0.047, "step": 18870 }, { "epoch": 12.033142128744423, "grad_norm": 0.677911102771759, "learning_rate": 9.995429923988263e-05, "loss": 0.0484, "step": 18880 }, { "epoch": 12.039515615041427, "grad_norm": 0.4545127749443054, "learning_rate": 9.995406334063062e-05, "loss": 0.0438, "step": 18890 }, { "epoch": 12.045889101338432, "grad_norm": 0.4634072482585907, "learning_rate": 9.995382683438958e-05, "loss": 0.064, "step": 18900 }, { "epoch": 12.052262587635436, "grad_norm": 0.5201743245124817, "learning_rate": 9.995358972116238e-05, "loss": 0.0466, "step": 18910 }, { "epoch": 12.05863607393244, "grad_norm": 0.7461873888969421, "learning_rate": 9.99533520009519e-05, "loss": 0.0449, "step": 18920 }, { "epoch": 12.065009560229445, "grad_norm": 0.5392987728118896, "learning_rate": 9.995311367376103e-05, "loss": 0.0491, "step": 18930 }, { "epoch": 12.07138304652645, "grad_norm": 0.7738301753997803, "learning_rate": 9.995287473959267e-05, "loss": 0.0515, "step": 18940 }, { "epoch": 12.077756532823454, "grad_norm": 0.7333182096481323, "learning_rate": 9.995263519844971e-05, "loss": 0.0426, "step": 18950 }, { "epoch": 12.084130019120458, "grad_norm": 0.7453256845474243, "learning_rate": 9.995239505033508e-05, "loss": 0.0409, "step": 18960 }, { "epoch": 12.090503505417463, "grad_norm": 0.6855975389480591, "learning_rate": 9.995215429525169e-05, "loss": 0.0542, "step": 18970 }, { "epoch": 12.096876991714467, "grad_norm": 0.35131192207336426, "learning_rate": 9.995191293320244e-05, "loss": 0.0532, "step": 18980 }, { "epoch": 12.103250478011473, "grad_norm": 0.5471280813217163, "learning_rate": 9.99516709641903e-05, "loss": 0.0464, "step": 18990 }, { "epoch": 12.109623964308478, "grad_norm": 0.5725041627883911, "learning_rate": 9.99514283882182e-05, "loss": 0.0531, "step": 19000 }, { "epoch": 12.115997450605482, "grad_norm": 0.4529055953025818, "learning_rate": 9.995118520528907e-05, "loss": 0.0507, "step": 19010 }, { "epoch": 12.122370936902486, "grad_norm": 0.8099885582923889, "learning_rate": 9.995094141540588e-05, "loss": 0.0496, "step": 19020 }, { "epoch": 12.12874442319949, "grad_norm": 0.37661561369895935, "learning_rate": 9.995069701857161e-05, "loss": 0.0471, "step": 19030 }, { "epoch": 12.135117909496495, "grad_norm": 0.5382967591285706, "learning_rate": 9.995045201478918e-05, "loss": 0.0595, "step": 19040 }, { "epoch": 12.1414913957935, "grad_norm": 0.5374127626419067, "learning_rate": 9.995020640406162e-05, "loss": 0.0439, "step": 19050 }, { "epoch": 12.147864882090504, "grad_norm": 0.27735188603401184, "learning_rate": 9.994996018639189e-05, "loss": 0.0458, "step": 19060 }, { "epoch": 12.154238368387508, "grad_norm": 0.6636207699775696, "learning_rate": 9.994971336178298e-05, "loss": 0.0477, "step": 19070 }, { "epoch": 12.160611854684513, "grad_norm": 0.6746484041213989, "learning_rate": 9.994946593023789e-05, "loss": 0.0471, "step": 19080 }, { "epoch": 12.166985340981517, "grad_norm": 0.32199937105178833, "learning_rate": 9.994921789175962e-05, "loss": 0.0509, "step": 19090 }, { "epoch": 12.173358827278522, "grad_norm": 0.6574446558952332, "learning_rate": 9.99489692463512e-05, "loss": 0.0636, "step": 19100 }, { "epoch": 12.179732313575526, "grad_norm": 0.7455927133560181, "learning_rate": 9.994871999401565e-05, "loss": 0.0415, "step": 19110 }, { "epoch": 12.18610579987253, "grad_norm": 0.6799868941307068, "learning_rate": 9.9948470134756e-05, "loss": 0.0504, "step": 19120 }, { "epoch": 12.192479286169535, "grad_norm": 0.3111349940299988, "learning_rate": 9.994821966857527e-05, "loss": 0.0478, "step": 19130 }, { "epoch": 12.19885277246654, "grad_norm": 0.4255180358886719, "learning_rate": 9.99479685954765e-05, "loss": 0.0604, "step": 19140 }, { "epoch": 12.205226258763544, "grad_norm": 0.6373127698898315, "learning_rate": 9.994771691546277e-05, "loss": 0.0404, "step": 19150 }, { "epoch": 12.211599745060548, "grad_norm": 0.7585551142692566, "learning_rate": 9.99474646285371e-05, "loss": 0.0459, "step": 19160 }, { "epoch": 12.217973231357552, "grad_norm": 0.5794774889945984, "learning_rate": 9.994721173470259e-05, "loss": 0.0457, "step": 19170 }, { "epoch": 12.224346717654557, "grad_norm": 0.5366019606590271, "learning_rate": 9.994695823396229e-05, "loss": 0.0562, "step": 19180 }, { "epoch": 12.230720203951561, "grad_norm": 0.7718575596809387, "learning_rate": 9.99467041263193e-05, "loss": 0.0578, "step": 19190 }, { "epoch": 12.237093690248566, "grad_norm": 0.4321316182613373, "learning_rate": 9.99464494117767e-05, "loss": 0.0435, "step": 19200 }, { "epoch": 12.24346717654557, "grad_norm": 0.5822618007659912, "learning_rate": 9.994619409033759e-05, "loss": 0.0535, "step": 19210 }, { "epoch": 12.249840662842574, "grad_norm": 0.6593590378761292, "learning_rate": 9.994593816200504e-05, "loss": 0.042, "step": 19220 }, { "epoch": 12.256214149139579, "grad_norm": 0.5769529342651367, "learning_rate": 9.994568162678219e-05, "loss": 0.0542, "step": 19230 }, { "epoch": 12.262587635436583, "grad_norm": 0.42017069458961487, "learning_rate": 9.994542448467216e-05, "loss": 0.0502, "step": 19240 }, { "epoch": 12.268961121733588, "grad_norm": 0.8386192321777344, "learning_rate": 9.994516673567807e-05, "loss": 0.0521, "step": 19250 }, { "epoch": 12.275334608030592, "grad_norm": 0.4258641004562378, "learning_rate": 9.994490837980305e-05, "loss": 0.0323, "step": 19260 }, { "epoch": 12.281708094327596, "grad_norm": 0.5408238172531128, "learning_rate": 9.994464941705022e-05, "loss": 0.036, "step": 19270 }, { "epoch": 12.2880815806246, "grad_norm": 0.423775851726532, "learning_rate": 9.994438984742275e-05, "loss": 0.0405, "step": 19280 }, { "epoch": 12.294455066921605, "grad_norm": 0.4911399185657501, "learning_rate": 9.99441296709238e-05, "loss": 0.0466, "step": 19290 }, { "epoch": 12.300828553218611, "grad_norm": 0.5672997832298279, "learning_rate": 9.994386888755651e-05, "loss": 0.0489, "step": 19300 }, { "epoch": 12.307202039515616, "grad_norm": 0.6688255071640015, "learning_rate": 9.994360749732407e-05, "loss": 0.0418, "step": 19310 }, { "epoch": 12.31357552581262, "grad_norm": 0.8784617185592651, "learning_rate": 9.994334550022964e-05, "loss": 0.0504, "step": 19320 }, { "epoch": 12.319949012109625, "grad_norm": 0.6308597922325134, "learning_rate": 9.994308289627639e-05, "loss": 0.0626, "step": 19330 }, { "epoch": 12.326322498406629, "grad_norm": 0.5725502371788025, "learning_rate": 9.994281968546755e-05, "loss": 0.0591, "step": 19340 }, { "epoch": 12.332695984703633, "grad_norm": 0.6846663951873779, "learning_rate": 9.994255586780629e-05, "loss": 0.0693, "step": 19350 }, { "epoch": 12.339069471000638, "grad_norm": 0.6701339483261108, "learning_rate": 9.99422914432958e-05, "loss": 0.0443, "step": 19360 }, { "epoch": 12.345442957297642, "grad_norm": 0.618817925453186, "learning_rate": 9.994202641193935e-05, "loss": 0.045, "step": 19370 }, { "epoch": 12.351816443594647, "grad_norm": 0.5576841831207275, "learning_rate": 9.994176077374012e-05, "loss": 0.0372, "step": 19380 }, { "epoch": 12.358189929891651, "grad_norm": 0.8428722620010376, "learning_rate": 9.994149452870133e-05, "loss": 0.0697, "step": 19390 }, { "epoch": 12.364563416188656, "grad_norm": 0.930770218372345, "learning_rate": 9.994122767682623e-05, "loss": 0.0519, "step": 19400 }, { "epoch": 12.37093690248566, "grad_norm": 1.0066298246383667, "learning_rate": 9.994096021811808e-05, "loss": 0.0484, "step": 19410 }, { "epoch": 12.377310388782664, "grad_norm": 0.42919373512268066, "learning_rate": 9.994069215258008e-05, "loss": 0.0488, "step": 19420 }, { "epoch": 12.383683875079669, "grad_norm": 0.5092325806617737, "learning_rate": 9.994042348021556e-05, "loss": 0.0602, "step": 19430 }, { "epoch": 12.390057361376673, "grad_norm": 0.5449823141098022, "learning_rate": 9.994015420102771e-05, "loss": 0.0655, "step": 19440 }, { "epoch": 12.396430847673678, "grad_norm": 0.5374361872673035, "learning_rate": 9.993988431501983e-05, "loss": 0.0463, "step": 19450 }, { "epoch": 12.402804333970682, "grad_norm": 0.6612887382507324, "learning_rate": 9.993961382219523e-05, "loss": 0.0533, "step": 19460 }, { "epoch": 12.409177820267686, "grad_norm": 0.5763975381851196, "learning_rate": 9.993934272255715e-05, "loss": 0.0492, "step": 19470 }, { "epoch": 12.41555130656469, "grad_norm": 0.46687546372413635, "learning_rate": 9.993907101610893e-05, "loss": 0.0537, "step": 19480 }, { "epoch": 12.421924792861695, "grad_norm": 0.6873669028282166, "learning_rate": 9.993879870285383e-05, "loss": 0.0477, "step": 19490 }, { "epoch": 12.4282982791587, "grad_norm": 0.793414294719696, "learning_rate": 9.993852578279518e-05, "loss": 0.0648, "step": 19500 }, { "epoch": 12.434671765455704, "grad_norm": 0.38007280230522156, "learning_rate": 9.993825225593629e-05, "loss": 0.0558, "step": 19510 }, { "epoch": 12.441045251752708, "grad_norm": 0.39757072925567627, "learning_rate": 9.993797812228048e-05, "loss": 0.0463, "step": 19520 }, { "epoch": 12.447418738049713, "grad_norm": 0.5992366075515747, "learning_rate": 9.99377033818311e-05, "loss": 0.0492, "step": 19530 }, { "epoch": 12.453792224346717, "grad_norm": 0.6468036770820618, "learning_rate": 9.993742803459146e-05, "loss": 0.0485, "step": 19540 }, { "epoch": 12.460165710643722, "grad_norm": 0.8821713328361511, "learning_rate": 9.993715208056493e-05, "loss": 0.053, "step": 19550 }, { "epoch": 12.466539196940726, "grad_norm": 3.29703950881958, "learning_rate": 9.993687551975485e-05, "loss": 0.0622, "step": 19560 }, { "epoch": 12.47291268323773, "grad_norm": 0.6634793281555176, "learning_rate": 9.99365983521646e-05, "loss": 0.0607, "step": 19570 }, { "epoch": 12.479286169534735, "grad_norm": 0.628406822681427, "learning_rate": 9.993632057779752e-05, "loss": 0.0526, "step": 19580 }, { "epoch": 12.48565965583174, "grad_norm": 0.41636964678764343, "learning_rate": 9.993604219665698e-05, "loss": 0.0444, "step": 19590 }, { "epoch": 12.492033142128744, "grad_norm": 0.7125439643859863, "learning_rate": 9.99357632087464e-05, "loss": 0.0574, "step": 19600 }, { "epoch": 12.498406628425748, "grad_norm": 0.37637677788734436, "learning_rate": 9.993548361406914e-05, "loss": 0.0467, "step": 19610 }, { "epoch": 12.504780114722752, "grad_norm": 0.4394405484199524, "learning_rate": 9.99352034126286e-05, "loss": 0.0483, "step": 19620 }, { "epoch": 12.511153601019759, "grad_norm": 0.5288838744163513, "learning_rate": 9.993492260442822e-05, "loss": 0.0607, "step": 19630 }, { "epoch": 12.517527087316763, "grad_norm": 0.5851248502731323, "learning_rate": 9.993464118947137e-05, "loss": 0.0349, "step": 19640 }, { "epoch": 12.523900573613767, "grad_norm": 0.663963258266449, "learning_rate": 9.993435916776149e-05, "loss": 0.0535, "step": 19650 }, { "epoch": 12.530274059910772, "grad_norm": 0.6779718399047852, "learning_rate": 9.993407653930198e-05, "loss": 0.0556, "step": 19660 }, { "epoch": 12.536647546207776, "grad_norm": 0.881489634513855, "learning_rate": 9.993379330409632e-05, "loss": 0.0498, "step": 19670 }, { "epoch": 12.54302103250478, "grad_norm": 0.5240219235420227, "learning_rate": 9.99335094621479e-05, "loss": 0.0413, "step": 19680 }, { "epoch": 12.549394518801785, "grad_norm": 0.5868619680404663, "learning_rate": 9.993322501346021e-05, "loss": 0.0538, "step": 19690 }, { "epoch": 12.55576800509879, "grad_norm": 0.7972108721733093, "learning_rate": 9.993293995803669e-05, "loss": 0.0508, "step": 19700 }, { "epoch": 12.562141491395794, "grad_norm": 0.739250659942627, "learning_rate": 9.993265429588079e-05, "loss": 0.0543, "step": 19710 }, { "epoch": 12.568514977692798, "grad_norm": 0.6294760704040527, "learning_rate": 9.993236802699601e-05, "loss": 0.055, "step": 19720 }, { "epoch": 12.574888463989803, "grad_norm": 0.6474262475967407, "learning_rate": 9.993208115138581e-05, "loss": 0.0474, "step": 19730 }, { "epoch": 12.581261950286807, "grad_norm": 0.49055972695350647, "learning_rate": 9.99317936690537e-05, "loss": 0.0554, "step": 19740 }, { "epoch": 12.587635436583811, "grad_norm": 0.5933214426040649, "learning_rate": 9.993150558000313e-05, "loss": 0.0365, "step": 19750 }, { "epoch": 12.594008922880816, "grad_norm": 0.6509645581245422, "learning_rate": 9.993121688423764e-05, "loss": 0.0509, "step": 19760 }, { "epoch": 12.60038240917782, "grad_norm": 0.588777482509613, "learning_rate": 9.99309275817607e-05, "loss": 0.055, "step": 19770 }, { "epoch": 12.606755895474825, "grad_norm": 0.5462532639503479, "learning_rate": 9.993063767257586e-05, "loss": 0.0506, "step": 19780 }, { "epoch": 12.613129381771829, "grad_norm": 0.6147275567054749, "learning_rate": 9.993034715668664e-05, "loss": 0.0534, "step": 19790 }, { "epoch": 12.619502868068833, "grad_norm": 0.687687873840332, "learning_rate": 9.993005603409655e-05, "loss": 0.0459, "step": 19800 }, { "epoch": 12.625876354365838, "grad_norm": 0.6746910810470581, "learning_rate": 9.992976430480914e-05, "loss": 0.054, "step": 19810 }, { "epoch": 12.632249840662842, "grad_norm": 0.7588403820991516, "learning_rate": 9.992947196882796e-05, "loss": 0.043, "step": 19820 }, { "epoch": 12.638623326959847, "grad_norm": 0.4694322943687439, "learning_rate": 9.992917902615654e-05, "loss": 0.0445, "step": 19830 }, { "epoch": 12.644996813256851, "grad_norm": 0.36447465419769287, "learning_rate": 9.992888547679846e-05, "loss": 0.0415, "step": 19840 }, { "epoch": 12.651370299553856, "grad_norm": 0.5364222526550293, "learning_rate": 9.992859132075728e-05, "loss": 0.0565, "step": 19850 }, { "epoch": 12.65774378585086, "grad_norm": 0.5305570960044861, "learning_rate": 9.992829655803657e-05, "loss": 0.0423, "step": 19860 }, { "epoch": 12.664117272147864, "grad_norm": 0.6378289461135864, "learning_rate": 9.992800118863992e-05, "loss": 0.0553, "step": 19870 }, { "epoch": 12.670490758444869, "grad_norm": 0.5443289875984192, "learning_rate": 9.992770521257092e-05, "loss": 0.0628, "step": 19880 }, { "epoch": 12.676864244741873, "grad_norm": 0.7340859174728394, "learning_rate": 9.992740862983316e-05, "loss": 0.0515, "step": 19890 }, { "epoch": 12.683237731038878, "grad_norm": 0.4966719448566437, "learning_rate": 9.992711144043024e-05, "loss": 0.0453, "step": 19900 }, { "epoch": 12.689611217335882, "grad_norm": 0.7671192288398743, "learning_rate": 9.992681364436578e-05, "loss": 0.0662, "step": 19910 }, { "epoch": 12.695984703632888, "grad_norm": 0.41733086109161377, "learning_rate": 9.992651524164341e-05, "loss": 0.0329, "step": 19920 }, { "epoch": 12.702358189929893, "grad_norm": 0.5677781701087952, "learning_rate": 9.992621623226672e-05, "loss": 0.0574, "step": 19930 }, { "epoch": 12.708731676226897, "grad_norm": 0.637280285358429, "learning_rate": 9.992591661623937e-05, "loss": 0.0483, "step": 19940 }, { "epoch": 12.715105162523901, "grad_norm": 0.5569368600845337, "learning_rate": 9.992561639356502e-05, "loss": 0.0458, "step": 19950 }, { "epoch": 12.721478648820906, "grad_norm": 0.556477963924408, "learning_rate": 9.992531556424726e-05, "loss": 0.0544, "step": 19960 }, { "epoch": 12.72785213511791, "grad_norm": 0.5600087642669678, "learning_rate": 9.992501412828978e-05, "loss": 0.0579, "step": 19970 }, { "epoch": 12.734225621414915, "grad_norm": 0.6145752668380737, "learning_rate": 9.992471208569626e-05, "loss": 0.0516, "step": 19980 }, { "epoch": 12.740599107711919, "grad_norm": 0.38638541102409363, "learning_rate": 9.992440943647035e-05, "loss": 0.0557, "step": 19990 }, { "epoch": 12.746972594008923, "grad_norm": 0.6183023452758789, "learning_rate": 9.992410618061572e-05, "loss": 0.0551, "step": 20000 }, { "epoch": 12.753346080305928, "grad_norm": 0.6596343517303467, "learning_rate": 9.992380231813607e-05, "loss": 0.0601, "step": 20010 }, { "epoch": 12.759719566602932, "grad_norm": 0.7001132965087891, "learning_rate": 9.992349784903508e-05, "loss": 0.0427, "step": 20020 }, { "epoch": 12.766093052899937, "grad_norm": 0.4830794632434845, "learning_rate": 9.992319277331646e-05, "loss": 0.0522, "step": 20030 }, { "epoch": 12.772466539196941, "grad_norm": 0.8132880330085754, "learning_rate": 9.992288709098389e-05, "loss": 0.0561, "step": 20040 }, { "epoch": 12.778840025493945, "grad_norm": 0.8195405006408691, "learning_rate": 9.992258080204113e-05, "loss": 0.0504, "step": 20050 }, { "epoch": 12.78521351179095, "grad_norm": 0.6251553297042847, "learning_rate": 9.99222739064919e-05, "loss": 0.0474, "step": 20060 }, { "epoch": 12.791586998087954, "grad_norm": 0.38529151678085327, "learning_rate": 9.992196640433986e-05, "loss": 0.0425, "step": 20070 }, { "epoch": 12.797960484384959, "grad_norm": 0.7417767643928528, "learning_rate": 9.992165829558881e-05, "loss": 0.063, "step": 20080 }, { "epoch": 12.804333970681963, "grad_norm": 0.8575668334960938, "learning_rate": 9.99213495802425e-05, "loss": 0.0565, "step": 20090 }, { "epoch": 12.810707456978967, "grad_norm": 0.5553948879241943, "learning_rate": 9.992104025830465e-05, "loss": 0.0422, "step": 20100 }, { "epoch": 12.817080943275972, "grad_norm": 0.41762441396713257, "learning_rate": 9.992073032977903e-05, "loss": 0.0461, "step": 20110 }, { "epoch": 12.823454429572976, "grad_norm": 0.5332140922546387, "learning_rate": 9.992041979466937e-05, "loss": 0.0499, "step": 20120 }, { "epoch": 12.82982791586998, "grad_norm": 0.6717177629470825, "learning_rate": 9.992010865297952e-05, "loss": 0.0499, "step": 20130 }, { "epoch": 12.836201402166985, "grad_norm": 0.5670310258865356, "learning_rate": 9.991979690471319e-05, "loss": 0.0427, "step": 20140 }, { "epoch": 12.84257488846399, "grad_norm": 0.5739524960517883, "learning_rate": 9.991948454987421e-05, "loss": 0.0538, "step": 20150 }, { "epoch": 12.848948374760994, "grad_norm": 0.5475471019744873, "learning_rate": 9.991917158846635e-05, "loss": 0.0419, "step": 20160 }, { "epoch": 12.855321861057998, "grad_norm": 0.7085890173912048, "learning_rate": 9.991885802049343e-05, "loss": 0.0581, "step": 20170 }, { "epoch": 12.861695347355003, "grad_norm": 0.9820379018783569, "learning_rate": 9.991854384595926e-05, "loss": 0.0481, "step": 20180 }, { "epoch": 12.868068833652007, "grad_norm": 0.2957818806171417, "learning_rate": 9.991822906486763e-05, "loss": 0.051, "step": 20190 }, { "epoch": 12.874442319949011, "grad_norm": 0.6987794041633606, "learning_rate": 9.991791367722241e-05, "loss": 0.0514, "step": 20200 }, { "epoch": 12.880815806246016, "grad_norm": 0.4746705889701843, "learning_rate": 9.991759768302739e-05, "loss": 0.0597, "step": 20210 }, { "epoch": 12.88718929254302, "grad_norm": 0.6064462661743164, "learning_rate": 9.991728108228645e-05, "loss": 0.0557, "step": 20220 }, { "epoch": 12.893562778840025, "grad_norm": 0.8060480356216431, "learning_rate": 9.99169638750034e-05, "loss": 0.0404, "step": 20230 }, { "epoch": 12.899936265137029, "grad_norm": 0.4107417166233063, "learning_rate": 9.991664606118212e-05, "loss": 0.0589, "step": 20240 }, { "epoch": 12.906309751434035, "grad_norm": 0.42253613471984863, "learning_rate": 9.991632764082644e-05, "loss": 0.0564, "step": 20250 }, { "epoch": 12.91268323773104, "grad_norm": 0.39724990725517273, "learning_rate": 9.991600861394026e-05, "loss": 0.0346, "step": 20260 }, { "epoch": 12.919056724028044, "grad_norm": 0.6950128674507141, "learning_rate": 9.991568898052745e-05, "loss": 0.0507, "step": 20270 }, { "epoch": 12.925430210325048, "grad_norm": 0.6304432153701782, "learning_rate": 9.99153687405919e-05, "loss": 0.0404, "step": 20280 }, { "epoch": 12.931803696622053, "grad_norm": 0.6180141568183899, "learning_rate": 9.991504789413748e-05, "loss": 0.0604, "step": 20290 }, { "epoch": 12.938177182919057, "grad_norm": 0.5379759669303894, "learning_rate": 9.99147264411681e-05, "loss": 0.0483, "step": 20300 }, { "epoch": 12.944550669216062, "grad_norm": 0.5783708691596985, "learning_rate": 9.991440438168767e-05, "loss": 0.0463, "step": 20310 }, { "epoch": 12.950924155513066, "grad_norm": 0.48275479674339294, "learning_rate": 9.99140817157001e-05, "loss": 0.0452, "step": 20320 }, { "epoch": 12.95729764181007, "grad_norm": 0.4784011244773865, "learning_rate": 9.991375844320931e-05, "loss": 0.0467, "step": 20330 }, { "epoch": 12.963671128107075, "grad_norm": 1.0397515296936035, "learning_rate": 9.991343456421923e-05, "loss": 0.0517, "step": 20340 }, { "epoch": 12.97004461440408, "grad_norm": 0.3007799983024597, "learning_rate": 9.99131100787338e-05, "loss": 0.0532, "step": 20350 }, { "epoch": 12.976418100701084, "grad_norm": 0.6348592042922974, "learning_rate": 9.991278498675694e-05, "loss": 0.0567, "step": 20360 }, { "epoch": 12.982791586998088, "grad_norm": 0.8217516541481018, "learning_rate": 9.991245928829262e-05, "loss": 0.0585, "step": 20370 }, { "epoch": 12.989165073295093, "grad_norm": 0.5914630889892578, "learning_rate": 9.991213298334481e-05, "loss": 0.054, "step": 20380 }, { "epoch": 12.995538559592097, "grad_norm": 0.6380867958068848, "learning_rate": 9.991180607191745e-05, "loss": 0.0722, "step": 20390 }, { "epoch": 13.001912045889101, "grad_norm": 0.6516058444976807, "learning_rate": 9.991147855401453e-05, "loss": 0.0449, "step": 20400 }, { "epoch": 13.008285532186106, "grad_norm": 0.6741889119148254, "learning_rate": 9.991115042964002e-05, "loss": 0.0488, "step": 20410 }, { "epoch": 13.01465901848311, "grad_norm": 0.6972917318344116, "learning_rate": 9.991082169879788e-05, "loss": 0.0532, "step": 20420 }, { "epoch": 13.021032504780115, "grad_norm": 0.5252955555915833, "learning_rate": 9.991049236149216e-05, "loss": 0.0419, "step": 20430 }, { "epoch": 13.027405991077119, "grad_norm": 0.5976403951644897, "learning_rate": 9.991016241772683e-05, "loss": 0.0523, "step": 20440 }, { "epoch": 13.033779477374123, "grad_norm": 0.7129591107368469, "learning_rate": 9.99098318675059e-05, "loss": 0.0552, "step": 20450 }, { "epoch": 13.040152963671128, "grad_norm": 0.541470468044281, "learning_rate": 9.99095007108334e-05, "loss": 0.0608, "step": 20460 }, { "epoch": 13.046526449968132, "grad_norm": 0.4547019898891449, "learning_rate": 9.990916894771334e-05, "loss": 0.0557, "step": 20470 }, { "epoch": 13.052899936265137, "grad_norm": 0.5617836713790894, "learning_rate": 9.990883657814975e-05, "loss": 0.0472, "step": 20480 }, { "epoch": 13.059273422562141, "grad_norm": 0.9097099304199219, "learning_rate": 9.990850360214667e-05, "loss": 0.044, "step": 20490 }, { "epoch": 13.065646908859145, "grad_norm": 0.421392560005188, "learning_rate": 9.990817001970816e-05, "loss": 0.0434, "step": 20500 }, { "epoch": 13.07202039515615, "grad_norm": 0.574134349822998, "learning_rate": 9.990783583083825e-05, "loss": 0.0433, "step": 20510 }, { "epoch": 13.078393881453154, "grad_norm": 0.6722016930580139, "learning_rate": 9.990750103554102e-05, "loss": 0.048, "step": 20520 }, { "epoch": 13.084767367750159, "grad_norm": 0.6953365206718445, "learning_rate": 9.990716563382055e-05, "loss": 0.0494, "step": 20530 }, { "epoch": 13.091140854047163, "grad_norm": 0.4717896580696106, "learning_rate": 9.990682962568087e-05, "loss": 0.0369, "step": 20540 }, { "epoch": 13.097514340344167, "grad_norm": 0.5848617553710938, "learning_rate": 9.990649301112611e-05, "loss": 0.0479, "step": 20550 }, { "epoch": 13.103887826641174, "grad_norm": 0.7929871082305908, "learning_rate": 9.990615579016033e-05, "loss": 0.0469, "step": 20560 }, { "epoch": 13.110261312938178, "grad_norm": 0.7508995532989502, "learning_rate": 9.990581796278764e-05, "loss": 0.0497, "step": 20570 }, { "epoch": 13.116634799235182, "grad_norm": 0.6095130443572998, "learning_rate": 9.990547952901215e-05, "loss": 0.0559, "step": 20580 }, { "epoch": 13.123008285532187, "grad_norm": 0.6503527164459229, "learning_rate": 9.990514048883794e-05, "loss": 0.0453, "step": 20590 }, { "epoch": 13.129381771829191, "grad_norm": 0.5425077676773071, "learning_rate": 9.990480084226919e-05, "loss": 0.0428, "step": 20600 }, { "epoch": 13.135755258126196, "grad_norm": 1.0947233438491821, "learning_rate": 9.990446058930996e-05, "loss": 0.0582, "step": 20610 }, { "epoch": 13.1421287444232, "grad_norm": 0.7393929362297058, "learning_rate": 9.990411972996443e-05, "loss": 0.051, "step": 20620 }, { "epoch": 13.148502230720204, "grad_norm": 0.5818952918052673, "learning_rate": 9.990377826423673e-05, "loss": 0.0481, "step": 20630 }, { "epoch": 13.154875717017209, "grad_norm": 0.42319175601005554, "learning_rate": 9.990343619213101e-05, "loss": 0.0478, "step": 20640 }, { "epoch": 13.161249203314213, "grad_norm": 0.5444415807723999, "learning_rate": 9.990309351365141e-05, "loss": 0.044, "step": 20650 }, { "epoch": 13.167622689611218, "grad_norm": 0.4666403532028198, "learning_rate": 9.990275022880213e-05, "loss": 0.035, "step": 20660 }, { "epoch": 13.173996175908222, "grad_norm": 0.31161317229270935, "learning_rate": 9.990240633758731e-05, "loss": 0.0374, "step": 20670 }, { "epoch": 13.180369662205226, "grad_norm": 0.6608639359474182, "learning_rate": 9.990206184001113e-05, "loss": 0.0455, "step": 20680 }, { "epoch": 13.18674314850223, "grad_norm": 0.3203983008861542, "learning_rate": 9.990171673607777e-05, "loss": 0.0356, "step": 20690 }, { "epoch": 13.193116634799235, "grad_norm": 0.4987826645374298, "learning_rate": 9.990137102579146e-05, "loss": 0.0436, "step": 20700 }, { "epoch": 13.19949012109624, "grad_norm": 0.8196752071380615, "learning_rate": 9.990102470915637e-05, "loss": 0.0511, "step": 20710 }, { "epoch": 13.205863607393244, "grad_norm": 0.6709547638893127, "learning_rate": 9.990067778617673e-05, "loss": 0.0503, "step": 20720 }, { "epoch": 13.212237093690248, "grad_norm": 0.513703465461731, "learning_rate": 9.990033025685672e-05, "loss": 0.0438, "step": 20730 }, { "epoch": 13.218610579987253, "grad_norm": 0.49383261799812317, "learning_rate": 9.98999821212006e-05, "loss": 0.0524, "step": 20740 }, { "epoch": 13.224984066284257, "grad_norm": 0.5461325645446777, "learning_rate": 9.989963337921256e-05, "loss": 0.0616, "step": 20750 }, { "epoch": 13.231357552581262, "grad_norm": 0.34099167585372925, "learning_rate": 9.989928403089688e-05, "loss": 0.0506, "step": 20760 }, { "epoch": 13.237731038878266, "grad_norm": 0.5841230154037476, "learning_rate": 9.989893407625779e-05, "loss": 0.0579, "step": 20770 }, { "epoch": 13.24410452517527, "grad_norm": 0.7240725159645081, "learning_rate": 9.989858351529953e-05, "loss": 0.0658, "step": 20780 }, { "epoch": 13.250478011472275, "grad_norm": 0.4097904860973358, "learning_rate": 9.989823234802636e-05, "loss": 0.0559, "step": 20790 }, { "epoch": 13.25685149776928, "grad_norm": 0.5467256307601929, "learning_rate": 9.989788057444256e-05, "loss": 0.0573, "step": 20800 }, { "epoch": 13.263224984066284, "grad_norm": 0.605908989906311, "learning_rate": 9.989752819455241e-05, "loss": 0.0524, "step": 20810 }, { "epoch": 13.269598470363288, "grad_norm": 0.3435722589492798, "learning_rate": 9.989717520836017e-05, "loss": 0.0529, "step": 20820 }, { "epoch": 13.275971956660293, "grad_norm": 0.6609158515930176, "learning_rate": 9.989682161587015e-05, "loss": 0.0516, "step": 20830 }, { "epoch": 13.282345442957297, "grad_norm": 1.0408082008361816, "learning_rate": 9.989646741708662e-05, "loss": 0.0555, "step": 20840 }, { "epoch": 13.288718929254301, "grad_norm": 0.620171308517456, "learning_rate": 9.989611261201393e-05, "loss": 0.0472, "step": 20850 }, { "epoch": 13.295092415551306, "grad_norm": 0.5598809719085693, "learning_rate": 9.989575720065633e-05, "loss": 0.0462, "step": 20860 }, { "epoch": 13.30146590184831, "grad_norm": 0.6512726545333862, "learning_rate": 9.98954011830182e-05, "loss": 0.0438, "step": 20870 }, { "epoch": 13.307839388145316, "grad_norm": 0.6691747307777405, "learning_rate": 9.989504455910381e-05, "loss": 0.0564, "step": 20880 }, { "epoch": 13.31421287444232, "grad_norm": 0.4432419240474701, "learning_rate": 9.989468732891754e-05, "loss": 0.0448, "step": 20890 }, { "epoch": 13.320586360739325, "grad_norm": 0.7473397254943848, "learning_rate": 9.98943294924637e-05, "loss": 0.0459, "step": 20900 }, { "epoch": 13.32695984703633, "grad_norm": 0.8737781643867493, "learning_rate": 9.989397104974665e-05, "loss": 0.0478, "step": 20910 }, { "epoch": 13.333333333333334, "grad_norm": 0.5254886150360107, "learning_rate": 9.989361200077074e-05, "loss": 0.0544, "step": 20920 }, { "epoch": 13.339706819630338, "grad_norm": 0.836776614189148, "learning_rate": 9.989325234554034e-05, "loss": 0.0445, "step": 20930 }, { "epoch": 13.346080305927343, "grad_norm": 0.8743926286697388, "learning_rate": 9.989289208405983e-05, "loss": 0.0532, "step": 20940 }, { "epoch": 13.352453792224347, "grad_norm": 0.4669855833053589, "learning_rate": 9.989253121633356e-05, "loss": 0.0457, "step": 20950 }, { "epoch": 13.358827278521352, "grad_norm": 0.4714788794517517, "learning_rate": 9.989216974236594e-05, "loss": 0.066, "step": 20960 }, { "epoch": 13.365200764818356, "grad_norm": 0.4179537296295166, "learning_rate": 9.989180766216134e-05, "loss": 0.045, "step": 20970 }, { "epoch": 13.37157425111536, "grad_norm": 0.508573055267334, "learning_rate": 9.989144497572418e-05, "loss": 0.0499, "step": 20980 }, { "epoch": 13.377947737412365, "grad_norm": 0.450468510389328, "learning_rate": 9.989108168305885e-05, "loss": 0.0481, "step": 20990 }, { "epoch": 13.38432122370937, "grad_norm": 0.3956815004348755, "learning_rate": 9.989071778416977e-05, "loss": 0.0382, "step": 21000 }, { "epoch": 13.390694710006374, "grad_norm": 0.6461054086685181, "learning_rate": 9.989035327906138e-05, "loss": 0.0493, "step": 21010 }, { "epoch": 13.397068196303378, "grad_norm": 0.6376568078994751, "learning_rate": 9.988998816773808e-05, "loss": 0.0483, "step": 21020 }, { "epoch": 13.403441682600382, "grad_norm": 0.4341847002506256, "learning_rate": 9.988962245020433e-05, "loss": 0.0407, "step": 21030 }, { "epoch": 13.409815168897387, "grad_norm": 0.5661268830299377, "learning_rate": 9.988925612646455e-05, "loss": 0.0591, "step": 21040 }, { "epoch": 13.416188655194391, "grad_norm": 1.0918251276016235, "learning_rate": 9.988888919652322e-05, "loss": 0.0512, "step": 21050 }, { "epoch": 13.422562141491396, "grad_norm": 0.6603458523750305, "learning_rate": 9.988852166038477e-05, "loss": 0.0584, "step": 21060 }, { "epoch": 13.4289356277884, "grad_norm": 0.4041292071342468, "learning_rate": 9.988815351805369e-05, "loss": 0.055, "step": 21070 }, { "epoch": 13.435309114085404, "grad_norm": 0.8125314116477966, "learning_rate": 9.988778476953443e-05, "loss": 0.0388, "step": 21080 }, { "epoch": 13.441682600382409, "grad_norm": 1.019276738166809, "learning_rate": 9.988741541483148e-05, "loss": 0.0569, "step": 21090 }, { "epoch": 13.448056086679413, "grad_norm": 0.7024986743927002, "learning_rate": 9.988704545394936e-05, "loss": 0.0638, "step": 21100 }, { "epoch": 13.454429572976418, "grad_norm": 0.9689950942993164, "learning_rate": 9.988667488689252e-05, "loss": 0.0516, "step": 21110 }, { "epoch": 13.460803059273422, "grad_norm": 0.707746684551239, "learning_rate": 9.988630371366548e-05, "loss": 0.0425, "step": 21120 }, { "epoch": 13.467176545570426, "grad_norm": 0.5330038666725159, "learning_rate": 9.988593193427274e-05, "loss": 0.06, "step": 21130 }, { "epoch": 13.47355003186743, "grad_norm": 0.5291359424591064, "learning_rate": 9.988555954871884e-05, "loss": 0.0518, "step": 21140 }, { "epoch": 13.479923518164435, "grad_norm": 0.5407653450965881, "learning_rate": 9.988518655700828e-05, "loss": 0.0593, "step": 21150 }, { "epoch": 13.48629700446144, "grad_norm": 0.5795421004295349, "learning_rate": 9.988481295914561e-05, "loss": 0.0509, "step": 21160 }, { "epoch": 13.492670490758444, "grad_norm": 1.0260026454925537, "learning_rate": 9.988443875513535e-05, "loss": 0.0569, "step": 21170 }, { "epoch": 13.49904397705545, "grad_norm": 0.4161250591278076, "learning_rate": 9.988406394498209e-05, "loss": 0.0465, "step": 21180 }, { "epoch": 13.505417463352455, "grad_norm": 0.734157919883728, "learning_rate": 9.988368852869034e-05, "loss": 0.055, "step": 21190 }, { "epoch": 13.511790949649459, "grad_norm": 0.3861832320690155, "learning_rate": 9.988331250626467e-05, "loss": 0.0569, "step": 21200 }, { "epoch": 13.518164435946463, "grad_norm": 0.3449721038341522, "learning_rate": 9.988293587770966e-05, "loss": 0.0455, "step": 21210 }, { "epoch": 13.524537922243468, "grad_norm": 0.7911856174468994, "learning_rate": 9.988255864302988e-05, "loss": 0.0453, "step": 21220 }, { "epoch": 13.530911408540472, "grad_norm": 0.7439214587211609, "learning_rate": 9.988218080222992e-05, "loss": 0.0483, "step": 21230 }, { "epoch": 13.537284894837477, "grad_norm": 0.2943658232688904, "learning_rate": 9.988180235531437e-05, "loss": 0.0461, "step": 21240 }, { "epoch": 13.543658381134481, "grad_norm": 0.5325548648834229, "learning_rate": 9.988142330228781e-05, "loss": 0.0476, "step": 21250 }, { "epoch": 13.550031867431485, "grad_norm": 0.9646657109260559, "learning_rate": 9.988104364315487e-05, "loss": 0.0494, "step": 21260 }, { "epoch": 13.55640535372849, "grad_norm": 0.6305964589118958, "learning_rate": 9.988066337792015e-05, "loss": 0.0658, "step": 21270 }, { "epoch": 13.562778840025494, "grad_norm": 0.39909276366233826, "learning_rate": 9.988028250658829e-05, "loss": 0.0388, "step": 21280 }, { "epoch": 13.569152326322499, "grad_norm": 0.35210394859313965, "learning_rate": 9.987990102916389e-05, "loss": 0.0558, "step": 21290 }, { "epoch": 13.575525812619503, "grad_norm": 0.4933224022388458, "learning_rate": 9.98795189456516e-05, "loss": 0.0503, "step": 21300 }, { "epoch": 13.581899298916507, "grad_norm": 0.5310596227645874, "learning_rate": 9.987913625605606e-05, "loss": 0.044, "step": 21310 }, { "epoch": 13.588272785213512, "grad_norm": 0.6643886566162109, "learning_rate": 9.987875296038192e-05, "loss": 0.0424, "step": 21320 }, { "epoch": 13.594646271510516, "grad_norm": 0.4461294412612915, "learning_rate": 9.987836905863384e-05, "loss": 0.044, "step": 21330 }, { "epoch": 13.60101975780752, "grad_norm": 0.731680691242218, "learning_rate": 9.987798455081648e-05, "loss": 0.051, "step": 21340 }, { "epoch": 13.607393244104525, "grad_norm": 0.4990619421005249, "learning_rate": 9.987759943693452e-05, "loss": 0.0543, "step": 21350 }, { "epoch": 13.61376673040153, "grad_norm": 0.6637498736381531, "learning_rate": 9.987721371699264e-05, "loss": 0.0402, "step": 21360 }, { "epoch": 13.620140216698534, "grad_norm": 0.5051818490028381, "learning_rate": 9.987682739099551e-05, "loss": 0.0548, "step": 21370 }, { "epoch": 13.626513702995538, "grad_norm": 0.47658732533454895, "learning_rate": 9.987644045894784e-05, "loss": 0.04, "step": 21380 }, { "epoch": 13.632887189292543, "grad_norm": 0.9316199421882629, "learning_rate": 9.987605292085432e-05, "loss": 0.0412, "step": 21390 }, { "epoch": 13.639260675589547, "grad_norm": 0.6131792664527893, "learning_rate": 9.987566477671968e-05, "loss": 0.0662, "step": 21400 }, { "epoch": 13.645634161886552, "grad_norm": 0.7972255349159241, "learning_rate": 9.987527602654861e-05, "loss": 0.047, "step": 21410 }, { "epoch": 13.652007648183556, "grad_norm": 0.34587010741233826, "learning_rate": 9.987488667034586e-05, "loss": 0.0682, "step": 21420 }, { "epoch": 13.65838113448056, "grad_norm": 0.5820200443267822, "learning_rate": 9.987449670811612e-05, "loss": 0.0534, "step": 21430 }, { "epoch": 13.664754620777565, "grad_norm": 0.6251576542854309, "learning_rate": 9.987410613986419e-05, "loss": 0.0469, "step": 21440 }, { "epoch": 13.67112810707457, "grad_norm": 0.6655545234680176, "learning_rate": 9.987371496559477e-05, "loss": 0.0496, "step": 21450 }, { "epoch": 13.677501593371574, "grad_norm": 0.8633976578712463, "learning_rate": 9.987332318531262e-05, "loss": 0.0533, "step": 21460 }, { "epoch": 13.683875079668578, "grad_norm": 0.6545677185058594, "learning_rate": 9.98729307990225e-05, "loss": 0.0684, "step": 21470 }, { "epoch": 13.690248565965582, "grad_norm": 0.3638930916786194, "learning_rate": 9.987253780672919e-05, "loss": 0.0498, "step": 21480 }, { "epoch": 13.696622052262587, "grad_norm": 0.7899186611175537, "learning_rate": 9.987214420843746e-05, "loss": 0.0564, "step": 21490 }, { "epoch": 13.702995538559591, "grad_norm": 0.6777027249336243, "learning_rate": 9.987175000415208e-05, "loss": 0.0661, "step": 21500 }, { "epoch": 13.709369024856597, "grad_norm": 0.5900900959968567, "learning_rate": 9.987135519387787e-05, "loss": 0.0597, "step": 21510 }, { "epoch": 13.715742511153602, "grad_norm": 0.47955411672592163, "learning_rate": 9.98709597776196e-05, "loss": 0.0503, "step": 21520 }, { "epoch": 13.722115997450606, "grad_norm": 0.8849825263023376, "learning_rate": 9.987056375538206e-05, "loss": 0.0448, "step": 21530 }, { "epoch": 13.72848948374761, "grad_norm": 0.5264932513237, "learning_rate": 9.987016712717011e-05, "loss": 0.0436, "step": 21540 }, { "epoch": 13.734862970044615, "grad_norm": 0.6136640310287476, "learning_rate": 9.986976989298854e-05, "loss": 0.0493, "step": 21550 }, { "epoch": 13.74123645634162, "grad_norm": 0.6094130873680115, "learning_rate": 9.986937205284216e-05, "loss": 0.0506, "step": 21560 }, { "epoch": 13.747609942638624, "grad_norm": 0.6434550285339355, "learning_rate": 9.986897360673585e-05, "loss": 0.0473, "step": 21570 }, { "epoch": 13.753983428935628, "grad_norm": 0.4452361762523651, "learning_rate": 9.986857455467443e-05, "loss": 0.0425, "step": 21580 }, { "epoch": 13.760356915232633, "grad_norm": 0.5930310487747192, "learning_rate": 9.986817489666273e-05, "loss": 0.0474, "step": 21590 }, { "epoch": 13.766730401529637, "grad_norm": 0.5597618222236633, "learning_rate": 9.986777463270563e-05, "loss": 0.0571, "step": 21600 }, { "epoch": 13.773103887826641, "grad_norm": 0.6503089070320129, "learning_rate": 9.986737376280799e-05, "loss": 0.045, "step": 21610 }, { "epoch": 13.779477374123646, "grad_norm": 0.5297825932502747, "learning_rate": 9.986697228697467e-05, "loss": 0.0597, "step": 21620 }, { "epoch": 13.78585086042065, "grad_norm": 0.3579823970794678, "learning_rate": 9.986657020521056e-05, "loss": 0.0629, "step": 21630 }, { "epoch": 13.792224346717655, "grad_norm": 0.5140018463134766, "learning_rate": 9.986616751752052e-05, "loss": 0.0446, "step": 21640 }, { "epoch": 13.798597833014659, "grad_norm": 1.2022029161453247, "learning_rate": 9.986576422390948e-05, "loss": 0.0474, "step": 21650 }, { "epoch": 13.804971319311663, "grad_norm": 0.6232237815856934, "learning_rate": 9.986536032438232e-05, "loss": 0.0457, "step": 21660 }, { "epoch": 13.811344805608668, "grad_norm": 0.7051699161529541, "learning_rate": 9.986495581894395e-05, "loss": 0.0479, "step": 21670 }, { "epoch": 13.817718291905672, "grad_norm": 0.6805098652839661, "learning_rate": 9.986455070759929e-05, "loss": 0.0488, "step": 21680 }, { "epoch": 13.824091778202677, "grad_norm": 0.5452792644500732, "learning_rate": 9.986414499035327e-05, "loss": 0.0403, "step": 21690 }, { "epoch": 13.830465264499681, "grad_norm": 0.5022963285446167, "learning_rate": 9.986373866721079e-05, "loss": 0.0508, "step": 21700 }, { "epoch": 13.836838750796685, "grad_norm": 0.3037646412849426, "learning_rate": 9.986333173817681e-05, "loss": 0.0526, "step": 21710 }, { "epoch": 13.84321223709369, "grad_norm": 0.753787100315094, "learning_rate": 9.986292420325627e-05, "loss": 0.055, "step": 21720 }, { "epoch": 13.849585723390694, "grad_norm": 0.38492351770401, "learning_rate": 9.986251606245412e-05, "loss": 0.0507, "step": 21730 }, { "epoch": 13.855959209687699, "grad_norm": 0.5820807814598083, "learning_rate": 9.986210731577534e-05, "loss": 0.0392, "step": 21740 }, { "epoch": 13.862332695984703, "grad_norm": 0.40073642134666443, "learning_rate": 9.986169796322487e-05, "loss": 0.0366, "step": 21750 }, { "epoch": 13.868706182281707, "grad_norm": 0.4889290928840637, "learning_rate": 9.986128800480768e-05, "loss": 0.0506, "step": 21760 }, { "epoch": 13.875079668578712, "grad_norm": 0.7286251783370972, "learning_rate": 9.986087744052876e-05, "loss": 0.0491, "step": 21770 }, { "epoch": 13.881453154875716, "grad_norm": 0.31427904963493347, "learning_rate": 9.986046627039311e-05, "loss": 0.0426, "step": 21780 }, { "epoch": 13.88782664117272, "grad_norm": 0.6496139168739319, "learning_rate": 9.986005449440574e-05, "loss": 0.0569, "step": 21790 }, { "epoch": 13.894200127469727, "grad_norm": 0.7005490660667419, "learning_rate": 9.985964211257162e-05, "loss": 0.048, "step": 21800 }, { "epoch": 13.900573613766731, "grad_norm": 0.5604730844497681, "learning_rate": 9.985922912489575e-05, "loss": 0.0659, "step": 21810 }, { "epoch": 13.906947100063736, "grad_norm": 0.5922862887382507, "learning_rate": 9.98588155313832e-05, "loss": 0.0377, "step": 21820 }, { "epoch": 13.91332058636074, "grad_norm": 0.5051929950714111, "learning_rate": 9.985840133203896e-05, "loss": 0.0511, "step": 21830 }, { "epoch": 13.919694072657744, "grad_norm": 0.5479230880737305, "learning_rate": 9.985798652686807e-05, "loss": 0.0533, "step": 21840 }, { "epoch": 13.926067558954749, "grad_norm": 0.424010306596756, "learning_rate": 9.985757111587556e-05, "loss": 0.0384, "step": 21850 }, { "epoch": 13.932441045251753, "grad_norm": 0.5242854952812195, "learning_rate": 9.985715509906649e-05, "loss": 0.0451, "step": 21860 }, { "epoch": 13.938814531548758, "grad_norm": 0.6178439855575562, "learning_rate": 9.985673847644592e-05, "loss": 0.0417, "step": 21870 }, { "epoch": 13.945188017845762, "grad_norm": 0.46104398369789124, "learning_rate": 9.98563212480189e-05, "loss": 0.0351, "step": 21880 }, { "epoch": 13.951561504142767, "grad_norm": 0.5755635499954224, "learning_rate": 9.985590341379051e-05, "loss": 0.0526, "step": 21890 }, { "epoch": 13.957934990439771, "grad_norm": 0.5781133770942688, "learning_rate": 9.985548497376582e-05, "loss": 0.037, "step": 21900 }, { "epoch": 13.964308476736775, "grad_norm": 0.5582672357559204, "learning_rate": 9.98550659279499e-05, "loss": 0.0444, "step": 21910 }, { "epoch": 13.97068196303378, "grad_norm": 0.5559388399124146, "learning_rate": 9.985464627634789e-05, "loss": 0.0575, "step": 21920 }, { "epoch": 13.977055449330784, "grad_norm": 0.5852295160293579, "learning_rate": 9.985422601896484e-05, "loss": 0.0433, "step": 21930 }, { "epoch": 13.983428935627789, "grad_norm": 0.48233315348625183, "learning_rate": 9.985380515580587e-05, "loss": 0.0389, "step": 21940 }, { "epoch": 13.989802421924793, "grad_norm": 0.760404109954834, "learning_rate": 9.985338368687609e-05, "loss": 0.0446, "step": 21950 }, { "epoch": 13.996175908221797, "grad_norm": 0.4950686991214752, "learning_rate": 9.985296161218064e-05, "loss": 0.0479, "step": 21960 }, { "epoch": 14.002549394518802, "grad_norm": 0.7322604656219482, "learning_rate": 9.985253893172464e-05, "loss": 0.0448, "step": 21970 }, { "epoch": 14.008922880815806, "grad_norm": 0.5258325934410095, "learning_rate": 9.985211564551322e-05, "loss": 0.0426, "step": 21980 }, { "epoch": 14.01529636711281, "grad_norm": 0.5011958479881287, "learning_rate": 9.985169175355153e-05, "loss": 0.0378, "step": 21990 }, { "epoch": 14.021669853409815, "grad_norm": 0.4282132685184479, "learning_rate": 9.985126725584471e-05, "loss": 0.0469, "step": 22000 }, { "epoch": 14.02804333970682, "grad_norm": 0.40375763177871704, "learning_rate": 9.985084215239792e-05, "loss": 0.0473, "step": 22010 }, { "epoch": 14.034416826003824, "grad_norm": 0.443429172039032, "learning_rate": 9.985041644321635e-05, "loss": 0.0485, "step": 22020 }, { "epoch": 14.040790312300828, "grad_norm": 0.5961761474609375, "learning_rate": 9.984999012830514e-05, "loss": 0.0607, "step": 22030 }, { "epoch": 14.047163798597833, "grad_norm": 0.5841917395591736, "learning_rate": 9.984956320766949e-05, "loss": 0.0372, "step": 22040 }, { "epoch": 14.053537284894837, "grad_norm": 0.5729540586471558, "learning_rate": 9.984913568131457e-05, "loss": 0.0376, "step": 22050 }, { "epoch": 14.059910771191841, "grad_norm": 0.6272742748260498, "learning_rate": 9.98487075492456e-05, "loss": 0.0403, "step": 22060 }, { "epoch": 14.066284257488846, "grad_norm": 0.7974969148635864, "learning_rate": 9.984827881146777e-05, "loss": 0.0416, "step": 22070 }, { "epoch": 14.07265774378585, "grad_norm": 0.4142572581768036, "learning_rate": 9.984784946798628e-05, "loss": 0.0292, "step": 22080 }, { "epoch": 14.079031230082855, "grad_norm": 0.4337879717350006, "learning_rate": 9.984741951880635e-05, "loss": 0.0537, "step": 22090 }, { "epoch": 14.085404716379859, "grad_norm": 0.6759001612663269, "learning_rate": 9.984698896393321e-05, "loss": 0.0491, "step": 22100 }, { "epoch": 14.091778202676863, "grad_norm": 0.5249837636947632, "learning_rate": 9.984655780337212e-05, "loss": 0.0414, "step": 22110 }, { "epoch": 14.098151688973868, "grad_norm": 0.570059597492218, "learning_rate": 9.984612603712825e-05, "loss": 0.0597, "step": 22120 }, { "epoch": 14.104525175270874, "grad_norm": 0.39427268505096436, "learning_rate": 9.98456936652069e-05, "loss": 0.0468, "step": 22130 }, { "epoch": 14.110898661567878, "grad_norm": 0.37906205654144287, "learning_rate": 9.98452606876133e-05, "loss": 0.0539, "step": 22140 }, { "epoch": 14.117272147864883, "grad_norm": 0.2791944444179535, "learning_rate": 9.984482710435274e-05, "loss": 0.0436, "step": 22150 }, { "epoch": 14.123645634161887, "grad_norm": 0.6249826550483704, "learning_rate": 9.984439291543046e-05, "loss": 0.0383, "step": 22160 }, { "epoch": 14.130019120458892, "grad_norm": 0.4677196741104126, "learning_rate": 9.984395812085175e-05, "loss": 0.0413, "step": 22170 }, { "epoch": 14.136392606755896, "grad_norm": 0.6214688420295715, "learning_rate": 9.98435227206219e-05, "loss": 0.0365, "step": 22180 }, { "epoch": 14.1427660930529, "grad_norm": 0.5297069549560547, "learning_rate": 9.984308671474616e-05, "loss": 0.0441, "step": 22190 }, { "epoch": 14.149139579349905, "grad_norm": 0.4164693057537079, "learning_rate": 9.984265010322987e-05, "loss": 0.0379, "step": 22200 }, { "epoch": 14.15551306564691, "grad_norm": 0.7650898098945618, "learning_rate": 9.984221288607834e-05, "loss": 0.0489, "step": 22210 }, { "epoch": 14.161886551943914, "grad_norm": 0.6391487121582031, "learning_rate": 9.984177506329684e-05, "loss": 0.0514, "step": 22220 }, { "epoch": 14.168260038240918, "grad_norm": 0.5873070359230042, "learning_rate": 9.984133663489072e-05, "loss": 0.039, "step": 22230 }, { "epoch": 14.174633524537922, "grad_norm": 0.48284444212913513, "learning_rate": 9.98408976008653e-05, "loss": 0.0455, "step": 22240 }, { "epoch": 14.181007010834927, "grad_norm": 0.4807731509208679, "learning_rate": 9.984045796122591e-05, "loss": 0.0455, "step": 22250 }, { "epoch": 14.187380497131931, "grad_norm": 0.4245636761188507, "learning_rate": 9.984001771597792e-05, "loss": 0.0318, "step": 22260 }, { "epoch": 14.193753983428936, "grad_norm": 0.5932088494300842, "learning_rate": 9.983957686512663e-05, "loss": 0.0542, "step": 22270 }, { "epoch": 14.20012746972594, "grad_norm": 1.124269962310791, "learning_rate": 9.983913540867747e-05, "loss": 0.0393, "step": 22280 }, { "epoch": 14.206500956022944, "grad_norm": 0.7120072841644287, "learning_rate": 9.983869334663571e-05, "loss": 0.0616, "step": 22290 }, { "epoch": 14.212874442319949, "grad_norm": 0.3253026306629181, "learning_rate": 9.983825067900681e-05, "loss": 0.0371, "step": 22300 }, { "epoch": 14.219247928616953, "grad_norm": 0.42797449231147766, "learning_rate": 9.983780740579608e-05, "loss": 0.0477, "step": 22310 }, { "epoch": 14.225621414913958, "grad_norm": 0.4663847088813782, "learning_rate": 9.983736352700894e-05, "loss": 0.0419, "step": 22320 }, { "epoch": 14.231994901210962, "grad_norm": 0.5582894086837769, "learning_rate": 9.983691904265079e-05, "loss": 0.0398, "step": 22330 }, { "epoch": 14.238368387507967, "grad_norm": 0.4541654884815216, "learning_rate": 9.983647395272701e-05, "loss": 0.0524, "step": 22340 }, { "epoch": 14.244741873804971, "grad_norm": 0.6301718354225159, "learning_rate": 9.983602825724303e-05, "loss": 0.0372, "step": 22350 }, { "epoch": 14.251115360101975, "grad_norm": 0.5594810843467712, "learning_rate": 9.983558195620424e-05, "loss": 0.0468, "step": 22360 }, { "epoch": 14.25748884639898, "grad_norm": 0.6863843202590942, "learning_rate": 9.983513504961608e-05, "loss": 0.0383, "step": 22370 }, { "epoch": 14.263862332695984, "grad_norm": 0.9575958847999573, "learning_rate": 9.983468753748397e-05, "loss": 0.0587, "step": 22380 }, { "epoch": 14.270235818992989, "grad_norm": 0.40769726037979126, "learning_rate": 9.983423941981338e-05, "loss": 0.0341, "step": 22390 }, { "epoch": 14.276609305289993, "grad_norm": 0.39516618847846985, "learning_rate": 9.983379069660971e-05, "loss": 0.0411, "step": 22400 }, { "epoch": 14.282982791586997, "grad_norm": 0.6157066822052002, "learning_rate": 9.983334136787843e-05, "loss": 0.0678, "step": 22410 }, { "epoch": 14.289356277884002, "grad_norm": 0.583670973777771, "learning_rate": 9.983289143362503e-05, "loss": 0.0547, "step": 22420 }, { "epoch": 14.295729764181006, "grad_norm": 0.8311648368835449, "learning_rate": 9.98324408938549e-05, "loss": 0.0499, "step": 22430 }, { "epoch": 14.302103250478012, "grad_norm": 0.6779030561447144, "learning_rate": 9.98319897485736e-05, "loss": 0.0453, "step": 22440 }, { "epoch": 14.308476736775017, "grad_norm": 0.6761898398399353, "learning_rate": 9.983153799778658e-05, "loss": 0.0516, "step": 22450 }, { "epoch": 14.314850223072021, "grad_norm": 0.42458826303482056, "learning_rate": 9.98310856414993e-05, "loss": 0.0404, "step": 22460 }, { "epoch": 14.321223709369026, "grad_norm": 0.464655339717865, "learning_rate": 9.983063267971729e-05, "loss": 0.0415, "step": 22470 }, { "epoch": 14.32759719566603, "grad_norm": 0.5490202307701111, "learning_rate": 9.983017911244604e-05, "loss": 0.0465, "step": 22480 }, { "epoch": 14.333970681963034, "grad_norm": 0.6828319430351257, "learning_rate": 9.982972493969108e-05, "loss": 0.0536, "step": 22490 }, { "epoch": 14.340344168260039, "grad_norm": 0.9010695219039917, "learning_rate": 9.982927016145791e-05, "loss": 0.0432, "step": 22500 }, { "epoch": 14.346717654557043, "grad_norm": 0.6490613222122192, "learning_rate": 9.982881477775206e-05, "loss": 0.0412, "step": 22510 }, { "epoch": 14.353091140854048, "grad_norm": 0.5261130332946777, "learning_rate": 9.982835878857905e-05, "loss": 0.0377, "step": 22520 }, { "epoch": 14.359464627151052, "grad_norm": 0.2353847175836563, "learning_rate": 9.982790219394445e-05, "loss": 0.0595, "step": 22530 }, { "epoch": 14.365838113448056, "grad_norm": 0.39870527386665344, "learning_rate": 9.98274449938538e-05, "loss": 0.0473, "step": 22540 }, { "epoch": 14.37221159974506, "grad_norm": 0.8602566719055176, "learning_rate": 9.982698718831262e-05, "loss": 0.0481, "step": 22550 }, { "epoch": 14.378585086042065, "grad_norm": 0.5441707968711853, "learning_rate": 9.982652877732653e-05, "loss": 0.0464, "step": 22560 }, { "epoch": 14.38495857233907, "grad_norm": 0.42394423484802246, "learning_rate": 9.982606976090107e-05, "loss": 0.0494, "step": 22570 }, { "epoch": 14.391332058636074, "grad_norm": 0.5326185822486877, "learning_rate": 9.98256101390418e-05, "loss": 0.0399, "step": 22580 }, { "epoch": 14.397705544933078, "grad_norm": 0.33314624428749084, "learning_rate": 9.982514991175433e-05, "loss": 0.0426, "step": 22590 }, { "epoch": 14.404079031230083, "grad_norm": 0.8561475872993469, "learning_rate": 9.982468907904426e-05, "loss": 0.0408, "step": 22600 }, { "epoch": 14.410452517527087, "grad_norm": 0.635107696056366, "learning_rate": 9.982422764091717e-05, "loss": 0.0335, "step": 22610 }, { "epoch": 14.416826003824092, "grad_norm": 0.4229608178138733, "learning_rate": 9.982376559737866e-05, "loss": 0.049, "step": 22620 }, { "epoch": 14.423199490121096, "grad_norm": 0.6325505971908569, "learning_rate": 9.982330294843436e-05, "loss": 0.0345, "step": 22630 }, { "epoch": 14.4295729764181, "grad_norm": 0.4519380033016205, "learning_rate": 9.98228396940899e-05, "loss": 0.0695, "step": 22640 }, { "epoch": 14.435946462715105, "grad_norm": 0.5471426248550415, "learning_rate": 9.982237583435089e-05, "loss": 0.0514, "step": 22650 }, { "epoch": 14.44231994901211, "grad_norm": 0.5457211136817932, "learning_rate": 9.982191136922296e-05, "loss": 0.0504, "step": 22660 }, { "epoch": 14.448693435309114, "grad_norm": 0.6787260174751282, "learning_rate": 9.982144629871179e-05, "loss": 0.0448, "step": 22670 }, { "epoch": 14.455066921606118, "grad_norm": 0.46625715494155884, "learning_rate": 9.982098062282299e-05, "loss": 0.0491, "step": 22680 }, { "epoch": 14.461440407903122, "grad_norm": 1.2135145664215088, "learning_rate": 9.982051434156225e-05, "loss": 0.0526, "step": 22690 }, { "epoch": 14.467813894200127, "grad_norm": 0.8382881283760071, "learning_rate": 9.982004745493523e-05, "loss": 0.0506, "step": 22700 }, { "epoch": 14.474187380497131, "grad_norm": 0.5003844499588013, "learning_rate": 9.981957996294757e-05, "loss": 0.0487, "step": 22710 }, { "epoch": 14.480560866794136, "grad_norm": 0.45827680826187134, "learning_rate": 9.9819111865605e-05, "loss": 0.0541, "step": 22720 }, { "epoch": 14.48693435309114, "grad_norm": 0.9332799911499023, "learning_rate": 9.981864316291316e-05, "loss": 0.0494, "step": 22730 }, { "epoch": 14.493307839388144, "grad_norm": 0.7069278359413147, "learning_rate": 9.98181738548778e-05, "loss": 0.0446, "step": 22740 }, { "epoch": 14.499681325685149, "grad_norm": 0.630792498588562, "learning_rate": 9.981770394150456e-05, "loss": 0.0468, "step": 22750 }, { "epoch": 14.506054811982153, "grad_norm": 0.5244588851928711, "learning_rate": 9.98172334227992e-05, "loss": 0.066, "step": 22760 }, { "epoch": 14.51242829827916, "grad_norm": 0.42685291171073914, "learning_rate": 9.981676229876741e-05, "loss": 0.0488, "step": 22770 }, { "epoch": 14.518801784576164, "grad_norm": 0.778305172920227, "learning_rate": 9.981629056941493e-05, "loss": 0.0507, "step": 22780 }, { "epoch": 14.525175270873168, "grad_norm": 0.622222900390625, "learning_rate": 9.981581823474749e-05, "loss": 0.042, "step": 22790 }, { "epoch": 14.531548757170173, "grad_norm": 0.3114643692970276, "learning_rate": 9.98153452947708e-05, "loss": 0.038, "step": 22800 }, { "epoch": 14.537922243467177, "grad_norm": 0.434457391500473, "learning_rate": 9.981487174949065e-05, "loss": 0.0376, "step": 22810 }, { "epoch": 14.544295729764182, "grad_norm": 0.4828667938709259, "learning_rate": 9.981439759891278e-05, "loss": 0.0571, "step": 22820 }, { "epoch": 14.550669216061186, "grad_norm": 0.5315593481063843, "learning_rate": 9.981392284304292e-05, "loss": 0.0456, "step": 22830 }, { "epoch": 14.55704270235819, "grad_norm": 0.5096555948257446, "learning_rate": 9.98134474818869e-05, "loss": 0.0476, "step": 22840 }, { "epoch": 14.563416188655195, "grad_norm": 0.6486849188804626, "learning_rate": 9.981297151545044e-05, "loss": 0.0625, "step": 22850 }, { "epoch": 14.5697896749522, "grad_norm": 0.7857367992401123, "learning_rate": 9.981249494373935e-05, "loss": 0.0538, "step": 22860 }, { "epoch": 14.576163161249204, "grad_norm": 0.6177077889442444, "learning_rate": 9.981201776675942e-05, "loss": 0.0403, "step": 22870 }, { "epoch": 14.582536647546208, "grad_norm": 0.601658284664154, "learning_rate": 9.981153998451645e-05, "loss": 0.0544, "step": 22880 }, { "epoch": 14.588910133843212, "grad_norm": 0.80660480260849, "learning_rate": 9.981106159701622e-05, "loss": 0.0541, "step": 22890 }, { "epoch": 14.595283620140217, "grad_norm": 0.5773524641990662, "learning_rate": 9.981058260426456e-05, "loss": 0.0586, "step": 22900 }, { "epoch": 14.601657106437221, "grad_norm": 0.5855287313461304, "learning_rate": 9.981010300626732e-05, "loss": 0.0657, "step": 22910 }, { "epoch": 14.608030592734226, "grad_norm": 0.690717875957489, "learning_rate": 9.980962280303028e-05, "loss": 0.0527, "step": 22920 }, { "epoch": 14.61440407903123, "grad_norm": 0.7315030097961426, "learning_rate": 9.980914199455929e-05, "loss": 0.0493, "step": 22930 }, { "epoch": 14.620777565328234, "grad_norm": 0.5379112958908081, "learning_rate": 9.980866058086021e-05, "loss": 0.0535, "step": 22940 }, { "epoch": 14.627151051625239, "grad_norm": 0.5226651430130005, "learning_rate": 9.980817856193887e-05, "loss": 0.0413, "step": 22950 }, { "epoch": 14.633524537922243, "grad_norm": 0.7389031052589417, "learning_rate": 9.980769593780115e-05, "loss": 0.0648, "step": 22960 }, { "epoch": 14.639898024219248, "grad_norm": 0.7381865978240967, "learning_rate": 9.980721270845288e-05, "loss": 0.0463, "step": 22970 }, { "epoch": 14.646271510516252, "grad_norm": 0.4846133887767792, "learning_rate": 9.980672887389994e-05, "loss": 0.0415, "step": 22980 }, { "epoch": 14.652644996813256, "grad_norm": 0.48646196722984314, "learning_rate": 9.980624443414824e-05, "loss": 0.0464, "step": 22990 }, { "epoch": 14.65901848311026, "grad_norm": 0.8300036191940308, "learning_rate": 9.980575938920364e-05, "loss": 0.0415, "step": 23000 }, { "epoch": 14.665391969407265, "grad_norm": 0.5487350225448608, "learning_rate": 9.980527373907204e-05, "loss": 0.0385, "step": 23010 }, { "epoch": 14.67176545570427, "grad_norm": 0.5324499011039734, "learning_rate": 9.980478748375935e-05, "loss": 0.0383, "step": 23020 }, { "epoch": 14.678138942001274, "grad_norm": 0.7573229074478149, "learning_rate": 9.980430062327144e-05, "loss": 0.0423, "step": 23030 }, { "epoch": 14.684512428298278, "grad_norm": 0.6033254861831665, "learning_rate": 9.980381315761426e-05, "loss": 0.0513, "step": 23040 }, { "epoch": 14.690885914595283, "grad_norm": 1.2765908241271973, "learning_rate": 9.980332508679375e-05, "loss": 0.0589, "step": 23050 }, { "epoch": 14.697259400892289, "grad_norm": 0.42224645614624023, "learning_rate": 9.98028364108158e-05, "loss": 0.0374, "step": 23060 }, { "epoch": 14.703632887189293, "grad_norm": 0.4443477392196655, "learning_rate": 9.980234712968638e-05, "loss": 0.0437, "step": 23070 }, { "epoch": 14.710006373486298, "grad_norm": 0.44286397099494934, "learning_rate": 9.980185724341142e-05, "loss": 0.0426, "step": 23080 }, { "epoch": 14.716379859783302, "grad_norm": 0.3904381990432739, "learning_rate": 9.980136675199686e-05, "loss": 0.0417, "step": 23090 }, { "epoch": 14.722753346080307, "grad_norm": 0.5800367593765259, "learning_rate": 9.980087565544869e-05, "loss": 0.0498, "step": 23100 }, { "epoch": 14.729126832377311, "grad_norm": 0.6309531927108765, "learning_rate": 9.980038395377284e-05, "loss": 0.0576, "step": 23110 }, { "epoch": 14.735500318674315, "grad_norm": 0.5551360249519348, "learning_rate": 9.979989164697531e-05, "loss": 0.034, "step": 23120 }, { "epoch": 14.74187380497132, "grad_norm": 0.4391493797302246, "learning_rate": 9.97993987350621e-05, "loss": 0.0439, "step": 23130 }, { "epoch": 14.748247291268324, "grad_norm": 0.4850173592567444, "learning_rate": 9.979890521803917e-05, "loss": 0.0336, "step": 23140 }, { "epoch": 14.754620777565329, "grad_norm": 0.7128845453262329, "learning_rate": 9.97984110959125e-05, "loss": 0.0627, "step": 23150 }, { "epoch": 14.760994263862333, "grad_norm": 0.6060341000556946, "learning_rate": 9.979791636868813e-05, "loss": 0.0537, "step": 23160 }, { "epoch": 14.767367750159337, "grad_norm": 1.037618637084961, "learning_rate": 9.979742103637206e-05, "loss": 0.0354, "step": 23170 }, { "epoch": 14.773741236456342, "grad_norm": 0.6059609651565552, "learning_rate": 9.979692509897032e-05, "loss": 0.0477, "step": 23180 }, { "epoch": 14.780114722753346, "grad_norm": 1.1166725158691406, "learning_rate": 9.979642855648893e-05, "loss": 0.0542, "step": 23190 }, { "epoch": 14.78648820905035, "grad_norm": 0.3213508427143097, "learning_rate": 9.979593140893388e-05, "loss": 0.0393, "step": 23200 }, { "epoch": 14.792861695347355, "grad_norm": 0.3468857407569885, "learning_rate": 9.979543365631128e-05, "loss": 0.0438, "step": 23210 }, { "epoch": 14.79923518164436, "grad_norm": 0.6471170783042908, "learning_rate": 9.979493529862714e-05, "loss": 0.043, "step": 23220 }, { "epoch": 14.805608667941364, "grad_norm": 0.5723495483398438, "learning_rate": 9.979443633588749e-05, "loss": 0.0412, "step": 23230 }, { "epoch": 14.811982154238368, "grad_norm": 0.6439129710197449, "learning_rate": 9.979393676809847e-05, "loss": 0.0453, "step": 23240 }, { "epoch": 14.818355640535373, "grad_norm": 0.4173094928264618, "learning_rate": 9.979343659526608e-05, "loss": 0.0606, "step": 23250 }, { "epoch": 14.824729126832377, "grad_norm": 0.5355782508850098, "learning_rate": 9.979293581739644e-05, "loss": 0.0405, "step": 23260 }, { "epoch": 14.831102613129381, "grad_norm": 0.5931435227394104, "learning_rate": 9.97924344344956e-05, "loss": 0.0375, "step": 23270 }, { "epoch": 14.837476099426386, "grad_norm": 0.5154598951339722, "learning_rate": 9.979193244656968e-05, "loss": 0.0489, "step": 23280 }, { "epoch": 14.84384958572339, "grad_norm": 0.9111235737800598, "learning_rate": 9.979142985362478e-05, "loss": 0.0434, "step": 23290 }, { "epoch": 14.850223072020395, "grad_norm": 0.8295645713806152, "learning_rate": 9.979092665566697e-05, "loss": 0.0525, "step": 23300 }, { "epoch": 14.8565965583174, "grad_norm": 0.38093316555023193, "learning_rate": 9.97904228527024e-05, "loss": 0.0491, "step": 23310 }, { "epoch": 14.862970044614404, "grad_norm": 0.7697515487670898, "learning_rate": 9.978991844473719e-05, "loss": 0.0538, "step": 23320 }, { "epoch": 14.869343530911408, "grad_norm": 0.5359776616096497, "learning_rate": 9.978941343177746e-05, "loss": 0.0512, "step": 23330 }, { "epoch": 14.875717017208412, "grad_norm": 0.632086992263794, "learning_rate": 9.978890781382935e-05, "loss": 0.0535, "step": 23340 }, { "epoch": 14.882090503505417, "grad_norm": 0.40464523434638977, "learning_rate": 9.978840159089899e-05, "loss": 0.0458, "step": 23350 }, { "epoch": 14.888463989802421, "grad_norm": 0.3309280276298523, "learning_rate": 9.978789476299255e-05, "loss": 0.0448, "step": 23360 }, { "epoch": 14.894837476099426, "grad_norm": 0.4444672167301178, "learning_rate": 9.97873873301162e-05, "loss": 0.0565, "step": 23370 }, { "epoch": 14.90121096239643, "grad_norm": 0.5160964727401733, "learning_rate": 9.978687929227606e-05, "loss": 0.0502, "step": 23380 }, { "epoch": 14.907584448693436, "grad_norm": 0.5577272176742554, "learning_rate": 9.978637064947834e-05, "loss": 0.0502, "step": 23390 }, { "epoch": 14.91395793499044, "grad_norm": 0.7256981134414673, "learning_rate": 9.97858614017292e-05, "loss": 0.0492, "step": 23400 }, { "epoch": 14.920331421287445, "grad_norm": 0.6995009183883667, "learning_rate": 9.978535154903484e-05, "loss": 0.0585, "step": 23410 }, { "epoch": 14.92670490758445, "grad_norm": 0.5200843214988708, "learning_rate": 9.978484109140147e-05, "loss": 0.0391, "step": 23420 }, { "epoch": 14.933078393881454, "grad_norm": 0.6342918872833252, "learning_rate": 9.978433002883527e-05, "loss": 0.042, "step": 23430 }, { "epoch": 14.939451880178458, "grad_norm": 0.45928239822387695, "learning_rate": 9.978381836134244e-05, "loss": 0.0388, "step": 23440 }, { "epoch": 14.945825366475463, "grad_norm": 0.45357558131217957, "learning_rate": 9.978330608892923e-05, "loss": 0.0376, "step": 23450 }, { "epoch": 14.952198852772467, "grad_norm": 0.5008507370948792, "learning_rate": 9.978279321160184e-05, "loss": 0.0428, "step": 23460 }, { "epoch": 14.958572339069471, "grad_norm": 0.5462914705276489, "learning_rate": 9.978227972936652e-05, "loss": 0.0446, "step": 23470 }, { "epoch": 14.964945825366476, "grad_norm": 0.9730573892593384, "learning_rate": 9.978176564222948e-05, "loss": 0.0544, "step": 23480 }, { "epoch": 14.97131931166348, "grad_norm": 0.37451326847076416, "learning_rate": 9.978125095019701e-05, "loss": 0.0389, "step": 23490 }, { "epoch": 14.977692797960485, "grad_norm": 0.37431076169013977, "learning_rate": 9.978073565327533e-05, "loss": 0.0432, "step": 23500 }, { "epoch": 14.984066284257489, "grad_norm": 0.49148106575012207, "learning_rate": 9.978021975147069e-05, "loss": 0.0635, "step": 23510 }, { "epoch": 14.990439770554493, "grad_norm": 0.4281140863895416, "learning_rate": 9.97797032447894e-05, "loss": 0.0526, "step": 23520 }, { "epoch": 14.996813256851498, "grad_norm": 0.6121013164520264, "learning_rate": 9.977918613323771e-05, "loss": 0.0539, "step": 23530 }, { "epoch": 15.003186743148502, "grad_norm": 0.36562085151672363, "learning_rate": 9.977866841682192e-05, "loss": 0.0473, "step": 23540 }, { "epoch": 15.009560229445507, "grad_norm": 0.4976012110710144, "learning_rate": 9.97781500955483e-05, "loss": 0.0468, "step": 23550 }, { "epoch": 15.015933715742511, "grad_norm": 0.6325187683105469, "learning_rate": 9.977763116942315e-05, "loss": 0.0391, "step": 23560 }, { "epoch": 15.022307202039515, "grad_norm": 0.3797138035297394, "learning_rate": 9.97771116384528e-05, "loss": 0.0455, "step": 23570 }, { "epoch": 15.02868068833652, "grad_norm": 0.4993521273136139, "learning_rate": 9.977659150264355e-05, "loss": 0.0347, "step": 23580 }, { "epoch": 15.035054174633524, "grad_norm": 1.0263770818710327, "learning_rate": 9.97760707620017e-05, "loss": 0.0446, "step": 23590 }, { "epoch": 15.041427660930529, "grad_norm": 0.5573161244392395, "learning_rate": 9.97755494165336e-05, "loss": 0.0519, "step": 23600 }, { "epoch": 15.047801147227533, "grad_norm": 0.923227846622467, "learning_rate": 9.977502746624559e-05, "loss": 0.0532, "step": 23610 }, { "epoch": 15.054174633524537, "grad_norm": 0.9130227565765381, "learning_rate": 9.977450491114398e-05, "loss": 0.041, "step": 23620 }, { "epoch": 15.060548119821542, "grad_norm": 0.36806720495224, "learning_rate": 9.977398175123515e-05, "loss": 0.0395, "step": 23630 }, { "epoch": 15.066921606118546, "grad_norm": 0.4509584903717041, "learning_rate": 9.977345798652545e-05, "loss": 0.04, "step": 23640 }, { "epoch": 15.07329509241555, "grad_norm": 0.7122308611869812, "learning_rate": 9.977293361702125e-05, "loss": 0.0573, "step": 23650 }, { "epoch": 15.079668578712555, "grad_norm": 0.410665363073349, "learning_rate": 9.97724086427289e-05, "loss": 0.0452, "step": 23660 }, { "epoch": 15.08604206500956, "grad_norm": 0.3860701024532318, "learning_rate": 9.97718830636548e-05, "loss": 0.0413, "step": 23670 }, { "epoch": 15.092415551306564, "grad_norm": 0.33533331751823425, "learning_rate": 9.977135687980531e-05, "loss": 0.0471, "step": 23680 }, { "epoch": 15.098789037603568, "grad_norm": 0.5211670994758606, "learning_rate": 9.977083009118686e-05, "loss": 0.051, "step": 23690 }, { "epoch": 15.105162523900574, "grad_norm": 0.3898063004016876, "learning_rate": 9.977030269780583e-05, "loss": 0.0326, "step": 23700 }, { "epoch": 15.111536010197579, "grad_norm": 0.42077144980430603, "learning_rate": 9.976977469966862e-05, "loss": 0.0427, "step": 23710 }, { "epoch": 15.117909496494583, "grad_norm": 0.4900611340999603, "learning_rate": 9.976924609678167e-05, "loss": 0.0383, "step": 23720 }, { "epoch": 15.124282982791588, "grad_norm": 0.49385595321655273, "learning_rate": 9.976871688915139e-05, "loss": 0.0364, "step": 23730 }, { "epoch": 15.130656469088592, "grad_norm": 0.587699830532074, "learning_rate": 9.97681870767842e-05, "loss": 0.053, "step": 23740 }, { "epoch": 15.137029955385596, "grad_norm": 0.6034412384033203, "learning_rate": 9.976765665968654e-05, "loss": 0.0388, "step": 23750 }, { "epoch": 15.1434034416826, "grad_norm": 0.5031957030296326, "learning_rate": 9.97671256378649e-05, "loss": 0.0479, "step": 23760 }, { "epoch": 15.149776927979605, "grad_norm": 0.44968873262405396, "learning_rate": 9.976659401132566e-05, "loss": 0.0414, "step": 23770 }, { "epoch": 15.15615041427661, "grad_norm": 0.4617210030555725, "learning_rate": 9.976606178007532e-05, "loss": 0.0368, "step": 23780 }, { "epoch": 15.162523900573614, "grad_norm": 1.0046274662017822, "learning_rate": 9.976552894412037e-05, "loss": 0.0488, "step": 23790 }, { "epoch": 15.168897386870619, "grad_norm": 0.665860116481781, "learning_rate": 9.976499550346723e-05, "loss": 0.0443, "step": 23800 }, { "epoch": 15.175270873167623, "grad_norm": 0.5657275915145874, "learning_rate": 9.976446145812242e-05, "loss": 0.0307, "step": 23810 }, { "epoch": 15.181644359464627, "grad_norm": 0.4063440263271332, "learning_rate": 9.976392680809242e-05, "loss": 0.0431, "step": 23820 }, { "epoch": 15.188017845761632, "grad_norm": 0.5577939748764038, "learning_rate": 9.976339155338371e-05, "loss": 0.0368, "step": 23830 }, { "epoch": 15.194391332058636, "grad_norm": 0.4961129426956177, "learning_rate": 9.976285569400281e-05, "loss": 0.0439, "step": 23840 }, { "epoch": 15.20076481835564, "grad_norm": 0.5882431864738464, "learning_rate": 9.976231922995624e-05, "loss": 0.0441, "step": 23850 }, { "epoch": 15.207138304652645, "grad_norm": 0.8021961450576782, "learning_rate": 9.97617821612505e-05, "loss": 0.0448, "step": 23860 }, { "epoch": 15.21351179094965, "grad_norm": 0.7422100901603699, "learning_rate": 9.976124448789214e-05, "loss": 0.0447, "step": 23870 }, { "epoch": 15.219885277246654, "grad_norm": 0.5307670831680298, "learning_rate": 9.976070620988767e-05, "loss": 0.0415, "step": 23880 }, { "epoch": 15.226258763543658, "grad_norm": 0.5787435173988342, "learning_rate": 9.976016732724362e-05, "loss": 0.0328, "step": 23890 }, { "epoch": 15.232632249840663, "grad_norm": 0.6310264468193054, "learning_rate": 9.975962783996657e-05, "loss": 0.0503, "step": 23900 }, { "epoch": 15.239005736137667, "grad_norm": 0.6409702301025391, "learning_rate": 9.975908774806307e-05, "loss": 0.0436, "step": 23910 }, { "epoch": 15.245379222434671, "grad_norm": 0.7289593815803528, "learning_rate": 9.975854705153965e-05, "loss": 0.0417, "step": 23920 }, { "epoch": 15.251752708731676, "grad_norm": 0.5920099020004272, "learning_rate": 9.975800575040291e-05, "loss": 0.0494, "step": 23930 }, { "epoch": 15.25812619502868, "grad_norm": 0.42223116755485535, "learning_rate": 9.975746384465943e-05, "loss": 0.0367, "step": 23940 }, { "epoch": 15.264499681325685, "grad_norm": 0.3632909655570984, "learning_rate": 9.975692133431579e-05, "loss": 0.0353, "step": 23950 }, { "epoch": 15.270873167622689, "grad_norm": 0.546451985836029, "learning_rate": 9.975637821937857e-05, "loss": 0.0431, "step": 23960 }, { "epoch": 15.277246653919693, "grad_norm": 0.5947668552398682, "learning_rate": 9.975583449985437e-05, "loss": 0.0534, "step": 23970 }, { "epoch": 15.283620140216698, "grad_norm": 0.8924487233161926, "learning_rate": 9.97552901757498e-05, "loss": 0.0435, "step": 23980 }, { "epoch": 15.289993626513702, "grad_norm": 0.5419203042984009, "learning_rate": 9.97547452470715e-05, "loss": 0.0456, "step": 23990 }, { "epoch": 15.296367112810707, "grad_norm": 0.46796733140945435, "learning_rate": 9.975419971382606e-05, "loss": 0.0476, "step": 24000 }, { "epoch": 15.302740599107711, "grad_norm": 0.5434220433235168, "learning_rate": 9.97536535760201e-05, "loss": 0.0372, "step": 24010 }, { "epoch": 15.309114085404717, "grad_norm": 0.5641089081764221, "learning_rate": 9.97531068336603e-05, "loss": 0.0387, "step": 24020 }, { "epoch": 15.315487571701722, "grad_norm": 0.48083871603012085, "learning_rate": 9.975255948675326e-05, "loss": 0.0379, "step": 24030 }, { "epoch": 15.321861057998726, "grad_norm": 0.5026900172233582, "learning_rate": 9.975201153530566e-05, "loss": 0.0353, "step": 24040 }, { "epoch": 15.32823454429573, "grad_norm": 0.5611559152603149, "learning_rate": 9.975146297932414e-05, "loss": 0.054, "step": 24050 }, { "epoch": 15.334608030592735, "grad_norm": 0.4676796495914459, "learning_rate": 9.975091381881536e-05, "loss": 0.0449, "step": 24060 }, { "epoch": 15.34098151688974, "grad_norm": 0.5819118022918701, "learning_rate": 9.975036405378602e-05, "loss": 0.0488, "step": 24070 }, { "epoch": 15.347355003186744, "grad_norm": 0.870334804058075, "learning_rate": 9.974981368424278e-05, "loss": 0.047, "step": 24080 }, { "epoch": 15.353728489483748, "grad_norm": 0.3387395441532135, "learning_rate": 9.974926271019234e-05, "loss": 0.0528, "step": 24090 }, { "epoch": 15.360101975780752, "grad_norm": 0.4962310194969177, "learning_rate": 9.974871113164137e-05, "loss": 0.0517, "step": 24100 }, { "epoch": 15.366475462077757, "grad_norm": 0.7542265057563782, "learning_rate": 9.97481589485966e-05, "loss": 0.0435, "step": 24110 }, { "epoch": 15.372848948374761, "grad_norm": 0.8023645281791687, "learning_rate": 9.974760616106473e-05, "loss": 0.0349, "step": 24120 }, { "epoch": 15.379222434671766, "grad_norm": 0.6609359383583069, "learning_rate": 9.974705276905248e-05, "loss": 0.055, "step": 24130 }, { "epoch": 15.38559592096877, "grad_norm": 0.4856813848018646, "learning_rate": 9.974649877256656e-05, "loss": 0.041, "step": 24140 }, { "epoch": 15.391969407265774, "grad_norm": 0.4110677242279053, "learning_rate": 9.974594417161371e-05, "loss": 0.0521, "step": 24150 }, { "epoch": 15.398342893562779, "grad_norm": 0.5060667395591736, "learning_rate": 9.974538896620067e-05, "loss": 0.0448, "step": 24160 }, { "epoch": 15.404716379859783, "grad_norm": 0.3165733516216278, "learning_rate": 9.97448331563342e-05, "loss": 0.0339, "step": 24170 }, { "epoch": 15.411089866156788, "grad_norm": 0.39315980672836304, "learning_rate": 9.974427674202102e-05, "loss": 0.0491, "step": 24180 }, { "epoch": 15.417463352453792, "grad_norm": 0.6105425953865051, "learning_rate": 9.974371972326792e-05, "loss": 0.0387, "step": 24190 }, { "epoch": 15.423836838750796, "grad_norm": 0.44018280506134033, "learning_rate": 9.974316210008167e-05, "loss": 0.0332, "step": 24200 }, { "epoch": 15.4302103250478, "grad_norm": 0.6559227705001831, "learning_rate": 9.974260387246903e-05, "loss": 0.0459, "step": 24210 }, { "epoch": 15.436583811344805, "grad_norm": 0.6357271671295166, "learning_rate": 9.974204504043679e-05, "loss": 0.0457, "step": 24220 }, { "epoch": 15.44295729764181, "grad_norm": 0.34703901410102844, "learning_rate": 9.974148560399173e-05, "loss": 0.0412, "step": 24230 }, { "epoch": 15.449330783938814, "grad_norm": 0.37697798013687134, "learning_rate": 9.974092556314065e-05, "loss": 0.0466, "step": 24240 }, { "epoch": 15.455704270235818, "grad_norm": 0.8878664970397949, "learning_rate": 9.974036491789038e-05, "loss": 0.045, "step": 24250 }, { "epoch": 15.462077756532823, "grad_norm": 0.44537651538848877, "learning_rate": 9.973980366824772e-05, "loss": 0.0466, "step": 24260 }, { "epoch": 15.468451242829827, "grad_norm": 0.42828378081321716, "learning_rate": 9.973924181421945e-05, "loss": 0.0437, "step": 24270 }, { "epoch": 15.474824729126832, "grad_norm": 0.4945051074028015, "learning_rate": 9.973867935581245e-05, "loss": 0.0473, "step": 24280 }, { "epoch": 15.481198215423836, "grad_norm": 0.4891318380832672, "learning_rate": 9.973811629303354e-05, "loss": 0.0377, "step": 24290 }, { "epoch": 15.48757170172084, "grad_norm": 0.4379228353500366, "learning_rate": 9.973755262588955e-05, "loss": 0.058, "step": 24300 }, { "epoch": 15.493945188017845, "grad_norm": 0.5696465373039246, "learning_rate": 9.973698835438733e-05, "loss": 0.0434, "step": 24310 }, { "epoch": 15.500318674314851, "grad_norm": 0.40631696581840515, "learning_rate": 9.973642347853374e-05, "loss": 0.0478, "step": 24320 }, { "epoch": 15.506692160611856, "grad_norm": 0.26752984523773193, "learning_rate": 9.973585799833568e-05, "loss": 0.0517, "step": 24330 }, { "epoch": 15.51306564690886, "grad_norm": 0.9819962382316589, "learning_rate": 9.973529191379993e-05, "loss": 0.0402, "step": 24340 }, { "epoch": 15.519439133205864, "grad_norm": 0.5621557235717773, "learning_rate": 9.973472522493347e-05, "loss": 0.0516, "step": 24350 }, { "epoch": 15.525812619502869, "grad_norm": 0.5048269629478455, "learning_rate": 9.973415793174313e-05, "loss": 0.0451, "step": 24360 }, { "epoch": 15.532186105799873, "grad_norm": 0.5996031761169434, "learning_rate": 9.97335900342358e-05, "loss": 0.0364, "step": 24370 }, { "epoch": 15.538559592096878, "grad_norm": 0.5816421508789062, "learning_rate": 9.973302153241841e-05, "loss": 0.0367, "step": 24380 }, { "epoch": 15.544933078393882, "grad_norm": 0.3382156193256378, "learning_rate": 9.973245242629786e-05, "loss": 0.0352, "step": 24390 }, { "epoch": 15.551306564690886, "grad_norm": 0.5436844229698181, "learning_rate": 9.973188271588103e-05, "loss": 0.0402, "step": 24400 }, { "epoch": 15.55768005098789, "grad_norm": 0.5301566123962402, "learning_rate": 9.973131240117489e-05, "loss": 0.0482, "step": 24410 }, { "epoch": 15.564053537284895, "grad_norm": 0.7537147998809814, "learning_rate": 9.973074148218636e-05, "loss": 0.0433, "step": 24420 }, { "epoch": 15.5704270235819, "grad_norm": 0.5716974139213562, "learning_rate": 9.973016995892234e-05, "loss": 0.0534, "step": 24430 }, { "epoch": 15.576800509878904, "grad_norm": 0.5674687623977661, "learning_rate": 9.972959783138981e-05, "loss": 0.043, "step": 24440 }, { "epoch": 15.583173996175908, "grad_norm": 0.8504915237426758, "learning_rate": 9.972902509959572e-05, "loss": 0.0438, "step": 24450 }, { "epoch": 15.589547482472913, "grad_norm": 0.493272066116333, "learning_rate": 9.972845176354703e-05, "loss": 0.0493, "step": 24460 }, { "epoch": 15.595920968769917, "grad_norm": 0.683525800704956, "learning_rate": 9.972787782325068e-05, "loss": 0.0425, "step": 24470 }, { "epoch": 15.602294455066922, "grad_norm": 0.32319724559783936, "learning_rate": 9.97273032787137e-05, "loss": 0.0404, "step": 24480 }, { "epoch": 15.608667941363926, "grad_norm": 0.6681970357894897, "learning_rate": 9.972672812994299e-05, "loss": 0.0476, "step": 24490 }, { "epoch": 15.61504142766093, "grad_norm": 0.7049742341041565, "learning_rate": 9.97261523769456e-05, "loss": 0.0438, "step": 24500 }, { "epoch": 15.621414913957935, "grad_norm": 0.34148627519607544, "learning_rate": 9.972557601972852e-05, "loss": 0.0418, "step": 24510 }, { "epoch": 15.62778840025494, "grad_norm": 0.4473116397857666, "learning_rate": 9.972499905829875e-05, "loss": 0.0486, "step": 24520 }, { "epoch": 15.634161886551944, "grad_norm": 0.4250977635383606, "learning_rate": 9.972442149266326e-05, "loss": 0.0367, "step": 24530 }, { "epoch": 15.640535372848948, "grad_norm": 0.3553280830383301, "learning_rate": 9.972384332282914e-05, "loss": 0.0475, "step": 24540 }, { "epoch": 15.646908859145952, "grad_norm": 0.4400824308395386, "learning_rate": 9.972326454880334e-05, "loss": 0.0366, "step": 24550 }, { "epoch": 15.653282345442957, "grad_norm": 0.49200350046157837, "learning_rate": 9.972268517059296e-05, "loss": 0.0407, "step": 24560 }, { "epoch": 15.659655831739961, "grad_norm": 0.4620760679244995, "learning_rate": 9.9722105188205e-05, "loss": 0.0362, "step": 24570 }, { "epoch": 15.666029318036966, "grad_norm": 0.36829373240470886, "learning_rate": 9.97215246016465e-05, "loss": 0.034, "step": 24580 }, { "epoch": 15.67240280433397, "grad_norm": 0.8021905422210693, "learning_rate": 9.972094341092455e-05, "loss": 0.0472, "step": 24590 }, { "epoch": 15.678776290630974, "grad_norm": 0.2785084545612335, "learning_rate": 9.972036161604618e-05, "loss": 0.0683, "step": 24600 }, { "epoch": 15.685149776927979, "grad_norm": 0.35500892996788025, "learning_rate": 9.971977921701848e-05, "loss": 0.0422, "step": 24610 }, { "epoch": 15.691523263224983, "grad_norm": 1.3361483812332153, "learning_rate": 9.971919621384851e-05, "loss": 0.049, "step": 24620 }, { "epoch": 15.697896749521988, "grad_norm": 0.5870903134346008, "learning_rate": 9.971861260654338e-05, "loss": 0.0439, "step": 24630 }, { "epoch": 15.704270235818992, "grad_norm": 0.7056443691253662, "learning_rate": 9.971802839511015e-05, "loss": 0.0565, "step": 24640 }, { "epoch": 15.710643722115998, "grad_norm": 0.4484405517578125, "learning_rate": 9.971744357955595e-05, "loss": 0.037, "step": 24650 }, { "epoch": 15.717017208413003, "grad_norm": 0.5729345083236694, "learning_rate": 9.971685815988786e-05, "loss": 0.0411, "step": 24660 }, { "epoch": 15.723390694710007, "grad_norm": 0.5862831473350525, "learning_rate": 9.9716272136113e-05, "loss": 0.0384, "step": 24670 }, { "epoch": 15.729764181007011, "grad_norm": 0.27969372272491455, "learning_rate": 9.971568550823849e-05, "loss": 0.0464, "step": 24680 }, { "epoch": 15.736137667304016, "grad_norm": 0.6362307667732239, "learning_rate": 9.971509827627147e-05, "loss": 0.0463, "step": 24690 }, { "epoch": 15.74251115360102, "grad_norm": 0.5723720788955688, "learning_rate": 9.971451044021906e-05, "loss": 0.0403, "step": 24700 }, { "epoch": 15.748884639898025, "grad_norm": 0.41518616676330566, "learning_rate": 9.971392200008842e-05, "loss": 0.0571, "step": 24710 }, { "epoch": 15.755258126195029, "grad_norm": 0.43931353092193604, "learning_rate": 9.971333295588667e-05, "loss": 0.0397, "step": 24720 }, { "epoch": 15.761631612492033, "grad_norm": 0.6646326184272766, "learning_rate": 9.971274330762102e-05, "loss": 0.0465, "step": 24730 }, { "epoch": 15.768005098789038, "grad_norm": 0.8012535572052002, "learning_rate": 9.971215305529858e-05, "loss": 0.0484, "step": 24740 }, { "epoch": 15.774378585086042, "grad_norm": 0.6311519742012024, "learning_rate": 9.971156219892654e-05, "loss": 0.0537, "step": 24750 }, { "epoch": 15.780752071383047, "grad_norm": 0.6525386571884155, "learning_rate": 9.97109707385121e-05, "loss": 0.0413, "step": 24760 }, { "epoch": 15.787125557680051, "grad_norm": 0.6937606334686279, "learning_rate": 9.971037867406241e-05, "loss": 0.0571, "step": 24770 }, { "epoch": 15.793499043977056, "grad_norm": 0.2991196811199188, "learning_rate": 9.97097860055847e-05, "loss": 0.0422, "step": 24780 }, { "epoch": 15.79987253027406, "grad_norm": 0.5672038793563843, "learning_rate": 9.970919273308615e-05, "loss": 0.0359, "step": 24790 }, { "epoch": 15.806246016571064, "grad_norm": 0.4974217116832733, "learning_rate": 9.970859885657399e-05, "loss": 0.0447, "step": 24800 }, { "epoch": 15.812619502868069, "grad_norm": 0.4844803512096405, "learning_rate": 9.970800437605541e-05, "loss": 0.0373, "step": 24810 }, { "epoch": 15.818992989165073, "grad_norm": 0.5815979242324829, "learning_rate": 9.970740929153763e-05, "loss": 0.0498, "step": 24820 }, { "epoch": 15.825366475462078, "grad_norm": 0.24050360918045044, "learning_rate": 9.970681360302791e-05, "loss": 0.0433, "step": 24830 }, { "epoch": 15.831739961759082, "grad_norm": 0.32661160826683044, "learning_rate": 9.970621731053348e-05, "loss": 0.0349, "step": 24840 }, { "epoch": 15.838113448056086, "grad_norm": 0.37772035598754883, "learning_rate": 9.970562041406158e-05, "loss": 0.052, "step": 24850 }, { "epoch": 15.84448693435309, "grad_norm": 0.6252054572105408, "learning_rate": 9.970502291361946e-05, "loss": 0.0541, "step": 24860 }, { "epoch": 15.850860420650095, "grad_norm": 0.6228477954864502, "learning_rate": 9.970442480921436e-05, "loss": 0.0363, "step": 24870 }, { "epoch": 15.8572339069471, "grad_norm": 0.6973903775215149, "learning_rate": 9.97038261008536e-05, "loss": 0.0499, "step": 24880 }, { "epoch": 15.863607393244104, "grad_norm": 0.413540244102478, "learning_rate": 9.97032267885444e-05, "loss": 0.0475, "step": 24890 }, { "epoch": 15.869980879541108, "grad_norm": 0.4371397793292999, "learning_rate": 9.970262687229408e-05, "loss": 0.0417, "step": 24900 }, { "epoch": 15.876354365838113, "grad_norm": 0.41675591468811035, "learning_rate": 9.970202635210992e-05, "loss": 0.0462, "step": 24910 }, { "epoch": 15.882727852135117, "grad_norm": 0.4219648540019989, "learning_rate": 9.970142522799921e-05, "loss": 0.0377, "step": 24920 }, { "epoch": 15.889101338432122, "grad_norm": 0.6078414916992188, "learning_rate": 9.970082349996925e-05, "loss": 0.0325, "step": 24930 }, { "epoch": 15.895474824729128, "grad_norm": 0.4558377265930176, "learning_rate": 9.970022116802737e-05, "loss": 0.0394, "step": 24940 }, { "epoch": 15.901848311026132, "grad_norm": 0.7134793996810913, "learning_rate": 9.969961823218087e-05, "loss": 0.0376, "step": 24950 }, { "epoch": 15.908221797323137, "grad_norm": 0.35169529914855957, "learning_rate": 9.969901469243708e-05, "loss": 0.043, "step": 24960 }, { "epoch": 15.914595283620141, "grad_norm": 0.5167200565338135, "learning_rate": 9.969841054880336e-05, "loss": 0.0385, "step": 24970 }, { "epoch": 15.920968769917145, "grad_norm": 0.6052188873291016, "learning_rate": 9.9697805801287e-05, "loss": 0.0434, "step": 24980 }, { "epoch": 15.92734225621415, "grad_norm": 0.6890076398849487, "learning_rate": 9.96972004498954e-05, "loss": 0.0431, "step": 24990 }, { "epoch": 15.933715742511154, "grad_norm": 0.8521940112113953, "learning_rate": 9.969659449463588e-05, "loss": 0.0443, "step": 25000 }, { "epoch": 15.940089228808159, "grad_norm": 0.6303184628486633, "learning_rate": 9.969598793551582e-05, "loss": 0.0484, "step": 25010 }, { "epoch": 15.946462715105163, "grad_norm": 0.6598095297813416, "learning_rate": 9.969538077254259e-05, "loss": 0.0392, "step": 25020 }, { "epoch": 15.952836201402167, "grad_norm": 0.4992329478263855, "learning_rate": 9.969477300572356e-05, "loss": 0.0527, "step": 25030 }, { "epoch": 15.959209687699172, "grad_norm": 0.45346710085868835, "learning_rate": 9.969416463506612e-05, "loss": 0.0496, "step": 25040 }, { "epoch": 15.965583173996176, "grad_norm": 0.5461167097091675, "learning_rate": 9.969355566057766e-05, "loss": 0.0479, "step": 25050 }, { "epoch": 15.97195666029318, "grad_norm": 0.60860675573349, "learning_rate": 9.969294608226556e-05, "loss": 0.0443, "step": 25060 }, { "epoch": 15.978330146590185, "grad_norm": 0.5992637276649475, "learning_rate": 9.969233590013727e-05, "loss": 0.0436, "step": 25070 }, { "epoch": 15.98470363288719, "grad_norm": 0.9202326536178589, "learning_rate": 9.969172511420017e-05, "loss": 0.0403, "step": 25080 }, { "epoch": 15.991077119184194, "grad_norm": 0.5138540267944336, "learning_rate": 9.969111372446171e-05, "loss": 0.0446, "step": 25090 }, { "epoch": 15.997450605481198, "grad_norm": 0.5955187678337097, "learning_rate": 9.969050173092928e-05, "loss": 0.0496, "step": 25100 }, { "epoch": 16.003824091778203, "grad_norm": 0.5476959347724915, "learning_rate": 9.968988913361035e-05, "loss": 0.0444, "step": 25110 }, { "epoch": 16.010197578075207, "grad_norm": 0.37764671444892883, "learning_rate": 9.968927593251235e-05, "loss": 0.0352, "step": 25120 }, { "epoch": 16.01657106437221, "grad_norm": 0.5696294903755188, "learning_rate": 9.968866212764274e-05, "loss": 0.0455, "step": 25130 }, { "epoch": 16.022944550669216, "grad_norm": 0.41611307859420776, "learning_rate": 9.968804771900897e-05, "loss": 0.0412, "step": 25140 }, { "epoch": 16.02931803696622, "grad_norm": 0.4574086368083954, "learning_rate": 9.96874327066185e-05, "loss": 0.0456, "step": 25150 }, { "epoch": 16.035691523263225, "grad_norm": 0.8568750619888306, "learning_rate": 9.968681709047881e-05, "loss": 0.0476, "step": 25160 }, { "epoch": 16.04206500956023, "grad_norm": 0.5008062124252319, "learning_rate": 9.968620087059738e-05, "loss": 0.0274, "step": 25170 }, { "epoch": 16.048438495857233, "grad_norm": 0.5712280869483948, "learning_rate": 9.968558404698171e-05, "loss": 0.0426, "step": 25180 }, { "epoch": 16.054811982154238, "grad_norm": 0.4087081849575043, "learning_rate": 9.968496661963926e-05, "loss": 0.0454, "step": 25190 }, { "epoch": 16.061185468451242, "grad_norm": 0.4734458327293396, "learning_rate": 9.968434858857758e-05, "loss": 0.0366, "step": 25200 }, { "epoch": 16.067558954748247, "grad_norm": 0.4051951467990875, "learning_rate": 9.968372995380415e-05, "loss": 0.0343, "step": 25210 }, { "epoch": 16.07393244104525, "grad_norm": 0.49625903367996216, "learning_rate": 9.968311071532647e-05, "loss": 0.0421, "step": 25220 }, { "epoch": 16.080305927342256, "grad_norm": 0.8067222833633423, "learning_rate": 9.968249087315212e-05, "loss": 0.0388, "step": 25230 }, { "epoch": 16.08667941363926, "grad_norm": 0.5366131067276001, "learning_rate": 9.968187042728857e-05, "loss": 0.0491, "step": 25240 }, { "epoch": 16.093052899936264, "grad_norm": 0.5047993659973145, "learning_rate": 9.968124937774339e-05, "loss": 0.0385, "step": 25250 }, { "epoch": 16.09942638623327, "grad_norm": 0.5189358592033386, "learning_rate": 9.968062772452413e-05, "loss": 0.0323, "step": 25260 }, { "epoch": 16.105799872530273, "grad_norm": 0.47940099239349365, "learning_rate": 9.968000546763834e-05, "loss": 0.0348, "step": 25270 }, { "epoch": 16.112173358827278, "grad_norm": 0.5000100135803223, "learning_rate": 9.967938260709357e-05, "loss": 0.0442, "step": 25280 }, { "epoch": 16.118546845124282, "grad_norm": 0.4666995704174042, "learning_rate": 9.96787591428974e-05, "loss": 0.0401, "step": 25290 }, { "epoch": 16.124920331421286, "grad_norm": 0.5550605654716492, "learning_rate": 9.967813507505741e-05, "loss": 0.0426, "step": 25300 }, { "epoch": 16.13129381771829, "grad_norm": 0.4802536964416504, "learning_rate": 9.967751040358116e-05, "loss": 0.031, "step": 25310 }, { "epoch": 16.137667304015295, "grad_norm": 0.7131120562553406, "learning_rate": 9.967688512847626e-05, "loss": 0.0358, "step": 25320 }, { "epoch": 16.1440407903123, "grad_norm": 0.7372891902923584, "learning_rate": 9.967625924975031e-05, "loss": 0.0462, "step": 25330 }, { "epoch": 16.150414276609304, "grad_norm": 0.5742558836936951, "learning_rate": 9.967563276741091e-05, "loss": 0.0379, "step": 25340 }, { "epoch": 16.15678776290631, "grad_norm": 0.6399430632591248, "learning_rate": 9.967500568146566e-05, "loss": 0.0401, "step": 25350 }, { "epoch": 16.163161249203313, "grad_norm": 0.7068963050842285, "learning_rate": 9.96743779919222e-05, "loss": 0.0402, "step": 25360 }, { "epoch": 16.169534735500317, "grad_norm": 0.7580180168151855, "learning_rate": 9.967374969878814e-05, "loss": 0.0409, "step": 25370 }, { "epoch": 16.17590822179732, "grad_norm": 1.066556453704834, "learning_rate": 9.967312080207112e-05, "loss": 0.0359, "step": 25380 }, { "epoch": 16.182281708094326, "grad_norm": 0.40891554951667786, "learning_rate": 9.967249130177878e-05, "loss": 0.0295, "step": 25390 }, { "epoch": 16.18865519439133, "grad_norm": 0.3703814744949341, "learning_rate": 9.967186119791878e-05, "loss": 0.0402, "step": 25400 }, { "epoch": 16.195028680688335, "grad_norm": 0.6365624666213989, "learning_rate": 9.967123049049877e-05, "loss": 0.0401, "step": 25410 }, { "epoch": 16.20140216698534, "grad_norm": 0.5917449593544006, "learning_rate": 9.96705991795264e-05, "loss": 0.0378, "step": 25420 }, { "epoch": 16.207775653282347, "grad_norm": 0.4612136483192444, "learning_rate": 9.966996726500936e-05, "loss": 0.0399, "step": 25430 }, { "epoch": 16.21414913957935, "grad_norm": 0.541172981262207, "learning_rate": 9.966933474695532e-05, "loss": 0.0453, "step": 25440 }, { "epoch": 16.220522625876356, "grad_norm": 0.2582317292690277, "learning_rate": 9.966870162537199e-05, "loss": 0.0416, "step": 25450 }, { "epoch": 16.22689611217336, "grad_norm": 0.45396140217781067, "learning_rate": 9.966806790026701e-05, "loss": 0.0382, "step": 25460 }, { "epoch": 16.233269598470365, "grad_norm": 0.5514276027679443, "learning_rate": 9.966743357164811e-05, "loss": 0.0382, "step": 25470 }, { "epoch": 16.23964308476737, "grad_norm": 0.9388208985328674, "learning_rate": 9.966679863952302e-05, "loss": 0.0411, "step": 25480 }, { "epoch": 16.246016571064374, "grad_norm": 0.45645856857299805, "learning_rate": 9.966616310389942e-05, "loss": 0.0361, "step": 25490 }, { "epoch": 16.252390057361378, "grad_norm": 0.2370976358652115, "learning_rate": 9.966552696478504e-05, "loss": 0.036, "step": 25500 }, { "epoch": 16.258763543658382, "grad_norm": 0.2736761271953583, "learning_rate": 9.966489022218763e-05, "loss": 0.0446, "step": 25510 }, { "epoch": 16.265137029955387, "grad_norm": 0.3377840518951416, "learning_rate": 9.96642528761149e-05, "loss": 0.033, "step": 25520 }, { "epoch": 16.27151051625239, "grad_norm": 0.4454249441623688, "learning_rate": 9.966361492657461e-05, "loss": 0.0484, "step": 25530 }, { "epoch": 16.277884002549396, "grad_norm": 0.4505033493041992, "learning_rate": 9.966297637357452e-05, "loss": 0.0475, "step": 25540 }, { "epoch": 16.2842574888464, "grad_norm": 0.6705693006515503, "learning_rate": 9.966233721712237e-05, "loss": 0.0421, "step": 25550 }, { "epoch": 16.290630975143404, "grad_norm": 0.5719900131225586, "learning_rate": 9.966169745722593e-05, "loss": 0.0469, "step": 25560 }, { "epoch": 16.29700446144041, "grad_norm": 0.3500453233718872, "learning_rate": 9.966105709389297e-05, "loss": 0.0461, "step": 25570 }, { "epoch": 16.303377947737413, "grad_norm": 0.5628396272659302, "learning_rate": 9.966041612713129e-05, "loss": 0.0433, "step": 25580 }, { "epoch": 16.309751434034418, "grad_norm": 0.423896849155426, "learning_rate": 9.965977455694867e-05, "loss": 0.0437, "step": 25590 }, { "epoch": 16.316124920331422, "grad_norm": 0.6233727931976318, "learning_rate": 9.96591323833529e-05, "loss": 0.0449, "step": 25600 }, { "epoch": 16.322498406628426, "grad_norm": 0.5242258310317993, "learning_rate": 9.965848960635179e-05, "loss": 0.0383, "step": 25610 }, { "epoch": 16.32887189292543, "grad_norm": 0.6171064972877502, "learning_rate": 9.965784622595313e-05, "loss": 0.0403, "step": 25620 }, { "epoch": 16.335245379222435, "grad_norm": 0.7143844962120056, "learning_rate": 9.965720224216476e-05, "loss": 0.0392, "step": 25630 }, { "epoch": 16.34161886551944, "grad_norm": 0.4803479313850403, "learning_rate": 9.96565576549945e-05, "loss": 0.047, "step": 25640 }, { "epoch": 16.347992351816444, "grad_norm": 0.9446491599082947, "learning_rate": 9.965591246445018e-05, "loss": 0.0414, "step": 25650 }, { "epoch": 16.35436583811345, "grad_norm": 0.9977034330368042, "learning_rate": 9.965526667053963e-05, "loss": 0.0539, "step": 25660 }, { "epoch": 16.360739324410453, "grad_norm": 0.5730120539665222, "learning_rate": 9.965462027327073e-05, "loss": 0.0426, "step": 25670 }, { "epoch": 16.367112810707457, "grad_norm": 0.8035280108451843, "learning_rate": 9.965397327265132e-05, "loss": 0.0536, "step": 25680 }, { "epoch": 16.37348629700446, "grad_norm": 1.0698623657226562, "learning_rate": 9.965332566868923e-05, "loss": 0.0438, "step": 25690 }, { "epoch": 16.379859783301466, "grad_norm": 0.5060622096061707, "learning_rate": 9.965267746139235e-05, "loss": 0.0447, "step": 25700 }, { "epoch": 16.38623326959847, "grad_norm": 0.3704836070537567, "learning_rate": 9.965202865076857e-05, "loss": 0.0489, "step": 25710 }, { "epoch": 16.392606755895475, "grad_norm": 0.836049497127533, "learning_rate": 9.965137923682577e-05, "loss": 0.0446, "step": 25720 }, { "epoch": 16.39898024219248, "grad_norm": 0.886762797832489, "learning_rate": 9.965072921957181e-05, "loss": 0.0415, "step": 25730 }, { "epoch": 16.405353728489484, "grad_norm": 0.6053810119628906, "learning_rate": 9.965007859901464e-05, "loss": 0.0411, "step": 25740 }, { "epoch": 16.411727214786488, "grad_norm": 0.8052540421485901, "learning_rate": 9.964942737516211e-05, "loss": 0.0333, "step": 25750 }, { "epoch": 16.418100701083493, "grad_norm": 0.5735458731651306, "learning_rate": 9.964877554802217e-05, "loss": 0.0492, "step": 25760 }, { "epoch": 16.424474187380497, "grad_norm": 0.31387919187545776, "learning_rate": 9.964812311760275e-05, "loss": 0.0443, "step": 25770 }, { "epoch": 16.4308476736775, "grad_norm": 0.4805189371109009, "learning_rate": 9.964747008391173e-05, "loss": 0.0368, "step": 25780 }, { "epoch": 16.437221159974506, "grad_norm": 0.3303167521953583, "learning_rate": 9.964681644695708e-05, "loss": 0.0343, "step": 25790 }, { "epoch": 16.44359464627151, "grad_norm": 0.3784751296043396, "learning_rate": 9.964616220674675e-05, "loss": 0.0397, "step": 25800 }, { "epoch": 16.449968132568515, "grad_norm": 0.35350027680397034, "learning_rate": 9.964550736328866e-05, "loss": 0.0396, "step": 25810 }, { "epoch": 16.45634161886552, "grad_norm": 0.38310375809669495, "learning_rate": 9.964485191659079e-05, "loss": 0.039, "step": 25820 }, { "epoch": 16.462715105162523, "grad_norm": 0.362540066242218, "learning_rate": 9.964419586666109e-05, "loss": 0.0363, "step": 25830 }, { "epoch": 16.469088591459528, "grad_norm": 0.5591185092926025, "learning_rate": 9.964353921350755e-05, "loss": 0.0435, "step": 25840 }, { "epoch": 16.475462077756532, "grad_norm": 0.9858211278915405, "learning_rate": 9.964288195713813e-05, "loss": 0.0386, "step": 25850 }, { "epoch": 16.481835564053537, "grad_norm": 0.766990602016449, "learning_rate": 9.964222409756083e-05, "loss": 0.0438, "step": 25860 }, { "epoch": 16.48820905035054, "grad_norm": 0.6600323915481567, "learning_rate": 9.964156563478364e-05, "loss": 0.0424, "step": 25870 }, { "epoch": 16.494582536647545, "grad_norm": 0.46155497431755066, "learning_rate": 9.964090656881455e-05, "loss": 0.0548, "step": 25880 }, { "epoch": 16.50095602294455, "grad_norm": 0.5408865809440613, "learning_rate": 9.964024689966158e-05, "loss": 0.0495, "step": 25890 }, { "epoch": 16.507329509241554, "grad_norm": 0.6564632058143616, "learning_rate": 9.963958662733274e-05, "loss": 0.0283, "step": 25900 }, { "epoch": 16.51370299553856, "grad_norm": 0.9540555477142334, "learning_rate": 9.963892575183604e-05, "loss": 0.0453, "step": 25910 }, { "epoch": 16.520076481835563, "grad_norm": 0.6405223608016968, "learning_rate": 9.963826427317955e-05, "loss": 0.0506, "step": 25920 }, { "epoch": 16.526449968132567, "grad_norm": 0.5533133149147034, "learning_rate": 9.963760219137127e-05, "loss": 0.0432, "step": 25930 }, { "epoch": 16.532823454429572, "grad_norm": 0.5807307958602905, "learning_rate": 9.963693950641926e-05, "loss": 0.0482, "step": 25940 }, { "epoch": 16.539196940726576, "grad_norm": 0.48776960372924805, "learning_rate": 9.963627621833157e-05, "loss": 0.0366, "step": 25950 }, { "epoch": 16.54557042702358, "grad_norm": 0.6296766996383667, "learning_rate": 9.963561232711626e-05, "loss": 0.0525, "step": 25960 }, { "epoch": 16.551943913320585, "grad_norm": 0.6705887913703918, "learning_rate": 9.963494783278141e-05, "loss": 0.0369, "step": 25970 }, { "epoch": 16.55831739961759, "grad_norm": 0.5784368515014648, "learning_rate": 9.963428273533506e-05, "loss": 0.0399, "step": 25980 }, { "epoch": 16.564690885914594, "grad_norm": 0.6684627532958984, "learning_rate": 9.963361703478531e-05, "loss": 0.0465, "step": 25990 }, { "epoch": 16.5710643722116, "grad_norm": 0.528326153755188, "learning_rate": 9.963295073114026e-05, "loss": 0.0321, "step": 26000 }, { "epoch": 16.577437858508603, "grad_norm": 0.5851904153823853, "learning_rate": 9.9632283824408e-05, "loss": 0.0404, "step": 26010 }, { "epoch": 16.583811344805607, "grad_norm": 0.6672981977462769, "learning_rate": 9.963161631459663e-05, "loss": 0.06, "step": 26020 }, { "epoch": 16.59018483110261, "grad_norm": 0.5629258751869202, "learning_rate": 9.963094820171426e-05, "loss": 0.0419, "step": 26030 }, { "epoch": 16.59655831739962, "grad_norm": 0.7017990350723267, "learning_rate": 9.963027948576902e-05, "loss": 0.0642, "step": 26040 }, { "epoch": 16.60293180369662, "grad_norm": 0.6739966869354248, "learning_rate": 9.9629610166769e-05, "loss": 0.0465, "step": 26050 }, { "epoch": 16.609305289993628, "grad_norm": 0.5004031658172607, "learning_rate": 9.962894024472237e-05, "loss": 0.0368, "step": 26060 }, { "epoch": 16.615678776290633, "grad_norm": 0.4587778151035309, "learning_rate": 9.962826971963728e-05, "loss": 0.0398, "step": 26070 }, { "epoch": 16.622052262587637, "grad_norm": 0.654233455657959, "learning_rate": 9.962759859152184e-05, "loss": 0.0525, "step": 26080 }, { "epoch": 16.62842574888464, "grad_norm": 0.4020244777202606, "learning_rate": 9.962692686038422e-05, "loss": 0.0573, "step": 26090 }, { "epoch": 16.634799235181646, "grad_norm": 0.49503394961357117, "learning_rate": 9.962625452623258e-05, "loss": 0.0562, "step": 26100 }, { "epoch": 16.64117272147865, "grad_norm": 0.6481123566627502, "learning_rate": 9.962558158907509e-05, "loss": 0.0498, "step": 26110 }, { "epoch": 16.647546207775655, "grad_norm": 0.582618236541748, "learning_rate": 9.962490804891993e-05, "loss": 0.0337, "step": 26120 }, { "epoch": 16.65391969407266, "grad_norm": 0.553983747959137, "learning_rate": 9.962423390577529e-05, "loss": 0.0483, "step": 26130 }, { "epoch": 16.660293180369663, "grad_norm": 0.7455496191978455, "learning_rate": 9.962355915964936e-05, "loss": 0.0472, "step": 26140 }, { "epoch": 16.666666666666668, "grad_norm": 0.4680227041244507, "learning_rate": 9.962288381055031e-05, "loss": 0.0507, "step": 26150 }, { "epoch": 16.673040152963672, "grad_norm": 0.5346328616142273, "learning_rate": 9.96222078584864e-05, "loss": 0.0485, "step": 26160 }, { "epoch": 16.679413639260677, "grad_norm": 0.3488355875015259, "learning_rate": 9.96215313034658e-05, "loss": 0.0306, "step": 26170 }, { "epoch": 16.68578712555768, "grad_norm": 0.38558948040008545, "learning_rate": 9.962085414549673e-05, "loss": 0.0443, "step": 26180 }, { "epoch": 16.692160611854685, "grad_norm": 0.575023889541626, "learning_rate": 9.962017638458746e-05, "loss": 0.0528, "step": 26190 }, { "epoch": 16.69853409815169, "grad_norm": 0.5307489037513733, "learning_rate": 9.961949802074618e-05, "loss": 0.0438, "step": 26200 }, { "epoch": 16.704907584448694, "grad_norm": 0.7933769226074219, "learning_rate": 9.961881905398117e-05, "loss": 0.046, "step": 26210 }, { "epoch": 16.7112810707457, "grad_norm": 0.5165237188339233, "learning_rate": 9.961813948430064e-05, "loss": 0.053, "step": 26220 }, { "epoch": 16.717654557042703, "grad_norm": 0.6027774810791016, "learning_rate": 9.961745931171287e-05, "loss": 0.0516, "step": 26230 }, { "epoch": 16.724028043339707, "grad_norm": 0.475593626499176, "learning_rate": 9.961677853622613e-05, "loss": 0.044, "step": 26240 }, { "epoch": 16.730401529636712, "grad_norm": 0.5969756245613098, "learning_rate": 9.961609715784868e-05, "loss": 0.0369, "step": 26250 }, { "epoch": 16.736775015933716, "grad_norm": 0.44374409317970276, "learning_rate": 9.96154151765888e-05, "loss": 0.0534, "step": 26260 }, { "epoch": 16.74314850223072, "grad_norm": 0.6982294917106628, "learning_rate": 9.961473259245478e-05, "loss": 0.0562, "step": 26270 }, { "epoch": 16.749521988527725, "grad_norm": 0.5568380951881409, "learning_rate": 9.961404940545492e-05, "loss": 0.0571, "step": 26280 }, { "epoch": 16.75589547482473, "grad_norm": 0.351503849029541, "learning_rate": 9.961336561559753e-05, "loss": 0.0532, "step": 26290 }, { "epoch": 16.762268961121734, "grad_norm": 0.5804415941238403, "learning_rate": 9.961268122289089e-05, "loss": 0.0633, "step": 26300 }, { "epoch": 16.76864244741874, "grad_norm": 0.6005134582519531, "learning_rate": 9.961199622734333e-05, "loss": 0.0562, "step": 26310 }, { "epoch": 16.775015933715743, "grad_norm": 0.3110552430152893, "learning_rate": 9.961131062896318e-05, "loss": 0.0582, "step": 26320 }, { "epoch": 16.781389420012747, "grad_norm": 0.5617906451225281, "learning_rate": 9.961062442775876e-05, "loss": 0.0458, "step": 26330 }, { "epoch": 16.78776290630975, "grad_norm": 0.7597768306732178, "learning_rate": 9.960993762373842e-05, "loss": 0.0432, "step": 26340 }, { "epoch": 16.794136392606756, "grad_norm": 0.6495307683944702, "learning_rate": 9.960925021691049e-05, "loss": 0.0441, "step": 26350 }, { "epoch": 16.80050987890376, "grad_norm": 0.4253827631473541, "learning_rate": 9.960856220728333e-05, "loss": 0.054, "step": 26360 }, { "epoch": 16.806883365200765, "grad_norm": 0.5670692920684814, "learning_rate": 9.960787359486531e-05, "loss": 0.0478, "step": 26370 }, { "epoch": 16.81325685149777, "grad_norm": 0.5977931618690491, "learning_rate": 9.960718437966479e-05, "loss": 0.0434, "step": 26380 }, { "epoch": 16.819630337794774, "grad_norm": 0.41143953800201416, "learning_rate": 9.960649456169014e-05, "loss": 0.0427, "step": 26390 }, { "epoch": 16.826003824091778, "grad_norm": 0.47518542408943176, "learning_rate": 9.960580414094973e-05, "loss": 0.0518, "step": 26400 }, { "epoch": 16.832377310388782, "grad_norm": 0.6711757183074951, "learning_rate": 9.960511311745198e-05, "loss": 0.0433, "step": 26410 }, { "epoch": 16.838750796685787, "grad_norm": 0.582792341709137, "learning_rate": 9.960442149120526e-05, "loss": 0.0357, "step": 26420 }, { "epoch": 16.84512428298279, "grad_norm": 0.7889184355735779, "learning_rate": 9.9603729262218e-05, "loss": 0.0499, "step": 26430 }, { "epoch": 16.851497769279796, "grad_norm": 0.4969421923160553, "learning_rate": 9.960303643049858e-05, "loss": 0.0587, "step": 26440 }, { "epoch": 16.8578712555768, "grad_norm": 0.9252558350563049, "learning_rate": 9.960234299605544e-05, "loss": 0.0454, "step": 26450 }, { "epoch": 16.864244741873804, "grad_norm": 0.5752676725387573, "learning_rate": 9.9601648958897e-05, "loss": 0.0336, "step": 26460 }, { "epoch": 16.87061822817081, "grad_norm": 0.42766857147216797, "learning_rate": 9.96009543190317e-05, "loss": 0.0486, "step": 26470 }, { "epoch": 16.876991714467813, "grad_norm": 0.8740835785865784, "learning_rate": 9.960025907646797e-05, "loss": 0.042, "step": 26480 }, { "epoch": 16.883365200764818, "grad_norm": 0.458851158618927, "learning_rate": 9.959956323121425e-05, "loss": 0.0374, "step": 26490 }, { "epoch": 16.889738687061822, "grad_norm": 0.35557204484939575, "learning_rate": 9.959886678327903e-05, "loss": 0.0366, "step": 26500 }, { "epoch": 16.896112173358826, "grad_norm": 0.6742868423461914, "learning_rate": 9.959816973267073e-05, "loss": 0.0675, "step": 26510 }, { "epoch": 16.90248565965583, "grad_norm": 0.6167845726013184, "learning_rate": 9.959747207939785e-05, "loss": 0.0437, "step": 26520 }, { "epoch": 16.908859145952835, "grad_norm": 0.636807918548584, "learning_rate": 9.959677382346884e-05, "loss": 0.0467, "step": 26530 }, { "epoch": 16.91523263224984, "grad_norm": 0.6801275610923767, "learning_rate": 9.959607496489223e-05, "loss": 0.0531, "step": 26540 }, { "epoch": 16.921606118546844, "grad_norm": 0.3105388581752777, "learning_rate": 9.959537550367646e-05, "loss": 0.04, "step": 26550 }, { "epoch": 16.92797960484385, "grad_norm": 0.6922913193702698, "learning_rate": 9.959467543983005e-05, "loss": 0.043, "step": 26560 }, { "epoch": 16.934353091140853, "grad_norm": 0.5376835465431213, "learning_rate": 9.959397477336152e-05, "loss": 0.0496, "step": 26570 }, { "epoch": 16.940726577437857, "grad_norm": 0.5774103403091431, "learning_rate": 9.959327350427937e-05, "loss": 0.0368, "step": 26580 }, { "epoch": 16.94710006373486, "grad_norm": 0.4151640832424164, "learning_rate": 9.959257163259212e-05, "loss": 0.0359, "step": 26590 }, { "epoch": 16.953473550031866, "grad_norm": 0.5616617202758789, "learning_rate": 9.95918691583083e-05, "loss": 0.0367, "step": 26600 }, { "epoch": 16.95984703632887, "grad_norm": 0.6855557560920715, "learning_rate": 9.959116608143646e-05, "loss": 0.0442, "step": 26610 }, { "epoch": 16.966220522625875, "grad_norm": 0.561180591583252, "learning_rate": 9.959046240198511e-05, "loss": 0.0406, "step": 26620 }, { "epoch": 16.97259400892288, "grad_norm": 0.4963124692440033, "learning_rate": 9.958975811996284e-05, "loss": 0.0578, "step": 26630 }, { "epoch": 16.978967495219884, "grad_norm": 0.6078842878341675, "learning_rate": 9.958905323537817e-05, "loss": 0.0433, "step": 26640 }, { "epoch": 16.985340981516888, "grad_norm": 0.5090770721435547, "learning_rate": 9.958834774823969e-05, "loss": 0.0333, "step": 26650 }, { "epoch": 16.991714467813892, "grad_norm": 0.676506519317627, "learning_rate": 9.958764165855598e-05, "loss": 0.0353, "step": 26660 }, { "epoch": 16.9980879541109, "grad_norm": 0.6919217109680176, "learning_rate": 9.95869349663356e-05, "loss": 0.0417, "step": 26670 }, { "epoch": 17.004461440407905, "grad_norm": 0.6133757829666138, "learning_rate": 9.958622767158713e-05, "loss": 0.0329, "step": 26680 }, { "epoch": 17.01083492670491, "grad_norm": 0.8568653464317322, "learning_rate": 9.958551977431918e-05, "loss": 0.0363, "step": 26690 }, { "epoch": 17.017208413001914, "grad_norm": 0.38847360014915466, "learning_rate": 9.958481127454036e-05, "loss": 0.0414, "step": 26700 }, { "epoch": 17.023581899298918, "grad_norm": 0.4781525731086731, "learning_rate": 9.958410217225925e-05, "loss": 0.0541, "step": 26710 }, { "epoch": 17.029955385595922, "grad_norm": 0.5136752724647522, "learning_rate": 9.95833924674845e-05, "loss": 0.0322, "step": 26720 }, { "epoch": 17.036328871892927, "grad_norm": 0.3825243413448334, "learning_rate": 9.958268216022471e-05, "loss": 0.0296, "step": 26730 }, { "epoch": 17.04270235818993, "grad_norm": 0.3564102053642273, "learning_rate": 9.958197125048852e-05, "loss": 0.0368, "step": 26740 }, { "epoch": 17.049075844486936, "grad_norm": 0.6764777302742004, "learning_rate": 9.958125973828454e-05, "loss": 0.0383, "step": 26750 }, { "epoch": 17.05544933078394, "grad_norm": 0.6747488975524902, "learning_rate": 9.958054762362148e-05, "loss": 0.0453, "step": 26760 }, { "epoch": 17.061822817080945, "grad_norm": 0.5109786987304688, "learning_rate": 9.957983490650793e-05, "loss": 0.0417, "step": 26770 }, { "epoch": 17.06819630337795, "grad_norm": 0.3799660801887512, "learning_rate": 9.95791215869526e-05, "loss": 0.0433, "step": 26780 }, { "epoch": 17.074569789674953, "grad_norm": 0.6049870252609253, "learning_rate": 9.957840766496411e-05, "loss": 0.0413, "step": 26790 }, { "epoch": 17.080943275971958, "grad_norm": 0.9162454009056091, "learning_rate": 9.957769314055117e-05, "loss": 0.0419, "step": 26800 }, { "epoch": 17.087316762268962, "grad_norm": 0.30667006969451904, "learning_rate": 9.957697801372243e-05, "loss": 0.0443, "step": 26810 }, { "epoch": 17.093690248565967, "grad_norm": 0.5703659057617188, "learning_rate": 9.957626228448662e-05, "loss": 0.04, "step": 26820 }, { "epoch": 17.10006373486297, "grad_norm": 0.7101849913597107, "learning_rate": 9.95755459528524e-05, "loss": 0.0392, "step": 26830 }, { "epoch": 17.106437221159975, "grad_norm": 0.36343079805374146, "learning_rate": 9.95748290188285e-05, "loss": 0.0369, "step": 26840 }, { "epoch": 17.11281070745698, "grad_norm": 0.6599883437156677, "learning_rate": 9.957411148242361e-05, "loss": 0.0543, "step": 26850 }, { "epoch": 17.119184193753984, "grad_norm": 0.35964056849479675, "learning_rate": 9.957339334364648e-05, "loss": 0.0376, "step": 26860 }, { "epoch": 17.12555768005099, "grad_norm": 0.35593265295028687, "learning_rate": 9.957267460250579e-05, "loss": 0.0325, "step": 26870 }, { "epoch": 17.131931166347993, "grad_norm": 0.5668849349021912, "learning_rate": 9.957195525901032e-05, "loss": 0.0382, "step": 26880 }, { "epoch": 17.138304652644997, "grad_norm": 0.6566433906555176, "learning_rate": 9.957123531316878e-05, "loss": 0.051, "step": 26890 }, { "epoch": 17.144678138942, "grad_norm": 0.5178101658821106, "learning_rate": 9.957051476498994e-05, "loss": 0.034, "step": 26900 }, { "epoch": 17.151051625239006, "grad_norm": 1.091646671295166, "learning_rate": 9.956979361448253e-05, "loss": 0.038, "step": 26910 }, { "epoch": 17.15742511153601, "grad_norm": 0.3322891592979431, "learning_rate": 9.956907186165532e-05, "loss": 0.038, "step": 26920 }, { "epoch": 17.163798597833015, "grad_norm": 0.6825461387634277, "learning_rate": 9.95683495065171e-05, "loss": 0.0374, "step": 26930 }, { "epoch": 17.17017208413002, "grad_norm": 0.4510245621204376, "learning_rate": 9.956762654907664e-05, "loss": 0.0425, "step": 26940 }, { "epoch": 17.176545570427024, "grad_norm": 0.739872395992279, "learning_rate": 9.95669029893427e-05, "loss": 0.0569, "step": 26950 }, { "epoch": 17.182919056724028, "grad_norm": 0.42860573530197144, "learning_rate": 9.956617882732409e-05, "loss": 0.0385, "step": 26960 }, { "epoch": 17.189292543021033, "grad_norm": 0.46240487694740295, "learning_rate": 9.956545406302962e-05, "loss": 0.0421, "step": 26970 }, { "epoch": 17.195666029318037, "grad_norm": 0.6773727536201477, "learning_rate": 9.956472869646807e-05, "loss": 0.0476, "step": 26980 }, { "epoch": 17.20203951561504, "grad_norm": 0.6244494915008545, "learning_rate": 9.956400272764829e-05, "loss": 0.0347, "step": 26990 }, { "epoch": 17.208413001912046, "grad_norm": 0.5109573602676392, "learning_rate": 9.956327615657907e-05, "loss": 0.0393, "step": 27000 }, { "epoch": 17.21478648820905, "grad_norm": 0.6456117033958435, "learning_rate": 9.956254898326924e-05, "loss": 0.0323, "step": 27010 }, { "epoch": 17.221159974506055, "grad_norm": 0.582136869430542, "learning_rate": 9.956182120772765e-05, "loss": 0.0305, "step": 27020 }, { "epoch": 17.22753346080306, "grad_norm": 0.6260345578193665, "learning_rate": 9.956109282996315e-05, "loss": 0.0356, "step": 27030 }, { "epoch": 17.233906947100063, "grad_norm": 0.564299464225769, "learning_rate": 9.956036384998457e-05, "loss": 0.0418, "step": 27040 }, { "epoch": 17.240280433397068, "grad_norm": 0.45445767045021057, "learning_rate": 9.95596342678008e-05, "loss": 0.0432, "step": 27050 }, { "epoch": 17.246653919694072, "grad_norm": 0.604080855846405, "learning_rate": 9.955890408342065e-05, "loss": 0.037, "step": 27060 }, { "epoch": 17.253027405991077, "grad_norm": 0.3245340883731842, "learning_rate": 9.955817329685303e-05, "loss": 0.0321, "step": 27070 }, { "epoch": 17.25940089228808, "grad_norm": 0.4026404917240143, "learning_rate": 9.955744190810682e-05, "loss": 0.0258, "step": 27080 }, { "epoch": 17.265774378585085, "grad_norm": 0.7042772173881531, "learning_rate": 9.95567099171909e-05, "loss": 0.0497, "step": 27090 }, { "epoch": 17.27214786488209, "grad_norm": 0.4526478350162506, "learning_rate": 9.955597732411417e-05, "loss": 0.0355, "step": 27100 }, { "epoch": 17.278521351179094, "grad_norm": 0.37130364775657654, "learning_rate": 9.95552441288855e-05, "loss": 0.0354, "step": 27110 }, { "epoch": 17.2848948374761, "grad_norm": 0.4773696959018707, "learning_rate": 9.955451033151387e-05, "loss": 0.0466, "step": 27120 }, { "epoch": 17.291268323773103, "grad_norm": 0.30388012528419495, "learning_rate": 9.955377593200813e-05, "loss": 0.0415, "step": 27130 }, { "epoch": 17.297641810070107, "grad_norm": 0.4556509256362915, "learning_rate": 9.955304093037721e-05, "loss": 0.0456, "step": 27140 }, { "epoch": 17.304015296367112, "grad_norm": 0.7264136075973511, "learning_rate": 9.955230532663007e-05, "loss": 0.0424, "step": 27150 }, { "epoch": 17.310388782664116, "grad_norm": 0.4282618761062622, "learning_rate": 9.955156912077565e-05, "loss": 0.0314, "step": 27160 }, { "epoch": 17.31676226896112, "grad_norm": 0.5929535627365112, "learning_rate": 9.955083231282286e-05, "loss": 0.0329, "step": 27170 }, { "epoch": 17.323135755258125, "grad_norm": 0.4965292811393738, "learning_rate": 9.955009490278068e-05, "loss": 0.0445, "step": 27180 }, { "epoch": 17.32950924155513, "grad_norm": 0.2812654972076416, "learning_rate": 9.954935689065808e-05, "loss": 0.0499, "step": 27190 }, { "epoch": 17.335882727852134, "grad_norm": 0.5554900169372559, "learning_rate": 9.954861827646398e-05, "loss": 0.0417, "step": 27200 }, { "epoch": 17.34225621414914, "grad_norm": 0.5372625589370728, "learning_rate": 9.95478790602074e-05, "loss": 0.0549, "step": 27210 }, { "epoch": 17.348629700446143, "grad_norm": 0.42561760544776917, "learning_rate": 9.954713924189731e-05, "loss": 0.033, "step": 27220 }, { "epoch": 17.355003186743147, "grad_norm": 0.5412494540214539, "learning_rate": 9.954639882154271e-05, "loss": 0.039, "step": 27230 }, { "epoch": 17.36137667304015, "grad_norm": 0.4809313416481018, "learning_rate": 9.954565779915257e-05, "loss": 0.0317, "step": 27240 }, { "epoch": 17.367750159337156, "grad_norm": 0.7674503922462463, "learning_rate": 9.954491617473591e-05, "loss": 0.048, "step": 27250 }, { "epoch": 17.37412364563416, "grad_norm": 0.5053706169128418, "learning_rate": 9.954417394830176e-05, "loss": 0.0362, "step": 27260 }, { "epoch": 17.380497131931165, "grad_norm": 0.40367111563682556, "learning_rate": 9.954343111985908e-05, "loss": 0.0322, "step": 27270 }, { "epoch": 17.38687061822817, "grad_norm": 0.5624961853027344, "learning_rate": 9.954268768941697e-05, "loss": 0.0264, "step": 27280 }, { "epoch": 17.393244104525174, "grad_norm": 0.44181734323501587, "learning_rate": 9.954194365698441e-05, "loss": 0.0348, "step": 27290 }, { "epoch": 17.39961759082218, "grad_norm": 0.743519127368927, "learning_rate": 9.954119902257047e-05, "loss": 0.0321, "step": 27300 }, { "epoch": 17.405991077119186, "grad_norm": 0.38629281520843506, "learning_rate": 9.954045378618418e-05, "loss": 0.0382, "step": 27310 }, { "epoch": 17.41236456341619, "grad_norm": 0.656932532787323, "learning_rate": 9.95397079478346e-05, "loss": 0.0481, "step": 27320 }, { "epoch": 17.418738049713195, "grad_norm": 0.628499448299408, "learning_rate": 9.953896150753082e-05, "loss": 0.035, "step": 27330 }, { "epoch": 17.4251115360102, "grad_norm": 0.5086801052093506, "learning_rate": 9.953821446528186e-05, "loss": 0.0286, "step": 27340 }, { "epoch": 17.431485022307204, "grad_norm": 0.7382622361183167, "learning_rate": 9.953746682109683e-05, "loss": 0.0444, "step": 27350 }, { "epoch": 17.437858508604208, "grad_norm": 0.30309590697288513, "learning_rate": 9.95367185749848e-05, "loss": 0.0465, "step": 27360 }, { "epoch": 17.444231994901212, "grad_norm": 0.7429732084274292, "learning_rate": 9.953596972695487e-05, "loss": 0.0378, "step": 27370 }, { "epoch": 17.450605481198217, "grad_norm": 0.9609372019767761, "learning_rate": 9.953522027701614e-05, "loss": 0.0381, "step": 27380 }, { "epoch": 17.45697896749522, "grad_norm": 0.735106885433197, "learning_rate": 9.953447022517772e-05, "loss": 0.044, "step": 27390 }, { "epoch": 17.463352453792226, "grad_norm": 0.5321285724639893, "learning_rate": 9.953371957144872e-05, "loss": 0.0351, "step": 27400 }, { "epoch": 17.46972594008923, "grad_norm": 0.7595342993736267, "learning_rate": 9.953296831583825e-05, "loss": 0.0373, "step": 27410 }, { "epoch": 17.476099426386234, "grad_norm": 0.6421709060668945, "learning_rate": 9.953221645835545e-05, "loss": 0.0425, "step": 27420 }, { "epoch": 17.48247291268324, "grad_norm": 0.5293576717376709, "learning_rate": 9.953146399900945e-05, "loss": 0.0324, "step": 27430 }, { "epoch": 17.488846398980243, "grad_norm": 0.5224354863166809, "learning_rate": 9.953071093780941e-05, "loss": 0.0533, "step": 27440 }, { "epoch": 17.495219885277248, "grad_norm": 0.5056937336921692, "learning_rate": 9.952995727476446e-05, "loss": 0.0495, "step": 27450 }, { "epoch": 17.501593371574252, "grad_norm": 0.6087573766708374, "learning_rate": 9.952920300988378e-05, "loss": 0.0395, "step": 27460 }, { "epoch": 17.507966857871256, "grad_norm": 0.7454321980476379, "learning_rate": 9.95284481431765e-05, "loss": 0.0296, "step": 27470 }, { "epoch": 17.51434034416826, "grad_norm": 0.4375068247318268, "learning_rate": 9.952769267465182e-05, "loss": 0.0609, "step": 27480 }, { "epoch": 17.520713830465265, "grad_norm": 0.49835067987442017, "learning_rate": 9.95269366043189e-05, "loss": 0.036, "step": 27490 }, { "epoch": 17.52708731676227, "grad_norm": 0.6412295699119568, "learning_rate": 9.952617993218696e-05, "loss": 0.044, "step": 27500 }, { "epoch": 17.533460803059274, "grad_norm": 0.6280710101127625, "learning_rate": 9.952542265826516e-05, "loss": 0.0655, "step": 27510 }, { "epoch": 17.53983428935628, "grad_norm": 0.4177946150302887, "learning_rate": 9.952466478256273e-05, "loss": 0.0352, "step": 27520 }, { "epoch": 17.546207775653283, "grad_norm": 0.370174765586853, "learning_rate": 9.952390630508884e-05, "loss": 0.0436, "step": 27530 }, { "epoch": 17.552581261950287, "grad_norm": 0.45558491349220276, "learning_rate": 9.952314722585274e-05, "loss": 0.0508, "step": 27540 }, { "epoch": 17.55895474824729, "grad_norm": 0.7363624572753906, "learning_rate": 9.952238754486365e-05, "loss": 0.0337, "step": 27550 }, { "epoch": 17.565328234544296, "grad_norm": 0.6047791242599487, "learning_rate": 9.952162726213078e-05, "loss": 0.0392, "step": 27560 }, { "epoch": 17.5717017208413, "grad_norm": 0.30739152431488037, "learning_rate": 9.95208663776634e-05, "loss": 0.0289, "step": 27570 }, { "epoch": 17.578075207138305, "grad_norm": 0.36529111862182617, "learning_rate": 9.952010489147074e-05, "loss": 0.038, "step": 27580 }, { "epoch": 17.58444869343531, "grad_norm": 0.27576979994773865, "learning_rate": 9.951934280356203e-05, "loss": 0.0382, "step": 27590 }, { "epoch": 17.590822179732314, "grad_norm": 0.4822746515274048, "learning_rate": 9.951858011394656e-05, "loss": 0.0383, "step": 27600 }, { "epoch": 17.597195666029318, "grad_norm": 0.5201099514961243, "learning_rate": 9.95178168226336e-05, "loss": 0.0552, "step": 27610 }, { "epoch": 17.603569152326322, "grad_norm": 0.33975285291671753, "learning_rate": 9.95170529296324e-05, "loss": 0.0527, "step": 27620 }, { "epoch": 17.609942638623327, "grad_norm": 0.39848917722702026, "learning_rate": 9.951628843495226e-05, "loss": 0.0374, "step": 27630 }, { "epoch": 17.61631612492033, "grad_norm": 0.4005988836288452, "learning_rate": 9.951552333860246e-05, "loss": 0.0494, "step": 27640 }, { "epoch": 17.622689611217336, "grad_norm": 0.4525968134403229, "learning_rate": 9.951475764059231e-05, "loss": 0.0449, "step": 27650 }, { "epoch": 17.62906309751434, "grad_norm": 0.731062650680542, "learning_rate": 9.95139913409311e-05, "loss": 0.0455, "step": 27660 }, { "epoch": 17.635436583811344, "grad_norm": 0.7559607625007629, "learning_rate": 9.951322443962815e-05, "loss": 0.0634, "step": 27670 }, { "epoch": 17.64181007010835, "grad_norm": 0.6006494164466858, "learning_rate": 9.951245693669277e-05, "loss": 0.0416, "step": 27680 }, { "epoch": 17.648183556405353, "grad_norm": 0.5289352536201477, "learning_rate": 9.951168883213428e-05, "loss": 0.0424, "step": 27690 }, { "epoch": 17.654557042702358, "grad_norm": 0.6971225142478943, "learning_rate": 9.951092012596204e-05, "loss": 0.0398, "step": 27700 }, { "epoch": 17.660930528999362, "grad_norm": 0.6305059790611267, "learning_rate": 9.951015081818537e-05, "loss": 0.0427, "step": 27710 }, { "epoch": 17.667304015296367, "grad_norm": 0.8132280111312866, "learning_rate": 9.950938090881362e-05, "loss": 0.0355, "step": 27720 }, { "epoch": 17.67367750159337, "grad_norm": 0.3476380705833435, "learning_rate": 9.950861039785614e-05, "loss": 0.0357, "step": 27730 }, { "epoch": 17.680050987890375, "grad_norm": 0.39315930008888245, "learning_rate": 9.950783928532232e-05, "loss": 0.0357, "step": 27740 }, { "epoch": 17.68642447418738, "grad_norm": 0.41412943601608276, "learning_rate": 9.950706757122151e-05, "loss": 0.0349, "step": 27750 }, { "epoch": 17.692797960484384, "grad_norm": 0.424165815114975, "learning_rate": 9.950629525556306e-05, "loss": 0.0455, "step": 27760 }, { "epoch": 17.69917144678139, "grad_norm": 0.3875099718570709, "learning_rate": 9.95055223383564e-05, "loss": 0.0466, "step": 27770 }, { "epoch": 17.705544933078393, "grad_norm": 0.5878488421440125, "learning_rate": 9.95047488196109e-05, "loss": 0.047, "step": 27780 }, { "epoch": 17.711918419375397, "grad_norm": 0.8434865474700928, "learning_rate": 9.950397469933597e-05, "loss": 0.0475, "step": 27790 }, { "epoch": 17.7182919056724, "grad_norm": 0.43690919876098633, "learning_rate": 9.950319997754098e-05, "loss": 0.0484, "step": 27800 }, { "epoch": 17.724665391969406, "grad_norm": 0.7100998163223267, "learning_rate": 9.950242465423541e-05, "loss": 0.0516, "step": 27810 }, { "epoch": 17.73103887826641, "grad_norm": 0.660385251045227, "learning_rate": 9.950164872942861e-05, "loss": 0.0491, "step": 27820 }, { "epoch": 17.737412364563415, "grad_norm": 0.8436694145202637, "learning_rate": 9.950087220313005e-05, "loss": 0.0445, "step": 27830 }, { "epoch": 17.74378585086042, "grad_norm": 0.5409918427467346, "learning_rate": 9.950009507534917e-05, "loss": 0.043, "step": 27840 }, { "epoch": 17.750159337157424, "grad_norm": 0.4436703324317932, "learning_rate": 9.949931734609538e-05, "loss": 0.0445, "step": 27850 }, { "epoch": 17.756532823454428, "grad_norm": 0.4926864206790924, "learning_rate": 9.949853901537815e-05, "loss": 0.0342, "step": 27860 }, { "epoch": 17.762906309751433, "grad_norm": 0.4110957384109497, "learning_rate": 9.949776008320695e-05, "loss": 0.0287, "step": 27870 }, { "epoch": 17.769279796048437, "grad_norm": 0.571541428565979, "learning_rate": 9.949698054959123e-05, "loss": 0.0411, "step": 27880 }, { "epoch": 17.77565328234544, "grad_norm": 1.0171386003494263, "learning_rate": 9.949620041454046e-05, "loss": 0.0625, "step": 27890 }, { "epoch": 17.782026768642446, "grad_norm": 0.44043421745300293, "learning_rate": 9.94954196780641e-05, "loss": 0.0314, "step": 27900 }, { "epoch": 17.78840025493945, "grad_norm": 0.5718318223953247, "learning_rate": 9.94946383401717e-05, "loss": 0.0412, "step": 27910 }, { "epoch": 17.794773741236455, "grad_norm": 0.5614866614341736, "learning_rate": 9.94938564008727e-05, "loss": 0.0387, "step": 27920 }, { "epoch": 17.801147227533463, "grad_norm": 0.6895573735237122, "learning_rate": 9.94930738601766e-05, "loss": 0.043, "step": 27930 }, { "epoch": 17.807520713830467, "grad_norm": 0.6153039932250977, "learning_rate": 9.949229071809293e-05, "loss": 0.0366, "step": 27940 }, { "epoch": 17.81389420012747, "grad_norm": 0.3587973415851593, "learning_rate": 9.949150697463121e-05, "loss": 0.0421, "step": 27950 }, { "epoch": 17.820267686424476, "grad_norm": 0.5704420208930969, "learning_rate": 9.949072262980094e-05, "loss": 0.0377, "step": 27960 }, { "epoch": 17.82664117272148, "grad_norm": 0.374412477016449, "learning_rate": 9.948993768361166e-05, "loss": 0.0455, "step": 27970 }, { "epoch": 17.833014659018485, "grad_norm": 0.6191907525062561, "learning_rate": 9.948915213607293e-05, "loss": 0.0397, "step": 27980 }, { "epoch": 17.83938814531549, "grad_norm": 0.4485732316970825, "learning_rate": 9.948836598719425e-05, "loss": 0.0374, "step": 27990 }, { "epoch": 17.845761631612493, "grad_norm": 0.3787826597690582, "learning_rate": 9.948757923698522e-05, "loss": 0.0433, "step": 28000 }, { "epoch": 17.852135117909498, "grad_norm": 0.44880014657974243, "learning_rate": 9.948679188545538e-05, "loss": 0.0549, "step": 28010 }, { "epoch": 17.858508604206502, "grad_norm": 0.9366286396980286, "learning_rate": 9.948600393261427e-05, "loss": 0.0409, "step": 28020 }, { "epoch": 17.864882090503507, "grad_norm": 0.5621974468231201, "learning_rate": 9.948521537847151e-05, "loss": 0.0503, "step": 28030 }, { "epoch": 17.87125557680051, "grad_norm": 0.6286352276802063, "learning_rate": 9.948442622303667e-05, "loss": 0.0391, "step": 28040 }, { "epoch": 17.877629063097515, "grad_norm": 0.7673023343086243, "learning_rate": 9.948363646631931e-05, "loss": 0.0523, "step": 28050 }, { "epoch": 17.88400254939452, "grad_norm": 0.8487848043441772, "learning_rate": 9.948284610832905e-05, "loss": 0.0437, "step": 28060 }, { "epoch": 17.890376035691524, "grad_norm": 0.5971744060516357, "learning_rate": 9.94820551490755e-05, "loss": 0.0527, "step": 28070 }, { "epoch": 17.89674952198853, "grad_norm": 0.6268824934959412, "learning_rate": 9.948126358856825e-05, "loss": 0.0427, "step": 28080 }, { "epoch": 17.903123008285533, "grad_norm": 0.6447497606277466, "learning_rate": 9.948047142681694e-05, "loss": 0.0339, "step": 28090 }, { "epoch": 17.909496494582537, "grad_norm": 0.5796855092048645, "learning_rate": 9.947967866383117e-05, "loss": 0.057, "step": 28100 }, { "epoch": 17.915869980879542, "grad_norm": 0.6518093943595886, "learning_rate": 9.947888529962061e-05, "loss": 0.036, "step": 28110 }, { "epoch": 17.922243467176546, "grad_norm": 0.6476771235466003, "learning_rate": 9.947809133419486e-05, "loss": 0.0389, "step": 28120 }, { "epoch": 17.92861695347355, "grad_norm": 0.6231204867362976, "learning_rate": 9.947729676756359e-05, "loss": 0.0375, "step": 28130 }, { "epoch": 17.934990439770555, "grad_norm": 0.5564508438110352, "learning_rate": 9.947650159973645e-05, "loss": 0.0341, "step": 28140 }, { "epoch": 17.94136392606756, "grad_norm": 0.6180081367492676, "learning_rate": 9.94757058307231e-05, "loss": 0.0304, "step": 28150 }, { "epoch": 17.947737412364564, "grad_norm": 0.7775779962539673, "learning_rate": 9.947490946053321e-05, "loss": 0.0567, "step": 28160 }, { "epoch": 17.95411089866157, "grad_norm": 0.4301908016204834, "learning_rate": 9.947411248917647e-05, "loss": 0.034, "step": 28170 }, { "epoch": 17.960484384958573, "grad_norm": 0.4990338683128357, "learning_rate": 9.947331491666255e-05, "loss": 0.0491, "step": 28180 }, { "epoch": 17.966857871255577, "grad_norm": 0.3363014757633209, "learning_rate": 9.947251674300113e-05, "loss": 0.0458, "step": 28190 }, { "epoch": 17.97323135755258, "grad_norm": 0.7903397083282471, "learning_rate": 9.947171796820192e-05, "loss": 0.0575, "step": 28200 }, { "epoch": 17.979604843849586, "grad_norm": 0.610996425151825, "learning_rate": 9.947091859227464e-05, "loss": 0.0517, "step": 28210 }, { "epoch": 17.98597833014659, "grad_norm": 0.4906894266605377, "learning_rate": 9.947011861522899e-05, "loss": 0.0275, "step": 28220 }, { "epoch": 17.992351816443595, "grad_norm": 0.5095095634460449, "learning_rate": 9.946931803707468e-05, "loss": 0.044, "step": 28230 }, { "epoch": 17.9987253027406, "grad_norm": 0.4036194980144501, "learning_rate": 9.946851685782147e-05, "loss": 0.045, "step": 28240 }, { "epoch": 18.005098789037604, "grad_norm": 0.42551663517951965, "learning_rate": 9.946771507747906e-05, "loss": 0.047, "step": 28250 }, { "epoch": 18.011472275334608, "grad_norm": 0.643452525138855, "learning_rate": 9.946691269605721e-05, "loss": 0.0497, "step": 28260 }, { "epoch": 18.017845761631612, "grad_norm": 0.5258587002754211, "learning_rate": 9.946610971356567e-05, "loss": 0.0392, "step": 28270 }, { "epoch": 18.024219247928617, "grad_norm": 0.32409507036209106, "learning_rate": 9.946530613001418e-05, "loss": 0.0347, "step": 28280 }, { "epoch": 18.03059273422562, "grad_norm": 0.48933738470077515, "learning_rate": 9.946450194541253e-05, "loss": 0.033, "step": 28290 }, { "epoch": 18.036966220522626, "grad_norm": 0.5729212760925293, "learning_rate": 9.946369715977047e-05, "loss": 0.0334, "step": 28300 }, { "epoch": 18.04333970681963, "grad_norm": 0.6380464434623718, "learning_rate": 9.94628917730978e-05, "loss": 0.0378, "step": 28310 }, { "epoch": 18.049713193116634, "grad_norm": 0.6099402904510498, "learning_rate": 9.946208578540428e-05, "loss": 0.0388, "step": 28320 }, { "epoch": 18.05608667941364, "grad_norm": 0.7217431664466858, "learning_rate": 9.946127919669972e-05, "loss": 0.0374, "step": 28330 }, { "epoch": 18.062460165710643, "grad_norm": 0.8351543545722961, "learning_rate": 9.946047200699393e-05, "loss": 0.048, "step": 28340 }, { "epoch": 18.068833652007648, "grad_norm": 0.507265031337738, "learning_rate": 9.94596642162967e-05, "loss": 0.038, "step": 28350 }, { "epoch": 18.075207138304652, "grad_norm": 0.38116517663002014, "learning_rate": 9.945885582461785e-05, "loss": 0.0356, "step": 28360 }, { "epoch": 18.081580624601656, "grad_norm": 0.6168778538703918, "learning_rate": 9.94580468319672e-05, "loss": 0.0355, "step": 28370 }, { "epoch": 18.08795411089866, "grad_norm": 0.5395082831382751, "learning_rate": 9.94572372383546e-05, "loss": 0.0375, "step": 28380 }, { "epoch": 18.094327597195665, "grad_norm": 0.5192626118659973, "learning_rate": 9.945642704378986e-05, "loss": 0.036, "step": 28390 }, { "epoch": 18.10070108349267, "grad_norm": 0.6316375136375427, "learning_rate": 9.945561624828285e-05, "loss": 0.0493, "step": 28400 }, { "epoch": 18.107074569789674, "grad_norm": 0.5782895684242249, "learning_rate": 9.94548048518434e-05, "loss": 0.0338, "step": 28410 }, { "epoch": 18.11344805608668, "grad_norm": 0.5730478167533875, "learning_rate": 9.945399285448137e-05, "loss": 0.0543, "step": 28420 }, { "epoch": 18.119821542383683, "grad_norm": 0.5736696124076843, "learning_rate": 9.945318025620665e-05, "loss": 0.0467, "step": 28430 }, { "epoch": 18.126195028680687, "grad_norm": 0.9262635707855225, "learning_rate": 9.945236705702908e-05, "loss": 0.0419, "step": 28440 }, { "epoch": 18.13256851497769, "grad_norm": 0.5693608522415161, "learning_rate": 9.945155325695858e-05, "loss": 0.0401, "step": 28450 }, { "epoch": 18.138942001274696, "grad_norm": 0.45460256934165955, "learning_rate": 9.9450738856005e-05, "loss": 0.0326, "step": 28460 }, { "epoch": 18.1453154875717, "grad_norm": 0.4986417293548584, "learning_rate": 9.944992385417826e-05, "loss": 0.0331, "step": 28470 }, { "epoch": 18.151688973868705, "grad_norm": 0.44941723346710205, "learning_rate": 9.944910825148826e-05, "loss": 0.0449, "step": 28480 }, { "epoch": 18.15806246016571, "grad_norm": 0.5503662824630737, "learning_rate": 9.94482920479449e-05, "loss": 0.0555, "step": 28490 }, { "epoch": 18.164435946462714, "grad_norm": 0.7120167016983032, "learning_rate": 9.944747524355811e-05, "loss": 0.0314, "step": 28500 }, { "epoch": 18.170809432759718, "grad_norm": 0.819176435470581, "learning_rate": 9.944665783833782e-05, "loss": 0.0474, "step": 28510 }, { "epoch": 18.177182919056722, "grad_norm": 0.5169119238853455, "learning_rate": 9.944583983229394e-05, "loss": 0.0481, "step": 28520 }, { "epoch": 18.183556405353727, "grad_norm": 0.5913349986076355, "learning_rate": 9.944502122543641e-05, "loss": 0.0402, "step": 28530 }, { "epoch": 18.18992989165073, "grad_norm": 0.8381208181381226, "learning_rate": 9.94442020177752e-05, "loss": 0.0516, "step": 28540 }, { "epoch": 18.196303377947736, "grad_norm": 0.9311365485191345, "learning_rate": 9.944338220932025e-05, "loss": 0.0477, "step": 28550 }, { "epoch": 18.202676864244744, "grad_norm": 0.8508256077766418, "learning_rate": 9.944256180008154e-05, "loss": 0.0374, "step": 28560 }, { "epoch": 18.209050350541748, "grad_norm": 1.5421631336212158, "learning_rate": 9.9441740790069e-05, "loss": 0.0826, "step": 28570 }, { "epoch": 18.215423836838752, "grad_norm": 0.5102139711380005, "learning_rate": 9.944091917929264e-05, "loss": 0.042, "step": 28580 }, { "epoch": 18.221797323135757, "grad_norm": 0.60739666223526, "learning_rate": 9.944009696776244e-05, "loss": 0.0454, "step": 28590 }, { "epoch": 18.22817080943276, "grad_norm": 0.6634073257446289, "learning_rate": 9.943927415548837e-05, "loss": 0.0418, "step": 28600 }, { "epoch": 18.234544295729766, "grad_norm": 0.6340003609657288, "learning_rate": 9.943845074248044e-05, "loss": 0.0534, "step": 28610 }, { "epoch": 18.24091778202677, "grad_norm": 0.3241119980812073, "learning_rate": 9.943762672874865e-05, "loss": 0.0435, "step": 28620 }, { "epoch": 18.247291268323774, "grad_norm": 0.4300916790962219, "learning_rate": 9.943680211430304e-05, "loss": 0.0414, "step": 28630 }, { "epoch": 18.25366475462078, "grad_norm": 0.310761034488678, "learning_rate": 9.943597689915361e-05, "loss": 0.0357, "step": 28640 }, { "epoch": 18.260038240917783, "grad_norm": 0.7176051139831543, "learning_rate": 9.943515108331036e-05, "loss": 0.0393, "step": 28650 }, { "epoch": 18.266411727214788, "grad_norm": 0.5310367941856384, "learning_rate": 9.943432466678336e-05, "loss": 0.0361, "step": 28660 }, { "epoch": 18.272785213511792, "grad_norm": 0.3363970220088959, "learning_rate": 9.943349764958266e-05, "loss": 0.0426, "step": 28670 }, { "epoch": 18.279158699808796, "grad_norm": 1.1641829013824463, "learning_rate": 9.943267003171825e-05, "loss": 0.0407, "step": 28680 }, { "epoch": 18.2855321861058, "grad_norm": 0.9257823824882507, "learning_rate": 9.943184181320026e-05, "loss": 0.0446, "step": 28690 }, { "epoch": 18.291905672402805, "grad_norm": 0.8590275645256042, "learning_rate": 9.943101299403871e-05, "loss": 0.0389, "step": 28700 }, { "epoch": 18.29827915869981, "grad_norm": 0.4272668659687042, "learning_rate": 9.943018357424369e-05, "loss": 0.0373, "step": 28710 }, { "epoch": 18.304652644996814, "grad_norm": 0.33628958463668823, "learning_rate": 9.942935355382525e-05, "loss": 0.0455, "step": 28720 }, { "epoch": 18.31102613129382, "grad_norm": 0.5181383490562439, "learning_rate": 9.942852293279351e-05, "loss": 0.0503, "step": 28730 }, { "epoch": 18.317399617590823, "grad_norm": 0.5718725919723511, "learning_rate": 9.942769171115856e-05, "loss": 0.0615, "step": 28740 }, { "epoch": 18.323773103887827, "grad_norm": 0.7419205904006958, "learning_rate": 9.942685988893045e-05, "loss": 0.0352, "step": 28750 }, { "epoch": 18.33014659018483, "grad_norm": 0.6127808094024658, "learning_rate": 9.942602746611936e-05, "loss": 0.0517, "step": 28760 }, { "epoch": 18.336520076481836, "grad_norm": 0.40552347898483276, "learning_rate": 9.942519444273537e-05, "loss": 0.0312, "step": 28770 }, { "epoch": 18.34289356277884, "grad_norm": 0.3745090067386627, "learning_rate": 9.942436081878859e-05, "loss": 0.03, "step": 28780 }, { "epoch": 18.349267049075845, "grad_norm": 0.4008125960826874, "learning_rate": 9.942352659428916e-05, "loss": 0.0398, "step": 28790 }, { "epoch": 18.35564053537285, "grad_norm": 0.6054003834724426, "learning_rate": 9.942269176924722e-05, "loss": 0.0464, "step": 28800 }, { "epoch": 18.362014021669854, "grad_norm": 0.47119346261024475, "learning_rate": 9.942185634367292e-05, "loss": 0.0341, "step": 28810 }, { "epoch": 18.368387507966858, "grad_norm": 0.6217157244682312, "learning_rate": 9.942102031757639e-05, "loss": 0.0334, "step": 28820 }, { "epoch": 18.374760994263863, "grad_norm": 0.6488839387893677, "learning_rate": 9.942018369096781e-05, "loss": 0.0328, "step": 28830 }, { "epoch": 18.381134480560867, "grad_norm": 0.5669146776199341, "learning_rate": 9.941934646385734e-05, "loss": 0.0361, "step": 28840 }, { "epoch": 18.38750796685787, "grad_norm": 0.8450927138328552, "learning_rate": 9.941850863625515e-05, "loss": 0.0348, "step": 28850 }, { "epoch": 18.393881453154876, "grad_norm": 0.5762073397636414, "learning_rate": 9.941767020817142e-05, "loss": 0.0549, "step": 28860 }, { "epoch": 18.40025493945188, "grad_norm": 0.490662544965744, "learning_rate": 9.941683117961634e-05, "loss": 0.0394, "step": 28870 }, { "epoch": 18.406628425748885, "grad_norm": 0.8585314154624939, "learning_rate": 9.941599155060011e-05, "loss": 0.0541, "step": 28880 }, { "epoch": 18.41300191204589, "grad_norm": 0.70564866065979, "learning_rate": 9.941515132113292e-05, "loss": 0.0391, "step": 28890 }, { "epoch": 18.419375398342893, "grad_norm": 0.5997070670127869, "learning_rate": 9.941431049122498e-05, "loss": 0.0425, "step": 28900 }, { "epoch": 18.425748884639898, "grad_norm": 0.4389268755912781, "learning_rate": 9.94134690608865e-05, "loss": 0.0424, "step": 28910 }, { "epoch": 18.432122370936902, "grad_norm": 0.694188117980957, "learning_rate": 9.941262703012775e-05, "loss": 0.0403, "step": 28920 }, { "epoch": 18.438495857233907, "grad_norm": 0.6166446208953857, "learning_rate": 9.941178439895891e-05, "loss": 0.0342, "step": 28930 }, { "epoch": 18.44486934353091, "grad_norm": 0.502869725227356, "learning_rate": 9.941094116739025e-05, "loss": 0.0365, "step": 28940 }, { "epoch": 18.451242829827915, "grad_norm": 0.46384647488594055, "learning_rate": 9.941009733543199e-05, "loss": 0.0431, "step": 28950 }, { "epoch": 18.45761631612492, "grad_norm": 0.6524788737297058, "learning_rate": 9.940925290309439e-05, "loss": 0.0441, "step": 28960 }, { "epoch": 18.463989802421924, "grad_norm": 0.6036084294319153, "learning_rate": 9.940840787038773e-05, "loss": 0.0306, "step": 28970 }, { "epoch": 18.47036328871893, "grad_norm": 0.387070894241333, "learning_rate": 9.940756223732226e-05, "loss": 0.0387, "step": 28980 }, { "epoch": 18.476736775015933, "grad_norm": 0.3576427102088928, "learning_rate": 9.940671600390828e-05, "loss": 0.0289, "step": 28990 }, { "epoch": 18.483110261312937, "grad_norm": 0.7115057706832886, "learning_rate": 9.940586917015602e-05, "loss": 0.0429, "step": 29000 }, { "epoch": 18.489483747609942, "grad_norm": 0.5900391936302185, "learning_rate": 9.940502173607582e-05, "loss": 0.0415, "step": 29010 }, { "epoch": 18.495857233906946, "grad_norm": 0.2714010775089264, "learning_rate": 9.940417370167796e-05, "loss": 0.0449, "step": 29020 }, { "epoch": 18.50223072020395, "grad_norm": 0.4475362300872803, "learning_rate": 9.940332506697274e-05, "loss": 0.0322, "step": 29030 }, { "epoch": 18.508604206500955, "grad_norm": 0.7675207257270813, "learning_rate": 9.940247583197047e-05, "loss": 0.0409, "step": 29040 }, { "epoch": 18.51497769279796, "grad_norm": 0.626508355140686, "learning_rate": 9.940162599668149e-05, "loss": 0.0387, "step": 29050 }, { "epoch": 18.521351179094964, "grad_norm": 0.2268603891134262, "learning_rate": 9.94007755611161e-05, "loss": 0.0392, "step": 29060 }, { "epoch": 18.52772466539197, "grad_norm": 0.4541628062725067, "learning_rate": 9.939992452528466e-05, "loss": 0.0448, "step": 29070 }, { "epoch": 18.534098151688973, "grad_norm": 0.6295639872550964, "learning_rate": 9.939907288919747e-05, "loss": 0.0316, "step": 29080 }, { "epoch": 18.540471637985977, "grad_norm": 0.6493943333625793, "learning_rate": 9.939822065286493e-05, "loss": 0.0334, "step": 29090 }, { "epoch": 18.54684512428298, "grad_norm": 0.5731953978538513, "learning_rate": 9.939736781629736e-05, "loss": 0.036, "step": 29100 }, { "epoch": 18.553218610579986, "grad_norm": 0.4753711223602295, "learning_rate": 9.939651437950512e-05, "loss": 0.0426, "step": 29110 }, { "epoch": 18.55959209687699, "grad_norm": 0.6338264346122742, "learning_rate": 9.939566034249861e-05, "loss": 0.0332, "step": 29120 }, { "epoch": 18.565965583173995, "grad_norm": 0.47118523716926575, "learning_rate": 9.939480570528819e-05, "loss": 0.0382, "step": 29130 }, { "epoch": 18.572339069471, "grad_norm": 0.7295191287994385, "learning_rate": 9.939395046788422e-05, "loss": 0.0261, "step": 29140 }, { "epoch": 18.578712555768004, "grad_norm": 0.4889342486858368, "learning_rate": 9.939309463029713e-05, "loss": 0.0411, "step": 29150 }, { "epoch": 18.585086042065008, "grad_norm": 0.6443138718605042, "learning_rate": 9.939223819253731e-05, "loss": 0.0415, "step": 29160 }, { "epoch": 18.591459528362012, "grad_norm": 0.5513317584991455, "learning_rate": 9.939138115461516e-05, "loss": 0.0398, "step": 29170 }, { "epoch": 18.597833014659017, "grad_norm": 0.6576837301254272, "learning_rate": 9.939052351654111e-05, "loss": 0.0587, "step": 29180 }, { "epoch": 18.604206500956025, "grad_norm": 0.48124411702156067, "learning_rate": 9.938966527832555e-05, "loss": 0.0348, "step": 29190 }, { "epoch": 18.61057998725303, "grad_norm": 0.9200423359870911, "learning_rate": 9.938880643997893e-05, "loss": 0.0438, "step": 29200 }, { "epoch": 18.616953473550033, "grad_norm": 0.7226598858833313, "learning_rate": 9.938794700151168e-05, "loss": 0.0392, "step": 29210 }, { "epoch": 18.623326959847038, "grad_norm": 0.5152060985565186, "learning_rate": 9.938708696293424e-05, "loss": 0.0454, "step": 29220 }, { "epoch": 18.629700446144042, "grad_norm": 0.41204386949539185, "learning_rate": 9.938622632425707e-05, "loss": 0.0463, "step": 29230 }, { "epoch": 18.636073932441047, "grad_norm": 0.6263889670372009, "learning_rate": 9.93853650854906e-05, "loss": 0.0498, "step": 29240 }, { "epoch": 18.64244741873805, "grad_norm": 0.4043098986148834, "learning_rate": 9.938450324664535e-05, "loss": 0.0297, "step": 29250 }, { "epoch": 18.648820905035056, "grad_norm": 0.47623881697654724, "learning_rate": 9.938364080773175e-05, "loss": 0.0359, "step": 29260 }, { "epoch": 18.65519439133206, "grad_norm": 0.49849221110343933, "learning_rate": 9.938277776876028e-05, "loss": 0.0391, "step": 29270 }, { "epoch": 18.661567877629064, "grad_norm": 0.3075621724128723, "learning_rate": 9.938191412974143e-05, "loss": 0.0381, "step": 29280 }, { "epoch": 18.66794136392607, "grad_norm": 0.7212318778038025, "learning_rate": 9.93810498906857e-05, "loss": 0.0391, "step": 29290 }, { "epoch": 18.674314850223073, "grad_norm": 0.44030001759529114, "learning_rate": 9.93801850516036e-05, "loss": 0.0359, "step": 29300 }, { "epoch": 18.680688336520078, "grad_norm": 0.324235737323761, "learning_rate": 9.937931961250564e-05, "loss": 0.0378, "step": 29310 }, { "epoch": 18.687061822817082, "grad_norm": 0.6790275573730469, "learning_rate": 9.93784535734023e-05, "loss": 0.0457, "step": 29320 }, { "epoch": 18.693435309114086, "grad_norm": 0.508127748966217, "learning_rate": 9.937758693430414e-05, "loss": 0.0323, "step": 29330 }, { "epoch": 18.69980879541109, "grad_norm": 0.569007933139801, "learning_rate": 9.937671969522169e-05, "loss": 0.0318, "step": 29340 }, { "epoch": 18.706182281708095, "grad_norm": 0.45339545607566833, "learning_rate": 9.937585185616546e-05, "loss": 0.0361, "step": 29350 }, { "epoch": 18.7125557680051, "grad_norm": 0.6050662994384766, "learning_rate": 9.937498341714602e-05, "loss": 0.0458, "step": 29360 }, { "epoch": 18.718929254302104, "grad_norm": 0.5710495710372925, "learning_rate": 9.93741143781739e-05, "loss": 0.041, "step": 29370 }, { "epoch": 18.72530274059911, "grad_norm": 0.4593247175216675, "learning_rate": 9.937324473925969e-05, "loss": 0.0367, "step": 29380 }, { "epoch": 18.731676226896113, "grad_norm": 0.6179113388061523, "learning_rate": 9.937237450041393e-05, "loss": 0.0425, "step": 29390 }, { "epoch": 18.738049713193117, "grad_norm": 0.7028769850730896, "learning_rate": 9.93715036616472e-05, "loss": 0.0343, "step": 29400 }, { "epoch": 18.74442319949012, "grad_norm": 0.4269255995750427, "learning_rate": 9.93706322229701e-05, "loss": 0.037, "step": 29410 }, { "epoch": 18.750796685787126, "grad_norm": 0.4630488157272339, "learning_rate": 9.936976018439321e-05, "loss": 0.041, "step": 29420 }, { "epoch": 18.75717017208413, "grad_norm": 0.43804648518562317, "learning_rate": 9.936888754592711e-05, "loss": 0.0302, "step": 29430 }, { "epoch": 18.763543658381135, "grad_norm": 0.7423017024993896, "learning_rate": 9.93680143075824e-05, "loss": 0.0425, "step": 29440 }, { "epoch": 18.76991714467814, "grad_norm": 0.6526033878326416, "learning_rate": 9.936714046936972e-05, "loss": 0.0429, "step": 29450 }, { "epoch": 18.776290630975144, "grad_norm": 0.8027708530426025, "learning_rate": 9.936626603129969e-05, "loss": 0.0431, "step": 29460 }, { "epoch": 18.782664117272148, "grad_norm": 0.4835878610610962, "learning_rate": 9.936539099338289e-05, "loss": 0.0277, "step": 29470 }, { "epoch": 18.789037603569152, "grad_norm": 0.5432160496711731, "learning_rate": 9.936451535562999e-05, "loss": 0.0497, "step": 29480 }, { "epoch": 18.795411089866157, "grad_norm": 0.5552000999450684, "learning_rate": 9.936363911805162e-05, "loss": 0.0337, "step": 29490 }, { "epoch": 18.80178457616316, "grad_norm": 0.2930489778518677, "learning_rate": 9.936276228065842e-05, "loss": 0.046, "step": 29500 }, { "epoch": 18.808158062460166, "grad_norm": 0.541323184967041, "learning_rate": 9.936188484346107e-05, "loss": 0.0472, "step": 29510 }, { "epoch": 18.81453154875717, "grad_norm": 0.4334292709827423, "learning_rate": 9.936100680647021e-05, "loss": 0.0328, "step": 29520 }, { "epoch": 18.820905035054174, "grad_norm": 0.6419758200645447, "learning_rate": 9.93601281696965e-05, "loss": 0.0344, "step": 29530 }, { "epoch": 18.82727852135118, "grad_norm": 0.44829273223876953, "learning_rate": 9.935924893315065e-05, "loss": 0.0416, "step": 29540 }, { "epoch": 18.833652007648183, "grad_norm": 0.3829452395439148, "learning_rate": 9.93583690968433e-05, "loss": 0.0447, "step": 29550 }, { "epoch": 18.840025493945188, "grad_norm": 0.4967927038669586, "learning_rate": 9.935748866078518e-05, "loss": 0.0357, "step": 29560 }, { "epoch": 18.846398980242192, "grad_norm": 0.7680585980415344, "learning_rate": 9.935660762498696e-05, "loss": 0.0333, "step": 29570 }, { "epoch": 18.852772466539196, "grad_norm": 0.4204786419868469, "learning_rate": 9.935572598945937e-05, "loss": 0.0454, "step": 29580 }, { "epoch": 18.8591459528362, "grad_norm": 0.6508587002754211, "learning_rate": 9.935484375421309e-05, "loss": 0.0428, "step": 29590 }, { "epoch": 18.865519439133205, "grad_norm": 0.3644666373729706, "learning_rate": 9.935396091925889e-05, "loss": 0.0302, "step": 29600 }, { "epoch": 18.87189292543021, "grad_norm": 0.5341609716415405, "learning_rate": 9.935307748460744e-05, "loss": 0.0337, "step": 29610 }, { "epoch": 18.878266411727214, "grad_norm": 0.38523614406585693, "learning_rate": 9.935219345026951e-05, "loss": 0.0219, "step": 29620 }, { "epoch": 18.88463989802422, "grad_norm": 0.39375802874565125, "learning_rate": 9.935130881625583e-05, "loss": 0.0418, "step": 29630 }, { "epoch": 18.891013384321223, "grad_norm": 0.35671064257621765, "learning_rate": 9.935042358257715e-05, "loss": 0.0314, "step": 29640 }, { "epoch": 18.897386870618227, "grad_norm": 0.525241494178772, "learning_rate": 9.934953774924424e-05, "loss": 0.0336, "step": 29650 }, { "epoch": 18.90376035691523, "grad_norm": 0.487211674451828, "learning_rate": 9.934865131626784e-05, "loss": 0.0406, "step": 29660 }, { "epoch": 18.910133843212236, "grad_norm": 0.32760319113731384, "learning_rate": 9.934776428365875e-05, "loss": 0.0475, "step": 29670 }, { "epoch": 18.91650732950924, "grad_norm": 1.010451078414917, "learning_rate": 9.934687665142773e-05, "loss": 0.0371, "step": 29680 }, { "epoch": 18.922880815806245, "grad_norm": 0.4842472970485687, "learning_rate": 9.934598841958555e-05, "loss": 0.0337, "step": 29690 }, { "epoch": 18.92925430210325, "grad_norm": 0.7316793203353882, "learning_rate": 9.934509958814304e-05, "loss": 0.0368, "step": 29700 }, { "epoch": 18.935627788400254, "grad_norm": 0.5729722380638123, "learning_rate": 9.934421015711097e-05, "loss": 0.0417, "step": 29710 }, { "epoch": 18.942001274697258, "grad_norm": 0.34655457735061646, "learning_rate": 9.934332012650017e-05, "loss": 0.0445, "step": 29720 }, { "epoch": 18.948374760994263, "grad_norm": 0.4603334367275238, "learning_rate": 9.934242949632143e-05, "loss": 0.0372, "step": 29730 }, { "epoch": 18.954748247291267, "grad_norm": 0.3202730119228363, "learning_rate": 9.934153826658557e-05, "loss": 0.0451, "step": 29740 }, { "epoch": 18.96112173358827, "grad_norm": 0.3960217833518982, "learning_rate": 9.934064643730345e-05, "loss": 0.0363, "step": 29750 }, { "epoch": 18.967495219885276, "grad_norm": 0.5772948861122131, "learning_rate": 9.93397540084859e-05, "loss": 0.0362, "step": 29760 }, { "epoch": 18.97386870618228, "grad_norm": 0.5629610419273376, "learning_rate": 9.933886098014374e-05, "loss": 0.0417, "step": 29770 }, { "epoch": 18.980242192479285, "grad_norm": 0.7069763541221619, "learning_rate": 9.933796735228785e-05, "loss": 0.0328, "step": 29780 }, { "epoch": 18.98661567877629, "grad_norm": 0.8418090343475342, "learning_rate": 9.933707312492906e-05, "loss": 0.0418, "step": 29790 }, { "epoch": 18.992989165073293, "grad_norm": 0.6206852793693542, "learning_rate": 9.933617829807825e-05, "loss": 0.032, "step": 29800 }, { "epoch": 18.999362651370298, "grad_norm": 0.9068872928619385, "learning_rate": 9.933528287174629e-05, "loss": 0.0359, "step": 29810 }, { "epoch": 19.005736137667306, "grad_norm": 0.5571058988571167, "learning_rate": 9.933438684594407e-05, "loss": 0.0353, "step": 29820 }, { "epoch": 19.01210962396431, "grad_norm": 0.7168715000152588, "learning_rate": 9.933349022068248e-05, "loss": 0.0412, "step": 29830 }, { "epoch": 19.018483110261315, "grad_norm": 0.45915642380714417, "learning_rate": 9.93325929959724e-05, "loss": 0.0587, "step": 29840 }, { "epoch": 19.02485659655832, "grad_norm": 0.8532593250274658, "learning_rate": 9.933169517182473e-05, "loss": 0.0415, "step": 29850 }, { "epoch": 19.031230082855323, "grad_norm": 0.6767350435256958, "learning_rate": 9.933079674825038e-05, "loss": 0.0539, "step": 29860 }, { "epoch": 19.037603569152328, "grad_norm": 0.4505847692489624, "learning_rate": 9.93298977252603e-05, "loss": 0.0402, "step": 29870 }, { "epoch": 19.043977055449332, "grad_norm": 0.40878984332084656, "learning_rate": 9.932899810286535e-05, "loss": 0.035, "step": 29880 }, { "epoch": 19.050350541746337, "grad_norm": 0.6644253134727478, "learning_rate": 9.932809788107653e-05, "loss": 0.0441, "step": 29890 }, { "epoch": 19.05672402804334, "grad_norm": 0.8221628069877625, "learning_rate": 9.932719705990474e-05, "loss": 0.0394, "step": 29900 }, { "epoch": 19.063097514340345, "grad_norm": 0.4734356105327606, "learning_rate": 9.932629563936094e-05, "loss": 0.0459, "step": 29910 }, { "epoch": 19.06947100063735, "grad_norm": 0.7842265367507935, "learning_rate": 9.932539361945606e-05, "loss": 0.0428, "step": 29920 }, { "epoch": 19.075844486934354, "grad_norm": 0.6345791816711426, "learning_rate": 9.932449100020107e-05, "loss": 0.0313, "step": 29930 }, { "epoch": 19.08221797323136, "grad_norm": 0.4702848196029663, "learning_rate": 9.932358778160697e-05, "loss": 0.0467, "step": 29940 }, { "epoch": 19.088591459528363, "grad_norm": 0.46002575755119324, "learning_rate": 9.932268396368469e-05, "loss": 0.0483, "step": 29950 }, { "epoch": 19.094964945825367, "grad_norm": 0.6662471890449524, "learning_rate": 9.932177954644524e-05, "loss": 0.0331, "step": 29960 }, { "epoch": 19.101338432122372, "grad_norm": 0.425302118062973, "learning_rate": 9.93208745298996e-05, "loss": 0.0374, "step": 29970 }, { "epoch": 19.107711918419376, "grad_norm": 0.5735731720924377, "learning_rate": 9.931996891405877e-05, "loss": 0.0371, "step": 29980 }, { "epoch": 19.11408540471638, "grad_norm": 0.4919684827327728, "learning_rate": 9.931906269893373e-05, "loss": 0.0411, "step": 29990 }, { "epoch": 19.120458891013385, "grad_norm": 0.4763234257698059, "learning_rate": 9.931815588453555e-05, "loss": 0.037, "step": 30000 }, { "epoch": 19.12683237731039, "grad_norm": 0.5986319780349731, "learning_rate": 9.93172484708752e-05, "loss": 0.0327, "step": 30010 }, { "epoch": 19.133205863607394, "grad_norm": 0.5059881806373596, "learning_rate": 9.93163404579637e-05, "loss": 0.0348, "step": 30020 }, { "epoch": 19.1395793499044, "grad_norm": 0.8306627869606018, "learning_rate": 9.931543184581211e-05, "loss": 0.0388, "step": 30030 }, { "epoch": 19.145952836201403, "grad_norm": 0.28949907422065735, "learning_rate": 9.931452263443146e-05, "loss": 0.026, "step": 30040 }, { "epoch": 19.152326322498407, "grad_norm": 0.6049891114234924, "learning_rate": 9.931361282383282e-05, "loss": 0.0407, "step": 30050 }, { "epoch": 19.15869980879541, "grad_norm": 0.7492976188659668, "learning_rate": 9.93127024140272e-05, "loss": 0.057, "step": 30060 }, { "epoch": 19.165073295092416, "grad_norm": 0.6889173984527588, "learning_rate": 9.931179140502569e-05, "loss": 0.036, "step": 30070 }, { "epoch": 19.17144678138942, "grad_norm": 0.6662917137145996, "learning_rate": 9.931087979683937e-05, "loss": 0.0403, "step": 30080 }, { "epoch": 19.177820267686425, "grad_norm": 0.6832749247550964, "learning_rate": 9.930996758947929e-05, "loss": 0.0317, "step": 30090 }, { "epoch": 19.18419375398343, "grad_norm": 0.3887522220611572, "learning_rate": 9.930905478295655e-05, "loss": 0.0325, "step": 30100 }, { "epoch": 19.190567240280433, "grad_norm": 0.49981802701950073, "learning_rate": 9.930814137728224e-05, "loss": 0.0403, "step": 30110 }, { "epoch": 19.196940726577438, "grad_norm": 0.5295567512512207, "learning_rate": 9.930722737246747e-05, "loss": 0.0534, "step": 30120 }, { "epoch": 19.203314212874442, "grad_norm": 0.7078784108161926, "learning_rate": 9.930631276852331e-05, "loss": 0.0375, "step": 30130 }, { "epoch": 19.209687699171447, "grad_norm": 0.45943304896354675, "learning_rate": 9.930539756546091e-05, "loss": 0.0362, "step": 30140 }, { "epoch": 19.21606118546845, "grad_norm": 0.622878909111023, "learning_rate": 9.930448176329139e-05, "loss": 0.0317, "step": 30150 }, { "epoch": 19.222434671765456, "grad_norm": 0.4926847517490387, "learning_rate": 9.930356536202585e-05, "loss": 0.0352, "step": 30160 }, { "epoch": 19.22880815806246, "grad_norm": 0.9150048494338989, "learning_rate": 9.930264836167544e-05, "loss": 0.0372, "step": 30170 }, { "epoch": 19.235181644359464, "grad_norm": 0.4262911081314087, "learning_rate": 9.93017307622513e-05, "loss": 0.0362, "step": 30180 }, { "epoch": 19.24155513065647, "grad_norm": 0.41691434383392334, "learning_rate": 9.930081256376461e-05, "loss": 0.0275, "step": 30190 }, { "epoch": 19.247928616953473, "grad_norm": 0.8108784556388855, "learning_rate": 9.929989376622646e-05, "loss": 0.0361, "step": 30200 }, { "epoch": 19.254302103250478, "grad_norm": 0.41191670298576355, "learning_rate": 9.929897436964807e-05, "loss": 0.0368, "step": 30210 }, { "epoch": 19.260675589547482, "grad_norm": 0.4684663414955139, "learning_rate": 9.92980543740406e-05, "loss": 0.0315, "step": 30220 }, { "epoch": 19.267049075844486, "grad_norm": 0.555474042892456, "learning_rate": 9.929713377941523e-05, "loss": 0.0308, "step": 30230 }, { "epoch": 19.27342256214149, "grad_norm": 0.43855366110801697, "learning_rate": 9.929621258578312e-05, "loss": 0.0361, "step": 30240 }, { "epoch": 19.279796048438495, "grad_norm": 0.6779509782791138, "learning_rate": 9.929529079315551e-05, "loss": 0.0462, "step": 30250 }, { "epoch": 19.2861695347355, "grad_norm": 0.6538847088813782, "learning_rate": 9.929436840154355e-05, "loss": 0.0373, "step": 30260 }, { "epoch": 19.292543021032504, "grad_norm": 0.48029494285583496, "learning_rate": 9.929344541095848e-05, "loss": 0.0259, "step": 30270 }, { "epoch": 19.29891650732951, "grad_norm": 0.41328009963035583, "learning_rate": 9.92925218214115e-05, "loss": 0.0445, "step": 30280 }, { "epoch": 19.305289993626513, "grad_norm": 0.44898107647895813, "learning_rate": 9.929159763291387e-05, "loss": 0.0513, "step": 30290 }, { "epoch": 19.311663479923517, "grad_norm": 0.48360171914100647, "learning_rate": 9.929067284547677e-05, "loss": 0.0424, "step": 30300 }, { "epoch": 19.31803696622052, "grad_norm": 0.5419996380805969, "learning_rate": 9.928974745911146e-05, "loss": 0.0305, "step": 30310 }, { "epoch": 19.324410452517526, "grad_norm": 0.4238566756248474, "learning_rate": 9.928882147382917e-05, "loss": 0.0511, "step": 30320 }, { "epoch": 19.33078393881453, "grad_norm": 0.44084325432777405, "learning_rate": 9.928789488964118e-05, "loss": 0.0418, "step": 30330 }, { "epoch": 19.337157425111535, "grad_norm": 0.48497602343559265, "learning_rate": 9.928696770655872e-05, "loss": 0.0459, "step": 30340 }, { "epoch": 19.34353091140854, "grad_norm": 0.734505832195282, "learning_rate": 9.928603992459306e-05, "loss": 0.0358, "step": 30350 }, { "epoch": 19.349904397705544, "grad_norm": 0.9555234313011169, "learning_rate": 9.928511154375551e-05, "loss": 0.0411, "step": 30360 }, { "epoch": 19.356277884002548, "grad_norm": 0.42453572154045105, "learning_rate": 9.928418256405729e-05, "loss": 0.0488, "step": 30370 }, { "epoch": 19.362651370299552, "grad_norm": 0.35866954922676086, "learning_rate": 9.928325298550973e-05, "loss": 0.0491, "step": 30380 }, { "epoch": 19.369024856596557, "grad_norm": 0.6244329810142517, "learning_rate": 9.928232280812413e-05, "loss": 0.0489, "step": 30390 }, { "epoch": 19.37539834289356, "grad_norm": 0.6762614250183105, "learning_rate": 9.928139203191176e-05, "loss": 0.0467, "step": 30400 }, { "epoch": 19.381771829190566, "grad_norm": 0.5204678177833557, "learning_rate": 9.928046065688396e-05, "loss": 0.0358, "step": 30410 }, { "epoch": 19.38814531548757, "grad_norm": 0.6249240040779114, "learning_rate": 9.927952868305203e-05, "loss": 0.0328, "step": 30420 }, { "epoch": 19.394518801784574, "grad_norm": 0.5215117931365967, "learning_rate": 9.927859611042729e-05, "loss": 0.0343, "step": 30430 }, { "epoch": 19.40089228808158, "grad_norm": 0.569284975528717, "learning_rate": 9.92776629390211e-05, "loss": 0.0354, "step": 30440 }, { "epoch": 19.407265774378587, "grad_norm": 0.7911409139633179, "learning_rate": 9.927672916884476e-05, "loss": 0.047, "step": 30450 }, { "epoch": 19.41363926067559, "grad_norm": 0.5741820931434631, "learning_rate": 9.927579479990964e-05, "loss": 0.0331, "step": 30460 }, { "epoch": 19.420012746972596, "grad_norm": 0.4142441749572754, "learning_rate": 9.927485983222709e-05, "loss": 0.0338, "step": 30470 }, { "epoch": 19.4263862332696, "grad_norm": 0.44358593225479126, "learning_rate": 9.927392426580847e-05, "loss": 0.0386, "step": 30480 }, { "epoch": 19.432759719566604, "grad_norm": 0.3871499300003052, "learning_rate": 9.927298810066514e-05, "loss": 0.0264, "step": 30490 }, { "epoch": 19.43913320586361, "grad_norm": 0.7500891089439392, "learning_rate": 9.92720513368085e-05, "loss": 0.0438, "step": 30500 }, { "epoch": 19.445506692160613, "grad_norm": 0.7373768091201782, "learning_rate": 9.927111397424991e-05, "loss": 0.0371, "step": 30510 }, { "epoch": 19.451880178457618, "grad_norm": 0.5835720896720886, "learning_rate": 9.927017601300077e-05, "loss": 0.0439, "step": 30520 }, { "epoch": 19.458253664754622, "grad_norm": 0.519656240940094, "learning_rate": 9.926923745307246e-05, "loss": 0.04, "step": 30530 }, { "epoch": 19.464627151051626, "grad_norm": 0.5708519220352173, "learning_rate": 9.926829829447641e-05, "loss": 0.0388, "step": 30540 }, { "epoch": 19.47100063734863, "grad_norm": 0.45617032051086426, "learning_rate": 9.926735853722401e-05, "loss": 0.0366, "step": 30550 }, { "epoch": 19.477374123645635, "grad_norm": 0.4663662016391754, "learning_rate": 9.926641818132669e-05, "loss": 0.0312, "step": 30560 }, { "epoch": 19.48374760994264, "grad_norm": 0.5140721797943115, "learning_rate": 9.926547722679586e-05, "loss": 0.0541, "step": 30570 }, { "epoch": 19.490121096239644, "grad_norm": 0.5515722036361694, "learning_rate": 9.9264535673643e-05, "loss": 0.0473, "step": 30580 }, { "epoch": 19.49649458253665, "grad_norm": 0.6542114615440369, "learning_rate": 9.92635935218795e-05, "loss": 0.0513, "step": 30590 }, { "epoch": 19.502868068833653, "grad_norm": 0.5618543028831482, "learning_rate": 9.926265077151682e-05, "loss": 0.0337, "step": 30600 }, { "epoch": 19.509241555130657, "grad_norm": 0.3177219033241272, "learning_rate": 9.926170742256643e-05, "loss": 0.0483, "step": 30610 }, { "epoch": 19.51561504142766, "grad_norm": 0.41662949323654175, "learning_rate": 9.926076347503979e-05, "loss": 0.035, "step": 30620 }, { "epoch": 19.521988527724666, "grad_norm": 0.5761436223983765, "learning_rate": 9.925981892894834e-05, "loss": 0.0446, "step": 30630 }, { "epoch": 19.52836201402167, "grad_norm": 0.6705889701843262, "learning_rate": 9.925887378430361e-05, "loss": 0.0337, "step": 30640 }, { "epoch": 19.534735500318675, "grad_norm": 0.4133850336074829, "learning_rate": 9.925792804111703e-05, "loss": 0.0305, "step": 30650 }, { "epoch": 19.54110898661568, "grad_norm": 0.69058758020401, "learning_rate": 9.925698169940013e-05, "loss": 0.0458, "step": 30660 }, { "epoch": 19.547482472912684, "grad_norm": 0.7333258390426636, "learning_rate": 9.925603475916439e-05, "loss": 0.0335, "step": 30670 }, { "epoch": 19.553855959209688, "grad_norm": 0.5526683330535889, "learning_rate": 9.925508722042134e-05, "loss": 0.0362, "step": 30680 }, { "epoch": 19.560229445506693, "grad_norm": 0.6082180738449097, "learning_rate": 9.925413908318247e-05, "loss": 0.0399, "step": 30690 }, { "epoch": 19.566602931803697, "grad_norm": 0.38242268562316895, "learning_rate": 9.925319034745928e-05, "loss": 0.0382, "step": 30700 }, { "epoch": 19.5729764181007, "grad_norm": 0.4181293547153473, "learning_rate": 9.925224101326335e-05, "loss": 0.0414, "step": 30710 }, { "epoch": 19.579349904397706, "grad_norm": 0.4419921040534973, "learning_rate": 9.925129108060618e-05, "loss": 0.0382, "step": 30720 }, { "epoch": 19.58572339069471, "grad_norm": 0.5485382080078125, "learning_rate": 9.925034054949933e-05, "loss": 0.0428, "step": 30730 }, { "epoch": 19.592096876991715, "grad_norm": 1.3604178428649902, "learning_rate": 9.924938941995434e-05, "loss": 0.0469, "step": 30740 }, { "epoch": 19.59847036328872, "grad_norm": 0.7106276750564575, "learning_rate": 9.924843769198276e-05, "loss": 0.0431, "step": 30750 }, { "epoch": 19.604843849585723, "grad_norm": 0.6182977557182312, "learning_rate": 9.924748536559618e-05, "loss": 0.0423, "step": 30760 }, { "epoch": 19.611217335882728, "grad_norm": 0.7213919758796692, "learning_rate": 9.924653244080614e-05, "loss": 0.036, "step": 30770 }, { "epoch": 19.617590822179732, "grad_norm": 0.4512563645839691, "learning_rate": 9.924557891762425e-05, "loss": 0.0374, "step": 30780 }, { "epoch": 19.623964308476737, "grad_norm": 0.8976032137870789, "learning_rate": 9.924462479606207e-05, "loss": 0.0396, "step": 30790 }, { "epoch": 19.63033779477374, "grad_norm": 0.44669491052627563, "learning_rate": 9.92436700761312e-05, "loss": 0.035, "step": 30800 }, { "epoch": 19.636711281070745, "grad_norm": 0.49275678396224976, "learning_rate": 9.924271475784325e-05, "loss": 0.0396, "step": 30810 }, { "epoch": 19.64308476736775, "grad_norm": 0.40826186537742615, "learning_rate": 9.924175884120983e-05, "loss": 0.0418, "step": 30820 }, { "epoch": 19.649458253664754, "grad_norm": 0.7369924783706665, "learning_rate": 9.924080232624254e-05, "loss": 0.033, "step": 30830 }, { "epoch": 19.65583173996176, "grad_norm": 0.5148134231567383, "learning_rate": 9.923984521295302e-05, "loss": 0.0342, "step": 30840 }, { "epoch": 19.662205226258763, "grad_norm": 0.734971284866333, "learning_rate": 9.923888750135289e-05, "loss": 0.0339, "step": 30850 }, { "epoch": 19.668578712555767, "grad_norm": 0.6121981739997864, "learning_rate": 9.923792919145377e-05, "loss": 0.031, "step": 30860 }, { "epoch": 19.674952198852772, "grad_norm": 0.49764755368232727, "learning_rate": 9.923697028326733e-05, "loss": 0.0441, "step": 30870 }, { "epoch": 19.681325685149776, "grad_norm": 0.5548027753829956, "learning_rate": 9.923601077680523e-05, "loss": 0.0362, "step": 30880 }, { "epoch": 19.68769917144678, "grad_norm": 0.8668046593666077, "learning_rate": 9.923505067207909e-05, "loss": 0.04, "step": 30890 }, { "epoch": 19.694072657743785, "grad_norm": 0.4896691143512726, "learning_rate": 9.92340899691006e-05, "loss": 0.0469, "step": 30900 }, { "epoch": 19.70044614404079, "grad_norm": 0.49435439705848694, "learning_rate": 9.923312866788145e-05, "loss": 0.0384, "step": 30910 }, { "epoch": 19.706819630337794, "grad_norm": 0.9273898601531982, "learning_rate": 9.923216676843329e-05, "loss": 0.0528, "step": 30920 }, { "epoch": 19.7131931166348, "grad_norm": 0.7938147187232971, "learning_rate": 9.923120427076782e-05, "loss": 0.041, "step": 30930 }, { "epoch": 19.719566602931803, "grad_norm": 0.4727533459663391, "learning_rate": 9.923024117489674e-05, "loss": 0.033, "step": 30940 }, { "epoch": 19.725940089228807, "grad_norm": 0.790953516960144, "learning_rate": 9.922927748083175e-05, "loss": 0.0432, "step": 30950 }, { "epoch": 19.73231357552581, "grad_norm": 0.3349590301513672, "learning_rate": 9.922831318858457e-05, "loss": 0.029, "step": 30960 }, { "epoch": 19.738687061822816, "grad_norm": 0.6204319000244141, "learning_rate": 9.922734829816688e-05, "loss": 0.044, "step": 30970 }, { "epoch": 19.74506054811982, "grad_norm": 0.5555143356323242, "learning_rate": 9.922638280959045e-05, "loss": 0.0315, "step": 30980 }, { "epoch": 19.751434034416825, "grad_norm": 0.36220285296440125, "learning_rate": 9.922541672286699e-05, "loss": 0.0336, "step": 30990 }, { "epoch": 19.75780752071383, "grad_norm": 0.49983927607536316, "learning_rate": 9.922445003800823e-05, "loss": 0.0423, "step": 31000 }, { "epoch": 19.764181007010833, "grad_norm": 0.6305146813392639, "learning_rate": 9.922348275502593e-05, "loss": 0.0393, "step": 31010 }, { "epoch": 19.770554493307838, "grad_norm": 0.4823150634765625, "learning_rate": 9.922251487393184e-05, "loss": 0.0446, "step": 31020 }, { "epoch": 19.776927979604842, "grad_norm": 0.2731166183948517, "learning_rate": 9.922154639473771e-05, "loss": 0.0274, "step": 31030 }, { "epoch": 19.783301465901847, "grad_norm": 0.3890707492828369, "learning_rate": 9.922057731745535e-05, "loss": 0.0336, "step": 31040 }, { "epoch": 19.78967495219885, "grad_norm": 0.33899155259132385, "learning_rate": 9.921960764209648e-05, "loss": 0.0323, "step": 31050 }, { "epoch": 19.79604843849586, "grad_norm": 0.5514169335365295, "learning_rate": 9.92186373686729e-05, "loss": 0.0322, "step": 31060 }, { "epoch": 19.80242192479286, "grad_norm": 0.5769526958465576, "learning_rate": 9.921766649719642e-05, "loss": 0.0316, "step": 31070 }, { "epoch": 19.808795411089868, "grad_norm": 0.41242578625679016, "learning_rate": 9.921669502767882e-05, "loss": 0.0327, "step": 31080 }, { "epoch": 19.815168897386872, "grad_norm": 0.8164010047912598, "learning_rate": 9.921572296013191e-05, "loss": 0.0351, "step": 31090 }, { "epoch": 19.821542383683877, "grad_norm": 0.4675033390522003, "learning_rate": 9.921475029456749e-05, "loss": 0.0426, "step": 31100 }, { "epoch": 19.82791586998088, "grad_norm": 0.6598914265632629, "learning_rate": 9.92137770309974e-05, "loss": 0.0341, "step": 31110 }, { "epoch": 19.834289356277885, "grad_norm": 0.6407289505004883, "learning_rate": 9.921280316943344e-05, "loss": 0.0456, "step": 31120 }, { "epoch": 19.84066284257489, "grad_norm": 0.2580413520336151, "learning_rate": 9.921182870988745e-05, "loss": 0.0338, "step": 31130 }, { "epoch": 19.847036328871894, "grad_norm": 0.7133857011795044, "learning_rate": 9.92108536523713e-05, "loss": 0.0433, "step": 31140 }, { "epoch": 19.8534098151689, "grad_norm": 0.5073107481002808, "learning_rate": 9.92098779968968e-05, "loss": 0.0324, "step": 31150 }, { "epoch": 19.859783301465903, "grad_norm": 0.6896771192550659, "learning_rate": 9.920890174347583e-05, "loss": 0.0329, "step": 31160 }, { "epoch": 19.866156787762907, "grad_norm": 0.4341394007205963, "learning_rate": 9.920792489212023e-05, "loss": 0.0395, "step": 31170 }, { "epoch": 19.872530274059912, "grad_norm": 0.5641313195228577, "learning_rate": 9.920694744284189e-05, "loss": 0.0466, "step": 31180 }, { "epoch": 19.878903760356916, "grad_norm": 0.5229228138923645, "learning_rate": 9.920596939565268e-05, "loss": 0.0379, "step": 31190 }, { "epoch": 19.88527724665392, "grad_norm": 0.4775829315185547, "learning_rate": 9.920499075056448e-05, "loss": 0.0355, "step": 31200 }, { "epoch": 19.891650732950925, "grad_norm": 0.46678051352500916, "learning_rate": 9.92040115075892e-05, "loss": 0.0316, "step": 31210 }, { "epoch": 19.89802421924793, "grad_norm": 0.4558700621128082, "learning_rate": 9.920303166673872e-05, "loss": 0.0375, "step": 31220 }, { "epoch": 19.904397705544934, "grad_norm": 0.4194766581058502, "learning_rate": 9.920205122802494e-05, "loss": 0.0394, "step": 31230 }, { "epoch": 19.91077119184194, "grad_norm": 0.4703217148780823, "learning_rate": 9.920107019145978e-05, "loss": 0.058, "step": 31240 }, { "epoch": 19.917144678138943, "grad_norm": 0.6007516384124756, "learning_rate": 9.920008855705519e-05, "loss": 0.032, "step": 31250 }, { "epoch": 19.923518164435947, "grad_norm": 0.5609459280967712, "learning_rate": 9.919910632482303e-05, "loss": 0.0452, "step": 31260 }, { "epoch": 19.92989165073295, "grad_norm": 0.5445705652236938, "learning_rate": 9.919812349477531e-05, "loss": 0.049, "step": 31270 }, { "epoch": 19.936265137029956, "grad_norm": 0.6331278085708618, "learning_rate": 9.919714006692393e-05, "loss": 0.0316, "step": 31280 }, { "epoch": 19.94263862332696, "grad_norm": 0.8607646822929382, "learning_rate": 9.919615604128084e-05, "loss": 0.0602, "step": 31290 }, { "epoch": 19.949012109623965, "grad_norm": 0.7605261206626892, "learning_rate": 9.919517141785802e-05, "loss": 0.0471, "step": 31300 }, { "epoch": 19.95538559592097, "grad_norm": 0.5033801198005676, "learning_rate": 9.91941861966674e-05, "loss": 0.0615, "step": 31310 }, { "epoch": 19.961759082217974, "grad_norm": 1.1008477210998535, "learning_rate": 9.919320037772098e-05, "loss": 0.0626, "step": 31320 }, { "epoch": 19.968132568514978, "grad_norm": 0.6704133749008179, "learning_rate": 9.919221396103073e-05, "loss": 0.0372, "step": 31330 }, { "epoch": 19.974506054811982, "grad_norm": 0.43298137187957764, "learning_rate": 9.919122694660863e-05, "loss": 0.0319, "step": 31340 }, { "epoch": 19.980879541108987, "grad_norm": 0.7401928901672363, "learning_rate": 9.919023933446669e-05, "loss": 0.0363, "step": 31350 }, { "epoch": 19.98725302740599, "grad_norm": 0.8605602979660034, "learning_rate": 9.918925112461688e-05, "loss": 0.0396, "step": 31360 }, { "epoch": 19.993626513702996, "grad_norm": 0.8451436758041382, "learning_rate": 9.918826231707126e-05, "loss": 0.0551, "step": 31370 }, { "epoch": 20.0, "grad_norm": 0.7706111669540405, "learning_rate": 9.918727291184178e-05, "loss": 0.0472, "step": 31380 }, { "epoch": 20.006373486297004, "grad_norm": 0.4842163920402527, "learning_rate": 9.918628290894049e-05, "loss": 0.0331, "step": 31390 }, { "epoch": 20.01274697259401, "grad_norm": 0.588696300983429, "learning_rate": 9.918529230837945e-05, "loss": 0.039, "step": 31400 }, { "epoch": 20.019120458891013, "grad_norm": 0.6585944294929504, "learning_rate": 9.918430111017066e-05, "loss": 0.0373, "step": 31410 }, { "epoch": 20.025493945188018, "grad_norm": 0.5827052593231201, "learning_rate": 9.918330931432616e-05, "loss": 0.035, "step": 31420 }, { "epoch": 20.031867431485022, "grad_norm": 0.6124544739723206, "learning_rate": 9.918231692085803e-05, "loss": 0.0437, "step": 31430 }, { "epoch": 20.038240917782026, "grad_norm": 0.41697749495506287, "learning_rate": 9.918132392977832e-05, "loss": 0.036, "step": 31440 }, { "epoch": 20.04461440407903, "grad_norm": 0.7197396755218506, "learning_rate": 9.918033034109908e-05, "loss": 0.0472, "step": 31450 }, { "epoch": 20.050987890376035, "grad_norm": 0.47255319356918335, "learning_rate": 9.917933615483238e-05, "loss": 0.0327, "step": 31460 }, { "epoch": 20.05736137667304, "grad_norm": 0.7539225816726685, "learning_rate": 9.917834137099032e-05, "loss": 0.0362, "step": 31470 }, { "epoch": 20.063734862970044, "grad_norm": 0.395638108253479, "learning_rate": 9.9177345989585e-05, "loss": 0.0233, "step": 31480 }, { "epoch": 20.07010834926705, "grad_norm": 0.5284397006034851, "learning_rate": 9.917635001062847e-05, "loss": 0.0387, "step": 31490 }, { "epoch": 20.076481835564053, "grad_norm": 0.42776015400886536, "learning_rate": 9.917535343413287e-05, "loss": 0.0469, "step": 31500 }, { "epoch": 20.082855321861057, "grad_norm": 0.4647214412689209, "learning_rate": 9.917435626011029e-05, "loss": 0.0305, "step": 31510 }, { "epoch": 20.08922880815806, "grad_norm": 0.5136334896087646, "learning_rate": 9.917335848857286e-05, "loss": 0.0334, "step": 31520 }, { "epoch": 20.095602294455066, "grad_norm": 0.6021164655685425, "learning_rate": 9.917236011953268e-05, "loss": 0.0501, "step": 31530 }, { "epoch": 20.10197578075207, "grad_norm": 0.47220155596733093, "learning_rate": 9.917136115300192e-05, "loss": 0.0482, "step": 31540 }, { "epoch": 20.108349267049075, "grad_norm": 0.5563257336616516, "learning_rate": 9.917036158899269e-05, "loss": 0.0298, "step": 31550 }, { "epoch": 20.11472275334608, "grad_norm": 0.44852110743522644, "learning_rate": 9.916936142751713e-05, "loss": 0.0278, "step": 31560 }, { "epoch": 20.121096239643084, "grad_norm": 0.4002387523651123, "learning_rate": 9.916836066858743e-05, "loss": 0.0316, "step": 31570 }, { "epoch": 20.127469725940088, "grad_norm": 0.6808817386627197, "learning_rate": 9.91673593122157e-05, "loss": 0.0382, "step": 31580 }, { "epoch": 20.133843212237093, "grad_norm": 0.47998136281967163, "learning_rate": 9.916635735841415e-05, "loss": 0.028, "step": 31590 }, { "epoch": 20.140216698534097, "grad_norm": 0.7633264064788818, "learning_rate": 9.916535480719493e-05, "loss": 0.0331, "step": 31600 }, { "epoch": 20.1465901848311, "grad_norm": 0.6634160280227661, "learning_rate": 9.916435165857023e-05, "loss": 0.0339, "step": 31610 }, { "epoch": 20.152963671128106, "grad_norm": 0.48326966166496277, "learning_rate": 9.916334791255224e-05, "loss": 0.0488, "step": 31620 }, { "epoch": 20.15933715742511, "grad_norm": 0.39859503507614136, "learning_rate": 9.916234356915315e-05, "loss": 0.0455, "step": 31630 }, { "epoch": 20.165710643722115, "grad_norm": 0.6801707148551941, "learning_rate": 9.916133862838517e-05, "loss": 0.0389, "step": 31640 }, { "epoch": 20.17208413001912, "grad_norm": 0.617017388343811, "learning_rate": 9.916033309026051e-05, "loss": 0.0456, "step": 31650 }, { "epoch": 20.178457616316123, "grad_norm": 0.79550701379776, "learning_rate": 9.91593269547914e-05, "loss": 0.048, "step": 31660 }, { "epoch": 20.184831102613128, "grad_norm": 0.7285943031311035, "learning_rate": 9.915832022199003e-05, "loss": 0.029, "step": 31670 }, { "epoch": 20.191204588910132, "grad_norm": 0.49208498001098633, "learning_rate": 9.915731289186866e-05, "loss": 0.0355, "step": 31680 }, { "epoch": 20.197578075207137, "grad_norm": 0.5854390859603882, "learning_rate": 9.915630496443955e-05, "loss": 0.0369, "step": 31690 }, { "epoch": 20.203951561504145, "grad_norm": 0.7893449068069458, "learning_rate": 9.91552964397149e-05, "loss": 0.0406, "step": 31700 }, { "epoch": 20.21032504780115, "grad_norm": 0.5641257762908936, "learning_rate": 9.9154287317707e-05, "loss": 0.037, "step": 31710 }, { "epoch": 20.216698534098153, "grad_norm": 0.37174034118652344, "learning_rate": 9.91532775984281e-05, "loss": 0.0388, "step": 31720 }, { "epoch": 20.223072020395158, "grad_norm": 0.4521788954734802, "learning_rate": 9.915226728189047e-05, "loss": 0.047, "step": 31730 }, { "epoch": 20.229445506692162, "grad_norm": 0.43040764331817627, "learning_rate": 9.915125636810638e-05, "loss": 0.0293, "step": 31740 }, { "epoch": 20.235818992989167, "grad_norm": 0.5899751782417297, "learning_rate": 9.915024485708812e-05, "loss": 0.038, "step": 31750 }, { "epoch": 20.24219247928617, "grad_norm": 0.5041904449462891, "learning_rate": 9.914923274884798e-05, "loss": 0.042, "step": 31760 }, { "epoch": 20.248565965583175, "grad_norm": 0.3741162419319153, "learning_rate": 9.914822004339825e-05, "loss": 0.0455, "step": 31770 }, { "epoch": 20.25493945188018, "grad_norm": 0.5811709761619568, "learning_rate": 9.914720674075125e-05, "loss": 0.0395, "step": 31780 }, { "epoch": 20.261312938177184, "grad_norm": 0.5764346718788147, "learning_rate": 9.914619284091929e-05, "loss": 0.0522, "step": 31790 }, { "epoch": 20.26768642447419, "grad_norm": 0.9789597392082214, "learning_rate": 9.914517834391468e-05, "loss": 0.0552, "step": 31800 }, { "epoch": 20.274059910771193, "grad_norm": 0.58121657371521, "learning_rate": 9.914416324974975e-05, "loss": 0.0504, "step": 31810 }, { "epoch": 20.280433397068197, "grad_norm": 0.6952957510948181, "learning_rate": 9.914314755843683e-05, "loss": 0.0434, "step": 31820 }, { "epoch": 20.2868068833652, "grad_norm": 0.6010002493858337, "learning_rate": 9.914213126998828e-05, "loss": 0.0421, "step": 31830 }, { "epoch": 20.293180369662206, "grad_norm": 0.728855311870575, "learning_rate": 9.914111438441644e-05, "loss": 0.0627, "step": 31840 }, { "epoch": 20.29955385595921, "grad_norm": 0.623874843120575, "learning_rate": 9.914009690173364e-05, "loss": 0.0442, "step": 31850 }, { "epoch": 20.305927342256215, "grad_norm": 0.411680668592453, "learning_rate": 9.91390788219523e-05, "loss": 0.0457, "step": 31860 }, { "epoch": 20.31230082855322, "grad_norm": 0.48180603981018066, "learning_rate": 9.913806014508474e-05, "loss": 0.0356, "step": 31870 }, { "epoch": 20.318674314850224, "grad_norm": 0.4155711233615875, "learning_rate": 9.913704087114335e-05, "loss": 0.0393, "step": 31880 }, { "epoch": 20.325047801147228, "grad_norm": 0.8110759258270264, "learning_rate": 9.913602100014053e-05, "loss": 0.0428, "step": 31890 }, { "epoch": 20.331421287444233, "grad_norm": 0.6050277352333069, "learning_rate": 9.913500053208866e-05, "loss": 0.0563, "step": 31900 }, { "epoch": 20.337794773741237, "grad_norm": 0.3016187846660614, "learning_rate": 9.913397946700014e-05, "loss": 0.0332, "step": 31910 }, { "epoch": 20.34416826003824, "grad_norm": 0.3538776934146881, "learning_rate": 9.913295780488738e-05, "loss": 0.0499, "step": 31920 }, { "epoch": 20.350541746335246, "grad_norm": 0.5012523531913757, "learning_rate": 9.91319355457628e-05, "loss": 0.0589, "step": 31930 }, { "epoch": 20.35691523263225, "grad_norm": 0.39226800203323364, "learning_rate": 9.913091268963881e-05, "loss": 0.0291, "step": 31940 }, { "epoch": 20.363288718929255, "grad_norm": 0.4774281978607178, "learning_rate": 9.912988923652784e-05, "loss": 0.0358, "step": 31950 }, { "epoch": 20.36966220522626, "grad_norm": 0.6373028755187988, "learning_rate": 9.912886518644234e-05, "loss": 0.0442, "step": 31960 }, { "epoch": 20.376035691523263, "grad_norm": 0.44271302223205566, "learning_rate": 9.912784053939473e-05, "loss": 0.0403, "step": 31970 }, { "epoch": 20.382409177820268, "grad_norm": 0.47209858894348145, "learning_rate": 9.912681529539748e-05, "loss": 0.0339, "step": 31980 }, { "epoch": 20.388782664117272, "grad_norm": 0.4574304223060608, "learning_rate": 9.912578945446305e-05, "loss": 0.0347, "step": 31990 }, { "epoch": 20.395156150414277, "grad_norm": 0.6760044693946838, "learning_rate": 9.912476301660387e-05, "loss": 0.0495, "step": 32000 }, { "epoch": 20.40152963671128, "grad_norm": 0.4937826097011566, "learning_rate": 9.912373598183246e-05, "loss": 0.0375, "step": 32010 }, { "epoch": 20.407903123008285, "grad_norm": 0.4969538748264313, "learning_rate": 9.912270835016126e-05, "loss": 0.0375, "step": 32020 }, { "epoch": 20.41427660930529, "grad_norm": 0.7108178734779358, "learning_rate": 9.912168012160278e-05, "loss": 0.0442, "step": 32030 }, { "epoch": 20.420650095602294, "grad_norm": 0.3456435203552246, "learning_rate": 9.912065129616952e-05, "loss": 0.0299, "step": 32040 }, { "epoch": 20.4270235818993, "grad_norm": 0.5581284165382385, "learning_rate": 9.911962187387397e-05, "loss": 0.0461, "step": 32050 }, { "epoch": 20.433397068196303, "grad_norm": 0.40117985010147095, "learning_rate": 9.911859185472862e-05, "loss": 0.036, "step": 32060 }, { "epoch": 20.439770554493307, "grad_norm": 0.8666346669197083, "learning_rate": 9.911756123874602e-05, "loss": 0.0381, "step": 32070 }, { "epoch": 20.446144040790312, "grad_norm": 0.397077739238739, "learning_rate": 9.911653002593867e-05, "loss": 0.0427, "step": 32080 }, { "epoch": 20.452517527087316, "grad_norm": 0.9644503593444824, "learning_rate": 9.91154982163191e-05, "loss": 0.0338, "step": 32090 }, { "epoch": 20.45889101338432, "grad_norm": 0.339598149061203, "learning_rate": 9.911446580989988e-05, "loss": 0.0366, "step": 32100 }, { "epoch": 20.465264499681325, "grad_norm": 0.7253049612045288, "learning_rate": 9.911343280669349e-05, "loss": 0.035, "step": 32110 }, { "epoch": 20.47163798597833, "grad_norm": 0.7502745389938354, "learning_rate": 9.911239920671254e-05, "loss": 0.0399, "step": 32120 }, { "epoch": 20.478011472275334, "grad_norm": 0.4695591330528259, "learning_rate": 9.911136500996956e-05, "loss": 0.0286, "step": 32130 }, { "epoch": 20.48438495857234, "grad_norm": 0.5460295081138611, "learning_rate": 9.911033021647715e-05, "loss": 0.0384, "step": 32140 }, { "epoch": 20.490758444869343, "grad_norm": 0.8266238570213318, "learning_rate": 9.910929482624784e-05, "loss": 0.0304, "step": 32150 }, { "epoch": 20.497131931166347, "grad_norm": 0.4289080798625946, "learning_rate": 9.910825883929424e-05, "loss": 0.0269, "step": 32160 }, { "epoch": 20.50350541746335, "grad_norm": 0.42191484570503235, "learning_rate": 9.910722225562891e-05, "loss": 0.0337, "step": 32170 }, { "epoch": 20.509878903760356, "grad_norm": 0.3428705334663391, "learning_rate": 9.910618507526448e-05, "loss": 0.0394, "step": 32180 }, { "epoch": 20.51625239005736, "grad_norm": 0.44297927618026733, "learning_rate": 9.910514729821354e-05, "loss": 0.0272, "step": 32190 }, { "epoch": 20.522625876354365, "grad_norm": 0.7043351531028748, "learning_rate": 9.91041089244887e-05, "loss": 0.0286, "step": 32200 }, { "epoch": 20.52899936265137, "grad_norm": 0.5189981460571289, "learning_rate": 9.910306995410257e-05, "loss": 0.041, "step": 32210 }, { "epoch": 20.535372848948374, "grad_norm": 0.7327574491500854, "learning_rate": 9.910203038706778e-05, "loss": 0.0361, "step": 32220 }, { "epoch": 20.541746335245378, "grad_norm": 0.5154539346694946, "learning_rate": 9.910099022339696e-05, "loss": 0.0375, "step": 32230 }, { "epoch": 20.548119821542382, "grad_norm": 0.42485809326171875, "learning_rate": 9.909994946310275e-05, "loss": 0.0438, "step": 32240 }, { "epoch": 20.554493307839387, "grad_norm": 0.46833348274230957, "learning_rate": 9.909890810619779e-05, "loss": 0.0362, "step": 32250 }, { "epoch": 20.56086679413639, "grad_norm": 0.44683268666267395, "learning_rate": 9.909786615269474e-05, "loss": 0.0463, "step": 32260 }, { "epoch": 20.567240280433396, "grad_norm": 0.39845260977745056, "learning_rate": 9.909682360260625e-05, "loss": 0.0333, "step": 32270 }, { "epoch": 20.5736137667304, "grad_norm": 0.3318699896335602, "learning_rate": 9.909578045594503e-05, "loss": 0.0304, "step": 32280 }, { "epoch": 20.579987253027404, "grad_norm": 0.46723198890686035, "learning_rate": 9.909473671272371e-05, "loss": 0.0258, "step": 32290 }, { "epoch": 20.58636073932441, "grad_norm": 0.2995181977748871, "learning_rate": 9.909369237295498e-05, "loss": 0.0318, "step": 32300 }, { "epoch": 20.592734225621413, "grad_norm": 1.0364646911621094, "learning_rate": 9.909264743665152e-05, "loss": 0.0389, "step": 32310 }, { "epoch": 20.59910771191842, "grad_norm": 0.6058845520019531, "learning_rate": 9.909160190382607e-05, "loss": 0.0493, "step": 32320 }, { "epoch": 20.605481198215422, "grad_norm": 0.6111282706260681, "learning_rate": 9.90905557744913e-05, "loss": 0.0314, "step": 32330 }, { "epoch": 20.61185468451243, "grad_norm": 0.5166767239570618, "learning_rate": 9.908950904865993e-05, "loss": 0.0378, "step": 32340 }, { "epoch": 20.618228170809434, "grad_norm": 0.3108362555503845, "learning_rate": 9.908846172634467e-05, "loss": 0.0363, "step": 32350 }, { "epoch": 20.62460165710644, "grad_norm": 0.3442677855491638, "learning_rate": 9.908741380755826e-05, "loss": 0.0319, "step": 32360 }, { "epoch": 20.630975143403443, "grad_norm": 0.3573359549045563, "learning_rate": 9.908636529231341e-05, "loss": 0.0324, "step": 32370 }, { "epoch": 20.637348629700448, "grad_norm": 0.3774619698524475, "learning_rate": 9.90853161806229e-05, "loss": 0.0397, "step": 32380 }, { "epoch": 20.643722115997452, "grad_norm": 0.351256400346756, "learning_rate": 9.908426647249942e-05, "loss": 0.0327, "step": 32390 }, { "epoch": 20.650095602294456, "grad_norm": 0.5157501101493835, "learning_rate": 9.90832161679558e-05, "loss": 0.0303, "step": 32400 }, { "epoch": 20.65646908859146, "grad_norm": 0.543287456035614, "learning_rate": 9.908216526700474e-05, "loss": 0.037, "step": 32410 }, { "epoch": 20.662842574888465, "grad_norm": 0.47728314995765686, "learning_rate": 9.908111376965903e-05, "loss": 0.0366, "step": 32420 }, { "epoch": 20.66921606118547, "grad_norm": 0.6593474745750427, "learning_rate": 9.908006167593145e-05, "loss": 0.0418, "step": 32430 }, { "epoch": 20.675589547482474, "grad_norm": 0.2726775109767914, "learning_rate": 9.907900898583477e-05, "loss": 0.0408, "step": 32440 }, { "epoch": 20.68196303377948, "grad_norm": 0.5829458832740784, "learning_rate": 9.90779556993818e-05, "loss": 0.0328, "step": 32450 }, { "epoch": 20.688336520076483, "grad_norm": 0.6999732255935669, "learning_rate": 9.907690181658534e-05, "loss": 0.0383, "step": 32460 }, { "epoch": 20.694710006373487, "grad_norm": 0.47810623049736023, "learning_rate": 9.907584733745818e-05, "loss": 0.0327, "step": 32470 }, { "epoch": 20.70108349267049, "grad_norm": 0.6252797245979309, "learning_rate": 9.907479226201315e-05, "loss": 0.0522, "step": 32480 }, { "epoch": 20.707456978967496, "grad_norm": 0.47271057963371277, "learning_rate": 9.907373659026305e-05, "loss": 0.0301, "step": 32490 }, { "epoch": 20.7138304652645, "grad_norm": 0.3314191997051239, "learning_rate": 9.907268032222071e-05, "loss": 0.0419, "step": 32500 }, { "epoch": 20.720203951561505, "grad_norm": 0.8567216992378235, "learning_rate": 9.907162345789898e-05, "loss": 0.042, "step": 32510 }, { "epoch": 20.72657743785851, "grad_norm": 0.6678280234336853, "learning_rate": 9.907056599731069e-05, "loss": 0.0335, "step": 32520 }, { "epoch": 20.732950924155514, "grad_norm": 0.4286373257637024, "learning_rate": 9.90695079404687e-05, "loss": 0.034, "step": 32530 }, { "epoch": 20.739324410452518, "grad_norm": 0.340762734413147, "learning_rate": 9.906844928738585e-05, "loss": 0.03, "step": 32540 }, { "epoch": 20.745697896749522, "grad_norm": 0.36485153436660767, "learning_rate": 9.906739003807503e-05, "loss": 0.051, "step": 32550 }, { "epoch": 20.752071383046527, "grad_norm": 0.3059370517730713, "learning_rate": 9.906633019254907e-05, "loss": 0.0523, "step": 32560 }, { "epoch": 20.75844486934353, "grad_norm": 0.5903476476669312, "learning_rate": 9.906526975082088e-05, "loss": 0.0455, "step": 32570 }, { "epoch": 20.764818355640536, "grad_norm": 0.932595431804657, "learning_rate": 9.906420871290335e-05, "loss": 0.0448, "step": 32580 }, { "epoch": 20.77119184193754, "grad_norm": 0.6965762376785278, "learning_rate": 9.906314707880935e-05, "loss": 0.0416, "step": 32590 }, { "epoch": 20.777565328234544, "grad_norm": 0.44755029678344727, "learning_rate": 9.906208484855179e-05, "loss": 0.034, "step": 32600 }, { "epoch": 20.78393881453155, "grad_norm": 0.42917704582214355, "learning_rate": 9.906102202214358e-05, "loss": 0.0418, "step": 32610 }, { "epoch": 20.790312300828553, "grad_norm": 0.37115615606307983, "learning_rate": 9.905995859959762e-05, "loss": 0.0399, "step": 32620 }, { "epoch": 20.796685787125558, "grad_norm": 0.44704997539520264, "learning_rate": 9.905889458092686e-05, "loss": 0.0424, "step": 32630 }, { "epoch": 20.803059273422562, "grad_norm": 0.3460333049297333, "learning_rate": 9.90578299661442e-05, "loss": 0.0311, "step": 32640 }, { "epoch": 20.809432759719567, "grad_norm": 0.7044121026992798, "learning_rate": 9.90567647552626e-05, "loss": 0.0318, "step": 32650 }, { "epoch": 20.81580624601657, "grad_norm": 0.5258469581604004, "learning_rate": 9.905569894829497e-05, "loss": 0.0443, "step": 32660 }, { "epoch": 20.822179732313575, "grad_norm": 0.33766207098960876, "learning_rate": 9.905463254525428e-05, "loss": 0.0448, "step": 32670 }, { "epoch": 20.82855321861058, "grad_norm": 0.4253661334514618, "learning_rate": 9.905356554615353e-05, "loss": 0.0273, "step": 32680 }, { "epoch": 20.834926704907584, "grad_norm": 0.39972105622291565, "learning_rate": 9.905249795100561e-05, "loss": 0.0372, "step": 32690 }, { "epoch": 20.84130019120459, "grad_norm": 0.4974951148033142, "learning_rate": 9.905142975982354e-05, "loss": 0.0394, "step": 32700 }, { "epoch": 20.847673677501593, "grad_norm": 0.7446993589401245, "learning_rate": 9.905036097262027e-05, "loss": 0.0421, "step": 32710 }, { "epoch": 20.854047163798597, "grad_norm": 0.6133078932762146, "learning_rate": 9.904929158940881e-05, "loss": 0.0371, "step": 32720 }, { "epoch": 20.8604206500956, "grad_norm": 0.6737104058265686, "learning_rate": 9.904822161020216e-05, "loss": 0.0376, "step": 32730 }, { "epoch": 20.866794136392606, "grad_norm": 0.4237561523914337, "learning_rate": 9.904715103501329e-05, "loss": 0.0302, "step": 32740 }, { "epoch": 20.87316762268961, "grad_norm": 0.560462236404419, "learning_rate": 9.904607986385523e-05, "loss": 0.0329, "step": 32750 }, { "epoch": 20.879541108986615, "grad_norm": 0.49120181798934937, "learning_rate": 9.9045008096741e-05, "loss": 0.0415, "step": 32760 }, { "epoch": 20.88591459528362, "grad_norm": 0.5374075174331665, "learning_rate": 9.90439357336836e-05, "loss": 0.0356, "step": 32770 }, { "epoch": 20.892288081580624, "grad_norm": 0.29945650696754456, "learning_rate": 9.904286277469609e-05, "loss": 0.035, "step": 32780 }, { "epoch": 20.898661567877628, "grad_norm": 0.6709789633750916, "learning_rate": 9.90417892197915e-05, "loss": 0.0442, "step": 32790 }, { "epoch": 20.905035054174633, "grad_norm": 0.4916990399360657, "learning_rate": 9.904071506898285e-05, "loss": 0.0294, "step": 32800 }, { "epoch": 20.911408540471637, "grad_norm": 0.39094769954681396, "learning_rate": 9.903964032228323e-05, "loss": 0.0458, "step": 32810 }, { "epoch": 20.91778202676864, "grad_norm": 0.32712340354919434, "learning_rate": 9.903856497970568e-05, "loss": 0.0294, "step": 32820 }, { "epoch": 20.924155513065646, "grad_norm": 0.7401004433631897, "learning_rate": 9.903748904126325e-05, "loss": 0.0501, "step": 32830 }, { "epoch": 20.93052899936265, "grad_norm": 0.6848212480545044, "learning_rate": 9.903641250696902e-05, "loss": 0.049, "step": 32840 }, { "epoch": 20.936902485659655, "grad_norm": 0.9070122838020325, "learning_rate": 9.903533537683609e-05, "loss": 0.0443, "step": 32850 }, { "epoch": 20.94327597195666, "grad_norm": 0.6189029812812805, "learning_rate": 9.903425765087755e-05, "loss": 0.0377, "step": 32860 }, { "epoch": 20.949649458253663, "grad_norm": 0.4983058571815491, "learning_rate": 9.903317932910647e-05, "loss": 0.0499, "step": 32870 }, { "epoch": 20.956022944550668, "grad_norm": 0.5379558801651001, "learning_rate": 9.903210041153596e-05, "loss": 0.0448, "step": 32880 }, { "epoch": 20.962396430847672, "grad_norm": 0.29845455288887024, "learning_rate": 9.903102089817915e-05, "loss": 0.0388, "step": 32890 }, { "epoch": 20.968769917144677, "grad_norm": 0.38562801480293274, "learning_rate": 9.902994078904915e-05, "loss": 0.0292, "step": 32900 }, { "epoch": 20.97514340344168, "grad_norm": 0.2758936882019043, "learning_rate": 9.902886008415906e-05, "loss": 0.0482, "step": 32910 }, { "epoch": 20.981516889738685, "grad_norm": 0.573087215423584, "learning_rate": 9.902777878352203e-05, "loss": 0.0469, "step": 32920 }, { "epoch": 20.98789037603569, "grad_norm": 0.48148173093795776, "learning_rate": 9.90266968871512e-05, "loss": 0.0495, "step": 32930 }, { "epoch": 20.994263862332694, "grad_norm": 0.5030044913291931, "learning_rate": 9.902561439505973e-05, "loss": 0.0335, "step": 32940 }, { "epoch": 21.000637348629702, "grad_norm": 0.35676664113998413, "learning_rate": 9.902453130726075e-05, "loss": 0.029, "step": 32950 }, { "epoch": 21.007010834926707, "grad_norm": 0.7540311813354492, "learning_rate": 9.902344762376742e-05, "loss": 0.0499, "step": 32960 }, { "epoch": 21.01338432122371, "grad_norm": 0.5658154487609863, "learning_rate": 9.902236334459291e-05, "loss": 0.0375, "step": 32970 }, { "epoch": 21.019757807520715, "grad_norm": 0.43763744831085205, "learning_rate": 9.90212784697504e-05, "loss": 0.0295, "step": 32980 }, { "epoch": 21.02613129381772, "grad_norm": 0.6033055782318115, "learning_rate": 9.902019299925309e-05, "loss": 0.0415, "step": 32990 }, { "epoch": 21.032504780114724, "grad_norm": 0.6296896934509277, "learning_rate": 9.901910693311415e-05, "loss": 0.0297, "step": 33000 }, { "epoch": 21.03887826641173, "grad_norm": 0.6170443892478943, "learning_rate": 9.901802027134677e-05, "loss": 0.0404, "step": 33010 }, { "epoch": 21.045251752708733, "grad_norm": 0.5801374912261963, "learning_rate": 9.901693301396417e-05, "loss": 0.0359, "step": 33020 }, { "epoch": 21.051625239005737, "grad_norm": 0.6196602582931519, "learning_rate": 9.901584516097954e-05, "loss": 0.0357, "step": 33030 }, { "epoch": 21.057998725302742, "grad_norm": 0.4480137526988983, "learning_rate": 9.901475671240613e-05, "loss": 0.0291, "step": 33040 }, { "epoch": 21.064372211599746, "grad_norm": 0.5306084156036377, "learning_rate": 9.901366766825714e-05, "loss": 0.0334, "step": 33050 }, { "epoch": 21.07074569789675, "grad_norm": 0.525765597820282, "learning_rate": 9.901257802854582e-05, "loss": 0.0533, "step": 33060 }, { "epoch": 21.077119184193755, "grad_norm": 1.007337212562561, "learning_rate": 9.90114877932854e-05, "loss": 0.0443, "step": 33070 }, { "epoch": 21.08349267049076, "grad_norm": 0.7523496150970459, "learning_rate": 9.90103969624891e-05, "loss": 0.0296, "step": 33080 }, { "epoch": 21.089866156787764, "grad_norm": 0.37494054436683655, "learning_rate": 9.900930553617024e-05, "loss": 0.0588, "step": 33090 }, { "epoch": 21.09623964308477, "grad_norm": 0.6175310611724854, "learning_rate": 9.900821351434203e-05, "loss": 0.0337, "step": 33100 }, { "epoch": 21.102613129381773, "grad_norm": 2.121645450592041, "learning_rate": 9.900712089701776e-05, "loss": 0.0456, "step": 33110 }, { "epoch": 21.108986615678777, "grad_norm": 0.5083878040313721, "learning_rate": 9.90060276842107e-05, "loss": 0.0264, "step": 33120 }, { "epoch": 21.11536010197578, "grad_norm": 0.4423936903476715, "learning_rate": 9.900493387593412e-05, "loss": 0.0505, "step": 33130 }, { "epoch": 21.121733588272786, "grad_norm": 0.4551222622394562, "learning_rate": 9.900383947220133e-05, "loss": 0.0242, "step": 33140 }, { "epoch": 21.12810707456979, "grad_norm": 0.7340670228004456, "learning_rate": 9.900274447302564e-05, "loss": 0.0468, "step": 33150 }, { "epoch": 21.134480560866795, "grad_norm": 0.2980981469154358, "learning_rate": 9.900164887842032e-05, "loss": 0.0247, "step": 33160 }, { "epoch": 21.1408540471638, "grad_norm": 0.3659701645374298, "learning_rate": 9.900055268839872e-05, "loss": 0.0393, "step": 33170 }, { "epoch": 21.147227533460804, "grad_norm": 0.5428755283355713, "learning_rate": 9.899945590297412e-05, "loss": 0.0429, "step": 33180 }, { "epoch": 21.153601019757808, "grad_norm": 0.6634163856506348, "learning_rate": 9.899835852215988e-05, "loss": 0.0414, "step": 33190 }, { "epoch": 21.159974506054812, "grad_norm": 0.5601595044136047, "learning_rate": 9.899726054596932e-05, "loss": 0.0277, "step": 33200 }, { "epoch": 21.166347992351817, "grad_norm": 0.3913784623146057, "learning_rate": 9.899616197441579e-05, "loss": 0.0315, "step": 33210 }, { "epoch": 21.17272147864882, "grad_norm": 0.42655080556869507, "learning_rate": 9.89950628075126e-05, "loss": 0.035, "step": 33220 }, { "epoch": 21.179094964945826, "grad_norm": 0.5679159164428711, "learning_rate": 9.899396304527317e-05, "loss": 0.04, "step": 33230 }, { "epoch": 21.18546845124283, "grad_norm": 0.2692587375640869, "learning_rate": 9.899286268771084e-05, "loss": 0.0461, "step": 33240 }, { "epoch": 21.191841937539834, "grad_norm": 0.21358618140220642, "learning_rate": 9.899176173483894e-05, "loss": 0.0444, "step": 33250 }, { "epoch": 21.19821542383684, "grad_norm": 0.3576653301715851, "learning_rate": 9.899066018667089e-05, "loss": 0.042, "step": 33260 }, { "epoch": 21.204588910133843, "grad_norm": 0.3487298786640167, "learning_rate": 9.898955804322007e-05, "loss": 0.0387, "step": 33270 }, { "epoch": 21.210962396430848, "grad_norm": 0.5706133842468262, "learning_rate": 9.898845530449987e-05, "loss": 0.0414, "step": 33280 }, { "epoch": 21.217335882727852, "grad_norm": 0.460977703332901, "learning_rate": 9.898735197052367e-05, "loss": 0.0322, "step": 33290 }, { "epoch": 21.223709369024856, "grad_norm": 0.5471218228340149, "learning_rate": 9.898624804130489e-05, "loss": 0.0323, "step": 33300 }, { "epoch": 21.23008285532186, "grad_norm": 0.37643781304359436, "learning_rate": 9.898514351685697e-05, "loss": 0.0381, "step": 33310 }, { "epoch": 21.236456341618865, "grad_norm": 0.35387441515922546, "learning_rate": 9.898403839719329e-05, "loss": 0.047, "step": 33320 }, { "epoch": 21.24282982791587, "grad_norm": 0.5888198614120483, "learning_rate": 9.898293268232731e-05, "loss": 0.0359, "step": 33330 }, { "epoch": 21.249203314212874, "grad_norm": 0.48150134086608887, "learning_rate": 9.898182637227243e-05, "loss": 0.0403, "step": 33340 }, { "epoch": 21.25557680050988, "grad_norm": 0.22532965242862701, "learning_rate": 9.898071946704211e-05, "loss": 0.0346, "step": 33350 }, { "epoch": 21.261950286806883, "grad_norm": 0.34886297583580017, "learning_rate": 9.897961196664982e-05, "loss": 0.0377, "step": 33360 }, { "epoch": 21.268323773103887, "grad_norm": 0.325795978307724, "learning_rate": 9.8978503871109e-05, "loss": 0.0339, "step": 33370 }, { "epoch": 21.27469725940089, "grad_norm": 0.5327252745628357, "learning_rate": 9.89773951804331e-05, "loss": 0.0427, "step": 33380 }, { "epoch": 21.281070745697896, "grad_norm": 0.685027003288269, "learning_rate": 9.897628589463561e-05, "loss": 0.0307, "step": 33390 }, { "epoch": 21.2874442319949, "grad_norm": 0.37561291456222534, "learning_rate": 9.897517601373002e-05, "loss": 0.0379, "step": 33400 }, { "epoch": 21.293817718291905, "grad_norm": 0.34865817427635193, "learning_rate": 9.897406553772979e-05, "loss": 0.0214, "step": 33410 }, { "epoch": 21.30019120458891, "grad_norm": 0.4685518145561218, "learning_rate": 9.897295446664843e-05, "loss": 0.0344, "step": 33420 }, { "epoch": 21.306564690885914, "grad_norm": 0.7396595478057861, "learning_rate": 9.897184280049943e-05, "loss": 0.039, "step": 33430 }, { "epoch": 21.312938177182918, "grad_norm": 0.6175336837768555, "learning_rate": 9.89707305392963e-05, "loss": 0.0401, "step": 33440 }, { "epoch": 21.319311663479922, "grad_norm": 0.4183218479156494, "learning_rate": 9.896961768305255e-05, "loss": 0.032, "step": 33450 }, { "epoch": 21.325685149776927, "grad_norm": 0.5580785274505615, "learning_rate": 9.896850423178173e-05, "loss": 0.0272, "step": 33460 }, { "epoch": 21.33205863607393, "grad_norm": 0.2957943379878998, "learning_rate": 9.896739018549735e-05, "loss": 0.0275, "step": 33470 }, { "epoch": 21.338432122370936, "grad_norm": 0.48510852456092834, "learning_rate": 9.896627554421292e-05, "loss": 0.036, "step": 33480 }, { "epoch": 21.34480560866794, "grad_norm": 0.42194828391075134, "learning_rate": 9.896516030794204e-05, "loss": 0.0284, "step": 33490 }, { "epoch": 21.351179094964944, "grad_norm": 0.7790687680244446, "learning_rate": 9.896404447669823e-05, "loss": 0.0304, "step": 33500 }, { "epoch": 21.35755258126195, "grad_norm": 0.7340450286865234, "learning_rate": 9.896292805049503e-05, "loss": 0.0435, "step": 33510 }, { "epoch": 21.363926067558953, "grad_norm": 0.4435712695121765, "learning_rate": 9.896181102934603e-05, "loss": 0.0492, "step": 33520 }, { "epoch": 21.370299553855958, "grad_norm": 0.7536898255348206, "learning_rate": 9.896069341326481e-05, "loss": 0.031, "step": 33530 }, { "epoch": 21.376673040152962, "grad_norm": 0.5052396059036255, "learning_rate": 9.895957520226493e-05, "loss": 0.043, "step": 33540 }, { "epoch": 21.383046526449967, "grad_norm": 0.3777838349342346, "learning_rate": 9.895845639635999e-05, "loss": 0.0321, "step": 33550 }, { "epoch": 21.38942001274697, "grad_norm": 0.3997042775154114, "learning_rate": 9.895733699556357e-05, "loss": 0.0373, "step": 33560 }, { "epoch": 21.395793499043975, "grad_norm": 0.2579830288887024, "learning_rate": 9.895621699988928e-05, "loss": 0.0299, "step": 33570 }, { "epoch": 21.402166985340983, "grad_norm": 0.42862001061439514, "learning_rate": 9.895509640935075e-05, "loss": 0.0511, "step": 33580 }, { "epoch": 21.408540471637988, "grad_norm": 0.5525429844856262, "learning_rate": 9.895397522396158e-05, "loss": 0.0246, "step": 33590 }, { "epoch": 21.414913957934992, "grad_norm": 0.5243450403213501, "learning_rate": 9.895285344373537e-05, "loss": 0.031, "step": 33600 }, { "epoch": 21.421287444231996, "grad_norm": 0.7823039293289185, "learning_rate": 9.895173106868578e-05, "loss": 0.0271, "step": 33610 }, { "epoch": 21.427660930529, "grad_norm": 0.6277959942817688, "learning_rate": 9.895060809882644e-05, "loss": 0.0376, "step": 33620 }, { "epoch": 21.434034416826005, "grad_norm": 0.37120676040649414, "learning_rate": 9.894948453417098e-05, "loss": 0.0433, "step": 33630 }, { "epoch": 21.44040790312301, "grad_norm": 0.2807099521160126, "learning_rate": 9.89483603747331e-05, "loss": 0.0321, "step": 33640 }, { "epoch": 21.446781389420014, "grad_norm": 0.5695929527282715, "learning_rate": 9.894723562052641e-05, "loss": 0.0386, "step": 33650 }, { "epoch": 21.45315487571702, "grad_norm": 0.4366040527820587, "learning_rate": 9.894611027156458e-05, "loss": 0.0308, "step": 33660 }, { "epoch": 21.459528362014023, "grad_norm": 0.46178221702575684, "learning_rate": 9.894498432786132e-05, "loss": 0.0291, "step": 33670 }, { "epoch": 21.465901848311027, "grad_norm": 0.625882625579834, "learning_rate": 9.894385778943027e-05, "loss": 0.0432, "step": 33680 }, { "epoch": 21.47227533460803, "grad_norm": 0.529136061668396, "learning_rate": 9.894273065628514e-05, "loss": 0.0306, "step": 33690 }, { "epoch": 21.478648820905036, "grad_norm": 0.49732890725135803, "learning_rate": 9.894160292843964e-05, "loss": 0.034, "step": 33700 }, { "epoch": 21.48502230720204, "grad_norm": 0.4490704834461212, "learning_rate": 9.894047460590744e-05, "loss": 0.0317, "step": 33710 }, { "epoch": 21.491395793499045, "grad_norm": 0.5355561375617981, "learning_rate": 9.893934568870227e-05, "loss": 0.0284, "step": 33720 }, { "epoch": 21.49776927979605, "grad_norm": 0.5296640396118164, "learning_rate": 9.893821617683784e-05, "loss": 0.041, "step": 33730 }, { "epoch": 21.504142766093054, "grad_norm": 0.6401536464691162, "learning_rate": 9.893708607032787e-05, "loss": 0.0312, "step": 33740 }, { "epoch": 21.510516252390058, "grad_norm": 0.7757630348205566, "learning_rate": 9.893595536918612e-05, "loss": 0.0325, "step": 33750 }, { "epoch": 21.516889738687063, "grad_norm": 0.36943888664245605, "learning_rate": 9.893482407342632e-05, "loss": 0.0258, "step": 33760 }, { "epoch": 21.523263224984067, "grad_norm": 0.6172603368759155, "learning_rate": 9.893369218306217e-05, "loss": 0.0438, "step": 33770 }, { "epoch": 21.52963671128107, "grad_norm": 0.4445973336696625, "learning_rate": 9.893255969810748e-05, "loss": 0.0307, "step": 33780 }, { "epoch": 21.536010197578076, "grad_norm": 0.45456793904304504, "learning_rate": 9.893142661857597e-05, "loss": 0.0358, "step": 33790 }, { "epoch": 21.54238368387508, "grad_norm": 0.4152179956436157, "learning_rate": 9.893029294448145e-05, "loss": 0.0283, "step": 33800 }, { "epoch": 21.548757170172085, "grad_norm": 0.3421584665775299, "learning_rate": 9.892915867583767e-05, "loss": 0.0367, "step": 33810 }, { "epoch": 21.55513065646909, "grad_norm": 0.346899151802063, "learning_rate": 9.89280238126584e-05, "loss": 0.0234, "step": 33820 }, { "epoch": 21.561504142766093, "grad_norm": 0.5675371885299683, "learning_rate": 9.892688835495747e-05, "loss": 0.0314, "step": 33830 }, { "epoch": 21.567877629063098, "grad_norm": 0.24062864482402802, "learning_rate": 9.892575230274863e-05, "loss": 0.0251, "step": 33840 }, { "epoch": 21.574251115360102, "grad_norm": 0.44688624143600464, "learning_rate": 9.892461565604572e-05, "loss": 0.0371, "step": 33850 }, { "epoch": 21.580624601657107, "grad_norm": 0.6995373368263245, "learning_rate": 9.892347841486252e-05, "loss": 0.0356, "step": 33860 }, { "epoch": 21.58699808795411, "grad_norm": 0.44560351967811584, "learning_rate": 9.892234057921288e-05, "loss": 0.0388, "step": 33870 }, { "epoch": 21.593371574251115, "grad_norm": 0.46044862270355225, "learning_rate": 9.89212021491106e-05, "loss": 0.0315, "step": 33880 }, { "epoch": 21.59974506054812, "grad_norm": 0.4010196626186371, "learning_rate": 9.892006312456954e-05, "loss": 0.0324, "step": 33890 }, { "epoch": 21.606118546845124, "grad_norm": 0.5219210982322693, "learning_rate": 9.891892350560352e-05, "loss": 0.0386, "step": 33900 }, { "epoch": 21.61249203314213, "grad_norm": 0.6276299357414246, "learning_rate": 9.891778329222638e-05, "loss": 0.0347, "step": 33910 }, { "epoch": 21.618865519439133, "grad_norm": 0.4902973473072052, "learning_rate": 9.8916642484452e-05, "loss": 0.0485, "step": 33920 }, { "epoch": 21.625239005736137, "grad_norm": 0.3681551516056061, "learning_rate": 9.891550108229423e-05, "loss": 0.0476, "step": 33930 }, { "epoch": 21.631612492033142, "grad_norm": 0.39696118235588074, "learning_rate": 9.891435908576692e-05, "loss": 0.0317, "step": 33940 }, { "epoch": 21.637985978330146, "grad_norm": 0.8397636413574219, "learning_rate": 9.891321649488398e-05, "loss": 0.0381, "step": 33950 }, { "epoch": 21.64435946462715, "grad_norm": 0.6202315092086792, "learning_rate": 9.891207330965928e-05, "loss": 0.0527, "step": 33960 }, { "epoch": 21.650732950924155, "grad_norm": 0.8681958317756653, "learning_rate": 9.89109295301067e-05, "loss": 0.0276, "step": 33970 }, { "epoch": 21.65710643722116, "grad_norm": 0.5664786696434021, "learning_rate": 9.890978515624017e-05, "loss": 0.0279, "step": 33980 }, { "epoch": 21.663479923518164, "grad_norm": 0.5184953808784485, "learning_rate": 9.890864018807354e-05, "loss": 0.0389, "step": 33990 }, { "epoch": 21.66985340981517, "grad_norm": 0.5251326560974121, "learning_rate": 9.890749462562077e-05, "loss": 0.0313, "step": 34000 }, { "epoch": 21.676226896112173, "grad_norm": 0.5055752396583557, "learning_rate": 9.890634846889577e-05, "loss": 0.0436, "step": 34010 }, { "epoch": 21.682600382409177, "grad_norm": 0.3840938210487366, "learning_rate": 9.890520171791245e-05, "loss": 0.0325, "step": 34020 }, { "epoch": 21.68897386870618, "grad_norm": 0.5885211229324341, "learning_rate": 9.890405437268475e-05, "loss": 0.0356, "step": 34030 }, { "epoch": 21.695347355003186, "grad_norm": 0.5757576823234558, "learning_rate": 9.890290643322661e-05, "loss": 0.0336, "step": 34040 }, { "epoch": 21.70172084130019, "grad_norm": 0.7059485912322998, "learning_rate": 9.8901757899552e-05, "loss": 0.0392, "step": 34050 }, { "epoch": 21.708094327597195, "grad_norm": 0.3947299122810364, "learning_rate": 9.890060877167487e-05, "loss": 0.0351, "step": 34060 }, { "epoch": 21.7144678138942, "grad_norm": 0.6324647068977356, "learning_rate": 9.889945904960916e-05, "loss": 0.0321, "step": 34070 }, { "epoch": 21.720841300191204, "grad_norm": 0.375550776720047, "learning_rate": 9.889830873336885e-05, "loss": 0.0358, "step": 34080 }, { "epoch": 21.727214786488208, "grad_norm": 0.7949948906898499, "learning_rate": 9.889715782296794e-05, "loss": 0.036, "step": 34090 }, { "epoch": 21.733588272785212, "grad_norm": 0.3773329555988312, "learning_rate": 9.889600631842038e-05, "loss": 0.0325, "step": 34100 }, { "epoch": 21.739961759082217, "grad_norm": 0.4015154242515564, "learning_rate": 9.889485421974017e-05, "loss": 0.034, "step": 34110 }, { "epoch": 21.74633524537922, "grad_norm": 0.7220259308815002, "learning_rate": 9.889370152694133e-05, "loss": 0.0344, "step": 34120 }, { "epoch": 21.752708731676226, "grad_norm": 0.5984421968460083, "learning_rate": 9.889254824003785e-05, "loss": 0.0274, "step": 34130 }, { "epoch": 21.75908221797323, "grad_norm": 0.4008987545967102, "learning_rate": 9.889139435904375e-05, "loss": 0.0317, "step": 34140 }, { "epoch": 21.765455704270234, "grad_norm": 0.4927680194377899, "learning_rate": 9.889023988397303e-05, "loss": 0.0249, "step": 34150 }, { "epoch": 21.77182919056724, "grad_norm": 0.41582080721855164, "learning_rate": 9.888908481483976e-05, "loss": 0.0331, "step": 34160 }, { "epoch": 21.778202676864243, "grad_norm": 0.7995622754096985, "learning_rate": 9.888792915165795e-05, "loss": 0.0376, "step": 34170 }, { "epoch": 21.784576163161248, "grad_norm": 0.45375895500183105, "learning_rate": 9.888677289444163e-05, "loss": 0.0367, "step": 34180 }, { "epoch": 21.790949649458252, "grad_norm": 0.5788255929946899, "learning_rate": 9.888561604320486e-05, "loss": 0.0325, "step": 34190 }, { "epoch": 21.797323135755256, "grad_norm": 0.8217025995254517, "learning_rate": 9.888445859796171e-05, "loss": 0.0348, "step": 34200 }, { "epoch": 21.803696622052264, "grad_norm": 0.553433895111084, "learning_rate": 9.888330055872623e-05, "loss": 0.0381, "step": 34210 }, { "epoch": 21.81007010834927, "grad_norm": 0.35488733649253845, "learning_rate": 9.888214192551249e-05, "loss": 0.0413, "step": 34220 }, { "epoch": 21.816443594646273, "grad_norm": 0.37866947054862976, "learning_rate": 9.888098269833457e-05, "loss": 0.0305, "step": 34230 }, { "epoch": 21.822817080943278, "grad_norm": 0.600410521030426, "learning_rate": 9.887982287720657e-05, "loss": 0.0468, "step": 34240 }, { "epoch": 21.829190567240282, "grad_norm": 0.5504175424575806, "learning_rate": 9.887866246214257e-05, "loss": 0.03, "step": 34250 }, { "epoch": 21.835564053537286, "grad_norm": 0.8275286555290222, "learning_rate": 9.887750145315668e-05, "loss": 0.039, "step": 34260 }, { "epoch": 21.84193753983429, "grad_norm": 0.698927640914917, "learning_rate": 9.887633985026298e-05, "loss": 0.0356, "step": 34270 }, { "epoch": 21.848311026131295, "grad_norm": 0.3693823218345642, "learning_rate": 9.887517765347562e-05, "loss": 0.0462, "step": 34280 }, { "epoch": 21.8546845124283, "grad_norm": 0.345838725566864, "learning_rate": 9.88740148628087e-05, "loss": 0.0375, "step": 34290 }, { "epoch": 21.861057998725304, "grad_norm": 0.9541895389556885, "learning_rate": 9.887285147827634e-05, "loss": 0.036, "step": 34300 }, { "epoch": 21.86743148502231, "grad_norm": 0.5108804106712341, "learning_rate": 9.88716874998927e-05, "loss": 0.0353, "step": 34310 }, { "epoch": 21.873804971319313, "grad_norm": 0.7709117531776428, "learning_rate": 9.887052292767193e-05, "loss": 0.0418, "step": 34320 }, { "epoch": 21.880178457616317, "grad_norm": 0.5951215028762817, "learning_rate": 9.886935776162816e-05, "loss": 0.0341, "step": 34330 }, { "epoch": 21.88655194391332, "grad_norm": 0.6434561610221863, "learning_rate": 9.886819200177554e-05, "loss": 0.0388, "step": 34340 }, { "epoch": 21.892925430210326, "grad_norm": 0.7685309052467346, "learning_rate": 9.886702564812825e-05, "loss": 0.033, "step": 34350 }, { "epoch": 21.89929891650733, "grad_norm": 0.3714256286621094, "learning_rate": 9.886585870070047e-05, "loss": 0.0416, "step": 34360 }, { "epoch": 21.905672402804335, "grad_norm": 0.5191555619239807, "learning_rate": 9.886469115950636e-05, "loss": 0.0351, "step": 34370 }, { "epoch": 21.91204588910134, "grad_norm": 0.5067389011383057, "learning_rate": 9.886352302456012e-05, "loss": 0.0379, "step": 34380 }, { "epoch": 21.918419375398344, "grad_norm": 0.5698462128639221, "learning_rate": 9.886235429587594e-05, "loss": 0.0541, "step": 34390 }, { "epoch": 21.924792861695348, "grad_norm": 0.42807474732398987, "learning_rate": 9.886118497346803e-05, "loss": 0.0434, "step": 34400 }, { "epoch": 21.931166347992352, "grad_norm": 0.733651340007782, "learning_rate": 9.886001505735059e-05, "loss": 0.0405, "step": 34410 }, { "epoch": 21.937539834289357, "grad_norm": 0.6504823565483093, "learning_rate": 9.885884454753781e-05, "loss": 0.037, "step": 34420 }, { "epoch": 21.94391332058636, "grad_norm": 0.6256840229034424, "learning_rate": 9.885767344404396e-05, "loss": 0.0433, "step": 34430 }, { "epoch": 21.950286806883366, "grad_norm": 0.3689855635166168, "learning_rate": 9.885650174688326e-05, "loss": 0.038, "step": 34440 }, { "epoch": 21.95666029318037, "grad_norm": 0.7795689702033997, "learning_rate": 9.885532945606992e-05, "loss": 0.0443, "step": 34450 }, { "epoch": 21.963033779477374, "grad_norm": 0.40495598316192627, "learning_rate": 9.885415657161818e-05, "loss": 0.0447, "step": 34460 }, { "epoch": 21.96940726577438, "grad_norm": 0.35698962211608887, "learning_rate": 9.885298309354234e-05, "loss": 0.0416, "step": 34470 }, { "epoch": 21.975780752071383, "grad_norm": 0.435452938079834, "learning_rate": 9.885180902185662e-05, "loss": 0.0406, "step": 34480 }, { "epoch": 21.982154238368388, "grad_norm": 0.4199587106704712, "learning_rate": 9.885063435657532e-05, "loss": 0.0259, "step": 34490 }, { "epoch": 21.988527724665392, "grad_norm": 0.608440637588501, "learning_rate": 9.884945909771266e-05, "loss": 0.0356, "step": 34500 }, { "epoch": 21.994901210962396, "grad_norm": 0.48252299427986145, "learning_rate": 9.884828324528295e-05, "loss": 0.0238, "step": 34510 }, { "epoch": 22.0012746972594, "grad_norm": 0.4837629795074463, "learning_rate": 9.88471067993005e-05, "loss": 0.032, "step": 34520 }, { "epoch": 22.007648183556405, "grad_norm": 0.6715797185897827, "learning_rate": 9.884592975977956e-05, "loss": 0.0514, "step": 34530 }, { "epoch": 22.01402166985341, "grad_norm": 1.0237489938735962, "learning_rate": 9.884475212673447e-05, "loss": 0.046, "step": 34540 }, { "epoch": 22.020395156150414, "grad_norm": 0.43757814168930054, "learning_rate": 9.884357390017954e-05, "loss": 0.0245, "step": 34550 }, { "epoch": 22.02676864244742, "grad_norm": 0.4704092741012573, "learning_rate": 9.884239508012904e-05, "loss": 0.0383, "step": 34560 }, { "epoch": 22.033142128744423, "grad_norm": 0.5576807856559753, "learning_rate": 9.884121566659733e-05, "loss": 0.0264, "step": 34570 }, { "epoch": 22.039515615041427, "grad_norm": 0.6145924925804138, "learning_rate": 9.884003565959874e-05, "loss": 0.0451, "step": 34580 }, { "epoch": 22.04588910133843, "grad_norm": 0.4765033721923828, "learning_rate": 9.883885505914761e-05, "loss": 0.0377, "step": 34590 }, { "epoch": 22.052262587635436, "grad_norm": 0.5439814925193787, "learning_rate": 9.883767386525829e-05, "loss": 0.0435, "step": 34600 }, { "epoch": 22.05863607393244, "grad_norm": 0.64704430103302, "learning_rate": 9.88364920779451e-05, "loss": 0.0341, "step": 34610 }, { "epoch": 22.065009560229445, "grad_norm": 0.442614883184433, "learning_rate": 9.883530969722243e-05, "loss": 0.0411, "step": 34620 }, { "epoch": 22.07138304652645, "grad_norm": 0.40172383189201355, "learning_rate": 9.883412672310463e-05, "loss": 0.0296, "step": 34630 }, { "epoch": 22.077756532823454, "grad_norm": 0.3768358528614044, "learning_rate": 9.88329431556061e-05, "loss": 0.033, "step": 34640 }, { "epoch": 22.084130019120458, "grad_norm": 0.3362719714641571, "learning_rate": 9.883175899474119e-05, "loss": 0.038, "step": 34650 }, { "epoch": 22.090503505417463, "grad_norm": 0.24823905527591705, "learning_rate": 9.883057424052433e-05, "loss": 0.0344, "step": 34660 }, { "epoch": 22.096876991714467, "grad_norm": 0.5325015187263489, "learning_rate": 9.882938889296987e-05, "loss": 0.0244, "step": 34670 }, { "epoch": 22.10325047801147, "grad_norm": 0.6017010807991028, "learning_rate": 9.882820295209223e-05, "loss": 0.0315, "step": 34680 }, { "epoch": 22.109623964308476, "grad_norm": 0.7105106115341187, "learning_rate": 9.882701641790582e-05, "loss": 0.0367, "step": 34690 }, { "epoch": 22.11599745060548, "grad_norm": 0.5900437831878662, "learning_rate": 9.882582929042507e-05, "loss": 0.0353, "step": 34700 }, { "epoch": 22.122370936902485, "grad_norm": 0.697341799736023, "learning_rate": 9.88246415696644e-05, "loss": 0.0382, "step": 34710 }, { "epoch": 22.12874442319949, "grad_norm": 0.40661507844924927, "learning_rate": 9.882345325563822e-05, "loss": 0.0333, "step": 34720 }, { "epoch": 22.135117909496493, "grad_norm": 0.517068088054657, "learning_rate": 9.882226434836099e-05, "loss": 0.0466, "step": 34730 }, { "epoch": 22.141491395793498, "grad_norm": 0.4185454547405243, "learning_rate": 9.882107484784715e-05, "loss": 0.024, "step": 34740 }, { "epoch": 22.147864882090502, "grad_norm": 0.6745854020118713, "learning_rate": 9.881988475411116e-05, "loss": 0.0385, "step": 34750 }, { "epoch": 22.154238368387507, "grad_norm": 0.567823052406311, "learning_rate": 9.881869406716748e-05, "loss": 0.032, "step": 34760 }, { "epoch": 22.16061185468451, "grad_norm": 0.5652419328689575, "learning_rate": 9.881750278703057e-05, "loss": 0.0301, "step": 34770 }, { "epoch": 22.166985340981515, "grad_norm": 0.461007684469223, "learning_rate": 9.881631091371491e-05, "loss": 0.0234, "step": 34780 }, { "epoch": 22.17335882727852, "grad_norm": 0.3966669738292694, "learning_rate": 9.881511844723499e-05, "loss": 0.041, "step": 34790 }, { "epoch": 22.179732313575524, "grad_norm": 0.29091978073120117, "learning_rate": 9.881392538760529e-05, "loss": 0.0264, "step": 34800 }, { "epoch": 22.18610579987253, "grad_norm": 0.6113274097442627, "learning_rate": 9.88127317348403e-05, "loss": 0.0309, "step": 34810 }, { "epoch": 22.192479286169533, "grad_norm": 0.5449618697166443, "learning_rate": 9.881153748895455e-05, "loss": 0.0281, "step": 34820 }, { "epoch": 22.198852772466537, "grad_norm": 0.3119959235191345, "learning_rate": 9.881034264996252e-05, "loss": 0.0303, "step": 34830 }, { "epoch": 22.205226258763545, "grad_norm": 0.4033069312572479, "learning_rate": 9.880914721787872e-05, "loss": 0.0264, "step": 34840 }, { "epoch": 22.21159974506055, "grad_norm": 0.42602333426475525, "learning_rate": 9.880795119271773e-05, "loss": 0.029, "step": 34850 }, { "epoch": 22.217973231357554, "grad_norm": 0.625603973865509, "learning_rate": 9.880675457449404e-05, "loss": 0.0366, "step": 34860 }, { "epoch": 22.22434671765456, "grad_norm": 0.5646366477012634, "learning_rate": 9.880555736322221e-05, "loss": 0.0379, "step": 34870 }, { "epoch": 22.230720203951563, "grad_norm": 0.34572529792785645, "learning_rate": 9.880435955891676e-05, "loss": 0.0334, "step": 34880 }, { "epoch": 22.237093690248567, "grad_norm": 0.516228199005127, "learning_rate": 9.880316116159229e-05, "loss": 0.0368, "step": 34890 }, { "epoch": 22.243467176545572, "grad_norm": 0.4655141532421112, "learning_rate": 9.880196217126332e-05, "loss": 0.0504, "step": 34900 }, { "epoch": 22.249840662842576, "grad_norm": 0.5334170460700989, "learning_rate": 9.880076258794441e-05, "loss": 0.0269, "step": 34910 }, { "epoch": 22.25621414913958, "grad_norm": 0.493969202041626, "learning_rate": 9.87995624116502e-05, "loss": 0.0454, "step": 34920 }, { "epoch": 22.262587635436585, "grad_norm": 0.4887004494667053, "learning_rate": 9.87983616423952e-05, "loss": 0.0346, "step": 34930 }, { "epoch": 22.26896112173359, "grad_norm": 0.4698292315006256, "learning_rate": 9.879716028019405e-05, "loss": 0.0317, "step": 34940 }, { "epoch": 22.275334608030594, "grad_norm": 0.7241379022598267, "learning_rate": 9.879595832506132e-05, "loss": 0.0386, "step": 34950 }, { "epoch": 22.2817080943276, "grad_norm": 0.6733170747756958, "learning_rate": 9.879475577701163e-05, "loss": 0.0488, "step": 34960 }, { "epoch": 22.288081580624603, "grad_norm": 0.7493279576301575, "learning_rate": 9.879355263605958e-05, "loss": 0.0445, "step": 34970 }, { "epoch": 22.294455066921607, "grad_norm": 0.5190390944480896, "learning_rate": 9.879234890221982e-05, "loss": 0.0466, "step": 34980 }, { "epoch": 22.30082855321861, "grad_norm": 0.4218645989894867, "learning_rate": 9.879114457550693e-05, "loss": 0.0343, "step": 34990 }, { "epoch": 22.307202039515616, "grad_norm": 0.5315443873405457, "learning_rate": 9.878993965593557e-05, "loss": 0.0309, "step": 35000 }, { "epoch": 22.31357552581262, "grad_norm": 0.5222482085227966, "learning_rate": 9.878873414352039e-05, "loss": 0.0404, "step": 35010 }, { "epoch": 22.319949012109625, "grad_norm": 0.3544315695762634, "learning_rate": 9.878752803827602e-05, "loss": 0.0448, "step": 35020 }, { "epoch": 22.32632249840663, "grad_norm": 0.7255458235740662, "learning_rate": 9.878632134021712e-05, "loss": 0.0383, "step": 35030 }, { "epoch": 22.332695984703633, "grad_norm": 0.5964975953102112, "learning_rate": 9.878511404935835e-05, "loss": 0.0452, "step": 35040 }, { "epoch": 22.339069471000638, "grad_norm": 0.7118129730224609, "learning_rate": 9.878390616571438e-05, "loss": 0.0471, "step": 35050 }, { "epoch": 22.345442957297642, "grad_norm": 0.3512742519378662, "learning_rate": 9.87826976892999e-05, "loss": 0.0403, "step": 35060 }, { "epoch": 22.351816443594647, "grad_norm": 0.6572921276092529, "learning_rate": 9.878148862012957e-05, "loss": 0.0338, "step": 35070 }, { "epoch": 22.35818992989165, "grad_norm": 0.37757009267807007, "learning_rate": 9.878027895821812e-05, "loss": 0.0373, "step": 35080 }, { "epoch": 22.364563416188656, "grad_norm": 1.0088146924972534, "learning_rate": 9.877906870358022e-05, "loss": 0.0401, "step": 35090 }, { "epoch": 22.37093690248566, "grad_norm": 0.6984426975250244, "learning_rate": 9.877785785623056e-05, "loss": 0.0427, "step": 35100 }, { "epoch": 22.377310388782664, "grad_norm": 0.30489569902420044, "learning_rate": 9.877664641618389e-05, "loss": 0.0475, "step": 35110 }, { "epoch": 22.38368387507967, "grad_norm": 0.40591955184936523, "learning_rate": 9.877543438345491e-05, "loss": 0.0434, "step": 35120 }, { "epoch": 22.390057361376673, "grad_norm": 0.822650134563446, "learning_rate": 9.877422175805835e-05, "loss": 0.0425, "step": 35130 }, { "epoch": 22.396430847673678, "grad_norm": 0.6844353079795837, "learning_rate": 9.877300854000893e-05, "loss": 0.048, "step": 35140 }, { "epoch": 22.402804333970682, "grad_norm": 0.547126054763794, "learning_rate": 9.877179472932143e-05, "loss": 0.0606, "step": 35150 }, { "epoch": 22.409177820267686, "grad_norm": 0.5891717076301575, "learning_rate": 9.877058032601056e-05, "loss": 0.0316, "step": 35160 }, { "epoch": 22.41555130656469, "grad_norm": 0.5831828117370605, "learning_rate": 9.876936533009112e-05, "loss": 0.0369, "step": 35170 }, { "epoch": 22.421924792861695, "grad_norm": 0.38758018612861633, "learning_rate": 9.876814974157781e-05, "loss": 0.0349, "step": 35180 }, { "epoch": 22.4282982791587, "grad_norm": 0.5421608686447144, "learning_rate": 9.876693356048545e-05, "loss": 0.0436, "step": 35190 }, { "epoch": 22.434671765455704, "grad_norm": 0.34861406683921814, "learning_rate": 9.87657167868288e-05, "loss": 0.0283, "step": 35200 }, { "epoch": 22.44104525175271, "grad_norm": 0.7452977895736694, "learning_rate": 9.876449942062266e-05, "loss": 0.0599, "step": 35210 }, { "epoch": 22.447418738049713, "grad_norm": 0.5051636099815369, "learning_rate": 9.876328146188181e-05, "loss": 0.042, "step": 35220 }, { "epoch": 22.453792224346717, "grad_norm": 0.3058321177959442, "learning_rate": 9.876206291062104e-05, "loss": 0.0298, "step": 35230 }, { "epoch": 22.46016571064372, "grad_norm": 0.4911765456199646, "learning_rate": 9.876084376685517e-05, "loss": 0.0381, "step": 35240 }, { "epoch": 22.466539196940726, "grad_norm": 0.6163874268531799, "learning_rate": 9.8759624030599e-05, "loss": 0.0322, "step": 35250 }, { "epoch": 22.47291268323773, "grad_norm": 0.41075339913368225, "learning_rate": 9.875840370186738e-05, "loss": 0.0307, "step": 35260 }, { "epoch": 22.479286169534735, "grad_norm": 0.6550604104995728, "learning_rate": 9.87571827806751e-05, "loss": 0.032, "step": 35270 }, { "epoch": 22.48565965583174, "grad_norm": 0.6580571532249451, "learning_rate": 9.875596126703704e-05, "loss": 0.0359, "step": 35280 }, { "epoch": 22.492033142128744, "grad_norm": 0.4544246196746826, "learning_rate": 9.8754739160968e-05, "loss": 0.0504, "step": 35290 }, { "epoch": 22.498406628425748, "grad_norm": 1.0186104774475098, "learning_rate": 9.875351646248284e-05, "loss": 0.0398, "step": 35300 }, { "epoch": 22.504780114722752, "grad_norm": 0.6584447622299194, "learning_rate": 9.875229317159645e-05, "loss": 0.0325, "step": 35310 }, { "epoch": 22.511153601019757, "grad_norm": 0.3832549750804901, "learning_rate": 9.875106928832364e-05, "loss": 0.0441, "step": 35320 }, { "epoch": 22.51752708731676, "grad_norm": 0.6077988743782043, "learning_rate": 9.874984481267933e-05, "loss": 0.0388, "step": 35330 }, { "epoch": 22.523900573613766, "grad_norm": 0.4665283262729645, "learning_rate": 9.874861974467836e-05, "loss": 0.0342, "step": 35340 }, { "epoch": 22.53027405991077, "grad_norm": 0.4998858869075775, "learning_rate": 9.874739408433565e-05, "loss": 0.0321, "step": 35350 }, { "epoch": 22.536647546207774, "grad_norm": 0.5833396315574646, "learning_rate": 9.874616783166607e-05, "loss": 0.0366, "step": 35360 }, { "epoch": 22.54302103250478, "grad_norm": 0.3186429738998413, "learning_rate": 9.874494098668452e-05, "loss": 0.0226, "step": 35370 }, { "epoch": 22.549394518801783, "grad_norm": 0.3945305645465851, "learning_rate": 9.874371354940592e-05, "loss": 0.0259, "step": 35380 }, { "epoch": 22.555768005098788, "grad_norm": 0.8879762887954712, "learning_rate": 9.874248551984519e-05, "loss": 0.0441, "step": 35390 }, { "epoch": 22.562141491395792, "grad_norm": 0.3650534749031067, "learning_rate": 9.874125689801723e-05, "loss": 0.0395, "step": 35400 }, { "epoch": 22.568514977692796, "grad_norm": 0.3361234664916992, "learning_rate": 9.874002768393698e-05, "loss": 0.04, "step": 35410 }, { "epoch": 22.5748884639898, "grad_norm": 0.5994387865066528, "learning_rate": 9.873879787761938e-05, "loss": 0.0352, "step": 35420 }, { "epoch": 22.581261950286805, "grad_norm": 0.7112413048744202, "learning_rate": 9.873756747907935e-05, "loss": 0.0527, "step": 35430 }, { "epoch": 22.58763543658381, "grad_norm": 0.5254963040351868, "learning_rate": 9.873633648833187e-05, "loss": 0.032, "step": 35440 }, { "epoch": 22.594008922880818, "grad_norm": 0.4152343273162842, "learning_rate": 9.873510490539189e-05, "loss": 0.0381, "step": 35450 }, { "epoch": 22.60038240917782, "grad_norm": 0.5694442391395569, "learning_rate": 9.873387273027438e-05, "loss": 0.0317, "step": 35460 }, { "epoch": 22.606755895474826, "grad_norm": 0.1981181502342224, "learning_rate": 9.873263996299429e-05, "loss": 0.0288, "step": 35470 }, { "epoch": 22.61312938177183, "grad_norm": 0.6542591452598572, "learning_rate": 9.873140660356663e-05, "loss": 0.0337, "step": 35480 }, { "epoch": 22.619502868068835, "grad_norm": 0.8989945650100708, "learning_rate": 9.873017265200635e-05, "loss": 0.0337, "step": 35490 }, { "epoch": 22.62587635436584, "grad_norm": 0.7037742137908936, "learning_rate": 9.872893810832847e-05, "loss": 0.0297, "step": 35500 }, { "epoch": 22.632249840662844, "grad_norm": 0.5993256568908691, "learning_rate": 9.872770297254799e-05, "loss": 0.0362, "step": 35510 }, { "epoch": 22.63862332695985, "grad_norm": 0.4735136330127716, "learning_rate": 9.872646724467991e-05, "loss": 0.0368, "step": 35520 }, { "epoch": 22.644996813256853, "grad_norm": 0.2969745397567749, "learning_rate": 9.872523092473924e-05, "loss": 0.0417, "step": 35530 }, { "epoch": 22.651370299553857, "grad_norm": 0.4498269855976105, "learning_rate": 9.872399401274101e-05, "loss": 0.0333, "step": 35540 }, { "epoch": 22.65774378585086, "grad_norm": 0.6345143914222717, "learning_rate": 9.872275650870027e-05, "loss": 0.0402, "step": 35550 }, { "epoch": 22.664117272147866, "grad_norm": 0.6988514065742493, "learning_rate": 9.872151841263201e-05, "loss": 0.0405, "step": 35560 }, { "epoch": 22.67049075844487, "grad_norm": 0.7707237601280212, "learning_rate": 9.872027972455132e-05, "loss": 0.0364, "step": 35570 }, { "epoch": 22.676864244741875, "grad_norm": 0.6449145078659058, "learning_rate": 9.871904044447322e-05, "loss": 0.0431, "step": 35580 }, { "epoch": 22.68323773103888, "grad_norm": 0.5047450065612793, "learning_rate": 9.87178005724128e-05, "loss": 0.0315, "step": 35590 }, { "epoch": 22.689611217335884, "grad_norm": 0.594368040561676, "learning_rate": 9.871656010838508e-05, "loss": 0.0429, "step": 35600 }, { "epoch": 22.695984703632888, "grad_norm": 0.2974255084991455, "learning_rate": 9.871531905240517e-05, "loss": 0.0437, "step": 35610 }, { "epoch": 22.702358189929893, "grad_norm": 0.36203864216804504, "learning_rate": 9.871407740448815e-05, "loss": 0.0386, "step": 35620 }, { "epoch": 22.708731676226897, "grad_norm": 0.3179348409175873, "learning_rate": 9.871283516464908e-05, "loss": 0.0224, "step": 35630 }, { "epoch": 22.7151051625239, "grad_norm": 0.3596300482749939, "learning_rate": 9.871159233290307e-05, "loss": 0.029, "step": 35640 }, { "epoch": 22.721478648820906, "grad_norm": 0.6995904445648193, "learning_rate": 9.871034890926523e-05, "loss": 0.0421, "step": 35650 }, { "epoch": 22.72785213511791, "grad_norm": 0.5614590048789978, "learning_rate": 9.870910489375065e-05, "loss": 0.0377, "step": 35660 }, { "epoch": 22.734225621414915, "grad_norm": 0.616037905216217, "learning_rate": 9.870786028637446e-05, "loss": 0.0423, "step": 35670 }, { "epoch": 22.74059910771192, "grad_norm": 0.6043813824653625, "learning_rate": 9.870661508715178e-05, "loss": 0.0361, "step": 35680 }, { "epoch": 22.746972594008923, "grad_norm": 0.613956093788147, "learning_rate": 9.870536929609776e-05, "loss": 0.0312, "step": 35690 }, { "epoch": 22.753346080305928, "grad_norm": 0.48165464401245117, "learning_rate": 9.87041229132275e-05, "loss": 0.0341, "step": 35700 }, { "epoch": 22.759719566602932, "grad_norm": 0.3860820233821869, "learning_rate": 9.870287593855616e-05, "loss": 0.0358, "step": 35710 }, { "epoch": 22.766093052899937, "grad_norm": 0.5796486139297485, "learning_rate": 9.870162837209889e-05, "loss": 0.0284, "step": 35720 }, { "epoch": 22.77246653919694, "grad_norm": 0.7215161919593811, "learning_rate": 9.870038021387086e-05, "loss": 0.0273, "step": 35730 }, { "epoch": 22.778840025493945, "grad_norm": 0.6009196639060974, "learning_rate": 9.869913146388723e-05, "loss": 0.0333, "step": 35740 }, { "epoch": 22.78521351179095, "grad_norm": 0.7514901757240295, "learning_rate": 9.869788212216318e-05, "loss": 0.0375, "step": 35750 }, { "epoch": 22.791586998087954, "grad_norm": 0.7864081859588623, "learning_rate": 9.869663218871387e-05, "loss": 0.0359, "step": 35760 }, { "epoch": 22.79796048438496, "grad_norm": 0.2704887092113495, "learning_rate": 9.869538166355451e-05, "loss": 0.0384, "step": 35770 }, { "epoch": 22.804333970681963, "grad_norm": 0.5436603426933289, "learning_rate": 9.869413054670028e-05, "loss": 0.0461, "step": 35780 }, { "epoch": 22.810707456978967, "grad_norm": 0.519588053226471, "learning_rate": 9.869287883816641e-05, "loss": 0.031, "step": 35790 }, { "epoch": 22.817080943275972, "grad_norm": 0.4137604832649231, "learning_rate": 9.869162653796807e-05, "loss": 0.0376, "step": 35800 }, { "epoch": 22.823454429572976, "grad_norm": 0.29036200046539307, "learning_rate": 9.869037364612048e-05, "loss": 0.0403, "step": 35810 }, { "epoch": 22.82982791586998, "grad_norm": 0.5087034106254578, "learning_rate": 9.86891201626389e-05, "loss": 0.0248, "step": 35820 }, { "epoch": 22.836201402166985, "grad_norm": 0.35461094975471497, "learning_rate": 9.868786608753853e-05, "loss": 0.0325, "step": 35830 }, { "epoch": 22.84257488846399, "grad_norm": 0.2711033225059509, "learning_rate": 9.868661142083464e-05, "loss": 0.0261, "step": 35840 }, { "epoch": 22.848948374760994, "grad_norm": 0.536163330078125, "learning_rate": 9.868535616254243e-05, "loss": 0.0459, "step": 35850 }, { "epoch": 22.855321861058, "grad_norm": 0.5534836649894714, "learning_rate": 9.868410031267718e-05, "loss": 0.0429, "step": 35860 }, { "epoch": 22.861695347355003, "grad_norm": 0.4618101119995117, "learning_rate": 9.868284387125415e-05, "loss": 0.0293, "step": 35870 }, { "epoch": 22.868068833652007, "grad_norm": 0.6351956725120544, "learning_rate": 9.868158683828862e-05, "loss": 0.034, "step": 35880 }, { "epoch": 22.87444231994901, "grad_norm": 0.409598171710968, "learning_rate": 9.868032921379582e-05, "loss": 0.0361, "step": 35890 }, { "epoch": 22.880815806246016, "grad_norm": 0.3513404428958893, "learning_rate": 9.867907099779106e-05, "loss": 0.0328, "step": 35900 }, { "epoch": 22.88718929254302, "grad_norm": 0.46665483713150024, "learning_rate": 9.867781219028964e-05, "loss": 0.0254, "step": 35910 }, { "epoch": 22.893562778840025, "grad_norm": 0.6560744047164917, "learning_rate": 9.867655279130683e-05, "loss": 0.0531, "step": 35920 }, { "epoch": 22.89993626513703, "grad_norm": 0.4701434075832367, "learning_rate": 9.867529280085795e-05, "loss": 0.0261, "step": 35930 }, { "epoch": 22.906309751434033, "grad_norm": 0.5362236499786377, "learning_rate": 9.86740322189583e-05, "loss": 0.034, "step": 35940 }, { "epoch": 22.912683237731038, "grad_norm": 0.43048354983329773, "learning_rate": 9.867277104562323e-05, "loss": 0.0428, "step": 35950 }, { "epoch": 22.919056724028042, "grad_norm": 0.4683257043361664, "learning_rate": 9.867150928086801e-05, "loss": 0.0388, "step": 35960 }, { "epoch": 22.925430210325047, "grad_norm": 0.2844640612602234, "learning_rate": 9.8670246924708e-05, "loss": 0.0342, "step": 35970 }, { "epoch": 22.93180369662205, "grad_norm": 0.68386310338974, "learning_rate": 9.866898397715855e-05, "loss": 0.0412, "step": 35980 }, { "epoch": 22.938177182919055, "grad_norm": 0.5814060568809509, "learning_rate": 9.866772043823498e-05, "loss": 0.0475, "step": 35990 }, { "epoch": 22.94455066921606, "grad_norm": 0.49509575963020325, "learning_rate": 9.866645630795267e-05, "loss": 0.0342, "step": 36000 }, { "epoch": 22.950924155513064, "grad_norm": 0.24639937281608582, "learning_rate": 9.866519158632697e-05, "loss": 0.0394, "step": 36010 }, { "epoch": 22.95729764181007, "grad_norm": 0.26937276124954224, "learning_rate": 9.866392627337322e-05, "loss": 0.0292, "step": 36020 }, { "epoch": 22.963671128107073, "grad_norm": 0.587815523147583, "learning_rate": 9.866266036910683e-05, "loss": 0.0341, "step": 36030 }, { "epoch": 22.970044614404078, "grad_norm": 0.41701415181159973, "learning_rate": 9.866139387354318e-05, "loss": 0.0263, "step": 36040 }, { "epoch": 22.976418100701082, "grad_norm": 0.4268590807914734, "learning_rate": 9.866012678669764e-05, "loss": 0.0383, "step": 36050 }, { "epoch": 22.982791586998086, "grad_norm": 0.5448164343833923, "learning_rate": 9.865885910858564e-05, "loss": 0.0259, "step": 36060 }, { "epoch": 22.98916507329509, "grad_norm": 0.5478262901306152, "learning_rate": 9.865759083922253e-05, "loss": 0.0249, "step": 36070 }, { "epoch": 22.9955385595921, "grad_norm": 0.3121899366378784, "learning_rate": 9.865632197862377e-05, "loss": 0.0282, "step": 36080 }, { "epoch": 23.001912045889103, "grad_norm": 0.45514288544654846, "learning_rate": 9.865505252680476e-05, "loss": 0.0309, "step": 36090 }, { "epoch": 23.008285532186108, "grad_norm": 0.31081923842430115, "learning_rate": 9.865378248378091e-05, "loss": 0.0309, "step": 36100 }, { "epoch": 23.014659018483112, "grad_norm": 0.656318724155426, "learning_rate": 9.865251184956767e-05, "loss": 0.0498, "step": 36110 }, { "epoch": 23.021032504780116, "grad_norm": 0.42957475781440735, "learning_rate": 9.865124062418048e-05, "loss": 0.0208, "step": 36120 }, { "epoch": 23.02740599107712, "grad_norm": 0.5568720698356628, "learning_rate": 9.864996880763478e-05, "loss": 0.0248, "step": 36130 }, { "epoch": 23.033779477374125, "grad_norm": 0.2583651840686798, "learning_rate": 9.864869639994603e-05, "loss": 0.0328, "step": 36140 }, { "epoch": 23.04015296367113, "grad_norm": 0.5028584599494934, "learning_rate": 9.864742340112969e-05, "loss": 0.0318, "step": 36150 }, { "epoch": 23.046526449968134, "grad_norm": 0.8908265233039856, "learning_rate": 9.864614981120122e-05, "loss": 0.0349, "step": 36160 }, { "epoch": 23.05289993626514, "grad_norm": 0.6324557662010193, "learning_rate": 9.864487563017609e-05, "loss": 0.0337, "step": 36170 }, { "epoch": 23.059273422562143, "grad_norm": 0.5461339950561523, "learning_rate": 9.864360085806981e-05, "loss": 0.036, "step": 36180 }, { "epoch": 23.065646908859147, "grad_norm": 0.6034703254699707, "learning_rate": 9.864232549489782e-05, "loss": 0.0273, "step": 36190 }, { "epoch": 23.07202039515615, "grad_norm": 0.9316211938858032, "learning_rate": 9.864104954067567e-05, "loss": 0.0387, "step": 36200 }, { "epoch": 23.078393881453156, "grad_norm": 0.4064961373806, "learning_rate": 9.863977299541887e-05, "loss": 0.0243, "step": 36210 }, { "epoch": 23.08476736775016, "grad_norm": 0.4518246054649353, "learning_rate": 9.863849585914286e-05, "loss": 0.0499, "step": 36220 }, { "epoch": 23.091140854047165, "grad_norm": 0.7394051551818848, "learning_rate": 9.863721813186324e-05, "loss": 0.0354, "step": 36230 }, { "epoch": 23.09751434034417, "grad_norm": 0.3792189359664917, "learning_rate": 9.863593981359548e-05, "loss": 0.0345, "step": 36240 }, { "epoch": 23.103887826641174, "grad_norm": 0.6214410066604614, "learning_rate": 9.863466090435513e-05, "loss": 0.0222, "step": 36250 }, { "epoch": 23.110261312938178, "grad_norm": 0.34806138277053833, "learning_rate": 9.863338140415774e-05, "loss": 0.0322, "step": 36260 }, { "epoch": 23.116634799235182, "grad_norm": 0.434895396232605, "learning_rate": 9.863210131301883e-05, "loss": 0.0347, "step": 36270 }, { "epoch": 23.123008285532187, "grad_norm": 0.5578153133392334, "learning_rate": 9.863082063095399e-05, "loss": 0.0328, "step": 36280 }, { "epoch": 23.12938177182919, "grad_norm": 0.6369719505310059, "learning_rate": 9.862953935797876e-05, "loss": 0.0438, "step": 36290 }, { "epoch": 23.135755258126196, "grad_norm": 0.3536309003829956, "learning_rate": 9.862825749410871e-05, "loss": 0.042, "step": 36300 }, { "epoch": 23.1421287444232, "grad_norm": 0.5308420658111572, "learning_rate": 9.862697503935942e-05, "loss": 0.0247, "step": 36310 }, { "epoch": 23.148502230720204, "grad_norm": 0.7654992341995239, "learning_rate": 9.862569199374648e-05, "loss": 0.0301, "step": 36320 }, { "epoch": 23.15487571701721, "grad_norm": 0.6247915625572205, "learning_rate": 9.862440835728547e-05, "loss": 0.0402, "step": 36330 }, { "epoch": 23.161249203314213, "grad_norm": 0.36276599764823914, "learning_rate": 9.862312412999198e-05, "loss": 0.0314, "step": 36340 }, { "epoch": 23.167622689611218, "grad_norm": 0.4394279420375824, "learning_rate": 9.862183931188162e-05, "loss": 0.0284, "step": 36350 }, { "epoch": 23.173996175908222, "grad_norm": 0.5732686519622803, "learning_rate": 9.862055390297003e-05, "loss": 0.0392, "step": 36360 }, { "epoch": 23.180369662205226, "grad_norm": 0.4227307438850403, "learning_rate": 9.861926790327278e-05, "loss": 0.0313, "step": 36370 }, { "epoch": 23.18674314850223, "grad_norm": 0.6741951704025269, "learning_rate": 9.861798131280553e-05, "loss": 0.038, "step": 36380 }, { "epoch": 23.193116634799235, "grad_norm": 0.2974992096424103, "learning_rate": 9.861669413158391e-05, "loss": 0.0342, "step": 36390 }, { "epoch": 23.19949012109624, "grad_norm": 0.3867451548576355, "learning_rate": 9.861540635962354e-05, "loss": 0.0405, "step": 36400 }, { "epoch": 23.205863607393244, "grad_norm": 0.4575144052505493, "learning_rate": 9.861411799694009e-05, "loss": 0.035, "step": 36410 }, { "epoch": 23.21223709369025, "grad_norm": 0.5691337585449219, "learning_rate": 9.861282904354922e-05, "loss": 0.0283, "step": 36420 }, { "epoch": 23.218610579987253, "grad_norm": 0.5980804562568665, "learning_rate": 9.861153949946655e-05, "loss": 0.0356, "step": 36430 }, { "epoch": 23.224984066284257, "grad_norm": 0.5748704671859741, "learning_rate": 9.86102493647078e-05, "loss": 0.0463, "step": 36440 }, { "epoch": 23.23135755258126, "grad_norm": 0.752718448638916, "learning_rate": 9.860895863928863e-05, "loss": 0.04, "step": 36450 }, { "epoch": 23.237731038878266, "grad_norm": 0.4540983736515045, "learning_rate": 9.86076673232247e-05, "loss": 0.0421, "step": 36460 }, { "epoch": 23.24410452517527, "grad_norm": 0.8507975339889526, "learning_rate": 9.860637541653172e-05, "loss": 0.0428, "step": 36470 }, { "epoch": 23.250478011472275, "grad_norm": 0.5903734564781189, "learning_rate": 9.860508291922541e-05, "loss": 0.0328, "step": 36480 }, { "epoch": 23.25685149776928, "grad_norm": 0.4271882176399231, "learning_rate": 9.860378983132143e-05, "loss": 0.0281, "step": 36490 }, { "epoch": 23.263224984066284, "grad_norm": 0.3872644007205963, "learning_rate": 9.860249615283552e-05, "loss": 0.0307, "step": 36500 }, { "epoch": 23.269598470363288, "grad_norm": 0.4694727659225464, "learning_rate": 9.860120188378339e-05, "loss": 0.0271, "step": 36510 }, { "epoch": 23.275971956660293, "grad_norm": 0.7661231756210327, "learning_rate": 9.859990702418077e-05, "loss": 0.0299, "step": 36520 }, { "epoch": 23.282345442957297, "grad_norm": 0.5404526591300964, "learning_rate": 9.85986115740434e-05, "loss": 0.0304, "step": 36530 }, { "epoch": 23.2887189292543, "grad_norm": 0.5391700863838196, "learning_rate": 9.859731553338701e-05, "loss": 0.0323, "step": 36540 }, { "epoch": 23.295092415551306, "grad_norm": 0.5226530432701111, "learning_rate": 9.859601890222735e-05, "loss": 0.0359, "step": 36550 }, { "epoch": 23.30146590184831, "grad_norm": 0.4017866551876068, "learning_rate": 9.859472168058018e-05, "loss": 0.0281, "step": 36560 }, { "epoch": 23.307839388145315, "grad_norm": 0.5931110382080078, "learning_rate": 9.859342386846127e-05, "loss": 0.0359, "step": 36570 }, { "epoch": 23.31421287444232, "grad_norm": 0.39134860038757324, "learning_rate": 9.859212546588636e-05, "loss": 0.0236, "step": 36580 }, { "epoch": 23.320586360739323, "grad_norm": 0.5923264026641846, "learning_rate": 9.859082647287126e-05, "loss": 0.047, "step": 36590 }, { "epoch": 23.326959847036328, "grad_norm": 0.49127089977264404, "learning_rate": 9.858952688943174e-05, "loss": 0.0443, "step": 36600 }, { "epoch": 23.333333333333332, "grad_norm": 0.48142293095588684, "learning_rate": 9.85882267155836e-05, "loss": 0.0376, "step": 36610 }, { "epoch": 23.339706819630337, "grad_norm": 0.6156603097915649, "learning_rate": 9.858692595134263e-05, "loss": 0.0286, "step": 36620 }, { "epoch": 23.34608030592734, "grad_norm": 0.6895135641098022, "learning_rate": 9.858562459672462e-05, "loss": 0.032, "step": 36630 }, { "epoch": 23.352453792224345, "grad_norm": 0.45453163981437683, "learning_rate": 9.858432265174541e-05, "loss": 0.0419, "step": 36640 }, { "epoch": 23.35882727852135, "grad_norm": 0.603053867816925, "learning_rate": 9.85830201164208e-05, "loss": 0.0291, "step": 36650 }, { "epoch": 23.365200764818354, "grad_norm": 0.41972616314888, "learning_rate": 9.858171699076664e-05, "loss": 0.0347, "step": 36660 }, { "epoch": 23.37157425111536, "grad_norm": 0.5342583656311035, "learning_rate": 9.858041327479875e-05, "loss": 0.0324, "step": 36670 }, { "epoch": 23.377947737412363, "grad_norm": 0.35054877400398254, "learning_rate": 9.857910896853295e-05, "loss": 0.0427, "step": 36680 }, { "epoch": 23.384321223709367, "grad_norm": 0.49216216802597046, "learning_rate": 9.857780407198512e-05, "loss": 0.035, "step": 36690 }, { "epoch": 23.390694710006372, "grad_norm": 0.39326122403144836, "learning_rate": 9.85764985851711e-05, "loss": 0.0351, "step": 36700 }, { "epoch": 23.39706819630338, "grad_norm": 0.6145009994506836, "learning_rate": 9.857519250810678e-05, "loss": 0.0212, "step": 36710 }, { "epoch": 23.403441682600384, "grad_norm": 0.4676506519317627, "learning_rate": 9.8573885840808e-05, "loss": 0.0363, "step": 36720 }, { "epoch": 23.40981516889739, "grad_norm": 0.3322041928768158, "learning_rate": 9.857257858329063e-05, "loss": 0.0359, "step": 36730 }, { "epoch": 23.416188655194393, "grad_norm": 0.6079815030097961, "learning_rate": 9.857127073557057e-05, "loss": 0.039, "step": 36740 }, { "epoch": 23.422562141491397, "grad_norm": 0.5640106201171875, "learning_rate": 9.856996229766374e-05, "loss": 0.02, "step": 36750 }, { "epoch": 23.4289356277884, "grad_norm": 0.7194762229919434, "learning_rate": 9.856865326958599e-05, "loss": 0.0318, "step": 36760 }, { "epoch": 23.435309114085406, "grad_norm": 0.4033360183238983, "learning_rate": 9.856734365135324e-05, "loss": 0.0333, "step": 36770 }, { "epoch": 23.44168260038241, "grad_norm": 0.5370861887931824, "learning_rate": 9.856603344298143e-05, "loss": 0.0384, "step": 36780 }, { "epoch": 23.448056086679415, "grad_norm": 0.45059478282928467, "learning_rate": 9.856472264448645e-05, "loss": 0.0331, "step": 36790 }, { "epoch": 23.45442957297642, "grad_norm": 0.6513437628746033, "learning_rate": 9.856341125588423e-05, "loss": 0.0406, "step": 36800 }, { "epoch": 23.460803059273424, "grad_norm": 0.3299298882484436, "learning_rate": 9.856209927719071e-05, "loss": 0.0291, "step": 36810 }, { "epoch": 23.467176545570428, "grad_norm": 0.5888707041740417, "learning_rate": 9.856078670842186e-05, "loss": 0.0365, "step": 36820 }, { "epoch": 23.473550031867433, "grad_norm": 0.7050340175628662, "learning_rate": 9.855947354959357e-05, "loss": 0.0448, "step": 36830 }, { "epoch": 23.479923518164437, "grad_norm": 0.4835750460624695, "learning_rate": 9.855815980072185e-05, "loss": 0.0289, "step": 36840 }, { "epoch": 23.48629700446144, "grad_norm": 0.6583870053291321, "learning_rate": 9.855684546182262e-05, "loss": 0.0324, "step": 36850 }, { "epoch": 23.492670490758446, "grad_norm": 0.8079191446304321, "learning_rate": 9.85555305329119e-05, "loss": 0.0515, "step": 36860 }, { "epoch": 23.49904397705545, "grad_norm": 0.38979196548461914, "learning_rate": 9.855421501400562e-05, "loss": 0.035, "step": 36870 }, { "epoch": 23.505417463352455, "grad_norm": 0.421710342168808, "learning_rate": 9.855289890511979e-05, "loss": 0.0393, "step": 36880 }, { "epoch": 23.51179094964946, "grad_norm": 0.35926273465156555, "learning_rate": 9.85515822062704e-05, "loss": 0.0441, "step": 36890 }, { "epoch": 23.518164435946463, "grad_norm": 0.571291983127594, "learning_rate": 9.855026491747344e-05, "loss": 0.0398, "step": 36900 }, { "epoch": 23.524537922243468, "grad_norm": 0.7023671269416809, "learning_rate": 9.854894703874493e-05, "loss": 0.0367, "step": 36910 }, { "epoch": 23.530911408540472, "grad_norm": 0.2783041298389435, "learning_rate": 9.854762857010087e-05, "loss": 0.0406, "step": 36920 }, { "epoch": 23.537284894837477, "grad_norm": 1.0094852447509766, "learning_rate": 9.85463095115573e-05, "loss": 0.0425, "step": 36930 }, { "epoch": 23.54365838113448, "grad_norm": 0.47917866706848145, "learning_rate": 9.854498986313021e-05, "loss": 0.0348, "step": 36940 }, { "epoch": 23.550031867431485, "grad_norm": 0.4936361312866211, "learning_rate": 9.854366962483568e-05, "loss": 0.0393, "step": 36950 }, { "epoch": 23.55640535372849, "grad_norm": 0.9955158233642578, "learning_rate": 9.854234879668971e-05, "loss": 0.0262, "step": 36960 }, { "epoch": 23.562778840025494, "grad_norm": 0.8160923719406128, "learning_rate": 9.854102737870838e-05, "loss": 0.0577, "step": 36970 }, { "epoch": 23.5691523263225, "grad_norm": 0.6121181845664978, "learning_rate": 9.853970537090775e-05, "loss": 0.0481, "step": 36980 }, { "epoch": 23.575525812619503, "grad_norm": 0.5603047013282776, "learning_rate": 9.853838277330385e-05, "loss": 0.0405, "step": 36990 }, { "epoch": 23.581899298916507, "grad_norm": 0.25304895639419556, "learning_rate": 9.853705958591279e-05, "loss": 0.0271, "step": 37000 }, { "epoch": 23.588272785213512, "grad_norm": 0.415301650762558, "learning_rate": 9.853573580875062e-05, "loss": 0.037, "step": 37010 }, { "epoch": 23.594646271510516, "grad_norm": 0.43318790197372437, "learning_rate": 9.853441144183345e-05, "loss": 0.0335, "step": 37020 }, { "epoch": 23.60101975780752, "grad_norm": 0.4673224091529846, "learning_rate": 9.853308648517732e-05, "loss": 0.0359, "step": 37030 }, { "epoch": 23.607393244104525, "grad_norm": 0.47700899839401245, "learning_rate": 9.85317609387984e-05, "loss": 0.0279, "step": 37040 }, { "epoch": 23.61376673040153, "grad_norm": 0.9069885015487671, "learning_rate": 9.853043480271276e-05, "loss": 0.036, "step": 37050 }, { "epoch": 23.620140216698534, "grad_norm": 1.0800111293792725, "learning_rate": 9.85291080769365e-05, "loss": 0.0415, "step": 37060 }, { "epoch": 23.62651370299554, "grad_norm": 0.37277042865753174, "learning_rate": 9.852778076148576e-05, "loss": 0.0271, "step": 37070 }, { "epoch": 23.632887189292543, "grad_norm": 0.6269493699073792, "learning_rate": 9.852645285637668e-05, "loss": 0.0416, "step": 37080 }, { "epoch": 23.639260675589547, "grad_norm": 0.2793123424053192, "learning_rate": 9.852512436162538e-05, "loss": 0.0317, "step": 37090 }, { "epoch": 23.64563416188655, "grad_norm": 0.34734880924224854, "learning_rate": 9.8523795277248e-05, "loss": 0.0293, "step": 37100 }, { "epoch": 23.652007648183556, "grad_norm": 0.661919891834259, "learning_rate": 9.852246560326069e-05, "loss": 0.0383, "step": 37110 }, { "epoch": 23.65838113448056, "grad_norm": 0.33034586906433105, "learning_rate": 9.852113533967961e-05, "loss": 0.0295, "step": 37120 }, { "epoch": 23.664754620777565, "grad_norm": 0.5846928954124451, "learning_rate": 9.851980448652092e-05, "loss": 0.0381, "step": 37130 }, { "epoch": 23.67112810707457, "grad_norm": 0.41564038395881653, "learning_rate": 9.85184730438008e-05, "loss": 0.0362, "step": 37140 }, { "epoch": 23.677501593371574, "grad_norm": 0.39438435435295105, "learning_rate": 9.851714101153542e-05, "loss": 0.0311, "step": 37150 }, { "epoch": 23.683875079668578, "grad_norm": 0.6188374161720276, "learning_rate": 9.851580838974097e-05, "loss": 0.0284, "step": 37160 }, { "epoch": 23.690248565965582, "grad_norm": 0.4564496874809265, "learning_rate": 9.851447517843366e-05, "loss": 0.0383, "step": 37170 }, { "epoch": 23.696622052262587, "grad_norm": 0.6168647408485413, "learning_rate": 9.851314137762965e-05, "loss": 0.0311, "step": 37180 }, { "epoch": 23.70299553855959, "grad_norm": 0.416768878698349, "learning_rate": 9.851180698734517e-05, "loss": 0.05, "step": 37190 }, { "epoch": 23.709369024856596, "grad_norm": 0.6282676458358765, "learning_rate": 9.851047200759645e-05, "loss": 0.0391, "step": 37200 }, { "epoch": 23.7157425111536, "grad_norm": 0.5219295620918274, "learning_rate": 9.850913643839967e-05, "loss": 0.028, "step": 37210 }, { "epoch": 23.722115997450604, "grad_norm": 0.4956703782081604, "learning_rate": 9.850780027977109e-05, "loss": 0.0295, "step": 37220 }, { "epoch": 23.72848948374761, "grad_norm": 0.6324825882911682, "learning_rate": 9.850646353172694e-05, "loss": 0.0371, "step": 37230 }, { "epoch": 23.734862970044613, "grad_norm": 0.5696253180503845, "learning_rate": 9.850512619428346e-05, "loss": 0.0455, "step": 37240 }, { "epoch": 23.741236456341618, "grad_norm": 0.5835239887237549, "learning_rate": 9.85037882674569e-05, "loss": 0.0412, "step": 37250 }, { "epoch": 23.747609942638622, "grad_norm": 0.6295267939567566, "learning_rate": 9.850244975126352e-05, "loss": 0.0326, "step": 37260 }, { "epoch": 23.753983428935626, "grad_norm": 0.6781933307647705, "learning_rate": 9.850111064571958e-05, "loss": 0.0277, "step": 37270 }, { "epoch": 23.76035691523263, "grad_norm": 0.8851192593574524, "learning_rate": 9.849977095084135e-05, "loss": 0.035, "step": 37280 }, { "epoch": 23.766730401529635, "grad_norm": 0.7456137537956238, "learning_rate": 9.849843066664512e-05, "loss": 0.0455, "step": 37290 }, { "epoch": 23.77310388782664, "grad_norm": 1.1779190301895142, "learning_rate": 9.849708979314716e-05, "loss": 0.0334, "step": 37300 }, { "epoch": 23.779477374123644, "grad_norm": 0.801834762096405, "learning_rate": 9.849574833036377e-05, "loss": 0.0375, "step": 37310 }, { "epoch": 23.78585086042065, "grad_norm": 0.7612029910087585, "learning_rate": 9.849440627831126e-05, "loss": 0.0391, "step": 37320 }, { "epoch": 23.792224346717653, "grad_norm": 0.49355363845825195, "learning_rate": 9.849306363700591e-05, "loss": 0.0411, "step": 37330 }, { "epoch": 23.79859783301466, "grad_norm": 0.4595854580402374, "learning_rate": 9.849172040646406e-05, "loss": 0.0264, "step": 37340 }, { "epoch": 23.80497131931166, "grad_norm": 0.2429874986410141, "learning_rate": 9.849037658670202e-05, "loss": 0.039, "step": 37350 }, { "epoch": 23.81134480560867, "grad_norm": 0.6151285767555237, "learning_rate": 9.848903217773613e-05, "loss": 0.0438, "step": 37360 }, { "epoch": 23.817718291905674, "grad_norm": 0.49741554260253906, "learning_rate": 9.848768717958272e-05, "loss": 0.0286, "step": 37370 }, { "epoch": 23.82409177820268, "grad_norm": 0.5290849208831787, "learning_rate": 9.848634159225812e-05, "loss": 0.0279, "step": 37380 }, { "epoch": 23.830465264499683, "grad_norm": 0.6803759336471558, "learning_rate": 9.848499541577871e-05, "loss": 0.0368, "step": 37390 }, { "epoch": 23.836838750796687, "grad_norm": 0.5212733745574951, "learning_rate": 9.84836486501608e-05, "loss": 0.0338, "step": 37400 }, { "epoch": 23.84321223709369, "grad_norm": 0.3238081932067871, "learning_rate": 9.848230129542079e-05, "loss": 0.033, "step": 37410 }, { "epoch": 23.849585723390696, "grad_norm": 0.4323550760746002, "learning_rate": 9.848095335157506e-05, "loss": 0.0375, "step": 37420 }, { "epoch": 23.8559592096877, "grad_norm": 0.3778531849384308, "learning_rate": 9.847960481863997e-05, "loss": 0.0349, "step": 37430 }, { "epoch": 23.862332695984705, "grad_norm": 0.4219679832458496, "learning_rate": 9.847825569663189e-05, "loss": 0.0303, "step": 37440 }, { "epoch": 23.86870618228171, "grad_norm": 0.680701494216919, "learning_rate": 9.847690598556726e-05, "loss": 0.0582, "step": 37450 }, { "epoch": 23.875079668578714, "grad_norm": 0.5344898104667664, "learning_rate": 9.847555568546243e-05, "loss": 0.0448, "step": 37460 }, { "epoch": 23.881453154875718, "grad_norm": 0.798287570476532, "learning_rate": 9.847420479633384e-05, "loss": 0.0487, "step": 37470 }, { "epoch": 23.887826641172722, "grad_norm": 0.5545568466186523, "learning_rate": 9.847285331819788e-05, "loss": 0.0365, "step": 37480 }, { "epoch": 23.894200127469727, "grad_norm": 0.6355661153793335, "learning_rate": 9.8471501251071e-05, "loss": 0.0323, "step": 37490 }, { "epoch": 23.90057361376673, "grad_norm": 0.6961902976036072, "learning_rate": 9.847014859496963e-05, "loss": 0.045, "step": 37500 }, { "epoch": 23.906947100063736, "grad_norm": 0.7367477416992188, "learning_rate": 9.846879534991018e-05, "loss": 0.0417, "step": 37510 }, { "epoch": 23.91332058636074, "grad_norm": 0.4929988384246826, "learning_rate": 9.84674415159091e-05, "loss": 0.0315, "step": 37520 }, { "epoch": 23.919694072657744, "grad_norm": 1.0386641025543213, "learning_rate": 9.846608709298284e-05, "loss": 0.0454, "step": 37530 }, { "epoch": 23.92606755895475, "grad_norm": 0.4639098048210144, "learning_rate": 9.846473208114786e-05, "loss": 0.0316, "step": 37540 }, { "epoch": 23.932441045251753, "grad_norm": 0.9023931622505188, "learning_rate": 9.846337648042063e-05, "loss": 0.0409, "step": 37550 }, { "epoch": 23.938814531548758, "grad_norm": 0.3718760013580322, "learning_rate": 9.846202029081763e-05, "loss": 0.0325, "step": 37560 }, { "epoch": 23.945188017845762, "grad_norm": 0.43446189165115356, "learning_rate": 9.846066351235532e-05, "loss": 0.0382, "step": 37570 }, { "epoch": 23.951561504142767, "grad_norm": 0.30794772505760193, "learning_rate": 9.845930614505019e-05, "loss": 0.0305, "step": 37580 }, { "epoch": 23.95793499043977, "grad_norm": 0.4514079988002777, "learning_rate": 9.845794818891875e-05, "loss": 0.026, "step": 37590 }, { "epoch": 23.964308476736775, "grad_norm": 0.5350214838981628, "learning_rate": 9.845658964397748e-05, "loss": 0.0339, "step": 37600 }, { "epoch": 23.97068196303378, "grad_norm": 0.5738411545753479, "learning_rate": 9.84552305102429e-05, "loss": 0.0392, "step": 37610 }, { "epoch": 23.977055449330784, "grad_norm": 0.7394959330558777, "learning_rate": 9.84538707877315e-05, "loss": 0.0411, "step": 37620 }, { "epoch": 23.98342893562779, "grad_norm": 0.632942259311676, "learning_rate": 9.845251047645983e-05, "loss": 0.0328, "step": 37630 }, { "epoch": 23.989802421924793, "grad_norm": 0.37757253646850586, "learning_rate": 9.845114957644442e-05, "loss": 0.0332, "step": 37640 }, { "epoch": 23.996175908221797, "grad_norm": 0.46853557229042053, "learning_rate": 9.84497880877018e-05, "loss": 0.0318, "step": 37650 }, { "epoch": 24.0025493945188, "grad_norm": 0.4802883565425873, "learning_rate": 9.844842601024851e-05, "loss": 0.032, "step": 37660 }, { "epoch": 24.008922880815806, "grad_norm": 0.5748679637908936, "learning_rate": 9.844706334410108e-05, "loss": 0.053, "step": 37670 }, { "epoch": 24.01529636711281, "grad_norm": 0.6048776507377625, "learning_rate": 9.844570008927611e-05, "loss": 0.0357, "step": 37680 }, { "epoch": 24.021669853409815, "grad_norm": 0.7143999934196472, "learning_rate": 9.844433624579014e-05, "loss": 0.0361, "step": 37690 }, { "epoch": 24.02804333970682, "grad_norm": 0.4235288202762604, "learning_rate": 9.844297181365974e-05, "loss": 0.0297, "step": 37700 }, { "epoch": 24.034416826003824, "grad_norm": 0.4564266502857208, "learning_rate": 9.84416067929015e-05, "loss": 0.0368, "step": 37710 }, { "epoch": 24.040790312300828, "grad_norm": 0.6228170394897461, "learning_rate": 9.8440241183532e-05, "loss": 0.031, "step": 37720 }, { "epoch": 24.047163798597833, "grad_norm": 0.40617209672927856, "learning_rate": 9.843887498556783e-05, "loss": 0.0312, "step": 37730 }, { "epoch": 24.053537284894837, "grad_norm": 0.39136525988578796, "learning_rate": 9.84375081990256e-05, "loss": 0.0309, "step": 37740 }, { "epoch": 24.05991077119184, "grad_norm": 0.8294902443885803, "learning_rate": 9.843614082392192e-05, "loss": 0.0283, "step": 37750 }, { "epoch": 24.066284257488846, "grad_norm": 0.640822172164917, "learning_rate": 9.843477286027338e-05, "loss": 0.0438, "step": 37760 }, { "epoch": 24.07265774378585, "grad_norm": 0.9119243621826172, "learning_rate": 9.843340430809663e-05, "loss": 0.0555, "step": 37770 }, { "epoch": 24.079031230082855, "grad_norm": 0.5937010049819946, "learning_rate": 9.843203516740829e-05, "loss": 0.0377, "step": 37780 }, { "epoch": 24.08540471637986, "grad_norm": 0.515120804309845, "learning_rate": 9.8430665438225e-05, "loss": 0.0337, "step": 37790 }, { "epoch": 24.091778202676863, "grad_norm": 0.4904628396034241, "learning_rate": 9.842929512056338e-05, "loss": 0.0342, "step": 37800 }, { "epoch": 24.098151688973868, "grad_norm": 0.18511086702346802, "learning_rate": 9.842792421444011e-05, "loss": 0.0292, "step": 37810 }, { "epoch": 24.104525175270872, "grad_norm": 0.27261674404144287, "learning_rate": 9.842655271987185e-05, "loss": 0.042, "step": 37820 }, { "epoch": 24.110898661567877, "grad_norm": 0.5782015323638916, "learning_rate": 9.842518063687524e-05, "loss": 0.0451, "step": 37830 }, { "epoch": 24.11727214786488, "grad_norm": 0.44219568371772766, "learning_rate": 9.842380796546697e-05, "loss": 0.0304, "step": 37840 }, { "epoch": 24.123645634161885, "grad_norm": 0.5537877082824707, "learning_rate": 9.84224347056637e-05, "loss": 0.0391, "step": 37850 }, { "epoch": 24.13001912045889, "grad_norm": 0.4428641200065613, "learning_rate": 9.842106085748215e-05, "loss": 0.0279, "step": 37860 }, { "epoch": 24.136392606755894, "grad_norm": 0.4276747703552246, "learning_rate": 9.841968642093898e-05, "loss": 0.0341, "step": 37870 }, { "epoch": 24.1427660930529, "grad_norm": 0.3970367908477783, "learning_rate": 9.84183113960509e-05, "loss": 0.0313, "step": 37880 }, { "epoch": 24.149139579349903, "grad_norm": 0.46533265709877014, "learning_rate": 9.841693578283463e-05, "loss": 0.0461, "step": 37890 }, { "epoch": 24.155513065646907, "grad_norm": 0.5321467518806458, "learning_rate": 9.841555958130688e-05, "loss": 0.0318, "step": 37900 }, { "epoch": 24.161886551943912, "grad_norm": 0.43100494146347046, "learning_rate": 9.841418279148435e-05, "loss": 0.0305, "step": 37910 }, { "epoch": 24.168260038240916, "grad_norm": 0.4807056486606598, "learning_rate": 9.841280541338382e-05, "loss": 0.0396, "step": 37920 }, { "epoch": 24.17463352453792, "grad_norm": 0.565575361251831, "learning_rate": 9.841142744702197e-05, "loss": 0.0335, "step": 37930 }, { "epoch": 24.181007010834925, "grad_norm": 0.5753377676010132, "learning_rate": 9.841004889241559e-05, "loss": 0.0412, "step": 37940 }, { "epoch": 24.18738049713193, "grad_norm": 0.4535566568374634, "learning_rate": 9.840866974958138e-05, "loss": 0.0343, "step": 37950 }, { "epoch": 24.193753983428934, "grad_norm": 0.5226954221725464, "learning_rate": 9.840729001853615e-05, "loss": 0.0569, "step": 37960 }, { "epoch": 24.200127469725942, "grad_norm": 0.6045295596122742, "learning_rate": 9.840590969929664e-05, "loss": 0.0363, "step": 37970 }, { "epoch": 24.206500956022946, "grad_norm": 0.8840661644935608, "learning_rate": 9.840452879187962e-05, "loss": 0.0391, "step": 37980 }, { "epoch": 24.21287444231995, "grad_norm": 0.4054742455482483, "learning_rate": 9.840314729630186e-05, "loss": 0.0276, "step": 37990 }, { "epoch": 24.219247928616955, "grad_norm": 0.6850738525390625, "learning_rate": 9.840176521258017e-05, "loss": 0.0281, "step": 38000 }, { "epoch": 24.22562141491396, "grad_norm": 0.33180710673332214, "learning_rate": 9.840038254073134e-05, "loss": 0.0598, "step": 38010 }, { "epoch": 24.231994901210964, "grad_norm": 0.6189227104187012, "learning_rate": 9.839899928077216e-05, "loss": 0.0487, "step": 38020 }, { "epoch": 24.23836838750797, "grad_norm": 0.5323652625083923, "learning_rate": 9.839761543271946e-05, "loss": 0.0317, "step": 38030 }, { "epoch": 24.244741873804973, "grad_norm": 0.3720780313014984, "learning_rate": 9.839623099659002e-05, "loss": 0.0301, "step": 38040 }, { "epoch": 24.251115360101977, "grad_norm": 0.5797668695449829, "learning_rate": 9.839484597240067e-05, "loss": 0.0263, "step": 38050 }, { "epoch": 24.25748884639898, "grad_norm": 0.4359554648399353, "learning_rate": 9.839346036016827e-05, "loss": 0.0432, "step": 38060 }, { "epoch": 24.263862332695986, "grad_norm": 0.4080764949321747, "learning_rate": 9.839207415990963e-05, "loss": 0.0398, "step": 38070 }, { "epoch": 24.27023581899299, "grad_norm": 0.5639428496360779, "learning_rate": 9.83906873716416e-05, "loss": 0.0247, "step": 38080 }, { "epoch": 24.276609305289995, "grad_norm": 0.42230623960494995, "learning_rate": 9.838929999538102e-05, "loss": 0.035, "step": 38090 }, { "epoch": 24.282982791587, "grad_norm": 0.4116472005844116, "learning_rate": 9.838791203114477e-05, "loss": 0.0295, "step": 38100 }, { "epoch": 24.289356277884004, "grad_norm": 0.7337201833724976, "learning_rate": 9.83865234789497e-05, "loss": 0.0425, "step": 38110 }, { "epoch": 24.295729764181008, "grad_norm": 0.638992428779602, "learning_rate": 9.838513433881268e-05, "loss": 0.0399, "step": 38120 }, { "epoch": 24.302103250478012, "grad_norm": 0.7952844500541687, "learning_rate": 9.838374461075059e-05, "loss": 0.0317, "step": 38130 }, { "epoch": 24.308476736775017, "grad_norm": 0.6632501482963562, "learning_rate": 9.838235429478033e-05, "loss": 0.0347, "step": 38140 }, { "epoch": 24.31485022307202, "grad_norm": 0.4192325472831726, "learning_rate": 9.838096339091877e-05, "loss": 0.0491, "step": 38150 }, { "epoch": 24.321223709369026, "grad_norm": 0.4519707262516022, "learning_rate": 9.837957189918285e-05, "loss": 0.0279, "step": 38160 }, { "epoch": 24.32759719566603, "grad_norm": 0.36190348863601685, "learning_rate": 9.837817981958945e-05, "loss": 0.0358, "step": 38170 }, { "epoch": 24.333970681963034, "grad_norm": 0.7802234292030334, "learning_rate": 9.837678715215548e-05, "loss": 0.0391, "step": 38180 }, { "epoch": 24.34034416826004, "grad_norm": 0.32212376594543457, "learning_rate": 9.837539389689787e-05, "loss": 0.0392, "step": 38190 }, { "epoch": 24.346717654557043, "grad_norm": 0.5439888834953308, "learning_rate": 9.837400005383354e-05, "loss": 0.0362, "step": 38200 }, { "epoch": 24.353091140854048, "grad_norm": 0.549466609954834, "learning_rate": 9.837260562297946e-05, "loss": 0.0332, "step": 38210 }, { "epoch": 24.359464627151052, "grad_norm": 1.0089901685714722, "learning_rate": 9.837121060435254e-05, "loss": 0.0347, "step": 38220 }, { "epoch": 24.365838113448056, "grad_norm": 0.3550252914428711, "learning_rate": 9.836981499796974e-05, "loss": 0.0362, "step": 38230 }, { "epoch": 24.37221159974506, "grad_norm": 0.4326470196247101, "learning_rate": 9.836841880384802e-05, "loss": 0.0489, "step": 38240 }, { "epoch": 24.378585086042065, "grad_norm": 0.492059588432312, "learning_rate": 9.836702202200434e-05, "loss": 0.0314, "step": 38250 }, { "epoch": 24.38495857233907, "grad_norm": 0.4955512285232544, "learning_rate": 9.83656246524557e-05, "loss": 0.0298, "step": 38260 }, { "epoch": 24.391332058636074, "grad_norm": 0.7034401893615723, "learning_rate": 9.836422669521902e-05, "loss": 0.0355, "step": 38270 }, { "epoch": 24.39770554493308, "grad_norm": 0.594855785369873, "learning_rate": 9.836282815031134e-05, "loss": 0.0372, "step": 38280 }, { "epoch": 24.404079031230083, "grad_norm": 0.5855913758277893, "learning_rate": 9.836142901774963e-05, "loss": 0.0351, "step": 38290 }, { "epoch": 24.410452517527087, "grad_norm": 0.656150758266449, "learning_rate": 9.83600292975509e-05, "loss": 0.0267, "step": 38300 }, { "epoch": 24.41682600382409, "grad_norm": 0.49173814058303833, "learning_rate": 9.835862898973216e-05, "loss": 0.038, "step": 38310 }, { "epoch": 24.423199490121096, "grad_norm": 0.42426395416259766, "learning_rate": 9.83572280943104e-05, "loss": 0.0385, "step": 38320 }, { "epoch": 24.4295729764181, "grad_norm": 0.3419315814971924, "learning_rate": 9.83558266113027e-05, "loss": 0.0321, "step": 38330 }, { "epoch": 24.435946462715105, "grad_norm": 0.5565381050109863, "learning_rate": 9.8354424540726e-05, "loss": 0.0339, "step": 38340 }, { "epoch": 24.44231994901211, "grad_norm": 0.6314712762832642, "learning_rate": 9.835302188259743e-05, "loss": 0.0409, "step": 38350 }, { "epoch": 24.448693435309114, "grad_norm": 1.8308894634246826, "learning_rate": 9.835161863693398e-05, "loss": 0.0257, "step": 38360 }, { "epoch": 24.455066921606118, "grad_norm": 0.37085020542144775, "learning_rate": 9.83502148037527e-05, "loss": 0.0212, "step": 38370 }, { "epoch": 24.461440407903122, "grad_norm": 0.5847166180610657, "learning_rate": 9.834881038307068e-05, "loss": 0.0398, "step": 38380 }, { "epoch": 24.467813894200127, "grad_norm": 0.5794952511787415, "learning_rate": 9.834740537490495e-05, "loss": 0.0263, "step": 38390 }, { "epoch": 24.47418738049713, "grad_norm": 0.8091599345207214, "learning_rate": 9.83459997792726e-05, "loss": 0.0293, "step": 38400 }, { "epoch": 24.480560866794136, "grad_norm": 0.3204199969768524, "learning_rate": 9.834459359619071e-05, "loss": 0.0371, "step": 38410 }, { "epoch": 24.48693435309114, "grad_norm": 0.6389208436012268, "learning_rate": 9.834318682567635e-05, "loss": 0.0348, "step": 38420 }, { "epoch": 24.493307839388144, "grad_norm": 0.6860884428024292, "learning_rate": 9.834177946774665e-05, "loss": 0.0375, "step": 38430 }, { "epoch": 24.49968132568515, "grad_norm": 0.7466098666191101, "learning_rate": 9.834037152241868e-05, "loss": 0.0319, "step": 38440 }, { "epoch": 24.506054811982153, "grad_norm": 0.6468111276626587, "learning_rate": 9.833896298970955e-05, "loss": 0.0385, "step": 38450 }, { "epoch": 24.512428298279158, "grad_norm": 0.8163811564445496, "learning_rate": 9.833755386963637e-05, "loss": 0.0466, "step": 38460 }, { "epoch": 24.518801784576162, "grad_norm": 0.37384459376335144, "learning_rate": 9.833614416221628e-05, "loss": 0.0332, "step": 38470 }, { "epoch": 24.525175270873167, "grad_norm": 0.5225015878677368, "learning_rate": 9.833473386746641e-05, "loss": 0.037, "step": 38480 }, { "epoch": 24.53154875717017, "grad_norm": 0.872281551361084, "learning_rate": 9.833332298540388e-05, "loss": 0.041, "step": 38490 }, { "epoch": 24.537922243467175, "grad_norm": 0.6988385319709778, "learning_rate": 9.833191151604583e-05, "loss": 0.0462, "step": 38500 }, { "epoch": 24.54429572976418, "grad_norm": 0.42815154790878296, "learning_rate": 9.833049945940945e-05, "loss": 0.0348, "step": 38510 }, { "epoch": 24.550669216061184, "grad_norm": 0.7160035371780396, "learning_rate": 9.832908681551185e-05, "loss": 0.0348, "step": 38520 }, { "epoch": 24.55704270235819, "grad_norm": 0.44611823558807373, "learning_rate": 9.832767358437023e-05, "loss": 0.0396, "step": 38530 }, { "epoch": 24.563416188655193, "grad_norm": 0.5921107530593872, "learning_rate": 9.832625976600173e-05, "loss": 0.0339, "step": 38540 }, { "epoch": 24.569789674952197, "grad_norm": 0.354353129863739, "learning_rate": 9.832484536042355e-05, "loss": 0.0434, "step": 38550 }, { "epoch": 24.5761631612492, "grad_norm": 0.3932753801345825, "learning_rate": 9.832343036765288e-05, "loss": 0.0279, "step": 38560 }, { "epoch": 24.582536647546206, "grad_norm": 0.5793641209602356, "learning_rate": 9.83220147877069e-05, "loss": 0.0477, "step": 38570 }, { "epoch": 24.58891013384321, "grad_norm": 0.5384972095489502, "learning_rate": 9.832059862060282e-05, "loss": 0.0362, "step": 38580 }, { "epoch": 24.595283620140215, "grad_norm": 0.5239253044128418, "learning_rate": 9.831918186635785e-05, "loss": 0.0356, "step": 38590 }, { "epoch": 24.601657106437223, "grad_norm": 0.46745747327804565, "learning_rate": 9.831776452498919e-05, "loss": 0.0338, "step": 38600 }, { "epoch": 24.608030592734227, "grad_norm": 0.3033801317214966, "learning_rate": 9.831634659651407e-05, "loss": 0.0375, "step": 38610 }, { "epoch": 24.61440407903123, "grad_norm": 0.6810902953147888, "learning_rate": 9.831492808094973e-05, "loss": 0.0413, "step": 38620 }, { "epoch": 24.620777565328236, "grad_norm": 0.6167108416557312, "learning_rate": 9.83135089783134e-05, "loss": 0.0382, "step": 38630 }, { "epoch": 24.62715105162524, "grad_norm": 0.7084755301475525, "learning_rate": 9.83120892886223e-05, "loss": 0.0391, "step": 38640 }, { "epoch": 24.633524537922245, "grad_norm": 0.5848824977874756, "learning_rate": 9.831066901189372e-05, "loss": 0.0352, "step": 38650 }, { "epoch": 24.63989802421925, "grad_norm": 0.4761250615119934, "learning_rate": 9.830924814814488e-05, "loss": 0.0429, "step": 38660 }, { "epoch": 24.646271510516254, "grad_norm": 0.5063850283622742, "learning_rate": 9.830782669739308e-05, "loss": 0.0299, "step": 38670 }, { "epoch": 24.652644996813258, "grad_norm": 0.708981454372406, "learning_rate": 9.830640465965557e-05, "loss": 0.0325, "step": 38680 }, { "epoch": 24.659018483110263, "grad_norm": 0.4199800193309784, "learning_rate": 9.830498203494964e-05, "loss": 0.0359, "step": 38690 }, { "epoch": 24.665391969407267, "grad_norm": 0.6549193859100342, "learning_rate": 9.830355882329257e-05, "loss": 0.0389, "step": 38700 }, { "epoch": 24.67176545570427, "grad_norm": 0.9048917889595032, "learning_rate": 9.830213502470165e-05, "loss": 0.0456, "step": 38710 }, { "epoch": 24.678138942001276, "grad_norm": 0.47741028666496277, "learning_rate": 9.830071063919418e-05, "loss": 0.0266, "step": 38720 }, { "epoch": 24.68451242829828, "grad_norm": 0.4264077842235565, "learning_rate": 9.829928566678748e-05, "loss": 0.0396, "step": 38730 }, { "epoch": 24.690885914595285, "grad_norm": 0.32953059673309326, "learning_rate": 9.829786010749885e-05, "loss": 0.0326, "step": 38740 }, { "epoch": 24.69725940089229, "grad_norm": 0.3859044015407562, "learning_rate": 9.829643396134563e-05, "loss": 0.0247, "step": 38750 }, { "epoch": 24.703632887189293, "grad_norm": 0.816207766532898, "learning_rate": 9.829500722834513e-05, "loss": 0.0321, "step": 38760 }, { "epoch": 24.710006373486298, "grad_norm": 0.4959228038787842, "learning_rate": 9.829357990851468e-05, "loss": 0.0326, "step": 38770 }, { "epoch": 24.716379859783302, "grad_norm": 0.7457894086837769, "learning_rate": 9.829215200187167e-05, "loss": 0.0385, "step": 38780 }, { "epoch": 24.722753346080307, "grad_norm": 0.27765101194381714, "learning_rate": 9.82907235084334e-05, "loss": 0.0373, "step": 38790 }, { "epoch": 24.72912683237731, "grad_norm": 0.41058456897735596, "learning_rate": 9.828929442821725e-05, "loss": 0.0304, "step": 38800 }, { "epoch": 24.735500318674315, "grad_norm": 0.688689649105072, "learning_rate": 9.828786476124056e-05, "loss": 0.0329, "step": 38810 }, { "epoch": 24.74187380497132, "grad_norm": 0.5703933238983154, "learning_rate": 9.828643450752076e-05, "loss": 0.0408, "step": 38820 }, { "epoch": 24.748247291268324, "grad_norm": 0.48635345697402954, "learning_rate": 9.828500366707516e-05, "loss": 0.0401, "step": 38830 }, { "epoch": 24.75462077756533, "grad_norm": 0.6730559468269348, "learning_rate": 9.828357223992119e-05, "loss": 0.0396, "step": 38840 }, { "epoch": 24.760994263862333, "grad_norm": 0.4870309829711914, "learning_rate": 9.828214022607623e-05, "loss": 0.0399, "step": 38850 }, { "epoch": 24.767367750159337, "grad_norm": 1.1711093187332153, "learning_rate": 9.828070762555767e-05, "loss": 0.0524, "step": 38860 }, { "epoch": 24.773741236456342, "grad_norm": 0.3556297719478607, "learning_rate": 9.827927443838293e-05, "loss": 0.0215, "step": 38870 }, { "epoch": 24.780114722753346, "grad_norm": 0.40977638959884644, "learning_rate": 9.827784066456943e-05, "loss": 0.0302, "step": 38880 }, { "epoch": 24.78648820905035, "grad_norm": 0.42178672552108765, "learning_rate": 9.827640630413459e-05, "loss": 0.0387, "step": 38890 }, { "epoch": 24.792861695347355, "grad_norm": 0.4901495575904846, "learning_rate": 9.82749713570958e-05, "loss": 0.032, "step": 38900 }, { "epoch": 24.79923518164436, "grad_norm": 0.45226696133613586, "learning_rate": 9.827353582347057e-05, "loss": 0.0326, "step": 38910 }, { "epoch": 24.805608667941364, "grad_norm": 0.43765273690223694, "learning_rate": 9.827209970327629e-05, "loss": 0.0379, "step": 38920 }, { "epoch": 24.81198215423837, "grad_norm": 0.43268218636512756, "learning_rate": 9.82706629965304e-05, "loss": 0.0361, "step": 38930 }, { "epoch": 24.818355640535373, "grad_norm": 0.5586946606636047, "learning_rate": 9.82692257032504e-05, "loss": 0.0289, "step": 38940 }, { "epoch": 24.824729126832377, "grad_norm": 0.8792363405227661, "learning_rate": 9.826778782345374e-05, "loss": 0.0433, "step": 38950 }, { "epoch": 24.83110261312938, "grad_norm": 0.4022250175476074, "learning_rate": 9.826634935715788e-05, "loss": 0.0266, "step": 38960 }, { "epoch": 24.837476099426386, "grad_norm": 0.4029271900653839, "learning_rate": 9.82649103043803e-05, "loss": 0.0241, "step": 38970 }, { "epoch": 24.84384958572339, "grad_norm": 0.547328531742096, "learning_rate": 9.826347066513847e-05, "loss": 0.0329, "step": 38980 }, { "epoch": 24.850223072020395, "grad_norm": 0.480762243270874, "learning_rate": 9.826203043944994e-05, "loss": 0.0415, "step": 38990 }, { "epoch": 24.8565965583174, "grad_norm": 0.5345247387886047, "learning_rate": 9.826058962733216e-05, "loss": 0.036, "step": 39000 }, { "epoch": 24.862970044614404, "grad_norm": 0.6169059872627258, "learning_rate": 9.825914822880264e-05, "loss": 0.025, "step": 39010 }, { "epoch": 24.869343530911408, "grad_norm": 0.2947370111942291, "learning_rate": 9.825770624387891e-05, "loss": 0.0311, "step": 39020 }, { "epoch": 24.875717017208412, "grad_norm": 0.5213439464569092, "learning_rate": 9.82562636725785e-05, "loss": 0.0341, "step": 39030 }, { "epoch": 24.882090503505417, "grad_norm": 0.3553881049156189, "learning_rate": 9.82548205149189e-05, "loss": 0.0281, "step": 39040 }, { "epoch": 24.88846398980242, "grad_norm": 0.480663537979126, "learning_rate": 9.825337677091768e-05, "loss": 0.0327, "step": 39050 }, { "epoch": 24.894837476099426, "grad_norm": 0.7088951468467712, "learning_rate": 9.825193244059238e-05, "loss": 0.0461, "step": 39060 }, { "epoch": 24.90121096239643, "grad_norm": 0.5639773011207581, "learning_rate": 9.825048752396055e-05, "loss": 0.0312, "step": 39070 }, { "epoch": 24.907584448693434, "grad_norm": 0.6111724376678467, "learning_rate": 9.824904202103972e-05, "loss": 0.0296, "step": 39080 }, { "epoch": 24.91395793499044, "grad_norm": 0.876187801361084, "learning_rate": 9.824759593184749e-05, "loss": 0.029, "step": 39090 }, { "epoch": 24.920331421287443, "grad_norm": 0.4184933304786682, "learning_rate": 9.824614925640141e-05, "loss": 0.0339, "step": 39100 }, { "epoch": 24.926704907584448, "grad_norm": 0.3652113974094391, "learning_rate": 9.824470199471907e-05, "loss": 0.0328, "step": 39110 }, { "epoch": 24.933078393881452, "grad_norm": 0.3299172818660736, "learning_rate": 9.824325414681805e-05, "loss": 0.0352, "step": 39120 }, { "epoch": 24.939451880178456, "grad_norm": 0.608484148979187, "learning_rate": 9.824180571271594e-05, "loss": 0.0234, "step": 39130 }, { "epoch": 24.94582536647546, "grad_norm": 0.40828120708465576, "learning_rate": 9.824035669243035e-05, "loss": 0.03, "step": 39140 }, { "epoch": 24.952198852772465, "grad_norm": 0.6209337711334229, "learning_rate": 9.823890708597887e-05, "loss": 0.0285, "step": 39150 }, { "epoch": 24.95857233906947, "grad_norm": 0.6148239374160767, "learning_rate": 9.823745689337912e-05, "loss": 0.0516, "step": 39160 }, { "epoch": 24.964945825366474, "grad_norm": 0.5113382935523987, "learning_rate": 9.823600611464874e-05, "loss": 0.0316, "step": 39170 }, { "epoch": 24.97131931166348, "grad_norm": 0.8183391690254211, "learning_rate": 9.823455474980532e-05, "loss": 0.0265, "step": 39180 }, { "epoch": 24.977692797960483, "grad_norm": 0.43828994035720825, "learning_rate": 9.823310279886654e-05, "loss": 0.0259, "step": 39190 }, { "epoch": 24.984066284257487, "grad_norm": 0.49602243304252625, "learning_rate": 9.823165026185001e-05, "loss": 0.0326, "step": 39200 }, { "epoch": 24.99043977055449, "grad_norm": 0.5951496362686157, "learning_rate": 9.823019713877339e-05, "loss": 0.0243, "step": 39210 }, { "epoch": 24.996813256851496, "grad_norm": 0.6928075551986694, "learning_rate": 9.822874342965435e-05, "loss": 0.0402, "step": 39220 }, { "epoch": 25.003186743148504, "grad_norm": 0.623993456363678, "learning_rate": 9.822728913451053e-05, "loss": 0.032, "step": 39230 }, { "epoch": 25.00956022944551, "grad_norm": 0.5616344809532166, "learning_rate": 9.822583425335963e-05, "loss": 0.0401, "step": 39240 }, { "epoch": 25.015933715742513, "grad_norm": 0.45921823382377625, "learning_rate": 9.822437878621931e-05, "loss": 0.0328, "step": 39250 }, { "epoch": 25.022307202039517, "grad_norm": 0.7087714076042175, "learning_rate": 9.822292273310723e-05, "loss": 0.0261, "step": 39260 }, { "epoch": 25.02868068833652, "grad_norm": 0.5531269907951355, "learning_rate": 9.822146609404112e-05, "loss": 0.0353, "step": 39270 }, { "epoch": 25.035054174633526, "grad_norm": 0.7592794895172119, "learning_rate": 9.822000886903869e-05, "loss": 0.0349, "step": 39280 }, { "epoch": 25.04142766093053, "grad_norm": 0.38011661171913147, "learning_rate": 9.821855105811761e-05, "loss": 0.0387, "step": 39290 }, { "epoch": 25.047801147227535, "grad_norm": 0.6682592034339905, "learning_rate": 9.82170926612956e-05, "loss": 0.0573, "step": 39300 }, { "epoch": 25.05417463352454, "grad_norm": 0.6371383666992188, "learning_rate": 9.82156336785904e-05, "loss": 0.037, "step": 39310 }, { "epoch": 25.060548119821544, "grad_norm": 0.3597332835197449, "learning_rate": 9.821417411001971e-05, "loss": 0.0342, "step": 39320 }, { "epoch": 25.066921606118548, "grad_norm": 0.43058523535728455, "learning_rate": 9.82127139556013e-05, "loss": 0.0307, "step": 39330 }, { "epoch": 25.073295092415552, "grad_norm": 0.5836226344108582, "learning_rate": 9.82112532153529e-05, "loss": 0.0313, "step": 39340 }, { "epoch": 25.079668578712557, "grad_norm": 0.6565682888031006, "learning_rate": 9.820979188929225e-05, "loss": 0.0359, "step": 39350 }, { "epoch": 25.08604206500956, "grad_norm": 0.6128043532371521, "learning_rate": 9.82083299774371e-05, "loss": 0.0452, "step": 39360 }, { "epoch": 25.092415551306566, "grad_norm": 0.30971190333366394, "learning_rate": 9.820686747980525e-05, "loss": 0.0295, "step": 39370 }, { "epoch": 25.09878903760357, "grad_norm": 0.4215708374977112, "learning_rate": 9.820540439641441e-05, "loss": 0.0372, "step": 39380 }, { "epoch": 25.105162523900574, "grad_norm": 0.3842742145061493, "learning_rate": 9.820394072728243e-05, "loss": 0.0351, "step": 39390 }, { "epoch": 25.11153601019758, "grad_norm": 0.49108433723449707, "learning_rate": 9.820247647242703e-05, "loss": 0.0406, "step": 39400 }, { "epoch": 25.117909496494583, "grad_norm": 0.3654775023460388, "learning_rate": 9.820101163186603e-05, "loss": 0.0495, "step": 39410 }, { "epoch": 25.124282982791588, "grad_norm": 0.36959317326545715, "learning_rate": 9.819954620561725e-05, "loss": 0.0426, "step": 39420 }, { "epoch": 25.130656469088592, "grad_norm": 0.6964523792266846, "learning_rate": 9.819808019369847e-05, "loss": 0.0324, "step": 39430 }, { "epoch": 25.137029955385596, "grad_norm": 1.0358121395111084, "learning_rate": 9.81966135961275e-05, "loss": 0.0367, "step": 39440 }, { "epoch": 25.1434034416826, "grad_norm": 0.42308324575424194, "learning_rate": 9.819514641292217e-05, "loss": 0.0322, "step": 39450 }, { "epoch": 25.149776927979605, "grad_norm": 0.45675790309906006, "learning_rate": 9.819367864410032e-05, "loss": 0.0303, "step": 39460 }, { "epoch": 25.15615041427661, "grad_norm": 0.3328995406627655, "learning_rate": 9.819221028967975e-05, "loss": 0.0298, "step": 39470 }, { "epoch": 25.162523900573614, "grad_norm": 0.327166348695755, "learning_rate": 9.819074134967834e-05, "loss": 0.0279, "step": 39480 }, { "epoch": 25.16889738687062, "grad_norm": 0.8217535614967346, "learning_rate": 9.818927182411392e-05, "loss": 0.0379, "step": 39490 }, { "epoch": 25.175270873167623, "grad_norm": 0.4297102093696594, "learning_rate": 9.818780171300435e-05, "loss": 0.0411, "step": 39500 }, { "epoch": 25.181644359464627, "grad_norm": 0.34240713715553284, "learning_rate": 9.81863310163675e-05, "loss": 0.026, "step": 39510 }, { "epoch": 25.18801784576163, "grad_norm": 0.8828849196434021, "learning_rate": 9.818485973422121e-05, "loss": 0.0266, "step": 39520 }, { "epoch": 25.194391332058636, "grad_norm": 1.1191885471343994, "learning_rate": 9.818338786658339e-05, "loss": 0.0377, "step": 39530 }, { "epoch": 25.20076481835564, "grad_norm": 0.38518843054771423, "learning_rate": 9.818191541347193e-05, "loss": 0.0287, "step": 39540 }, { "epoch": 25.207138304652645, "grad_norm": 0.5294539928436279, "learning_rate": 9.818044237490469e-05, "loss": 0.0395, "step": 39550 }, { "epoch": 25.21351179094965, "grad_norm": 0.6108579039573669, "learning_rate": 9.817896875089959e-05, "loss": 0.0331, "step": 39560 }, { "epoch": 25.219885277246654, "grad_norm": 0.43169453740119934, "learning_rate": 9.817749454147454e-05, "loss": 0.0444, "step": 39570 }, { "epoch": 25.226258763543658, "grad_norm": 0.40731561183929443, "learning_rate": 9.817601974664744e-05, "loss": 0.0221, "step": 39580 }, { "epoch": 25.232632249840663, "grad_norm": 0.6495663523674011, "learning_rate": 9.81745443664362e-05, "loss": 0.042, "step": 39590 }, { "epoch": 25.239005736137667, "grad_norm": 0.7028456926345825, "learning_rate": 9.817306840085877e-05, "loss": 0.0406, "step": 39600 }, { "epoch": 25.24537922243467, "grad_norm": 0.2752927839756012, "learning_rate": 9.817159184993308e-05, "loss": 0.0267, "step": 39610 }, { "epoch": 25.251752708731676, "grad_norm": 0.4462147653102875, "learning_rate": 9.817011471367706e-05, "loss": 0.0324, "step": 39620 }, { "epoch": 25.25812619502868, "grad_norm": 0.46049419045448303, "learning_rate": 9.816863699210868e-05, "loss": 0.0281, "step": 39630 }, { "epoch": 25.264499681325685, "grad_norm": 0.3987809717655182, "learning_rate": 9.816715868524586e-05, "loss": 0.0369, "step": 39640 }, { "epoch": 25.27087316762269, "grad_norm": 2.0677452087402344, "learning_rate": 9.81656797931066e-05, "loss": 0.0446, "step": 39650 }, { "epoch": 25.277246653919693, "grad_norm": 0.4305398762226105, "learning_rate": 9.816420031570885e-05, "loss": 0.0489, "step": 39660 }, { "epoch": 25.283620140216698, "grad_norm": 0.5744912028312683, "learning_rate": 9.81627202530706e-05, "loss": 0.0243, "step": 39670 }, { "epoch": 25.289993626513702, "grad_norm": 0.6020693778991699, "learning_rate": 9.816123960520982e-05, "loss": 0.031, "step": 39680 }, { "epoch": 25.296367112810707, "grad_norm": 0.6658399701118469, "learning_rate": 9.815975837214451e-05, "loss": 0.0369, "step": 39690 }, { "epoch": 25.30274059910771, "grad_norm": 0.2149060070514679, "learning_rate": 9.815827655389267e-05, "loss": 0.0299, "step": 39700 }, { "epoch": 25.309114085404715, "grad_norm": 0.45360514521598816, "learning_rate": 9.815679415047228e-05, "loss": 0.0292, "step": 39710 }, { "epoch": 25.31548757170172, "grad_norm": 0.49588531255722046, "learning_rate": 9.815531116190139e-05, "loss": 0.0241, "step": 39720 }, { "epoch": 25.321861057998724, "grad_norm": 0.42829108238220215, "learning_rate": 9.8153827588198e-05, "loss": 0.0271, "step": 39730 }, { "epoch": 25.32823454429573, "grad_norm": 0.4518713653087616, "learning_rate": 9.815234342938014e-05, "loss": 0.055, "step": 39740 }, { "epoch": 25.334608030592733, "grad_norm": 0.8778572678565979, "learning_rate": 9.815085868546585e-05, "loss": 0.0406, "step": 39750 }, { "epoch": 25.340981516889737, "grad_norm": 0.6271068453788757, "learning_rate": 9.814937335647316e-05, "loss": 0.0308, "step": 39760 }, { "epoch": 25.347355003186742, "grad_norm": 0.393430233001709, "learning_rate": 9.814788744242012e-05, "loss": 0.0387, "step": 39770 }, { "epoch": 25.353728489483746, "grad_norm": 0.3414284884929657, "learning_rate": 9.814640094332478e-05, "loss": 0.038, "step": 39780 }, { "epoch": 25.36010197578075, "grad_norm": 0.29574844241142273, "learning_rate": 9.814491385920523e-05, "loss": 0.0365, "step": 39790 }, { "epoch": 25.366475462077755, "grad_norm": 0.3667440414428711, "learning_rate": 9.814342619007951e-05, "loss": 0.0315, "step": 39800 }, { "epoch": 25.37284894837476, "grad_norm": 0.742486834526062, "learning_rate": 9.81419379359657e-05, "loss": 0.0559, "step": 39810 }, { "epoch": 25.379222434671764, "grad_norm": 0.4952855706214905, "learning_rate": 9.814044909688191e-05, "loss": 0.0247, "step": 39820 }, { "epoch": 25.38559592096877, "grad_norm": 0.9497376680374146, "learning_rate": 9.813895967284621e-05, "loss": 0.0397, "step": 39830 }, { "epoch": 25.391969407265773, "grad_norm": 0.46765750646591187, "learning_rate": 9.813746966387669e-05, "loss": 0.0268, "step": 39840 }, { "epoch": 25.398342893562777, "grad_norm": 0.9075254797935486, "learning_rate": 9.813597906999146e-05, "loss": 0.0375, "step": 39850 }, { "epoch": 25.404716379859785, "grad_norm": 0.5852899551391602, "learning_rate": 9.813448789120864e-05, "loss": 0.0295, "step": 39860 }, { "epoch": 25.41108986615679, "grad_norm": 0.5586073994636536, "learning_rate": 9.813299612754634e-05, "loss": 0.0352, "step": 39870 }, { "epoch": 25.417463352453794, "grad_norm": 0.4361903965473175, "learning_rate": 9.81315037790227e-05, "loss": 0.0384, "step": 39880 }, { "epoch": 25.4238368387508, "grad_norm": 0.3150855600833893, "learning_rate": 9.813001084565585e-05, "loss": 0.0374, "step": 39890 }, { "epoch": 25.430210325047803, "grad_norm": 0.41138219833374023, "learning_rate": 9.812851732746391e-05, "loss": 0.0354, "step": 39900 }, { "epoch": 25.436583811344807, "grad_norm": 0.4973923861980438, "learning_rate": 9.812702322446505e-05, "loss": 0.0276, "step": 39910 }, { "epoch": 25.44295729764181, "grad_norm": 0.6138496398925781, "learning_rate": 9.812552853667742e-05, "loss": 0.042, "step": 39920 }, { "epoch": 25.449330783938816, "grad_norm": 0.5973079204559326, "learning_rate": 9.812403326411919e-05, "loss": 0.0339, "step": 39930 }, { "epoch": 25.45570427023582, "grad_norm": 0.43267983198165894, "learning_rate": 9.812253740680851e-05, "loss": 0.0351, "step": 39940 }, { "epoch": 25.462077756532825, "grad_norm": 0.4871577322483063, "learning_rate": 9.812104096476358e-05, "loss": 0.0325, "step": 39950 }, { "epoch": 25.46845124282983, "grad_norm": 0.5238043665885925, "learning_rate": 9.811954393800255e-05, "loss": 0.0271, "step": 39960 }, { "epoch": 25.474824729126833, "grad_norm": 0.9439226388931274, "learning_rate": 9.811804632654362e-05, "loss": 0.0315, "step": 39970 }, { "epoch": 25.481198215423838, "grad_norm": 0.22500838339328766, "learning_rate": 9.811654813040502e-05, "loss": 0.0346, "step": 39980 }, { "epoch": 25.487571701720842, "grad_norm": 0.5708286166191101, "learning_rate": 9.811504934960492e-05, "loss": 0.0473, "step": 39990 }, { "epoch": 25.493945188017847, "grad_norm": 0.7597689032554626, "learning_rate": 9.811354998416154e-05, "loss": 0.0318, "step": 40000 }, { "epoch": 25.50031867431485, "grad_norm": 0.5784670114517212, "learning_rate": 9.81120500340931e-05, "loss": 0.0394, "step": 40010 }, { "epoch": 25.506692160611856, "grad_norm": 0.5145370364189148, "learning_rate": 9.811054949941783e-05, "loss": 0.0412, "step": 40020 }, { "epoch": 25.51306564690886, "grad_norm": 0.281352698802948, "learning_rate": 9.810904838015397e-05, "loss": 0.0318, "step": 40030 }, { "epoch": 25.519439133205864, "grad_norm": 0.6966325044631958, "learning_rate": 9.810754667631973e-05, "loss": 0.0465, "step": 40040 }, { "epoch": 25.52581261950287, "grad_norm": 0.5496698617935181, "learning_rate": 9.810604438793339e-05, "loss": 0.0335, "step": 40050 }, { "epoch": 25.532186105799873, "grad_norm": 0.6323529481887817, "learning_rate": 9.810454151501319e-05, "loss": 0.0314, "step": 40060 }, { "epoch": 25.538559592096878, "grad_norm": 0.5735466480255127, "learning_rate": 9.810303805757739e-05, "loss": 0.0302, "step": 40070 }, { "epoch": 25.544933078393882, "grad_norm": 0.7447389364242554, "learning_rate": 9.810153401564426e-05, "loss": 0.0485, "step": 40080 }, { "epoch": 25.551306564690886, "grad_norm": 1.0027998685836792, "learning_rate": 9.810002938923208e-05, "loss": 0.0255, "step": 40090 }, { "epoch": 25.55768005098789, "grad_norm": 0.48476648330688477, "learning_rate": 9.809852417835912e-05, "loss": 0.0409, "step": 40100 }, { "epoch": 25.564053537284895, "grad_norm": 0.8056094646453857, "learning_rate": 9.809701838304368e-05, "loss": 0.0342, "step": 40110 }, { "epoch": 25.5704270235819, "grad_norm": 0.28897541761398315, "learning_rate": 9.809551200330407e-05, "loss": 0.0254, "step": 40120 }, { "epoch": 25.576800509878904, "grad_norm": 0.5787471532821655, "learning_rate": 9.809400503915857e-05, "loss": 0.028, "step": 40130 }, { "epoch": 25.58317399617591, "grad_norm": 0.304202139377594, "learning_rate": 9.80924974906255e-05, "loss": 0.0339, "step": 40140 }, { "epoch": 25.589547482472913, "grad_norm": 0.907636284828186, "learning_rate": 9.809098935772318e-05, "loss": 0.0487, "step": 40150 }, { "epoch": 25.595920968769917, "grad_norm": 0.7451756000518799, "learning_rate": 9.808948064046993e-05, "loss": 0.0377, "step": 40160 }, { "epoch": 25.60229445506692, "grad_norm": 1.0392460823059082, "learning_rate": 9.80879713388841e-05, "loss": 0.045, "step": 40170 }, { "epoch": 25.608667941363926, "grad_norm": 0.9809366464614868, "learning_rate": 9.808646145298399e-05, "loss": 0.0348, "step": 40180 }, { "epoch": 25.61504142766093, "grad_norm": 0.7227786183357239, "learning_rate": 9.808495098278799e-05, "loss": 0.0391, "step": 40190 }, { "epoch": 25.621414913957935, "grad_norm": 0.33519279956817627, "learning_rate": 9.808343992831443e-05, "loss": 0.034, "step": 40200 }, { "epoch": 25.62778840025494, "grad_norm": 0.4900830388069153, "learning_rate": 9.808192828958169e-05, "loss": 0.0302, "step": 40210 }, { "epoch": 25.634161886551944, "grad_norm": 0.7125679850578308, "learning_rate": 9.808041606660811e-05, "loss": 0.0401, "step": 40220 }, { "epoch": 25.640535372848948, "grad_norm": 0.7391586899757385, "learning_rate": 9.80789032594121e-05, "loss": 0.0428, "step": 40230 }, { "epoch": 25.646908859145952, "grad_norm": 0.5739883184432983, "learning_rate": 9.807738986801198e-05, "loss": 0.0356, "step": 40240 }, { "epoch": 25.653282345442957, "grad_norm": 0.6630788445472717, "learning_rate": 9.807587589242621e-05, "loss": 0.0325, "step": 40250 }, { "epoch": 25.65965583173996, "grad_norm": 0.3680783808231354, "learning_rate": 9.807436133267316e-05, "loss": 0.0279, "step": 40260 }, { "epoch": 25.666029318036966, "grad_norm": 0.44966721534729004, "learning_rate": 9.807284618877121e-05, "loss": 0.0395, "step": 40270 }, { "epoch": 25.67240280433397, "grad_norm": 0.6302251815795898, "learning_rate": 9.807133046073881e-05, "loss": 0.0403, "step": 40280 }, { "epoch": 25.678776290630974, "grad_norm": 0.29586124420166016, "learning_rate": 9.806981414859435e-05, "loss": 0.0262, "step": 40290 }, { "epoch": 25.68514977692798, "grad_norm": 0.3809783160686493, "learning_rate": 9.806829725235627e-05, "loss": 0.0357, "step": 40300 }, { "epoch": 25.691523263224983, "grad_norm": 0.746821939945221, "learning_rate": 9.806677977204298e-05, "loss": 0.028, "step": 40310 }, { "epoch": 25.697896749521988, "grad_norm": 0.7282025814056396, "learning_rate": 9.806526170767294e-05, "loss": 0.0438, "step": 40320 }, { "epoch": 25.704270235818992, "grad_norm": 0.31228116154670715, "learning_rate": 9.80637430592646e-05, "loss": 0.0218, "step": 40330 }, { "epoch": 25.710643722115996, "grad_norm": 0.46417829394340515, "learning_rate": 9.806222382683639e-05, "loss": 0.0295, "step": 40340 }, { "epoch": 25.717017208413, "grad_norm": 0.6913856863975525, "learning_rate": 9.806070401040678e-05, "loss": 0.0374, "step": 40350 }, { "epoch": 25.723390694710005, "grad_norm": 0.4596313238143921, "learning_rate": 9.805918360999424e-05, "loss": 0.0257, "step": 40360 }, { "epoch": 25.72976418100701, "grad_norm": 0.45668748021125793, "learning_rate": 9.805766262561726e-05, "loss": 0.0306, "step": 40370 }, { "epoch": 25.736137667304014, "grad_norm": 0.48825353384017944, "learning_rate": 9.805614105729429e-05, "loss": 0.0334, "step": 40380 }, { "epoch": 25.74251115360102, "grad_norm": 0.8489535450935364, "learning_rate": 9.805461890504384e-05, "loss": 0.0371, "step": 40390 }, { "epoch": 25.748884639898023, "grad_norm": 0.6208127737045288, "learning_rate": 9.805309616888438e-05, "loss": 0.0256, "step": 40400 }, { "epoch": 25.755258126195027, "grad_norm": 0.41087955236434937, "learning_rate": 9.805157284883446e-05, "loss": 0.0368, "step": 40410 }, { "epoch": 25.76163161249203, "grad_norm": 0.42125561833381653, "learning_rate": 9.805004894491255e-05, "loss": 0.0269, "step": 40420 }, { "epoch": 25.768005098789036, "grad_norm": 0.3624279797077179, "learning_rate": 9.804852445713716e-05, "loss": 0.0455, "step": 40430 }, { "epoch": 25.77437858508604, "grad_norm": 0.3836914002895355, "learning_rate": 9.804699938552687e-05, "loss": 0.0386, "step": 40440 }, { "epoch": 25.780752071383045, "grad_norm": 0.3899528682231903, "learning_rate": 9.804547373010015e-05, "loss": 0.0281, "step": 40450 }, { "epoch": 25.78712555768005, "grad_norm": 0.3634029030799866, "learning_rate": 9.804394749087555e-05, "loss": 0.032, "step": 40460 }, { "epoch": 25.793499043977054, "grad_norm": 0.6043806076049805, "learning_rate": 9.804242066787167e-05, "loss": 0.0365, "step": 40470 }, { "epoch": 25.799872530274058, "grad_norm": 0.3717891275882721, "learning_rate": 9.804089326110697e-05, "loss": 0.0289, "step": 40480 }, { "epoch": 25.806246016571066, "grad_norm": 0.5614397525787354, "learning_rate": 9.803936527060009e-05, "loss": 0.0442, "step": 40490 }, { "epoch": 25.81261950286807, "grad_norm": 0.46124744415283203, "learning_rate": 9.803783669636955e-05, "loss": 0.0256, "step": 40500 }, { "epoch": 25.818992989165075, "grad_norm": 0.5526819229125977, "learning_rate": 9.803630753843395e-05, "loss": 0.0291, "step": 40510 }, { "epoch": 25.82536647546208, "grad_norm": 0.50849848985672, "learning_rate": 9.803477779681185e-05, "loss": 0.0357, "step": 40520 }, { "epoch": 25.831739961759084, "grad_norm": 0.7440577745437622, "learning_rate": 9.803324747152185e-05, "loss": 0.0316, "step": 40530 }, { "epoch": 25.838113448056088, "grad_norm": 0.5509979724884033, "learning_rate": 9.803171656258254e-05, "loss": 0.0315, "step": 40540 }, { "epoch": 25.844486934353093, "grad_norm": 0.5559027194976807, "learning_rate": 9.803018507001253e-05, "loss": 0.0323, "step": 40550 }, { "epoch": 25.850860420650097, "grad_norm": 0.5763866901397705, "learning_rate": 9.802865299383043e-05, "loss": 0.0373, "step": 40560 }, { "epoch": 25.8572339069471, "grad_norm": 0.4962155222892761, "learning_rate": 9.802712033405483e-05, "loss": 0.0365, "step": 40570 }, { "epoch": 25.863607393244106, "grad_norm": 0.44195908308029175, "learning_rate": 9.802558709070439e-05, "loss": 0.0275, "step": 40580 }, { "epoch": 25.86998087954111, "grad_norm": 0.5724771022796631, "learning_rate": 9.802405326379772e-05, "loss": 0.0333, "step": 40590 }, { "epoch": 25.876354365838115, "grad_norm": 0.42344096302986145, "learning_rate": 9.802251885335345e-05, "loss": 0.0264, "step": 40600 }, { "epoch": 25.88272785213512, "grad_norm": 0.5523172616958618, "learning_rate": 9.802098385939025e-05, "loss": 0.0309, "step": 40610 }, { "epoch": 25.889101338432123, "grad_norm": 0.43028560280799866, "learning_rate": 9.801944828192675e-05, "loss": 0.0317, "step": 40620 }, { "epoch": 25.895474824729128, "grad_norm": 0.4124147891998291, "learning_rate": 9.801791212098162e-05, "loss": 0.0296, "step": 40630 }, { "epoch": 25.901848311026132, "grad_norm": 0.41405194997787476, "learning_rate": 9.801637537657351e-05, "loss": 0.0284, "step": 40640 }, { "epoch": 25.908221797323137, "grad_norm": 0.5136837959289551, "learning_rate": 9.801483804872112e-05, "loss": 0.0324, "step": 40650 }, { "epoch": 25.91459528362014, "grad_norm": 0.30281969904899597, "learning_rate": 9.80133001374431e-05, "loss": 0.0262, "step": 40660 }, { "epoch": 25.920968769917145, "grad_norm": 0.4209826588630676, "learning_rate": 9.801176164275816e-05, "loss": 0.0487, "step": 40670 }, { "epoch": 25.92734225621415, "grad_norm": 0.3522534668445587, "learning_rate": 9.801022256468498e-05, "loss": 0.0382, "step": 40680 }, { "epoch": 25.933715742511154, "grad_norm": 0.6347562074661255, "learning_rate": 9.800868290324227e-05, "loss": 0.0238, "step": 40690 }, { "epoch": 25.94008922880816, "grad_norm": 0.3325772285461426, "learning_rate": 9.800714265844872e-05, "loss": 0.0377, "step": 40700 }, { "epoch": 25.946462715105163, "grad_norm": 0.4902306795120239, "learning_rate": 9.800560183032308e-05, "loss": 0.0263, "step": 40710 }, { "epoch": 25.952836201402167, "grad_norm": 0.2938218116760254, "learning_rate": 9.800406041888405e-05, "loss": 0.0252, "step": 40720 }, { "epoch": 25.959209687699172, "grad_norm": 0.5713808536529541, "learning_rate": 9.800251842415036e-05, "loss": 0.0402, "step": 40730 }, { "epoch": 25.965583173996176, "grad_norm": 0.5097112655639648, "learning_rate": 9.800097584614075e-05, "loss": 0.037, "step": 40740 }, { "epoch": 25.97195666029318, "grad_norm": 0.617712676525116, "learning_rate": 9.799943268487397e-05, "loss": 0.0323, "step": 40750 }, { "epoch": 25.978330146590185, "grad_norm": 0.5795984864234924, "learning_rate": 9.799788894036875e-05, "loss": 0.0393, "step": 40760 }, { "epoch": 25.98470363288719, "grad_norm": 0.5161428451538086, "learning_rate": 9.799634461264387e-05, "loss": 0.0252, "step": 40770 }, { "epoch": 25.991077119184194, "grad_norm": 0.35265636444091797, "learning_rate": 9.799479970171808e-05, "loss": 0.0313, "step": 40780 }, { "epoch": 25.9974506054812, "grad_norm": 0.3866861164569855, "learning_rate": 9.799325420761016e-05, "loss": 0.0318, "step": 40790 }, { "epoch": 26.003824091778203, "grad_norm": 0.8604021668434143, "learning_rate": 9.799170813033889e-05, "loss": 0.0272, "step": 40800 }, { "epoch": 26.010197578075207, "grad_norm": 0.47888457775115967, "learning_rate": 9.799016146992306e-05, "loss": 0.0258, "step": 40810 }, { "epoch": 26.01657106437221, "grad_norm": 0.4438333809375763, "learning_rate": 9.798861422638147e-05, "loss": 0.0368, "step": 40820 }, { "epoch": 26.022944550669216, "grad_norm": 0.80049067735672, "learning_rate": 9.798706639973288e-05, "loss": 0.0319, "step": 40830 }, { "epoch": 26.02931803696622, "grad_norm": 0.685234546661377, "learning_rate": 9.798551798999613e-05, "loss": 0.0245, "step": 40840 }, { "epoch": 26.035691523263225, "grad_norm": 0.6471700072288513, "learning_rate": 9.798396899719005e-05, "loss": 0.0376, "step": 40850 }, { "epoch": 26.04206500956023, "grad_norm": 0.5088632702827454, "learning_rate": 9.798241942133342e-05, "loss": 0.026, "step": 40860 }, { "epoch": 26.048438495857233, "grad_norm": 0.511092483997345, "learning_rate": 9.798086926244511e-05, "loss": 0.0274, "step": 40870 }, { "epoch": 26.054811982154238, "grad_norm": 0.56766676902771, "learning_rate": 9.797931852054393e-05, "loss": 0.0351, "step": 40880 }, { "epoch": 26.061185468451242, "grad_norm": 0.5463147759437561, "learning_rate": 9.797776719564873e-05, "loss": 0.035, "step": 40890 }, { "epoch": 26.067558954748247, "grad_norm": 0.7147695422172546, "learning_rate": 9.797621528777837e-05, "loss": 0.0385, "step": 40900 }, { "epoch": 26.07393244104525, "grad_norm": 0.5646417737007141, "learning_rate": 9.797466279695168e-05, "loss": 0.0325, "step": 40910 }, { "epoch": 26.080305927342256, "grad_norm": 0.3981863260269165, "learning_rate": 9.797310972318755e-05, "loss": 0.0364, "step": 40920 }, { "epoch": 26.08667941363926, "grad_norm": 0.7347411513328552, "learning_rate": 9.797155606650483e-05, "loss": 0.028, "step": 40930 }, { "epoch": 26.093052899936264, "grad_norm": 0.3876502215862274, "learning_rate": 9.797000182692243e-05, "loss": 0.0308, "step": 40940 }, { "epoch": 26.09942638623327, "grad_norm": 1.1200830936431885, "learning_rate": 9.796844700445921e-05, "loss": 0.0278, "step": 40950 }, { "epoch": 26.105799872530273, "grad_norm": 0.3290374279022217, "learning_rate": 9.796689159913406e-05, "loss": 0.0327, "step": 40960 }, { "epoch": 26.112173358827278, "grad_norm": 0.298495888710022, "learning_rate": 9.79653356109659e-05, "loss": 0.0255, "step": 40970 }, { "epoch": 26.118546845124282, "grad_norm": 0.7288366556167603, "learning_rate": 9.796377903997363e-05, "loss": 0.0415, "step": 40980 }, { "epoch": 26.124920331421286, "grad_norm": 0.4295289218425751, "learning_rate": 9.796222188617615e-05, "loss": 0.024, "step": 40990 }, { "epoch": 26.13129381771829, "grad_norm": 0.7116581797599792, "learning_rate": 9.796066414959238e-05, "loss": 0.0243, "step": 41000 }, { "epoch": 26.137667304015295, "grad_norm": 0.5832488536834717, "learning_rate": 9.795910583024129e-05, "loss": 0.0306, "step": 41010 }, { "epoch": 26.1440407903123, "grad_norm": 0.42752605676651, "learning_rate": 9.795754692814176e-05, "loss": 0.0246, "step": 41020 }, { "epoch": 26.150414276609304, "grad_norm": 0.7192235589027405, "learning_rate": 9.795598744331275e-05, "loss": 0.0288, "step": 41030 }, { "epoch": 26.15678776290631, "grad_norm": 0.7591318488121033, "learning_rate": 9.795442737577323e-05, "loss": 0.0432, "step": 41040 }, { "epoch": 26.163161249203313, "grad_norm": 0.5937816500663757, "learning_rate": 9.795286672554213e-05, "loss": 0.0392, "step": 41050 }, { "epoch": 26.169534735500317, "grad_norm": 0.4404557943344116, "learning_rate": 9.795130549263843e-05, "loss": 0.04, "step": 41060 }, { "epoch": 26.17590822179732, "grad_norm": 0.4103528559207916, "learning_rate": 9.794974367708109e-05, "loss": 0.0251, "step": 41070 }, { "epoch": 26.182281708094326, "grad_norm": 0.25227975845336914, "learning_rate": 9.79481812788891e-05, "loss": 0.0335, "step": 41080 }, { "epoch": 26.18865519439133, "grad_norm": 0.456488698720932, "learning_rate": 9.794661829808142e-05, "loss": 0.0239, "step": 41090 }, { "epoch": 26.195028680688335, "grad_norm": 0.48274660110473633, "learning_rate": 9.794505473467708e-05, "loss": 0.0544, "step": 41100 }, { "epoch": 26.20140216698534, "grad_norm": 0.6885577440261841, "learning_rate": 9.794349058869506e-05, "loss": 0.0334, "step": 41110 }, { "epoch": 26.207775653282347, "grad_norm": 0.45659247040748596, "learning_rate": 9.794192586015433e-05, "loss": 0.0383, "step": 41120 }, { "epoch": 26.21414913957935, "grad_norm": 0.672415554523468, "learning_rate": 9.794036054907396e-05, "loss": 0.0481, "step": 41130 }, { "epoch": 26.220522625876356, "grad_norm": 0.5611081123352051, "learning_rate": 9.793879465547296e-05, "loss": 0.0349, "step": 41140 }, { "epoch": 26.22689611217336, "grad_norm": 0.718242347240448, "learning_rate": 9.793722817937032e-05, "loss": 0.0342, "step": 41150 }, { "epoch": 26.233269598470365, "grad_norm": 0.41502007842063904, "learning_rate": 9.793566112078509e-05, "loss": 0.0227, "step": 41160 }, { "epoch": 26.23964308476737, "grad_norm": 0.8946523666381836, "learning_rate": 9.793409347973635e-05, "loss": 0.0387, "step": 41170 }, { "epoch": 26.246016571064374, "grad_norm": 0.9620791077613831, "learning_rate": 9.79325252562431e-05, "loss": 0.042, "step": 41180 }, { "epoch": 26.252390057361378, "grad_norm": 0.7109879851341248, "learning_rate": 9.793095645032441e-05, "loss": 0.0331, "step": 41190 }, { "epoch": 26.258763543658382, "grad_norm": 0.566011905670166, "learning_rate": 9.792938706199934e-05, "loss": 0.034, "step": 41200 }, { "epoch": 26.265137029955387, "grad_norm": 0.4160345792770386, "learning_rate": 9.792781709128698e-05, "loss": 0.023, "step": 41210 }, { "epoch": 26.27151051625239, "grad_norm": 0.2870670258998871, "learning_rate": 9.792624653820638e-05, "loss": 0.0265, "step": 41220 }, { "epoch": 26.277884002549396, "grad_norm": 0.6686944365501404, "learning_rate": 9.792467540277663e-05, "loss": 0.029, "step": 41230 }, { "epoch": 26.2842574888464, "grad_norm": 0.4624597132205963, "learning_rate": 9.792310368501683e-05, "loss": 0.0589, "step": 41240 }, { "epoch": 26.290630975143404, "grad_norm": 0.4988868534564972, "learning_rate": 9.792153138494608e-05, "loss": 0.026, "step": 41250 }, { "epoch": 26.29700446144041, "grad_norm": 0.6210262179374695, "learning_rate": 9.791995850258348e-05, "loss": 0.0369, "step": 41260 }, { "epoch": 26.303377947737413, "grad_norm": 0.3958175480365753, "learning_rate": 9.791838503794812e-05, "loss": 0.0404, "step": 41270 }, { "epoch": 26.309751434034418, "grad_norm": 0.4863801598548889, "learning_rate": 9.791681099105915e-05, "loss": 0.0412, "step": 41280 }, { "epoch": 26.316124920331422, "grad_norm": 0.3420000374317169, "learning_rate": 9.791523636193571e-05, "loss": 0.0239, "step": 41290 }, { "epoch": 26.322498406628426, "grad_norm": 0.4110923111438751, "learning_rate": 9.791366115059689e-05, "loss": 0.0226, "step": 41300 }, { "epoch": 26.32887189292543, "grad_norm": 0.576371967792511, "learning_rate": 9.791208535706184e-05, "loss": 0.0235, "step": 41310 }, { "epoch": 26.335245379222435, "grad_norm": 0.7259501218795776, "learning_rate": 9.791050898134972e-05, "loss": 0.0422, "step": 41320 }, { "epoch": 26.34161886551944, "grad_norm": 0.34121406078338623, "learning_rate": 9.790893202347969e-05, "loss": 0.0297, "step": 41330 }, { "epoch": 26.347992351816444, "grad_norm": 0.6072424054145813, "learning_rate": 9.79073544834709e-05, "loss": 0.0249, "step": 41340 }, { "epoch": 26.35436583811345, "grad_norm": 0.6023910641670227, "learning_rate": 9.79057763613425e-05, "loss": 0.0302, "step": 41350 }, { "epoch": 26.360739324410453, "grad_norm": 0.36167219281196594, "learning_rate": 9.790419765711372e-05, "loss": 0.0386, "step": 41360 }, { "epoch": 26.367112810707457, "grad_norm": 0.6131445169448853, "learning_rate": 9.790261837080368e-05, "loss": 0.0257, "step": 41370 }, { "epoch": 26.37348629700446, "grad_norm": 0.4489663243293762, "learning_rate": 9.790103850243161e-05, "loss": 0.0335, "step": 41380 }, { "epoch": 26.379859783301466, "grad_norm": 0.3466247618198395, "learning_rate": 9.789945805201672e-05, "loss": 0.0286, "step": 41390 }, { "epoch": 26.38623326959847, "grad_norm": 0.39985668659210205, "learning_rate": 9.789787701957815e-05, "loss": 0.0224, "step": 41400 }, { "epoch": 26.392606755895475, "grad_norm": 0.5911824107170105, "learning_rate": 9.789629540513518e-05, "loss": 0.0328, "step": 41410 }, { "epoch": 26.39898024219248, "grad_norm": 0.7254574298858643, "learning_rate": 9.789471320870699e-05, "loss": 0.0239, "step": 41420 }, { "epoch": 26.405353728489484, "grad_norm": 0.5336246490478516, "learning_rate": 9.789313043031281e-05, "loss": 0.029, "step": 41430 }, { "epoch": 26.411727214786488, "grad_norm": 0.4649886190891266, "learning_rate": 9.789154706997187e-05, "loss": 0.031, "step": 41440 }, { "epoch": 26.418100701083493, "grad_norm": 0.4161438047885895, "learning_rate": 9.788996312770344e-05, "loss": 0.0222, "step": 41450 }, { "epoch": 26.424474187380497, "grad_norm": 0.7702319622039795, "learning_rate": 9.788837860352673e-05, "loss": 0.0365, "step": 41460 }, { "epoch": 26.4308476736775, "grad_norm": 0.7023902535438538, "learning_rate": 9.788679349746101e-05, "loss": 0.0351, "step": 41470 }, { "epoch": 26.437221159974506, "grad_norm": 0.6471976637840271, "learning_rate": 9.788520780952553e-05, "loss": 0.0298, "step": 41480 }, { "epoch": 26.44359464627151, "grad_norm": 0.8786109089851379, "learning_rate": 9.788362153973955e-05, "loss": 0.0433, "step": 41490 }, { "epoch": 26.449968132568515, "grad_norm": 0.4884776175022125, "learning_rate": 9.788203468812238e-05, "loss": 0.0334, "step": 41500 }, { "epoch": 26.45634161886552, "grad_norm": 0.39006826281547546, "learning_rate": 9.788044725469327e-05, "loss": 0.0285, "step": 41510 }, { "epoch": 26.462715105162523, "grad_norm": 0.41547033190727234, "learning_rate": 9.787885923947153e-05, "loss": 0.0288, "step": 41520 }, { "epoch": 26.469088591459528, "grad_norm": 0.7170929908752441, "learning_rate": 9.787727064247643e-05, "loss": 0.0426, "step": 41530 }, { "epoch": 26.475462077756532, "grad_norm": 0.4527498185634613, "learning_rate": 9.787568146372729e-05, "loss": 0.0279, "step": 41540 }, { "epoch": 26.481835564053537, "grad_norm": 0.48785221576690674, "learning_rate": 9.787409170324345e-05, "loss": 0.0294, "step": 41550 }, { "epoch": 26.48820905035054, "grad_norm": 0.3759062886238098, "learning_rate": 9.787250136104415e-05, "loss": 0.0301, "step": 41560 }, { "epoch": 26.494582536647545, "grad_norm": 0.37527981400489807, "learning_rate": 9.787091043714879e-05, "loss": 0.0265, "step": 41570 }, { "epoch": 26.50095602294455, "grad_norm": 0.8074070811271667, "learning_rate": 9.786931893157666e-05, "loss": 0.0354, "step": 41580 }, { "epoch": 26.507329509241554, "grad_norm": 0.5236986875534058, "learning_rate": 9.786772684434711e-05, "loss": 0.0447, "step": 41590 }, { "epoch": 26.51370299553856, "grad_norm": 0.4914737641811371, "learning_rate": 9.786613417547947e-05, "loss": 0.034, "step": 41600 }, { "epoch": 26.520076481835563, "grad_norm": 0.3342273533344269, "learning_rate": 9.786454092499314e-05, "loss": 0.0297, "step": 41610 }, { "epoch": 26.526449968132567, "grad_norm": 0.5309076905250549, "learning_rate": 9.786294709290741e-05, "loss": 0.0402, "step": 41620 }, { "epoch": 26.532823454429572, "grad_norm": 0.5091167688369751, "learning_rate": 9.78613526792417e-05, "loss": 0.0336, "step": 41630 }, { "epoch": 26.539196940726576, "grad_norm": 0.4147992432117462, "learning_rate": 9.785975768401537e-05, "loss": 0.0366, "step": 41640 }, { "epoch": 26.54557042702358, "grad_norm": 0.4643586277961731, "learning_rate": 9.785816210724779e-05, "loss": 0.0232, "step": 41650 }, { "epoch": 26.551943913320585, "grad_norm": 0.35512903332710266, "learning_rate": 9.785656594895836e-05, "loss": 0.0326, "step": 41660 }, { "epoch": 26.55831739961759, "grad_norm": 0.502058744430542, "learning_rate": 9.785496920916646e-05, "loss": 0.0333, "step": 41670 }, { "epoch": 26.564690885914594, "grad_norm": 0.44213804602622986, "learning_rate": 9.785337188789151e-05, "loss": 0.0254, "step": 41680 }, { "epoch": 26.5710643722116, "grad_norm": 0.5241914987564087, "learning_rate": 9.78517739851529e-05, "loss": 0.0284, "step": 41690 }, { "epoch": 26.577437858508603, "grad_norm": 0.3741266429424286, "learning_rate": 9.785017550097006e-05, "loss": 0.0314, "step": 41700 }, { "epoch": 26.583811344805607, "grad_norm": 0.36622586846351624, "learning_rate": 9.784857643536243e-05, "loss": 0.0233, "step": 41710 }, { "epoch": 26.59018483110261, "grad_norm": 0.37584641575813293, "learning_rate": 9.78469767883494e-05, "loss": 0.0223, "step": 41720 }, { "epoch": 26.59655831739962, "grad_norm": 0.3701398968696594, "learning_rate": 9.784537655995043e-05, "loss": 0.0353, "step": 41730 }, { "epoch": 26.60293180369662, "grad_norm": 0.49583032727241516, "learning_rate": 9.784377575018496e-05, "loss": 0.0489, "step": 41740 }, { "epoch": 26.609305289993628, "grad_norm": 0.39954572916030884, "learning_rate": 9.784217435907244e-05, "loss": 0.0281, "step": 41750 }, { "epoch": 26.615678776290633, "grad_norm": 0.4347255229949951, "learning_rate": 9.784057238663234e-05, "loss": 0.0387, "step": 41760 }, { "epoch": 26.622052262587637, "grad_norm": 0.46338048577308655, "learning_rate": 9.783896983288412e-05, "loss": 0.0333, "step": 41770 }, { "epoch": 26.62842574888464, "grad_norm": 0.5483663082122803, "learning_rate": 9.783736669784725e-05, "loss": 0.0313, "step": 41780 }, { "epoch": 26.634799235181646, "grad_norm": 0.6989869475364685, "learning_rate": 9.783576298154119e-05, "loss": 0.0338, "step": 41790 }, { "epoch": 26.64117272147865, "grad_norm": 0.45994454622268677, "learning_rate": 9.783415868398545e-05, "loss": 0.0425, "step": 41800 }, { "epoch": 26.647546207775655, "grad_norm": 0.550617516040802, "learning_rate": 9.783255380519953e-05, "loss": 0.035, "step": 41810 }, { "epoch": 26.65391969407266, "grad_norm": 0.2228291928768158, "learning_rate": 9.783094834520291e-05, "loss": 0.0287, "step": 41820 }, { "epoch": 26.660293180369663, "grad_norm": 0.39090222120285034, "learning_rate": 9.782934230401512e-05, "loss": 0.0207, "step": 41830 }, { "epoch": 26.666666666666668, "grad_norm": 0.6888470649719238, "learning_rate": 9.782773568165565e-05, "loss": 0.0395, "step": 41840 }, { "epoch": 26.673040152963672, "grad_norm": 0.47411271929740906, "learning_rate": 9.782612847814404e-05, "loss": 0.032, "step": 41850 }, { "epoch": 26.679413639260677, "grad_norm": 0.5727260708808899, "learning_rate": 9.782452069349981e-05, "loss": 0.0322, "step": 41860 }, { "epoch": 26.68578712555768, "grad_norm": 0.7682517766952515, "learning_rate": 9.78229123277425e-05, "loss": 0.045, "step": 41870 }, { "epoch": 26.692160611854685, "grad_norm": 0.3485850989818573, "learning_rate": 9.782130338089165e-05, "loss": 0.0234, "step": 41880 }, { "epoch": 26.69853409815169, "grad_norm": 0.3398924469947815, "learning_rate": 9.781969385296683e-05, "loss": 0.023, "step": 41890 }, { "epoch": 26.704907584448694, "grad_norm": 0.4345161020755768, "learning_rate": 9.781808374398757e-05, "loss": 0.0243, "step": 41900 }, { "epoch": 26.7112810707457, "grad_norm": 0.36500152945518494, "learning_rate": 9.781647305397343e-05, "loss": 0.0356, "step": 41910 }, { "epoch": 26.717654557042703, "grad_norm": 0.4141809940338135, "learning_rate": 9.781486178294402e-05, "loss": 0.0366, "step": 41920 }, { "epoch": 26.724028043339707, "grad_norm": 0.42421048879623413, "learning_rate": 9.781324993091889e-05, "loss": 0.0219, "step": 41930 }, { "epoch": 26.730401529636712, "grad_norm": 0.28780397772789, "learning_rate": 9.781163749791762e-05, "loss": 0.0309, "step": 41940 }, { "epoch": 26.736775015933716, "grad_norm": 0.7716911435127258, "learning_rate": 9.781002448395982e-05, "loss": 0.0236, "step": 41950 }, { "epoch": 26.74314850223072, "grad_norm": 0.39955684542655945, "learning_rate": 9.780841088906508e-05, "loss": 0.029, "step": 41960 }, { "epoch": 26.749521988527725, "grad_norm": 0.32886803150177, "learning_rate": 9.7806796713253e-05, "loss": 0.0328, "step": 41970 }, { "epoch": 26.75589547482473, "grad_norm": 0.28118640184402466, "learning_rate": 9.780518195654322e-05, "loss": 0.0283, "step": 41980 }, { "epoch": 26.762268961121734, "grad_norm": 0.5915328860282898, "learning_rate": 9.780356661895534e-05, "loss": 0.0289, "step": 41990 }, { "epoch": 26.76864244741874, "grad_norm": 0.6661473512649536, "learning_rate": 9.780195070050898e-05, "loss": 0.0392, "step": 42000 }, { "epoch": 26.775015933715743, "grad_norm": 0.3593858480453491, "learning_rate": 9.78003342012238e-05, "loss": 0.0278, "step": 42010 }, { "epoch": 26.781389420012747, "grad_norm": 0.23564191162586212, "learning_rate": 9.779871712111942e-05, "loss": 0.0262, "step": 42020 }, { "epoch": 26.78776290630975, "grad_norm": 0.4596879482269287, "learning_rate": 9.77970994602155e-05, "loss": 0.033, "step": 42030 }, { "epoch": 26.794136392606756, "grad_norm": 0.4411656856536865, "learning_rate": 9.779548121853169e-05, "loss": 0.0438, "step": 42040 }, { "epoch": 26.80050987890376, "grad_norm": 0.3380880355834961, "learning_rate": 9.779386239608765e-05, "loss": 0.0357, "step": 42050 }, { "epoch": 26.806883365200765, "grad_norm": 0.5125289559364319, "learning_rate": 9.779224299290308e-05, "loss": 0.0506, "step": 42060 }, { "epoch": 26.81325685149777, "grad_norm": 0.46288588643074036, "learning_rate": 9.779062300899762e-05, "loss": 0.0321, "step": 42070 }, { "epoch": 26.819630337794774, "grad_norm": 0.6414213180541992, "learning_rate": 9.778900244439097e-05, "loss": 0.0287, "step": 42080 }, { "epoch": 26.826003824091778, "grad_norm": 0.679557740688324, "learning_rate": 9.778738129910281e-05, "loss": 0.0365, "step": 42090 }, { "epoch": 26.832377310388782, "grad_norm": 0.41695597767829895, "learning_rate": 9.778575957315287e-05, "loss": 0.0358, "step": 42100 }, { "epoch": 26.838750796685787, "grad_norm": 0.484231561422348, "learning_rate": 9.778413726656081e-05, "loss": 0.0435, "step": 42110 }, { "epoch": 26.84512428298279, "grad_norm": 0.31692075729370117, "learning_rate": 9.778251437934637e-05, "loss": 0.0278, "step": 42120 }, { "epoch": 26.851497769279796, "grad_norm": 0.6634456515312195, "learning_rate": 9.778089091152929e-05, "loss": 0.0365, "step": 42130 }, { "epoch": 26.8578712555768, "grad_norm": 0.48582810163497925, "learning_rate": 9.777926686312926e-05, "loss": 0.0286, "step": 42140 }, { "epoch": 26.864244741873804, "grad_norm": 0.6180738210678101, "learning_rate": 9.777764223416602e-05, "loss": 0.0249, "step": 42150 }, { "epoch": 26.87061822817081, "grad_norm": 0.6959978938102722, "learning_rate": 9.777601702465933e-05, "loss": 0.0301, "step": 42160 }, { "epoch": 26.876991714467813, "grad_norm": 0.5285813212394714, "learning_rate": 9.77743912346289e-05, "loss": 0.0289, "step": 42170 }, { "epoch": 26.883365200764818, "grad_norm": 0.5362122058868408, "learning_rate": 9.777276486409454e-05, "loss": 0.0257, "step": 42180 }, { "epoch": 26.889738687061822, "grad_norm": 0.39523330330848694, "learning_rate": 9.777113791307598e-05, "loss": 0.0301, "step": 42190 }, { "epoch": 26.896112173358826, "grad_norm": 0.6104212999343872, "learning_rate": 9.776951038159296e-05, "loss": 0.0375, "step": 42200 }, { "epoch": 26.90248565965583, "grad_norm": 0.5991176962852478, "learning_rate": 9.776788226966531e-05, "loss": 0.0317, "step": 42210 }, { "epoch": 26.908859145952835, "grad_norm": 0.5169550776481628, "learning_rate": 9.776625357731279e-05, "loss": 0.0343, "step": 42220 }, { "epoch": 26.91523263224984, "grad_norm": 0.6992713809013367, "learning_rate": 9.776462430455519e-05, "loss": 0.0322, "step": 42230 }, { "epoch": 26.921606118546844, "grad_norm": 0.44037771224975586, "learning_rate": 9.776299445141231e-05, "loss": 0.0311, "step": 42240 }, { "epoch": 26.92797960484385, "grad_norm": 0.6163039803504944, "learning_rate": 9.776136401790393e-05, "loss": 0.0473, "step": 42250 }, { "epoch": 26.934353091140853, "grad_norm": 0.5443461537361145, "learning_rate": 9.77597330040499e-05, "loss": 0.0253, "step": 42260 }, { "epoch": 26.940726577437857, "grad_norm": 0.5473769307136536, "learning_rate": 9.775810140987003e-05, "loss": 0.0355, "step": 42270 }, { "epoch": 26.94710006373486, "grad_norm": 0.4330480694770813, "learning_rate": 9.775646923538411e-05, "loss": 0.0319, "step": 42280 }, { "epoch": 26.953473550031866, "grad_norm": 0.36358991265296936, "learning_rate": 9.775483648061203e-05, "loss": 0.0371, "step": 42290 }, { "epoch": 26.95984703632887, "grad_norm": 0.6562947034835815, "learning_rate": 9.775320314557358e-05, "loss": 0.0349, "step": 42300 }, { "epoch": 26.966220522625875, "grad_norm": 0.4184749126434326, "learning_rate": 9.775156923028863e-05, "loss": 0.0232, "step": 42310 }, { "epoch": 26.97259400892288, "grad_norm": 0.3798682689666748, "learning_rate": 9.774993473477702e-05, "loss": 0.0298, "step": 42320 }, { "epoch": 26.978967495219884, "grad_norm": 0.7274406552314758, "learning_rate": 9.774829965905863e-05, "loss": 0.0337, "step": 42330 }, { "epoch": 26.985340981516888, "grad_norm": 0.5030524134635925, "learning_rate": 9.774666400315333e-05, "loss": 0.0316, "step": 42340 }, { "epoch": 26.991714467813892, "grad_norm": 0.4737909734249115, "learning_rate": 9.774502776708096e-05, "loss": 0.035, "step": 42350 }, { "epoch": 26.9980879541109, "grad_norm": 0.4415510892868042, "learning_rate": 9.774339095086144e-05, "loss": 0.0345, "step": 42360 }, { "epoch": 27.004461440407905, "grad_norm": 0.45854583382606506, "learning_rate": 9.774175355451463e-05, "loss": 0.0359, "step": 42370 }, { "epoch": 27.01083492670491, "grad_norm": 0.41550180315971375, "learning_rate": 9.774011557806044e-05, "loss": 0.0421, "step": 42380 }, { "epoch": 27.017208413001914, "grad_norm": 0.5666798949241638, "learning_rate": 9.773847702151877e-05, "loss": 0.0312, "step": 42390 }, { "epoch": 27.023581899298918, "grad_norm": 0.8952369689941406, "learning_rate": 9.773683788490955e-05, "loss": 0.0481, "step": 42400 }, { "epoch": 27.029955385595922, "grad_norm": 0.6514202952384949, "learning_rate": 9.773519816825265e-05, "loss": 0.035, "step": 42410 }, { "epoch": 27.036328871892927, "grad_norm": 0.519882082939148, "learning_rate": 9.773355787156804e-05, "loss": 0.0298, "step": 42420 }, { "epoch": 27.04270235818993, "grad_norm": 0.6521331071853638, "learning_rate": 9.773191699487564e-05, "loss": 0.0332, "step": 42430 }, { "epoch": 27.049075844486936, "grad_norm": 0.5467392802238464, "learning_rate": 9.773027553819537e-05, "loss": 0.0415, "step": 42440 }, { "epoch": 27.05544933078394, "grad_norm": 0.6759063005447388, "learning_rate": 9.77286335015472e-05, "loss": 0.0392, "step": 42450 }, { "epoch": 27.061822817080945, "grad_norm": 0.6634697914123535, "learning_rate": 9.772699088495106e-05, "loss": 0.0392, "step": 42460 }, { "epoch": 27.06819630337795, "grad_norm": 0.42603620886802673, "learning_rate": 9.772534768842692e-05, "loss": 0.0357, "step": 42470 }, { "epoch": 27.074569789674953, "grad_norm": 0.5501759648323059, "learning_rate": 9.772370391199473e-05, "loss": 0.0385, "step": 42480 }, { "epoch": 27.080943275971958, "grad_norm": 0.5600732564926147, "learning_rate": 9.77220595556745e-05, "loss": 0.0242, "step": 42490 }, { "epoch": 27.087316762268962, "grad_norm": 0.6579861640930176, "learning_rate": 9.772041461948618e-05, "loss": 0.0388, "step": 42500 }, { "epoch": 27.093690248565967, "grad_norm": 0.7068892121315002, "learning_rate": 9.771876910344977e-05, "loss": 0.0403, "step": 42510 }, { "epoch": 27.10006373486297, "grad_norm": 0.45586320757865906, "learning_rate": 9.771712300758526e-05, "loss": 0.0371, "step": 42520 }, { "epoch": 27.106437221159975, "grad_norm": 0.4014410376548767, "learning_rate": 9.771547633191266e-05, "loss": 0.0326, "step": 42530 }, { "epoch": 27.11281070745698, "grad_norm": 0.6864639520645142, "learning_rate": 9.771382907645197e-05, "loss": 0.0313, "step": 42540 }, { "epoch": 27.119184193753984, "grad_norm": 0.4354401230812073, "learning_rate": 9.77121812412232e-05, "loss": 0.032, "step": 42550 }, { "epoch": 27.12555768005099, "grad_norm": 0.4531353712081909, "learning_rate": 9.771053282624638e-05, "loss": 0.0267, "step": 42560 }, { "epoch": 27.131931166347993, "grad_norm": 0.2705150842666626, "learning_rate": 9.770888383154156e-05, "loss": 0.028, "step": 42570 }, { "epoch": 27.138304652644997, "grad_norm": 0.5971609950065613, "learning_rate": 9.770723425712873e-05, "loss": 0.0263, "step": 42580 }, { "epoch": 27.144678138942, "grad_norm": 0.6712452173233032, "learning_rate": 9.770558410302796e-05, "loss": 0.0265, "step": 42590 }, { "epoch": 27.151051625239006, "grad_norm": 0.40640750527381897, "learning_rate": 9.770393336925932e-05, "loss": 0.03, "step": 42600 }, { "epoch": 27.15742511153601, "grad_norm": 0.3879620432853699, "learning_rate": 9.770228205584284e-05, "loss": 0.0247, "step": 42610 }, { "epoch": 27.163798597833015, "grad_norm": 0.4280916154384613, "learning_rate": 9.77006301627986e-05, "loss": 0.0296, "step": 42620 }, { "epoch": 27.17017208413002, "grad_norm": 0.42630475759506226, "learning_rate": 9.769897769014667e-05, "loss": 0.039, "step": 42630 }, { "epoch": 27.176545570427024, "grad_norm": 0.3668208420276642, "learning_rate": 9.76973246379071e-05, "loss": 0.0248, "step": 42640 }, { "epoch": 27.182919056724028, "grad_norm": 0.4459036886692047, "learning_rate": 9.769567100610003e-05, "loss": 0.0256, "step": 42650 }, { "epoch": 27.189292543021033, "grad_norm": 0.4663746953010559, "learning_rate": 9.769401679474553e-05, "loss": 0.0404, "step": 42660 }, { "epoch": 27.195666029318037, "grad_norm": 0.7490636110305786, "learning_rate": 9.769236200386367e-05, "loss": 0.0247, "step": 42670 }, { "epoch": 27.20203951561504, "grad_norm": 0.3479316830635071, "learning_rate": 9.769070663347459e-05, "loss": 0.0201, "step": 42680 }, { "epoch": 27.208413001912046, "grad_norm": 0.2896709740161896, "learning_rate": 9.768905068359841e-05, "loss": 0.0267, "step": 42690 }, { "epoch": 27.21478648820905, "grad_norm": 0.7208948731422424, "learning_rate": 9.768739415425523e-05, "loss": 0.0451, "step": 42700 }, { "epoch": 27.221159974506055, "grad_norm": 0.38189494609832764, "learning_rate": 9.768573704546517e-05, "loss": 0.0242, "step": 42710 }, { "epoch": 27.22753346080306, "grad_norm": 0.48520195484161377, "learning_rate": 9.768407935724842e-05, "loss": 0.0238, "step": 42720 }, { "epoch": 27.233906947100063, "grad_norm": 0.5111380219459534, "learning_rate": 9.768242108962506e-05, "loss": 0.0421, "step": 42730 }, { "epoch": 27.240280433397068, "grad_norm": 0.8313961625099182, "learning_rate": 9.768076224261526e-05, "loss": 0.021, "step": 42740 }, { "epoch": 27.246653919694072, "grad_norm": 0.2813982367515564, "learning_rate": 9.767910281623919e-05, "loss": 0.0286, "step": 42750 }, { "epoch": 27.253027405991077, "grad_norm": 0.6580824255943298, "learning_rate": 9.767744281051701e-05, "loss": 0.0331, "step": 42760 }, { "epoch": 27.25940089228808, "grad_norm": 0.4643847644329071, "learning_rate": 9.767578222546888e-05, "loss": 0.0296, "step": 42770 }, { "epoch": 27.265774378585085, "grad_norm": 0.6691859364509583, "learning_rate": 9.767412106111496e-05, "loss": 0.0227, "step": 42780 }, { "epoch": 27.27214786488209, "grad_norm": 0.5022182464599609, "learning_rate": 9.767245931747549e-05, "loss": 0.0388, "step": 42790 }, { "epoch": 27.278521351179094, "grad_norm": 0.4062817394733429, "learning_rate": 9.76707969945706e-05, "loss": 0.0288, "step": 42800 }, { "epoch": 27.2848948374761, "grad_norm": 0.7121409177780151, "learning_rate": 9.766913409242053e-05, "loss": 0.0284, "step": 42810 }, { "epoch": 27.291268323773103, "grad_norm": 0.4174536168575287, "learning_rate": 9.766747061104548e-05, "loss": 0.0274, "step": 42820 }, { "epoch": 27.297641810070107, "grad_norm": 0.6136816740036011, "learning_rate": 9.766580655046566e-05, "loss": 0.0253, "step": 42830 }, { "epoch": 27.304015296367112, "grad_norm": 0.6404725313186646, "learning_rate": 9.766414191070128e-05, "loss": 0.0315, "step": 42840 }, { "epoch": 27.310388782664116, "grad_norm": 0.374038428068161, "learning_rate": 9.766247669177258e-05, "loss": 0.0536, "step": 42850 }, { "epoch": 27.31676226896112, "grad_norm": 0.6502777338027954, "learning_rate": 9.766081089369979e-05, "loss": 0.0274, "step": 42860 }, { "epoch": 27.323135755258125, "grad_norm": 0.43991732597351074, "learning_rate": 9.765914451650313e-05, "loss": 0.0299, "step": 42870 }, { "epoch": 27.32950924155513, "grad_norm": 0.38164645433425903, "learning_rate": 9.765747756020287e-05, "loss": 0.025, "step": 42880 }, { "epoch": 27.335882727852134, "grad_norm": 0.325210303068161, "learning_rate": 9.765581002481927e-05, "loss": 0.0293, "step": 42890 }, { "epoch": 27.34225621414914, "grad_norm": 0.5042225122451782, "learning_rate": 9.765414191037257e-05, "loss": 0.0389, "step": 42900 }, { "epoch": 27.348629700446143, "grad_norm": 0.4349748492240906, "learning_rate": 9.765247321688307e-05, "loss": 0.0282, "step": 42910 }, { "epoch": 27.355003186743147, "grad_norm": 0.5858556032180786, "learning_rate": 9.765080394437102e-05, "loss": 0.024, "step": 42920 }, { "epoch": 27.36137667304015, "grad_norm": 0.3326181173324585, "learning_rate": 9.764913409285672e-05, "loss": 0.028, "step": 42930 }, { "epoch": 27.367750159337156, "grad_norm": 0.34276527166366577, "learning_rate": 9.764746366236044e-05, "loss": 0.0356, "step": 42940 }, { "epoch": 27.37412364563416, "grad_norm": 0.5838547348976135, "learning_rate": 9.764579265290251e-05, "loss": 0.032, "step": 42950 }, { "epoch": 27.380497131931165, "grad_norm": 0.3988277316093445, "learning_rate": 9.76441210645032e-05, "loss": 0.0234, "step": 42960 }, { "epoch": 27.38687061822817, "grad_norm": 0.41003066301345825, "learning_rate": 9.764244889718284e-05, "loss": 0.0402, "step": 42970 }, { "epoch": 27.393244104525174, "grad_norm": 0.5355158448219299, "learning_rate": 9.764077615096173e-05, "loss": 0.0333, "step": 42980 }, { "epoch": 27.39961759082218, "grad_norm": 0.352802574634552, "learning_rate": 9.763910282586022e-05, "loss": 0.0259, "step": 42990 }, { "epoch": 27.405991077119186, "grad_norm": 0.40623608231544495, "learning_rate": 9.763742892189865e-05, "loss": 0.0331, "step": 43000 }, { "epoch": 27.41236456341619, "grad_norm": 0.37196511030197144, "learning_rate": 9.763575443909732e-05, "loss": 0.029, "step": 43010 }, { "epoch": 27.418738049713195, "grad_norm": 0.5398632287979126, "learning_rate": 9.763407937747659e-05, "loss": 0.0258, "step": 43020 }, { "epoch": 27.4251115360102, "grad_norm": 0.7087506651878357, "learning_rate": 9.763240373705684e-05, "loss": 0.0293, "step": 43030 }, { "epoch": 27.431485022307204, "grad_norm": 0.4795816242694855, "learning_rate": 9.76307275178584e-05, "loss": 0.0391, "step": 43040 }, { "epoch": 27.437858508604208, "grad_norm": 0.4953254461288452, "learning_rate": 9.762905071990165e-05, "loss": 0.0365, "step": 43050 }, { "epoch": 27.444231994901212, "grad_norm": 0.6698494553565979, "learning_rate": 9.762737334320697e-05, "loss": 0.0265, "step": 43060 }, { "epoch": 27.450605481198217, "grad_norm": 0.5150022506713867, "learning_rate": 9.762569538779474e-05, "loss": 0.0361, "step": 43070 }, { "epoch": 27.45697896749522, "grad_norm": 0.9207808971405029, "learning_rate": 9.762401685368534e-05, "loss": 0.0355, "step": 43080 }, { "epoch": 27.463352453792226, "grad_norm": 0.7661358714103699, "learning_rate": 9.762233774089917e-05, "loss": 0.0334, "step": 43090 }, { "epoch": 27.46972594008923, "grad_norm": 0.41314250230789185, "learning_rate": 9.762065804945663e-05, "loss": 0.0302, "step": 43100 }, { "epoch": 27.476099426386234, "grad_norm": 0.48619672656059265, "learning_rate": 9.761897777937813e-05, "loss": 0.0344, "step": 43110 }, { "epoch": 27.48247291268324, "grad_norm": 0.3556061387062073, "learning_rate": 9.761729693068411e-05, "loss": 0.0354, "step": 43120 }, { "epoch": 27.488846398980243, "grad_norm": 0.473741352558136, "learning_rate": 9.761561550339496e-05, "loss": 0.025, "step": 43130 }, { "epoch": 27.495219885277248, "grad_norm": 0.6408193111419678, "learning_rate": 9.761393349753114e-05, "loss": 0.034, "step": 43140 }, { "epoch": 27.501593371574252, "grad_norm": 0.5366620421409607, "learning_rate": 9.761225091311305e-05, "loss": 0.0306, "step": 43150 }, { "epoch": 27.507966857871256, "grad_norm": 0.554911732673645, "learning_rate": 9.761056775016117e-05, "loss": 0.0236, "step": 43160 }, { "epoch": 27.51434034416826, "grad_norm": 0.45023760199546814, "learning_rate": 9.760888400869595e-05, "loss": 0.0496, "step": 43170 }, { "epoch": 27.520713830465265, "grad_norm": 0.4123894274234772, "learning_rate": 9.760719968873781e-05, "loss": 0.0267, "step": 43180 }, { "epoch": 27.52708731676227, "grad_norm": 0.5496741533279419, "learning_rate": 9.760551479030728e-05, "loss": 0.043, "step": 43190 }, { "epoch": 27.533460803059274, "grad_norm": 0.5389478206634521, "learning_rate": 9.760382931342477e-05, "loss": 0.0255, "step": 43200 }, { "epoch": 27.53983428935628, "grad_norm": 0.5101574659347534, "learning_rate": 9.76021432581108e-05, "loss": 0.0399, "step": 43210 }, { "epoch": 27.546207775653283, "grad_norm": 0.42027056217193604, "learning_rate": 9.760045662438585e-05, "loss": 0.0299, "step": 43220 }, { "epoch": 27.552581261950287, "grad_norm": 0.4567740857601166, "learning_rate": 9.75987694122704e-05, "loss": 0.0237, "step": 43230 }, { "epoch": 27.55895474824729, "grad_norm": 0.5957986116409302, "learning_rate": 9.759708162178497e-05, "loss": 0.0333, "step": 43240 }, { "epoch": 27.565328234544296, "grad_norm": 0.582748532295227, "learning_rate": 9.759539325295005e-05, "loss": 0.0278, "step": 43250 }, { "epoch": 27.5717017208413, "grad_norm": 0.31856614351272583, "learning_rate": 9.759370430578618e-05, "loss": 0.0337, "step": 43260 }, { "epoch": 27.578075207138305, "grad_norm": 0.29704105854034424, "learning_rate": 9.759201478031385e-05, "loss": 0.0371, "step": 43270 }, { "epoch": 27.58444869343531, "grad_norm": 0.23343893885612488, "learning_rate": 9.75903246765536e-05, "loss": 0.0242, "step": 43280 }, { "epoch": 27.590822179732314, "grad_norm": 0.5061424374580383, "learning_rate": 9.7588633994526e-05, "loss": 0.0306, "step": 43290 }, { "epoch": 27.597195666029318, "grad_norm": 0.37801432609558105, "learning_rate": 9.758694273425155e-05, "loss": 0.0361, "step": 43300 }, { "epoch": 27.603569152326322, "grad_norm": 0.5899465680122375, "learning_rate": 9.758525089575081e-05, "loss": 0.0355, "step": 43310 }, { "epoch": 27.609942638623327, "grad_norm": 0.6520706415176392, "learning_rate": 9.758355847904435e-05, "loss": 0.0388, "step": 43320 }, { "epoch": 27.61631612492033, "grad_norm": 0.5751564502716064, "learning_rate": 9.758186548415273e-05, "loss": 0.0336, "step": 43330 }, { "epoch": 27.622689611217336, "grad_norm": 0.6411465406417847, "learning_rate": 9.758017191109652e-05, "loss": 0.0339, "step": 43340 }, { "epoch": 27.62906309751434, "grad_norm": 0.5868284702301025, "learning_rate": 9.757847775989631e-05, "loss": 0.0414, "step": 43350 }, { "epoch": 27.635436583811344, "grad_norm": 0.8106486201286316, "learning_rate": 9.757678303057265e-05, "loss": 0.0347, "step": 43360 }, { "epoch": 27.64181007010835, "grad_norm": 0.5329729914665222, "learning_rate": 9.757508772314616e-05, "loss": 0.0365, "step": 43370 }, { "epoch": 27.648183556405353, "grad_norm": 0.6391049027442932, "learning_rate": 9.757339183763745e-05, "loss": 0.0381, "step": 43380 }, { "epoch": 27.654557042702358, "grad_norm": 0.35752376914024353, "learning_rate": 9.757169537406709e-05, "loss": 0.0385, "step": 43390 }, { "epoch": 27.660930528999362, "grad_norm": 0.36805009841918945, "learning_rate": 9.756999833245575e-05, "loss": 0.0276, "step": 43400 }, { "epoch": 27.667304015296367, "grad_norm": 0.4045107066631317, "learning_rate": 9.756830071282398e-05, "loss": 0.024, "step": 43410 }, { "epoch": 27.67367750159337, "grad_norm": 0.3612765669822693, "learning_rate": 9.756660251519248e-05, "loss": 0.028, "step": 43420 }, { "epoch": 27.680050987890375, "grad_norm": 0.5063760280609131, "learning_rate": 9.756490373958183e-05, "loss": 0.0253, "step": 43430 }, { "epoch": 27.68642447418738, "grad_norm": 0.3555130958557129, "learning_rate": 9.756320438601269e-05, "loss": 0.0362, "step": 43440 }, { "epoch": 27.692797960484384, "grad_norm": 0.541277289390564, "learning_rate": 9.75615044545057e-05, "loss": 0.0323, "step": 43450 }, { "epoch": 27.69917144678139, "grad_norm": 0.5074331164360046, "learning_rate": 9.755980394508155e-05, "loss": 0.0296, "step": 43460 }, { "epoch": 27.705544933078393, "grad_norm": 0.6467015743255615, "learning_rate": 9.755810285776086e-05, "loss": 0.0322, "step": 43470 }, { "epoch": 27.711918419375397, "grad_norm": 0.5056699514389038, "learning_rate": 9.755640119256432e-05, "loss": 0.0319, "step": 43480 }, { "epoch": 27.7182919056724, "grad_norm": 0.35040026903152466, "learning_rate": 9.755469894951261e-05, "loss": 0.0204, "step": 43490 }, { "epoch": 27.724665391969406, "grad_norm": 0.33973732590675354, "learning_rate": 9.75529961286264e-05, "loss": 0.0355, "step": 43500 }, { "epoch": 27.73103887826641, "grad_norm": 0.46022918820381165, "learning_rate": 9.75512927299264e-05, "loss": 0.0298, "step": 43510 }, { "epoch": 27.737412364563415, "grad_norm": 0.45593592524528503, "learning_rate": 9.754958875343329e-05, "loss": 0.0269, "step": 43520 }, { "epoch": 27.74378585086042, "grad_norm": 0.6009355783462524, "learning_rate": 9.754788419916779e-05, "loss": 0.0271, "step": 43530 }, { "epoch": 27.750159337157424, "grad_norm": 0.5592274069786072, "learning_rate": 9.75461790671506e-05, "loss": 0.0333, "step": 43540 }, { "epoch": 27.756532823454428, "grad_norm": 0.7361756563186646, "learning_rate": 9.754447335740244e-05, "loss": 0.0306, "step": 43550 }, { "epoch": 27.762906309751433, "grad_norm": 0.5715871453285217, "learning_rate": 9.754276706994406e-05, "loss": 0.0195, "step": 43560 }, { "epoch": 27.769279796048437, "grad_norm": 0.7657013535499573, "learning_rate": 9.754106020479614e-05, "loss": 0.0361, "step": 43570 }, { "epoch": 27.77565328234544, "grad_norm": 0.5092758536338806, "learning_rate": 9.753935276197947e-05, "loss": 0.0384, "step": 43580 }, { "epoch": 27.782026768642446, "grad_norm": 0.35640034079551697, "learning_rate": 9.753764474151477e-05, "loss": 0.0388, "step": 43590 }, { "epoch": 27.78840025493945, "grad_norm": 0.23954685032367706, "learning_rate": 9.753593614342282e-05, "loss": 0.019, "step": 43600 }, { "epoch": 27.794773741236455, "grad_norm": 0.43784767389297485, "learning_rate": 9.753422696772435e-05, "loss": 0.0316, "step": 43610 }, { "epoch": 27.801147227533463, "grad_norm": 0.6333953142166138, "learning_rate": 9.753251721444014e-05, "loss": 0.0309, "step": 43620 }, { "epoch": 27.807520713830467, "grad_norm": 0.31105631589889526, "learning_rate": 9.753080688359099e-05, "loss": 0.0314, "step": 43630 }, { "epoch": 27.81389420012747, "grad_norm": 0.41747790575027466, "learning_rate": 9.752909597519763e-05, "loss": 0.0334, "step": 43640 }, { "epoch": 27.820267686424476, "grad_norm": 0.7120935916900635, "learning_rate": 9.75273844892809e-05, "loss": 0.0288, "step": 43650 }, { "epoch": 27.82664117272148, "grad_norm": 0.40811508893966675, "learning_rate": 9.752567242586157e-05, "loss": 0.0252, "step": 43660 }, { "epoch": 27.833014659018485, "grad_norm": 0.35748323798179626, "learning_rate": 9.752395978496045e-05, "loss": 0.0206, "step": 43670 }, { "epoch": 27.83938814531549, "grad_norm": 0.5098660588264465, "learning_rate": 9.752224656659833e-05, "loss": 0.0281, "step": 43680 }, { "epoch": 27.845761631612493, "grad_norm": 0.5578731894493103, "learning_rate": 9.752053277079607e-05, "loss": 0.0193, "step": 43690 }, { "epoch": 27.852135117909498, "grad_norm": 0.7307285666465759, "learning_rate": 9.751881839757446e-05, "loss": 0.0432, "step": 43700 }, { "epoch": 27.858508604206502, "grad_norm": 0.811240017414093, "learning_rate": 9.751710344695435e-05, "loss": 0.0405, "step": 43710 }, { "epoch": 27.864882090503507, "grad_norm": 0.5275760889053345, "learning_rate": 9.751538791895656e-05, "loss": 0.0339, "step": 43720 }, { "epoch": 27.87125557680051, "grad_norm": 0.7013908624649048, "learning_rate": 9.751367181360195e-05, "loss": 0.036, "step": 43730 }, { "epoch": 27.877629063097515, "grad_norm": 0.4571201801300049, "learning_rate": 9.751195513091136e-05, "loss": 0.0216, "step": 43740 }, { "epoch": 27.88400254939452, "grad_norm": 0.5197249054908752, "learning_rate": 9.751023787090568e-05, "loss": 0.035, "step": 43750 }, { "epoch": 27.890376035691524, "grad_norm": 0.4787781536579132, "learning_rate": 9.750852003360573e-05, "loss": 0.0374, "step": 43760 }, { "epoch": 27.89674952198853, "grad_norm": 0.7194153666496277, "learning_rate": 9.750680161903241e-05, "loss": 0.026, "step": 43770 }, { "epoch": 27.903123008285533, "grad_norm": 0.6144666075706482, "learning_rate": 9.750508262720659e-05, "loss": 0.0405, "step": 43780 }, { "epoch": 27.909496494582537, "grad_norm": 0.6022862195968628, "learning_rate": 9.750336305814917e-05, "loss": 0.0338, "step": 43790 }, { "epoch": 27.915869980879542, "grad_norm": 0.7967201471328735, "learning_rate": 9.750164291188104e-05, "loss": 0.0291, "step": 43800 }, { "epoch": 27.922243467176546, "grad_norm": 0.27110058069229126, "learning_rate": 9.74999221884231e-05, "loss": 0.0265, "step": 43810 }, { "epoch": 27.92861695347355, "grad_norm": 0.5907347798347473, "learning_rate": 9.749820088779625e-05, "loss": 0.0283, "step": 43820 }, { "epoch": 27.934990439770555, "grad_norm": 0.4597783088684082, "learning_rate": 9.749647901002142e-05, "loss": 0.0281, "step": 43830 }, { "epoch": 27.94136392606756, "grad_norm": 0.40914514660835266, "learning_rate": 9.749475655511952e-05, "loss": 0.0352, "step": 43840 }, { "epoch": 27.947737412364564, "grad_norm": 0.6103016138076782, "learning_rate": 9.74930335231115e-05, "loss": 0.0335, "step": 43850 }, { "epoch": 27.95411089866157, "grad_norm": 0.44958028197288513, "learning_rate": 9.749130991401827e-05, "loss": 0.0355, "step": 43860 }, { "epoch": 27.960484384958573, "grad_norm": 0.5218605995178223, "learning_rate": 9.748958572786079e-05, "loss": 0.0301, "step": 43870 }, { "epoch": 27.966857871255577, "grad_norm": 0.3141953945159912, "learning_rate": 9.748786096466001e-05, "loss": 0.0285, "step": 43880 }, { "epoch": 27.97323135755258, "grad_norm": 0.5665042400360107, "learning_rate": 9.748613562443687e-05, "loss": 0.0263, "step": 43890 }, { "epoch": 27.979604843849586, "grad_norm": 0.343961626291275, "learning_rate": 9.748440970721236e-05, "loss": 0.0278, "step": 43900 }, { "epoch": 27.98597833014659, "grad_norm": 0.5238397121429443, "learning_rate": 9.748268321300742e-05, "loss": 0.0389, "step": 43910 }, { "epoch": 27.992351816443595, "grad_norm": 0.6671867370605469, "learning_rate": 9.748095614184308e-05, "loss": 0.036, "step": 43920 }, { "epoch": 27.9987253027406, "grad_norm": 0.4362097382545471, "learning_rate": 9.747922849374027e-05, "loss": 0.0269, "step": 43930 }, { "epoch": 28.005098789037604, "grad_norm": 0.39225107431411743, "learning_rate": 9.747750026872e-05, "loss": 0.0291, "step": 43940 }, { "epoch": 28.011472275334608, "grad_norm": 0.5118345022201538, "learning_rate": 9.747577146680329e-05, "loss": 0.0277, "step": 43950 }, { "epoch": 28.017845761631612, "grad_norm": 0.7160177826881409, "learning_rate": 9.747404208801115e-05, "loss": 0.0399, "step": 43960 }, { "epoch": 28.024219247928617, "grad_norm": 0.4097060561180115, "learning_rate": 9.747231213236454e-05, "loss": 0.0313, "step": 43970 }, { "epoch": 28.03059273422562, "grad_norm": 0.5713366270065308, "learning_rate": 9.747058159988453e-05, "loss": 0.0385, "step": 43980 }, { "epoch": 28.036966220522626, "grad_norm": 0.5803492069244385, "learning_rate": 9.746885049059213e-05, "loss": 0.0329, "step": 43990 }, { "epoch": 28.04333970681963, "grad_norm": 0.45776650309562683, "learning_rate": 9.746711880450838e-05, "loss": 0.0465, "step": 44000 }, { "epoch": 28.049713193116634, "grad_norm": 0.4027117192745209, "learning_rate": 9.746538654165433e-05, "loss": 0.0364, "step": 44010 }, { "epoch": 28.05608667941364, "grad_norm": 0.7454760670661926, "learning_rate": 9.746365370205101e-05, "loss": 0.0285, "step": 44020 }, { "epoch": 28.062460165710643, "grad_norm": 0.5461715459823608, "learning_rate": 9.74619202857195e-05, "loss": 0.0333, "step": 44030 }, { "epoch": 28.068833652007648, "grad_norm": 0.6563729643821716, "learning_rate": 9.746018629268083e-05, "loss": 0.0276, "step": 44040 }, { "epoch": 28.075207138304652, "grad_norm": 0.28457748889923096, "learning_rate": 9.745845172295609e-05, "loss": 0.0251, "step": 44050 }, { "epoch": 28.081580624601656, "grad_norm": 0.6931522488594055, "learning_rate": 9.745671657656634e-05, "loss": 0.0317, "step": 44060 }, { "epoch": 28.08795411089866, "grad_norm": 0.7118549346923828, "learning_rate": 9.74549808535327e-05, "loss": 0.0314, "step": 44070 }, { "epoch": 28.094327597195665, "grad_norm": 0.5831851959228516, "learning_rate": 9.745324455387623e-05, "loss": 0.0265, "step": 44080 }, { "epoch": 28.10070108349267, "grad_norm": 0.5146048665046692, "learning_rate": 9.745150767761804e-05, "loss": 0.03, "step": 44090 }, { "epoch": 28.107074569789674, "grad_norm": 0.4322667717933655, "learning_rate": 9.744977022477921e-05, "loss": 0.0413, "step": 44100 }, { "epoch": 28.11344805608668, "grad_norm": 0.4913763999938965, "learning_rate": 9.74480321953809e-05, "loss": 0.0418, "step": 44110 }, { "epoch": 28.119821542383683, "grad_norm": 0.5145254731178284, "learning_rate": 9.744629358944418e-05, "loss": 0.0311, "step": 44120 }, { "epoch": 28.126195028680687, "grad_norm": 0.3814544379711151, "learning_rate": 9.74445544069902e-05, "loss": 0.0309, "step": 44130 }, { "epoch": 28.13256851497769, "grad_norm": 0.4900464415550232, "learning_rate": 9.74428146480401e-05, "loss": 0.0298, "step": 44140 }, { "epoch": 28.138942001274696, "grad_norm": 0.43112340569496155, "learning_rate": 9.7441074312615e-05, "loss": 0.021, "step": 44150 }, { "epoch": 28.1453154875717, "grad_norm": 0.5536928772926331, "learning_rate": 9.743933340073606e-05, "loss": 0.0307, "step": 44160 }, { "epoch": 28.151688973868705, "grad_norm": 0.5054657459259033, "learning_rate": 9.743759191242443e-05, "loss": 0.0217, "step": 44170 }, { "epoch": 28.15806246016571, "grad_norm": 0.3926021456718445, "learning_rate": 9.743584984770125e-05, "loss": 0.0233, "step": 44180 }, { "epoch": 28.164435946462714, "grad_norm": 0.38819441199302673, "learning_rate": 9.743410720658773e-05, "loss": 0.0301, "step": 44190 }, { "epoch": 28.170809432759718, "grad_norm": 0.6325846314430237, "learning_rate": 9.743236398910501e-05, "loss": 0.0417, "step": 44200 }, { "epoch": 28.177182919056722, "grad_norm": 0.9727165102958679, "learning_rate": 9.74306201952743e-05, "loss": 0.0342, "step": 44210 }, { "epoch": 28.183556405353727, "grad_norm": 0.6439951062202454, "learning_rate": 9.742887582511675e-05, "loss": 0.0289, "step": 44220 }, { "epoch": 28.18992989165073, "grad_norm": 0.5638203024864197, "learning_rate": 9.742713087865358e-05, "loss": 0.0314, "step": 44230 }, { "epoch": 28.196303377947736, "grad_norm": 0.45663660764694214, "learning_rate": 9.742538535590601e-05, "loss": 0.0267, "step": 44240 }, { "epoch": 28.202676864244744, "grad_norm": 0.3631056547164917, "learning_rate": 9.742363925689521e-05, "loss": 0.0238, "step": 44250 }, { "epoch": 28.209050350541748, "grad_norm": 0.4587206542491913, "learning_rate": 9.742189258164243e-05, "loss": 0.0329, "step": 44260 }, { "epoch": 28.215423836838752, "grad_norm": 0.34747639298439026, "learning_rate": 9.742014533016888e-05, "loss": 0.044, "step": 44270 }, { "epoch": 28.221797323135757, "grad_norm": 0.2990148365497589, "learning_rate": 9.74183975024958e-05, "loss": 0.0288, "step": 44280 }, { "epoch": 28.22817080943276, "grad_norm": 0.3446245789527893, "learning_rate": 9.741664909864441e-05, "loss": 0.0301, "step": 44290 }, { "epoch": 28.234544295729766, "grad_norm": 0.701809287071228, "learning_rate": 9.741490011863597e-05, "loss": 0.0361, "step": 44300 }, { "epoch": 28.24091778202677, "grad_norm": 0.32744964957237244, "learning_rate": 9.741315056249171e-05, "loss": 0.0269, "step": 44310 }, { "epoch": 28.247291268323774, "grad_norm": 0.5077762007713318, "learning_rate": 9.741140043023294e-05, "loss": 0.0322, "step": 44320 }, { "epoch": 28.25366475462078, "grad_norm": 0.5891860723495483, "learning_rate": 9.740964972188086e-05, "loss": 0.0434, "step": 44330 }, { "epoch": 28.260038240917783, "grad_norm": 0.8284059166908264, "learning_rate": 9.74078984374568e-05, "loss": 0.0284, "step": 44340 }, { "epoch": 28.266411727214788, "grad_norm": 0.2561104893684387, "learning_rate": 9.740614657698198e-05, "loss": 0.0315, "step": 44350 }, { "epoch": 28.272785213511792, "grad_norm": 0.4391784071922302, "learning_rate": 9.740439414047773e-05, "loss": 0.0295, "step": 44360 }, { "epoch": 28.279158699808796, "grad_norm": 0.5876372456550598, "learning_rate": 9.740264112796536e-05, "loss": 0.0346, "step": 44370 }, { "epoch": 28.2855321861058, "grad_norm": 0.6304070949554443, "learning_rate": 9.740088753946612e-05, "loss": 0.0252, "step": 44380 }, { "epoch": 28.291905672402805, "grad_norm": 0.5483896136283875, "learning_rate": 9.739913337500134e-05, "loss": 0.0399, "step": 44390 }, { "epoch": 28.29827915869981, "grad_norm": 0.42039191722869873, "learning_rate": 9.739737863459235e-05, "loss": 0.0291, "step": 44400 }, { "epoch": 28.304652644996814, "grad_norm": 0.42290419340133667, "learning_rate": 9.739562331826045e-05, "loss": 0.0387, "step": 44410 }, { "epoch": 28.31102613129382, "grad_norm": 0.47661104798316956, "learning_rate": 9.739386742602698e-05, "loss": 0.0531, "step": 44420 }, { "epoch": 28.317399617590823, "grad_norm": 0.41108641028404236, "learning_rate": 9.739211095791329e-05, "loss": 0.0418, "step": 44430 }, { "epoch": 28.323773103887827, "grad_norm": 0.6814137101173401, "learning_rate": 9.739035391394069e-05, "loss": 0.0332, "step": 44440 }, { "epoch": 28.33014659018483, "grad_norm": 0.376516729593277, "learning_rate": 9.738859629413055e-05, "loss": 0.0312, "step": 44450 }, { "epoch": 28.336520076481836, "grad_norm": 0.6428983211517334, "learning_rate": 9.738683809850422e-05, "loss": 0.0344, "step": 44460 }, { "epoch": 28.34289356277884, "grad_norm": 0.5290582776069641, "learning_rate": 9.738507932708307e-05, "loss": 0.0307, "step": 44470 }, { "epoch": 28.349267049075845, "grad_norm": 0.7155981063842773, "learning_rate": 9.738331997988845e-05, "loss": 0.0418, "step": 44480 }, { "epoch": 28.35564053537285, "grad_norm": 0.6242151856422424, "learning_rate": 9.738156005694179e-05, "loss": 0.029, "step": 44490 }, { "epoch": 28.362014021669854, "grad_norm": 0.36684373021125793, "learning_rate": 9.737979955826442e-05, "loss": 0.0253, "step": 44500 }, { "epoch": 28.368387507966858, "grad_norm": 0.6103434562683105, "learning_rate": 9.737803848387775e-05, "loss": 0.0322, "step": 44510 }, { "epoch": 28.374760994263863, "grad_norm": 0.5094860196113586, "learning_rate": 9.737627683380317e-05, "loss": 0.0385, "step": 44520 }, { "epoch": 28.381134480560867, "grad_norm": 0.5378496050834656, "learning_rate": 9.73745146080621e-05, "loss": 0.0328, "step": 44530 }, { "epoch": 28.38750796685787, "grad_norm": 0.45685452222824097, "learning_rate": 9.737275180667597e-05, "loss": 0.0305, "step": 44540 }, { "epoch": 28.393881453154876, "grad_norm": 0.44867458939552307, "learning_rate": 9.737098842966617e-05, "loss": 0.0392, "step": 44550 }, { "epoch": 28.40025493945188, "grad_norm": 0.355293869972229, "learning_rate": 9.736922447705413e-05, "loss": 0.0337, "step": 44560 }, { "epoch": 28.406628425748885, "grad_norm": 0.4744146168231964, "learning_rate": 9.736745994886127e-05, "loss": 0.037, "step": 44570 }, { "epoch": 28.41300191204589, "grad_norm": 0.6998198628425598, "learning_rate": 9.736569484510908e-05, "loss": 0.0442, "step": 44580 }, { "epoch": 28.419375398342893, "grad_norm": 0.3674408793449402, "learning_rate": 9.736392916581896e-05, "loss": 0.0495, "step": 44590 }, { "epoch": 28.425748884639898, "grad_norm": 0.49607911705970764, "learning_rate": 9.736216291101239e-05, "loss": 0.0242, "step": 44600 }, { "epoch": 28.432122370936902, "grad_norm": 0.3356236517429352, "learning_rate": 9.736039608071081e-05, "loss": 0.0367, "step": 44610 }, { "epoch": 28.438495857233907, "grad_norm": 0.5102330446243286, "learning_rate": 9.735862867493572e-05, "loss": 0.0293, "step": 44620 }, { "epoch": 28.44486934353091, "grad_norm": 0.5678808689117432, "learning_rate": 9.735686069370857e-05, "loss": 0.0379, "step": 44630 }, { "epoch": 28.451242829827915, "grad_norm": 0.4358641803264618, "learning_rate": 9.735509213705086e-05, "loss": 0.0289, "step": 44640 }, { "epoch": 28.45761631612492, "grad_norm": 0.5033252835273743, "learning_rate": 9.735332300498405e-05, "loss": 0.0369, "step": 44650 }, { "epoch": 28.463989802421924, "grad_norm": 0.23804186284542084, "learning_rate": 9.735155329752966e-05, "loss": 0.0327, "step": 44660 }, { "epoch": 28.47036328871893, "grad_norm": 0.5551007390022278, "learning_rate": 9.734978301470919e-05, "loss": 0.0305, "step": 44670 }, { "epoch": 28.476736775015933, "grad_norm": 0.5479502081871033, "learning_rate": 9.734801215654417e-05, "loss": 0.0384, "step": 44680 }, { "epoch": 28.483110261312937, "grad_norm": 0.5125648975372314, "learning_rate": 9.734624072305607e-05, "loss": 0.0367, "step": 44690 }, { "epoch": 28.489483747609942, "grad_norm": 0.5217462778091431, "learning_rate": 9.734446871426645e-05, "loss": 0.033, "step": 44700 }, { "epoch": 28.495857233906946, "grad_norm": 0.5010473728179932, "learning_rate": 9.734269613019687e-05, "loss": 0.0377, "step": 44710 }, { "epoch": 28.50223072020395, "grad_norm": 0.4423927962779999, "learning_rate": 9.734092297086879e-05, "loss": 0.028, "step": 44720 }, { "epoch": 28.508604206500955, "grad_norm": 0.4629896581172943, "learning_rate": 9.733914923630381e-05, "loss": 0.016, "step": 44730 }, { "epoch": 28.51497769279796, "grad_norm": 0.5018481016159058, "learning_rate": 9.733737492652348e-05, "loss": 0.0288, "step": 44740 }, { "epoch": 28.521351179094964, "grad_norm": 0.3976377844810486, "learning_rate": 9.733560004154934e-05, "loss": 0.0331, "step": 44750 }, { "epoch": 28.52772466539197, "grad_norm": 0.7726106643676758, "learning_rate": 9.7333824581403e-05, "loss": 0.0387, "step": 44760 }, { "epoch": 28.534098151688973, "grad_norm": 0.4114859402179718, "learning_rate": 9.733204854610596e-05, "loss": 0.0356, "step": 44770 }, { "epoch": 28.540471637985977, "grad_norm": 0.40384554862976074, "learning_rate": 9.733027193567986e-05, "loss": 0.0349, "step": 44780 }, { "epoch": 28.54684512428298, "grad_norm": 0.22549287974834442, "learning_rate": 9.732849475014625e-05, "loss": 0.038, "step": 44790 }, { "epoch": 28.553218610579986, "grad_norm": 0.6069677472114563, "learning_rate": 9.732671698952676e-05, "loss": 0.0379, "step": 44800 }, { "epoch": 28.55959209687699, "grad_norm": 0.42942240834236145, "learning_rate": 9.732493865384299e-05, "loss": 0.0244, "step": 44810 }, { "epoch": 28.565965583173995, "grad_norm": 0.6117925643920898, "learning_rate": 9.73231597431165e-05, "loss": 0.0326, "step": 44820 }, { "epoch": 28.572339069471, "grad_norm": 0.42230334877967834, "learning_rate": 9.732138025736895e-05, "loss": 0.0613, "step": 44830 }, { "epoch": 28.578712555768004, "grad_norm": 0.27693191170692444, "learning_rate": 9.731960019662196e-05, "loss": 0.0351, "step": 44840 }, { "epoch": 28.585086042065008, "grad_norm": 0.6259333491325378, "learning_rate": 9.731781956089714e-05, "loss": 0.0387, "step": 44850 }, { "epoch": 28.591459528362012, "grad_norm": 0.31598782539367676, "learning_rate": 9.731603835021614e-05, "loss": 0.0293, "step": 44860 }, { "epoch": 28.597833014659017, "grad_norm": 0.5598121881484985, "learning_rate": 9.73142565646006e-05, "loss": 0.0303, "step": 44870 }, { "epoch": 28.604206500956025, "grad_norm": 0.7967960238456726, "learning_rate": 9.731247420407219e-05, "loss": 0.0435, "step": 44880 }, { "epoch": 28.61057998725303, "grad_norm": 0.47927552461624146, "learning_rate": 9.731069126865254e-05, "loss": 0.029, "step": 44890 }, { "epoch": 28.616953473550033, "grad_norm": 0.6913169026374817, "learning_rate": 9.73089077583633e-05, "loss": 0.0354, "step": 44900 }, { "epoch": 28.623326959847038, "grad_norm": 0.3500741720199585, "learning_rate": 9.730712367322618e-05, "loss": 0.0276, "step": 44910 }, { "epoch": 28.629700446144042, "grad_norm": 0.805134117603302, "learning_rate": 9.730533901326285e-05, "loss": 0.0348, "step": 44920 }, { "epoch": 28.636073932441047, "grad_norm": 0.4991244077682495, "learning_rate": 9.730355377849496e-05, "loss": 0.0408, "step": 44930 }, { "epoch": 28.64244741873805, "grad_norm": 0.7973124980926514, "learning_rate": 9.730176796894425e-05, "loss": 0.0467, "step": 44940 }, { "epoch": 28.648820905035056, "grad_norm": 0.5614928007125854, "learning_rate": 9.72999815846324e-05, "loss": 0.0302, "step": 44950 }, { "epoch": 28.65519439133206, "grad_norm": 0.36722517013549805, "learning_rate": 9.72981946255811e-05, "loss": 0.0309, "step": 44960 }, { "epoch": 28.661567877629064, "grad_norm": 0.6403418779373169, "learning_rate": 9.72964070918121e-05, "loss": 0.0253, "step": 44970 }, { "epoch": 28.66794136392607, "grad_norm": 0.5140339136123657, "learning_rate": 9.729461898334709e-05, "loss": 0.0455, "step": 44980 }, { "epoch": 28.674314850223073, "grad_norm": 0.530879557132721, "learning_rate": 9.72928303002078e-05, "loss": 0.0456, "step": 44990 }, { "epoch": 28.680688336520078, "grad_norm": 0.8292019963264465, "learning_rate": 9.729104104241598e-05, "loss": 0.032, "step": 45000 }, { "epoch": 28.687061822817082, "grad_norm": 0.6900933384895325, "learning_rate": 9.728925120999334e-05, "loss": 0.0313, "step": 45010 }, { "epoch": 28.693435309114086, "grad_norm": 0.42918214201927185, "learning_rate": 9.728746080296167e-05, "loss": 0.0279, "step": 45020 }, { "epoch": 28.69980879541109, "grad_norm": 0.3363021910190582, "learning_rate": 9.728566982134271e-05, "loss": 0.0394, "step": 45030 }, { "epoch": 28.706182281708095, "grad_norm": 0.3593839406967163, "learning_rate": 9.728387826515819e-05, "loss": 0.0299, "step": 45040 }, { "epoch": 28.7125557680051, "grad_norm": 0.5878353714942932, "learning_rate": 9.728208613442993e-05, "loss": 0.0324, "step": 45050 }, { "epoch": 28.718929254302104, "grad_norm": 0.4548419415950775, "learning_rate": 9.728029342917967e-05, "loss": 0.0276, "step": 45060 }, { "epoch": 28.72530274059911, "grad_norm": 0.6807811260223389, "learning_rate": 9.727850014942921e-05, "loss": 0.0422, "step": 45070 }, { "epoch": 28.731676226896113, "grad_norm": 0.9300608038902283, "learning_rate": 9.727670629520034e-05, "loss": 0.0286, "step": 45080 }, { "epoch": 28.738049713193117, "grad_norm": 0.6083965301513672, "learning_rate": 9.727491186651484e-05, "loss": 0.0261, "step": 45090 }, { "epoch": 28.74442319949012, "grad_norm": 0.9100776314735413, "learning_rate": 9.727311686339453e-05, "loss": 0.0308, "step": 45100 }, { "epoch": 28.750796685787126, "grad_norm": 0.4473556578159332, "learning_rate": 9.727132128586122e-05, "loss": 0.0324, "step": 45110 }, { "epoch": 28.75717017208413, "grad_norm": 0.4234658181667328, "learning_rate": 9.726952513393672e-05, "loss": 0.0371, "step": 45120 }, { "epoch": 28.763543658381135, "grad_norm": 0.47622305154800415, "learning_rate": 9.726772840764287e-05, "loss": 0.0383, "step": 45130 }, { "epoch": 28.76991714467814, "grad_norm": 0.5168135166168213, "learning_rate": 9.726593110700148e-05, "loss": 0.0289, "step": 45140 }, { "epoch": 28.776290630975144, "grad_norm": 0.6061549782752991, "learning_rate": 9.72641332320344e-05, "loss": 0.026, "step": 45150 }, { "epoch": 28.782664117272148, "grad_norm": 0.6520146727561951, "learning_rate": 9.726233478276347e-05, "loss": 0.0307, "step": 45160 }, { "epoch": 28.789037603569152, "grad_norm": 0.5421724915504456, "learning_rate": 9.726053575921057e-05, "loss": 0.0409, "step": 45170 }, { "epoch": 28.795411089866157, "grad_norm": 0.5494058132171631, "learning_rate": 9.725873616139753e-05, "loss": 0.0332, "step": 45180 }, { "epoch": 28.80178457616316, "grad_norm": 0.4381203055381775, "learning_rate": 9.725693598934622e-05, "loss": 0.0367, "step": 45190 }, { "epoch": 28.808158062460166, "grad_norm": 0.6294469833374023, "learning_rate": 9.725513524307853e-05, "loss": 0.0264, "step": 45200 }, { "epoch": 28.81453154875717, "grad_norm": 0.41329479217529297, "learning_rate": 9.725333392261634e-05, "loss": 0.0276, "step": 45210 }, { "epoch": 28.820905035054174, "grad_norm": 0.6800749897956848, "learning_rate": 9.725153202798149e-05, "loss": 0.0262, "step": 45220 }, { "epoch": 28.82727852135118, "grad_norm": 0.4792836010456085, "learning_rate": 9.724972955919595e-05, "loss": 0.0226, "step": 45230 }, { "epoch": 28.833652007648183, "grad_norm": 0.46982795000076294, "learning_rate": 9.724792651628156e-05, "loss": 0.0344, "step": 45240 }, { "epoch": 28.840025493945188, "grad_norm": 0.345822274684906, "learning_rate": 9.724612289926028e-05, "loss": 0.0387, "step": 45250 }, { "epoch": 28.846398980242192, "grad_norm": 0.8648724555969238, "learning_rate": 9.724431870815398e-05, "loss": 0.0338, "step": 45260 }, { "epoch": 28.852772466539196, "grad_norm": 0.5537001490592957, "learning_rate": 9.724251394298461e-05, "loss": 0.0252, "step": 45270 }, { "epoch": 28.8591459528362, "grad_norm": 0.41298604011535645, "learning_rate": 9.724070860377408e-05, "loss": 0.0361, "step": 45280 }, { "epoch": 28.865519439133205, "grad_norm": 0.977382481098175, "learning_rate": 9.723890269054434e-05, "loss": 0.0328, "step": 45290 }, { "epoch": 28.87189292543021, "grad_norm": 0.5002893805503845, "learning_rate": 9.723709620331733e-05, "loss": 0.0324, "step": 45300 }, { "epoch": 28.878266411727214, "grad_norm": 0.6122251749038696, "learning_rate": 9.723528914211502e-05, "loss": 0.0458, "step": 45310 }, { "epoch": 28.88463989802422, "grad_norm": 0.35261765122413635, "learning_rate": 9.723348150695934e-05, "loss": 0.0446, "step": 45320 }, { "epoch": 28.891013384321223, "grad_norm": 0.31689614057540894, "learning_rate": 9.723167329787227e-05, "loss": 0.0223, "step": 45330 }, { "epoch": 28.897386870618227, "grad_norm": 0.4325798451900482, "learning_rate": 9.722986451487574e-05, "loss": 0.0291, "step": 45340 }, { "epoch": 28.90376035691523, "grad_norm": 0.5852455496788025, "learning_rate": 9.72280551579918e-05, "loss": 0.0374, "step": 45350 }, { "epoch": 28.910133843212236, "grad_norm": 0.3591966927051544, "learning_rate": 9.72262452272424e-05, "loss": 0.0306, "step": 45360 }, { "epoch": 28.91650732950924, "grad_norm": 0.6264025568962097, "learning_rate": 9.722443472264951e-05, "loss": 0.0419, "step": 45370 }, { "epoch": 28.922880815806245, "grad_norm": 1.1702104806900024, "learning_rate": 9.722262364423517e-05, "loss": 0.0391, "step": 45380 }, { "epoch": 28.92925430210325, "grad_norm": 0.5118148326873779, "learning_rate": 9.722081199202136e-05, "loss": 0.0302, "step": 45390 }, { "epoch": 28.935627788400254, "grad_norm": 0.6835926175117493, "learning_rate": 9.72189997660301e-05, "loss": 0.0392, "step": 45400 }, { "epoch": 28.942001274697258, "grad_norm": 0.47925686836242676, "learning_rate": 9.721718696628341e-05, "loss": 0.0212, "step": 45410 }, { "epoch": 28.948374760994263, "grad_norm": 0.3401205837726593, "learning_rate": 9.721537359280331e-05, "loss": 0.0351, "step": 45420 }, { "epoch": 28.954748247291267, "grad_norm": 0.36152106523513794, "learning_rate": 9.721355964561185e-05, "loss": 0.0207, "step": 45430 }, { "epoch": 28.96112173358827, "grad_norm": 0.46962159872055054, "learning_rate": 9.721174512473107e-05, "loss": 0.0253, "step": 45440 }, { "epoch": 28.967495219885276, "grad_norm": 0.3419113755226135, "learning_rate": 9.7209930030183e-05, "loss": 0.0296, "step": 45450 }, { "epoch": 28.97386870618228, "grad_norm": 0.39809948205947876, "learning_rate": 9.720811436198971e-05, "loss": 0.0382, "step": 45460 }, { "epoch": 28.980242192479285, "grad_norm": 0.44267067313194275, "learning_rate": 9.720629812017327e-05, "loss": 0.0355, "step": 45470 }, { "epoch": 28.98661567877629, "grad_norm": 0.79180508852005, "learning_rate": 9.720448130475572e-05, "loss": 0.03, "step": 45480 }, { "epoch": 28.992989165073293, "grad_norm": 0.4299890995025635, "learning_rate": 9.720266391575916e-05, "loss": 0.0354, "step": 45490 }, { "epoch": 28.999362651370298, "grad_norm": 0.40014269948005676, "learning_rate": 9.720084595320567e-05, "loss": 0.0239, "step": 45500 }, { "epoch": 29.005736137667306, "grad_norm": 0.5993176102638245, "learning_rate": 9.719902741711735e-05, "loss": 0.0298, "step": 45510 }, { "epoch": 29.01210962396431, "grad_norm": 0.6219274997711182, "learning_rate": 9.719720830751627e-05, "loss": 0.0348, "step": 45520 }, { "epoch": 29.018483110261315, "grad_norm": 0.28383973240852356, "learning_rate": 9.719538862442455e-05, "loss": 0.0297, "step": 45530 }, { "epoch": 29.02485659655832, "grad_norm": 0.6099619269371033, "learning_rate": 9.719356836786429e-05, "loss": 0.023, "step": 45540 }, { "epoch": 29.031230082855323, "grad_norm": 0.368915319442749, "learning_rate": 9.719174753785761e-05, "loss": 0.0222, "step": 45550 }, { "epoch": 29.037603569152328, "grad_norm": 0.7281890511512756, "learning_rate": 9.718992613442666e-05, "loss": 0.0282, "step": 45560 }, { "epoch": 29.043977055449332, "grad_norm": 0.5120526552200317, "learning_rate": 9.718810415759354e-05, "loss": 0.0306, "step": 45570 }, { "epoch": 29.050350541746337, "grad_norm": 0.5037716031074524, "learning_rate": 9.718628160738041e-05, "loss": 0.0244, "step": 45580 }, { "epoch": 29.05672402804334, "grad_norm": 0.4021316170692444, "learning_rate": 9.71844584838094e-05, "loss": 0.0355, "step": 45590 }, { "epoch": 29.063097514340345, "grad_norm": 0.448818176984787, "learning_rate": 9.718263478690268e-05, "loss": 0.02, "step": 45600 }, { "epoch": 29.06947100063735, "grad_norm": 0.5142409801483154, "learning_rate": 9.718081051668239e-05, "loss": 0.032, "step": 45610 }, { "epoch": 29.075844486934354, "grad_norm": 1.1114367246627808, "learning_rate": 9.717898567317071e-05, "loss": 0.0336, "step": 45620 }, { "epoch": 29.08221797323136, "grad_norm": 0.5713164806365967, "learning_rate": 9.71771602563898e-05, "loss": 0.0449, "step": 45630 }, { "epoch": 29.088591459528363, "grad_norm": 0.5510401129722595, "learning_rate": 9.717533426636187e-05, "loss": 0.0292, "step": 45640 }, { "epoch": 29.094964945825367, "grad_norm": 0.5720933079719543, "learning_rate": 9.717350770310907e-05, "loss": 0.0311, "step": 45650 }, { "epoch": 29.101338432122372, "grad_norm": 0.874988853931427, "learning_rate": 9.717168056665362e-05, "loss": 0.0362, "step": 45660 }, { "epoch": 29.107711918419376, "grad_norm": 0.9443122148513794, "learning_rate": 9.716985285701772e-05, "loss": 0.0333, "step": 45670 }, { "epoch": 29.11408540471638, "grad_norm": 0.5816103219985962, "learning_rate": 9.716802457422356e-05, "loss": 0.0327, "step": 45680 }, { "epoch": 29.120458891013385, "grad_norm": 0.40928781032562256, "learning_rate": 9.716619571829338e-05, "loss": 0.0412, "step": 45690 }, { "epoch": 29.12683237731039, "grad_norm": 0.5306153297424316, "learning_rate": 9.716436628924937e-05, "loss": 0.032, "step": 45700 }, { "epoch": 29.133205863607394, "grad_norm": 0.39029309153556824, "learning_rate": 9.71625362871138e-05, "loss": 0.0265, "step": 45710 }, { "epoch": 29.1395793499044, "grad_norm": 0.6047243475914001, "learning_rate": 9.716070571190887e-05, "loss": 0.0319, "step": 45720 }, { "epoch": 29.145952836201403, "grad_norm": 0.3802611231803894, "learning_rate": 9.715887456365684e-05, "loss": 0.0301, "step": 45730 }, { "epoch": 29.152326322498407, "grad_norm": 0.472699373960495, "learning_rate": 9.715704284237997e-05, "loss": 0.0226, "step": 45740 }, { "epoch": 29.15869980879541, "grad_norm": 0.41487348079681396, "learning_rate": 9.715521054810051e-05, "loss": 0.0259, "step": 45750 }, { "epoch": 29.165073295092416, "grad_norm": 0.43770483136177063, "learning_rate": 9.71533776808407e-05, "loss": 0.0316, "step": 45760 }, { "epoch": 29.17144678138942, "grad_norm": 0.5364236831665039, "learning_rate": 9.715154424062284e-05, "loss": 0.03, "step": 45770 }, { "epoch": 29.177820267686425, "grad_norm": 0.40557777881622314, "learning_rate": 9.71497102274692e-05, "loss": 0.0229, "step": 45780 }, { "epoch": 29.18419375398343, "grad_norm": 0.6140701770782471, "learning_rate": 9.714787564140206e-05, "loss": 0.0329, "step": 45790 }, { "epoch": 29.190567240280433, "grad_norm": 0.5551201701164246, "learning_rate": 9.714604048244373e-05, "loss": 0.0385, "step": 45800 }, { "epoch": 29.196940726577438, "grad_norm": 0.435820609331131, "learning_rate": 9.714420475061648e-05, "loss": 0.0305, "step": 45810 }, { "epoch": 29.203314212874442, "grad_norm": 0.5099499225616455, "learning_rate": 9.714236844594263e-05, "loss": 0.0314, "step": 45820 }, { "epoch": 29.209687699171447, "grad_norm": 0.326946884393692, "learning_rate": 9.71405315684445e-05, "loss": 0.0322, "step": 45830 }, { "epoch": 29.21606118546845, "grad_norm": 0.5033048987388611, "learning_rate": 9.713869411814441e-05, "loss": 0.0488, "step": 45840 }, { "epoch": 29.222434671765456, "grad_norm": 0.5972860455513, "learning_rate": 9.713685609506467e-05, "loss": 0.0464, "step": 45850 }, { "epoch": 29.22880815806246, "grad_norm": 0.7451785206794739, "learning_rate": 9.713501749922762e-05, "loss": 0.0367, "step": 45860 }, { "epoch": 29.235181644359464, "grad_norm": 0.36853545904159546, "learning_rate": 9.713317833065561e-05, "loss": 0.0298, "step": 45870 }, { "epoch": 29.24155513065647, "grad_norm": 0.4983586072921753, "learning_rate": 9.7131338589371e-05, "loss": 0.0311, "step": 45880 }, { "epoch": 29.247928616953473, "grad_norm": 0.4712297320365906, "learning_rate": 9.712949827539611e-05, "loss": 0.0319, "step": 45890 }, { "epoch": 29.254302103250478, "grad_norm": 0.5224852561950684, "learning_rate": 9.712765738875332e-05, "loss": 0.0331, "step": 45900 }, { "epoch": 29.260675589547482, "grad_norm": 0.7474142909049988, "learning_rate": 9.712581592946499e-05, "loss": 0.0292, "step": 45910 }, { "epoch": 29.267049075844486, "grad_norm": 0.5282825827598572, "learning_rate": 9.71239738975535e-05, "loss": 0.0414, "step": 45920 }, { "epoch": 29.27342256214149, "grad_norm": 0.3555707335472107, "learning_rate": 9.712213129304124e-05, "loss": 0.0287, "step": 45930 }, { "epoch": 29.279796048438495, "grad_norm": 0.36459726095199585, "learning_rate": 9.712028811595059e-05, "loss": 0.024, "step": 45940 }, { "epoch": 29.2861695347355, "grad_norm": 0.4337528944015503, "learning_rate": 9.711844436630397e-05, "loss": 0.0298, "step": 45950 }, { "epoch": 29.292543021032504, "grad_norm": 0.5455673336982727, "learning_rate": 9.711660004412375e-05, "loss": 0.0356, "step": 45960 }, { "epoch": 29.29891650732951, "grad_norm": 0.4543514549732208, "learning_rate": 9.711475514943236e-05, "loss": 0.0243, "step": 45970 }, { "epoch": 29.305289993626513, "grad_norm": 0.3699268698692322, "learning_rate": 9.711290968225219e-05, "loss": 0.0278, "step": 45980 }, { "epoch": 29.311663479923517, "grad_norm": 0.5745609402656555, "learning_rate": 9.711106364260571e-05, "loss": 0.0296, "step": 45990 }, { "epoch": 29.31803696622052, "grad_norm": 0.5258733034133911, "learning_rate": 9.710921703051533e-05, "loss": 0.0418, "step": 46000 }, { "epoch": 29.324410452517526, "grad_norm": 0.4699323773384094, "learning_rate": 9.710736984600346e-05, "loss": 0.03, "step": 46010 }, { "epoch": 29.33078393881453, "grad_norm": 1.4567075967788696, "learning_rate": 9.710552208909258e-05, "loss": 0.0338, "step": 46020 }, { "epoch": 29.337157425111535, "grad_norm": 0.4261760115623474, "learning_rate": 9.710367375980513e-05, "loss": 0.025, "step": 46030 }, { "epoch": 29.34353091140854, "grad_norm": 0.7398717999458313, "learning_rate": 9.710182485816358e-05, "loss": 0.0319, "step": 46040 }, { "epoch": 29.349904397705544, "grad_norm": 0.431633859872818, "learning_rate": 9.709997538419038e-05, "loss": 0.0359, "step": 46050 }, { "epoch": 29.356277884002548, "grad_norm": 0.4918813705444336, "learning_rate": 9.7098125337908e-05, "loss": 0.0283, "step": 46060 }, { "epoch": 29.362651370299552, "grad_norm": 0.6754379272460938, "learning_rate": 9.709627471933896e-05, "loss": 0.0326, "step": 46070 }, { "epoch": 29.369024856596557, "grad_norm": 0.8517892956733704, "learning_rate": 9.709442352850569e-05, "loss": 0.0467, "step": 46080 }, { "epoch": 29.37539834289356, "grad_norm": 0.5752487778663635, "learning_rate": 9.709257176543071e-05, "loss": 0.0313, "step": 46090 }, { "epoch": 29.381771829190566, "grad_norm": 0.3474980592727661, "learning_rate": 9.709071943013652e-05, "loss": 0.0279, "step": 46100 }, { "epoch": 29.38814531548757, "grad_norm": 0.48432061076164246, "learning_rate": 9.708886652264563e-05, "loss": 0.0375, "step": 46110 }, { "epoch": 29.394518801784574, "grad_norm": 0.3884345293045044, "learning_rate": 9.708701304298056e-05, "loss": 0.0472, "step": 46120 }, { "epoch": 29.40089228808158, "grad_norm": 0.492495059967041, "learning_rate": 9.708515899116382e-05, "loss": 0.0305, "step": 46130 }, { "epoch": 29.407265774378587, "grad_norm": 0.6608692407608032, "learning_rate": 9.708330436721794e-05, "loss": 0.0294, "step": 46140 }, { "epoch": 29.41363926067559, "grad_norm": 0.6116952300071716, "learning_rate": 9.708144917116546e-05, "loss": 0.0289, "step": 46150 }, { "epoch": 29.420012746972596, "grad_norm": 0.5953435897827148, "learning_rate": 9.707959340302893e-05, "loss": 0.0313, "step": 46160 }, { "epoch": 29.4263862332696, "grad_norm": 0.6745482683181763, "learning_rate": 9.707773706283087e-05, "loss": 0.0374, "step": 46170 }, { "epoch": 29.432759719566604, "grad_norm": 0.6373438835144043, "learning_rate": 9.707588015059386e-05, "loss": 0.0342, "step": 46180 }, { "epoch": 29.43913320586361, "grad_norm": 0.6488941311836243, "learning_rate": 9.707402266634047e-05, "loss": 0.0357, "step": 46190 }, { "epoch": 29.445506692160613, "grad_norm": 0.4686749279499054, "learning_rate": 9.707216461009325e-05, "loss": 0.0288, "step": 46200 }, { "epoch": 29.451880178457618, "grad_norm": 0.3598555624485016, "learning_rate": 9.70703059818748e-05, "loss": 0.0362, "step": 46210 }, { "epoch": 29.458253664754622, "grad_norm": 0.5486948490142822, "learning_rate": 9.706844678170767e-05, "loss": 0.0353, "step": 46220 }, { "epoch": 29.464627151051626, "grad_norm": 0.45492681860923767, "learning_rate": 9.706658700961448e-05, "loss": 0.0302, "step": 46230 }, { "epoch": 29.47100063734863, "grad_norm": 0.446791410446167, "learning_rate": 9.706472666561783e-05, "loss": 0.0254, "step": 46240 }, { "epoch": 29.477374123645635, "grad_norm": 1.3344231843948364, "learning_rate": 9.70628657497403e-05, "loss": 0.0357, "step": 46250 }, { "epoch": 29.48374760994264, "grad_norm": 0.4148443937301636, "learning_rate": 9.706100426200452e-05, "loss": 0.0242, "step": 46260 }, { "epoch": 29.490121096239644, "grad_norm": 0.6448552012443542, "learning_rate": 9.705914220243313e-05, "loss": 0.0361, "step": 46270 }, { "epoch": 29.49649458253665, "grad_norm": 0.47051575779914856, "learning_rate": 9.705727957104871e-05, "loss": 0.0228, "step": 46280 }, { "epoch": 29.502868068833653, "grad_norm": 0.35242611169815063, "learning_rate": 9.70554163678739e-05, "loss": 0.0294, "step": 46290 }, { "epoch": 29.509241555130657, "grad_norm": 0.6139636635780334, "learning_rate": 9.705355259293137e-05, "loss": 0.0264, "step": 46300 }, { "epoch": 29.51561504142766, "grad_norm": 0.8718332052230835, "learning_rate": 9.705168824624376e-05, "loss": 0.0261, "step": 46310 }, { "epoch": 29.521988527724666, "grad_norm": 0.558094322681427, "learning_rate": 9.70498233278337e-05, "loss": 0.0266, "step": 46320 }, { "epoch": 29.52836201402167, "grad_norm": 0.5590869188308716, "learning_rate": 9.704795783772386e-05, "loss": 0.0277, "step": 46330 }, { "epoch": 29.534735500318675, "grad_norm": 0.37498053908348083, "learning_rate": 9.704609177593691e-05, "loss": 0.0325, "step": 46340 }, { "epoch": 29.54110898661568, "grad_norm": 0.6030009984970093, "learning_rate": 9.704422514249554e-05, "loss": 0.0406, "step": 46350 }, { "epoch": 29.547482472912684, "grad_norm": 0.46655571460723877, "learning_rate": 9.70423579374224e-05, "loss": 0.0394, "step": 46360 }, { "epoch": 29.553855959209688, "grad_norm": 0.5389651656150818, "learning_rate": 9.704049016074022e-05, "loss": 0.036, "step": 46370 }, { "epoch": 29.560229445506693, "grad_norm": 0.5369834899902344, "learning_rate": 9.703862181247166e-05, "loss": 0.0275, "step": 46380 }, { "epoch": 29.566602931803697, "grad_norm": 0.8305455446243286, "learning_rate": 9.703675289263942e-05, "loss": 0.0311, "step": 46390 }, { "epoch": 29.5729764181007, "grad_norm": 0.545233428478241, "learning_rate": 9.703488340126623e-05, "loss": 0.0353, "step": 46400 }, { "epoch": 29.579349904397706, "grad_norm": 0.4404902160167694, "learning_rate": 9.70330133383748e-05, "loss": 0.0298, "step": 46410 }, { "epoch": 29.58572339069471, "grad_norm": 0.6201958060264587, "learning_rate": 9.703114270398785e-05, "loss": 0.0345, "step": 46420 }, { "epoch": 29.592096876991715, "grad_norm": 0.5541747212409973, "learning_rate": 9.702927149812812e-05, "loss": 0.0326, "step": 46430 }, { "epoch": 29.59847036328872, "grad_norm": 0.3940418064594269, "learning_rate": 9.702739972081834e-05, "loss": 0.0277, "step": 46440 }, { "epoch": 29.604843849585723, "grad_norm": 0.4481523633003235, "learning_rate": 9.702552737208124e-05, "loss": 0.0253, "step": 46450 }, { "epoch": 29.611217335882728, "grad_norm": 0.5187958478927612, "learning_rate": 9.702365445193958e-05, "loss": 0.0226, "step": 46460 }, { "epoch": 29.617590822179732, "grad_norm": 0.5064092874526978, "learning_rate": 9.702178096041612e-05, "loss": 0.0262, "step": 46470 }, { "epoch": 29.623964308476737, "grad_norm": 0.2739316523075104, "learning_rate": 9.701990689753365e-05, "loss": 0.0335, "step": 46480 }, { "epoch": 29.63033779477374, "grad_norm": 0.3782416880130768, "learning_rate": 9.701803226331489e-05, "loss": 0.0374, "step": 46490 }, { "epoch": 29.636711281070745, "grad_norm": 0.47051769495010376, "learning_rate": 9.701615705778267e-05, "loss": 0.0262, "step": 46500 }, { "epoch": 29.64308476736775, "grad_norm": 0.7086570858955383, "learning_rate": 9.701428128095972e-05, "loss": 0.0308, "step": 46510 }, { "epoch": 29.649458253664754, "grad_norm": 0.4280010163784027, "learning_rate": 9.701240493286888e-05, "loss": 0.0319, "step": 46520 }, { "epoch": 29.65583173996176, "grad_norm": 0.3365243673324585, "learning_rate": 9.701052801353294e-05, "loss": 0.0236, "step": 46530 }, { "epoch": 29.662205226258763, "grad_norm": 0.4565931558609009, "learning_rate": 9.70086505229747e-05, "loss": 0.0566, "step": 46540 }, { "epoch": 29.668578712555767, "grad_norm": 0.79093998670578, "learning_rate": 9.700677246121695e-05, "loss": 0.0376, "step": 46550 }, { "epoch": 29.674952198852772, "grad_norm": 0.7757982015609741, "learning_rate": 9.700489382828256e-05, "loss": 0.0257, "step": 46560 }, { "epoch": 29.681325685149776, "grad_norm": 0.3275720775127411, "learning_rate": 9.70030146241943e-05, "loss": 0.0223, "step": 46570 }, { "epoch": 29.68769917144678, "grad_norm": 0.5628404021263123, "learning_rate": 9.700113484897506e-05, "loss": 0.0296, "step": 46580 }, { "epoch": 29.694072657743785, "grad_norm": 0.40180811285972595, "learning_rate": 9.699925450264765e-05, "loss": 0.0364, "step": 46590 }, { "epoch": 29.70044614404079, "grad_norm": 0.35429057478904724, "learning_rate": 9.699737358523492e-05, "loss": 0.0252, "step": 46600 }, { "epoch": 29.706819630337794, "grad_norm": 0.356086403131485, "learning_rate": 9.699549209675973e-05, "loss": 0.0274, "step": 46610 }, { "epoch": 29.7131931166348, "grad_norm": 0.4660834074020386, "learning_rate": 9.699361003724495e-05, "loss": 0.0321, "step": 46620 }, { "epoch": 29.719566602931803, "grad_norm": 0.40020546317100525, "learning_rate": 9.699172740671342e-05, "loss": 0.0234, "step": 46630 }, { "epoch": 29.725940089228807, "grad_norm": 0.30107176303863525, "learning_rate": 9.698984420518804e-05, "loss": 0.0268, "step": 46640 }, { "epoch": 29.73231357552581, "grad_norm": 0.36909517645835876, "learning_rate": 9.69879604326917e-05, "loss": 0.0239, "step": 46650 }, { "epoch": 29.738687061822816, "grad_norm": 0.548409104347229, "learning_rate": 9.698607608924726e-05, "loss": 0.0326, "step": 46660 }, { "epoch": 29.74506054811982, "grad_norm": 0.601775050163269, "learning_rate": 9.698419117487762e-05, "loss": 0.025, "step": 46670 }, { "epoch": 29.751434034416825, "grad_norm": 0.7909955978393555, "learning_rate": 9.698230568960574e-05, "loss": 0.0377, "step": 46680 }, { "epoch": 29.75780752071383, "grad_norm": 0.6439356207847595, "learning_rate": 9.698041963345446e-05, "loss": 0.031, "step": 46690 }, { "epoch": 29.764181007010833, "grad_norm": 0.36052584648132324, "learning_rate": 9.697853300644672e-05, "loss": 0.0347, "step": 46700 }, { "epoch": 29.770554493307838, "grad_norm": 0.4040693938732147, "learning_rate": 9.697664580860546e-05, "loss": 0.0343, "step": 46710 }, { "epoch": 29.776927979604842, "grad_norm": 0.6904430389404297, "learning_rate": 9.69747580399536e-05, "loss": 0.0369, "step": 46720 }, { "epoch": 29.783301465901847, "grad_norm": 0.56150221824646, "learning_rate": 9.697286970051407e-05, "loss": 0.0253, "step": 46730 }, { "epoch": 29.78967495219885, "grad_norm": 0.5458069443702698, "learning_rate": 9.697098079030982e-05, "loss": 0.0352, "step": 46740 }, { "epoch": 29.79604843849586, "grad_norm": 0.4143088459968567, "learning_rate": 9.696909130936382e-05, "loss": 0.0261, "step": 46750 }, { "epoch": 29.80242192479286, "grad_norm": 0.4481980800628662, "learning_rate": 9.6967201257699e-05, "loss": 0.0283, "step": 46760 }, { "epoch": 29.808795411089868, "grad_norm": 0.5536223649978638, "learning_rate": 9.696531063533835e-05, "loss": 0.0382, "step": 46770 }, { "epoch": 29.815168897386872, "grad_norm": 0.6695423126220703, "learning_rate": 9.696341944230484e-05, "loss": 0.04, "step": 46780 }, { "epoch": 29.821542383683877, "grad_norm": 0.7825644612312317, "learning_rate": 9.696152767862143e-05, "loss": 0.0335, "step": 46790 }, { "epoch": 29.82791586998088, "grad_norm": 0.5314565896987915, "learning_rate": 9.695963534431113e-05, "loss": 0.0293, "step": 46800 }, { "epoch": 29.834289356277885, "grad_norm": 0.49897485971450806, "learning_rate": 9.695774243939691e-05, "loss": 0.0379, "step": 46810 }, { "epoch": 29.84066284257489, "grad_norm": 0.5014762282371521, "learning_rate": 9.695584896390179e-05, "loss": 0.0538, "step": 46820 }, { "epoch": 29.847036328871894, "grad_norm": 0.6763722896575928, "learning_rate": 9.695395491784878e-05, "loss": 0.0322, "step": 46830 }, { "epoch": 29.8534098151689, "grad_norm": 0.6262781023979187, "learning_rate": 9.695206030126087e-05, "loss": 0.033, "step": 46840 }, { "epoch": 29.859783301465903, "grad_norm": 0.43103867769241333, "learning_rate": 9.695016511416111e-05, "loss": 0.0204, "step": 46850 }, { "epoch": 29.866156787762907, "grad_norm": 0.5196895599365234, "learning_rate": 9.694826935657252e-05, "loss": 0.0274, "step": 46860 }, { "epoch": 29.872530274059912, "grad_norm": 0.6107072234153748, "learning_rate": 9.694637302851813e-05, "loss": 0.0245, "step": 46870 }, { "epoch": 29.878903760356916, "grad_norm": 0.4301851689815521, "learning_rate": 9.694447613002096e-05, "loss": 0.0283, "step": 46880 }, { "epoch": 29.88527724665392, "grad_norm": 0.5175921320915222, "learning_rate": 9.69425786611041e-05, "loss": 0.025, "step": 46890 }, { "epoch": 29.891650732950925, "grad_norm": 0.6600993871688843, "learning_rate": 9.694068062179058e-05, "loss": 0.0286, "step": 46900 }, { "epoch": 29.89802421924793, "grad_norm": 0.5901431441307068, "learning_rate": 9.69387820121035e-05, "loss": 0.0238, "step": 46910 }, { "epoch": 29.904397705544934, "grad_norm": 0.5341496467590332, "learning_rate": 9.693688283206586e-05, "loss": 0.0324, "step": 46920 }, { "epoch": 29.91077119184194, "grad_norm": 0.3416377604007721, "learning_rate": 9.693498308170079e-05, "loss": 0.0302, "step": 46930 }, { "epoch": 29.917144678138943, "grad_norm": 0.43900632858276367, "learning_rate": 9.693308276103137e-05, "loss": 0.0275, "step": 46940 }, { "epoch": 29.923518164435947, "grad_norm": 0.4261423349380493, "learning_rate": 9.693118187008068e-05, "loss": 0.0351, "step": 46950 }, { "epoch": 29.92989165073295, "grad_norm": 0.49773746728897095, "learning_rate": 9.692928040887183e-05, "loss": 0.032, "step": 46960 }, { "epoch": 29.936265137029956, "grad_norm": 0.38618946075439453, "learning_rate": 9.692737837742789e-05, "loss": 0.0211, "step": 46970 }, { "epoch": 29.94263862332696, "grad_norm": 0.35614651441574097, "learning_rate": 9.692547577577202e-05, "loss": 0.0325, "step": 46980 }, { "epoch": 29.949012109623965, "grad_norm": 0.3810688853263855, "learning_rate": 9.69235726039273e-05, "loss": 0.0243, "step": 46990 }, { "epoch": 29.95538559592097, "grad_norm": 0.4994458258152008, "learning_rate": 9.692166886191686e-05, "loss": 0.0244, "step": 47000 }, { "epoch": 29.961759082217974, "grad_norm": 0.6641340255737305, "learning_rate": 9.691976454976387e-05, "loss": 0.0257, "step": 47010 }, { "epoch": 29.968132568514978, "grad_norm": 0.45964425802230835, "learning_rate": 9.691785966749143e-05, "loss": 0.0324, "step": 47020 }, { "epoch": 29.974506054811982, "grad_norm": 0.46489396691322327, "learning_rate": 9.691595421512269e-05, "loss": 0.0298, "step": 47030 }, { "epoch": 29.980879541108987, "grad_norm": 0.37213489413261414, "learning_rate": 9.691404819268082e-05, "loss": 0.033, "step": 47040 }, { "epoch": 29.98725302740599, "grad_norm": 0.33924904465675354, "learning_rate": 9.691214160018895e-05, "loss": 0.0332, "step": 47050 }, { "epoch": 29.993626513702996, "grad_norm": 0.7121188044548035, "learning_rate": 9.691023443767029e-05, "loss": 0.029, "step": 47060 }, { "epoch": 30.0, "grad_norm": 0.5272395014762878, "learning_rate": 9.690832670514798e-05, "loss": 0.0206, "step": 47070 }, { "epoch": 30.006373486297004, "grad_norm": 0.5339630842208862, "learning_rate": 9.690641840264522e-05, "loss": 0.0316, "step": 47080 }, { "epoch": 30.01274697259401, "grad_norm": 0.36145028471946716, "learning_rate": 9.690450953018518e-05, "loss": 0.0208, "step": 47090 }, { "epoch": 30.019120458891013, "grad_norm": 0.231989324092865, "learning_rate": 9.690260008779106e-05, "loss": 0.0254, "step": 47100 }, { "epoch": 30.025493945188018, "grad_norm": 0.44678598642349243, "learning_rate": 9.690069007548608e-05, "loss": 0.0358, "step": 47110 }, { "epoch": 30.031867431485022, "grad_norm": 0.24146738648414612, "learning_rate": 9.689877949329342e-05, "loss": 0.0219, "step": 47120 }, { "epoch": 30.038240917782026, "grad_norm": 0.48522791266441345, "learning_rate": 9.689686834123632e-05, "loss": 0.0241, "step": 47130 }, { "epoch": 30.04461440407903, "grad_norm": 0.393105685710907, "learning_rate": 9.6894956619338e-05, "loss": 0.0265, "step": 47140 }, { "epoch": 30.050987890376035, "grad_norm": 0.5010115504264832, "learning_rate": 9.689304432762167e-05, "loss": 0.0377, "step": 47150 }, { "epoch": 30.05736137667304, "grad_norm": 0.7131636142730713, "learning_rate": 9.689113146611057e-05, "loss": 0.0225, "step": 47160 }, { "epoch": 30.063734862970044, "grad_norm": 0.5025869011878967, "learning_rate": 9.688921803482796e-05, "loss": 0.0331, "step": 47170 }, { "epoch": 30.07010834926705, "grad_norm": 0.319277822971344, "learning_rate": 9.688730403379708e-05, "loss": 0.0359, "step": 47180 }, { "epoch": 30.076481835564053, "grad_norm": 0.45344534516334534, "learning_rate": 9.688538946304118e-05, "loss": 0.0297, "step": 47190 }, { "epoch": 30.082855321861057, "grad_norm": 0.16791537404060364, "learning_rate": 9.688347432258353e-05, "loss": 0.0217, "step": 47200 }, { "epoch": 30.08922880815806, "grad_norm": 0.4167848825454712, "learning_rate": 9.68815586124474e-05, "loss": 0.0239, "step": 47210 }, { "epoch": 30.095602294455066, "grad_norm": 0.4025765359401703, "learning_rate": 9.687964233265608e-05, "loss": 0.02, "step": 47220 }, { "epoch": 30.10197578075207, "grad_norm": 0.5122728943824768, "learning_rate": 9.687772548323284e-05, "loss": 0.0226, "step": 47230 }, { "epoch": 30.108349267049075, "grad_norm": 0.5051019787788391, "learning_rate": 9.687580806420098e-05, "loss": 0.0342, "step": 47240 }, { "epoch": 30.11472275334608, "grad_norm": 0.6490054130554199, "learning_rate": 9.687389007558379e-05, "loss": 0.0265, "step": 47250 }, { "epoch": 30.121096239643084, "grad_norm": 0.4030517637729645, "learning_rate": 9.687197151740457e-05, "loss": 0.0271, "step": 47260 }, { "epoch": 30.127469725940088, "grad_norm": 0.32583263516426086, "learning_rate": 9.687005238968665e-05, "loss": 0.0373, "step": 47270 }, { "epoch": 30.133843212237093, "grad_norm": 0.48657089471817017, "learning_rate": 9.686813269245333e-05, "loss": 0.0434, "step": 47280 }, { "epoch": 30.140216698534097, "grad_norm": 0.4019191265106201, "learning_rate": 9.686621242572797e-05, "loss": 0.0284, "step": 47290 }, { "epoch": 30.1465901848311, "grad_norm": 0.46040210127830505, "learning_rate": 9.686429158953388e-05, "loss": 0.032, "step": 47300 }, { "epoch": 30.152963671128106, "grad_norm": 0.2923794090747833, "learning_rate": 9.686237018389438e-05, "loss": 0.0257, "step": 47310 }, { "epoch": 30.15933715742511, "grad_norm": 0.49327313899993896, "learning_rate": 9.686044820883285e-05, "loss": 0.0253, "step": 47320 }, { "epoch": 30.165710643722115, "grad_norm": 0.5814868807792664, "learning_rate": 9.685852566437261e-05, "loss": 0.0241, "step": 47330 }, { "epoch": 30.17208413001912, "grad_norm": 0.5563207864761353, "learning_rate": 9.685660255053706e-05, "loss": 0.0229, "step": 47340 }, { "epoch": 30.178457616316123, "grad_norm": 0.6287422180175781, "learning_rate": 9.685467886734956e-05, "loss": 0.0275, "step": 47350 }, { "epoch": 30.184831102613128, "grad_norm": 0.6853327751159668, "learning_rate": 9.685275461483346e-05, "loss": 0.0286, "step": 47360 }, { "epoch": 30.191204588910132, "grad_norm": 0.4149956703186035, "learning_rate": 9.685082979301214e-05, "loss": 0.0256, "step": 47370 }, { "epoch": 30.197578075207137, "grad_norm": 0.6666845679283142, "learning_rate": 9.684890440190903e-05, "loss": 0.0431, "step": 47380 }, { "epoch": 30.203951561504145, "grad_norm": 0.7017353177070618, "learning_rate": 9.684697844154747e-05, "loss": 0.0351, "step": 47390 }, { "epoch": 30.21032504780115, "grad_norm": 0.2848077416419983, "learning_rate": 9.684505191195092e-05, "loss": 0.0244, "step": 47400 }, { "epoch": 30.216698534098153, "grad_norm": 0.3562880754470825, "learning_rate": 9.684312481314275e-05, "loss": 0.0313, "step": 47410 }, { "epoch": 30.223072020395158, "grad_norm": 0.3509465157985687, "learning_rate": 9.684119714514638e-05, "loss": 0.0242, "step": 47420 }, { "epoch": 30.229445506692162, "grad_norm": 0.5636443495750427, "learning_rate": 9.683926890798525e-05, "loss": 0.0336, "step": 47430 }, { "epoch": 30.235818992989167, "grad_norm": 0.3845672011375427, "learning_rate": 9.683734010168276e-05, "loss": 0.0279, "step": 47440 }, { "epoch": 30.24219247928617, "grad_norm": 0.6842441558837891, "learning_rate": 9.683541072626239e-05, "loss": 0.0429, "step": 47450 }, { "epoch": 30.248565965583175, "grad_norm": 0.5303451418876648, "learning_rate": 9.683348078174754e-05, "loss": 0.0276, "step": 47460 }, { "epoch": 30.25493945188018, "grad_norm": 0.601003885269165, "learning_rate": 9.683155026816168e-05, "loss": 0.0459, "step": 47470 }, { "epoch": 30.261312938177184, "grad_norm": 0.6152337789535522, "learning_rate": 9.682961918552828e-05, "loss": 0.0302, "step": 47480 }, { "epoch": 30.26768642447419, "grad_norm": 0.8760467767715454, "learning_rate": 9.682768753387079e-05, "loss": 0.041, "step": 47490 }, { "epoch": 30.274059910771193, "grad_norm": 0.3496536612510681, "learning_rate": 9.682575531321269e-05, "loss": 0.0195, "step": 47500 }, { "epoch": 30.280433397068197, "grad_norm": 0.6106433868408203, "learning_rate": 9.682382252357745e-05, "loss": 0.0332, "step": 47510 }, { "epoch": 30.2868068833652, "grad_norm": 0.680897057056427, "learning_rate": 9.682188916498855e-05, "loss": 0.0254, "step": 47520 }, { "epoch": 30.293180369662206, "grad_norm": 0.4990352988243103, "learning_rate": 9.681995523746948e-05, "loss": 0.0291, "step": 47530 }, { "epoch": 30.29955385595921, "grad_norm": 0.3918863832950592, "learning_rate": 9.681802074104376e-05, "loss": 0.0249, "step": 47540 }, { "epoch": 30.305927342256215, "grad_norm": 0.5724256038665771, "learning_rate": 9.681608567573487e-05, "loss": 0.0291, "step": 47550 }, { "epoch": 30.31230082855322, "grad_norm": 0.45661425590515137, "learning_rate": 9.681415004156638e-05, "loss": 0.0343, "step": 47560 }, { "epoch": 30.318674314850224, "grad_norm": 0.42636817693710327, "learning_rate": 9.681221383856173e-05, "loss": 0.0354, "step": 47570 }, { "epoch": 30.325047801147228, "grad_norm": 0.5731152296066284, "learning_rate": 9.68102770667445e-05, "loss": 0.0294, "step": 47580 }, { "epoch": 30.331421287444233, "grad_norm": 0.629697859287262, "learning_rate": 9.680833972613818e-05, "loss": 0.0381, "step": 47590 }, { "epoch": 30.337794773741237, "grad_norm": 0.6697096228599548, "learning_rate": 9.680640181676636e-05, "loss": 0.0222, "step": 47600 }, { "epoch": 30.34416826003824, "grad_norm": 0.5257476568222046, "learning_rate": 9.680446333865258e-05, "loss": 0.0547, "step": 47610 }, { "epoch": 30.350541746335246, "grad_norm": 0.6578401923179626, "learning_rate": 9.680252429182035e-05, "loss": 0.025, "step": 47620 }, { "epoch": 30.35691523263225, "grad_norm": 0.6949588656425476, "learning_rate": 9.680058467629328e-05, "loss": 0.0324, "step": 47630 }, { "epoch": 30.363288718929255, "grad_norm": 0.4209093749523163, "learning_rate": 9.679864449209491e-05, "loss": 0.0481, "step": 47640 }, { "epoch": 30.36966220522626, "grad_norm": 0.4853968024253845, "learning_rate": 9.679670373924883e-05, "loss": 0.0282, "step": 47650 }, { "epoch": 30.376035691523263, "grad_norm": 0.30764615535736084, "learning_rate": 9.67947624177786e-05, "loss": 0.031, "step": 47660 }, { "epoch": 30.382409177820268, "grad_norm": 0.4162934720516205, "learning_rate": 9.679282052770785e-05, "loss": 0.0238, "step": 47670 }, { "epoch": 30.388782664117272, "grad_norm": 0.2692345678806305, "learning_rate": 9.679087806906014e-05, "loss": 0.0221, "step": 47680 }, { "epoch": 30.395156150414277, "grad_norm": 0.523062527179718, "learning_rate": 9.678893504185908e-05, "loss": 0.0315, "step": 47690 }, { "epoch": 30.40152963671128, "grad_norm": 0.4525114893913269, "learning_rate": 9.678699144612828e-05, "loss": 0.0275, "step": 47700 }, { "epoch": 30.407903123008285, "grad_norm": 0.42778345942497253, "learning_rate": 9.678504728189138e-05, "loss": 0.0178, "step": 47710 }, { "epoch": 30.41427660930529, "grad_norm": 0.6708573698997498, "learning_rate": 9.678310254917197e-05, "loss": 0.022, "step": 47720 }, { "epoch": 30.420650095602294, "grad_norm": 0.45474258065223694, "learning_rate": 9.678115724799369e-05, "loss": 0.0292, "step": 47730 }, { "epoch": 30.4270235818993, "grad_norm": 0.6631261706352234, "learning_rate": 9.677921137838018e-05, "loss": 0.0276, "step": 47740 }, { "epoch": 30.433397068196303, "grad_norm": 0.5106426477432251, "learning_rate": 9.677726494035508e-05, "loss": 0.0388, "step": 47750 }, { "epoch": 30.439770554493307, "grad_norm": 0.4893629550933838, "learning_rate": 9.677531793394205e-05, "loss": 0.0254, "step": 47760 }, { "epoch": 30.446144040790312, "grad_norm": 0.5849584937095642, "learning_rate": 9.677337035916477e-05, "loss": 0.0294, "step": 47770 }, { "epoch": 30.452517527087316, "grad_norm": 0.6654514670372009, "learning_rate": 9.677142221604685e-05, "loss": 0.0266, "step": 47780 }, { "epoch": 30.45889101338432, "grad_norm": 0.5773518681526184, "learning_rate": 9.676947350461199e-05, "loss": 0.0419, "step": 47790 }, { "epoch": 30.465264499681325, "grad_norm": 0.41494619846343994, "learning_rate": 9.676752422488388e-05, "loss": 0.0193, "step": 47800 }, { "epoch": 30.47163798597833, "grad_norm": 0.5592244267463684, "learning_rate": 9.67655743768862e-05, "loss": 0.0266, "step": 47810 }, { "epoch": 30.478011472275334, "grad_norm": 0.29879140853881836, "learning_rate": 9.676362396064263e-05, "loss": 0.0211, "step": 47820 }, { "epoch": 30.48438495857234, "grad_norm": 0.47534358501434326, "learning_rate": 9.676167297617686e-05, "loss": 0.0185, "step": 47830 }, { "epoch": 30.490758444869343, "grad_norm": 0.3677673637866974, "learning_rate": 9.675972142351264e-05, "loss": 0.0446, "step": 47840 }, { "epoch": 30.497131931166347, "grad_norm": 0.5857642889022827, "learning_rate": 9.675776930267364e-05, "loss": 0.0283, "step": 47850 }, { "epoch": 30.50350541746335, "grad_norm": 0.3543374538421631, "learning_rate": 9.67558166136836e-05, "loss": 0.0477, "step": 47860 }, { "epoch": 30.509878903760356, "grad_norm": 0.3732214570045471, "learning_rate": 9.675386335656624e-05, "loss": 0.0291, "step": 47870 }, { "epoch": 30.51625239005736, "grad_norm": 0.43065503239631653, "learning_rate": 9.67519095313453e-05, "loss": 0.0183, "step": 47880 }, { "epoch": 30.522625876354365, "grad_norm": 0.6613799929618835, "learning_rate": 9.674995513804452e-05, "loss": 0.0299, "step": 47890 }, { "epoch": 30.52899936265137, "grad_norm": 0.3799893856048584, "learning_rate": 9.674800017668765e-05, "loss": 0.0304, "step": 47900 }, { "epoch": 30.535372848948374, "grad_norm": 0.6712312698364258, "learning_rate": 9.674604464729844e-05, "loss": 0.0361, "step": 47910 }, { "epoch": 30.541746335245378, "grad_norm": 0.5331726670265198, "learning_rate": 9.674408854990064e-05, "loss": 0.0285, "step": 47920 }, { "epoch": 30.548119821542382, "grad_norm": 0.7846921682357788, "learning_rate": 9.674213188451804e-05, "loss": 0.0327, "step": 47930 }, { "epoch": 30.554493307839387, "grad_norm": 0.414529025554657, "learning_rate": 9.67401746511744e-05, "loss": 0.0283, "step": 47940 }, { "epoch": 30.56086679413639, "grad_norm": 0.5281225442886353, "learning_rate": 9.673821684989352e-05, "loss": 0.0289, "step": 47950 }, { "epoch": 30.567240280433396, "grad_norm": 0.6101946234703064, "learning_rate": 9.673625848069916e-05, "loss": 0.0224, "step": 47960 }, { "epoch": 30.5736137667304, "grad_norm": 0.4984627068042755, "learning_rate": 9.673429954361514e-05, "loss": 0.0316, "step": 47970 }, { "epoch": 30.579987253027404, "grad_norm": 0.525843620300293, "learning_rate": 9.673234003866526e-05, "loss": 0.0291, "step": 47980 }, { "epoch": 30.58636073932441, "grad_norm": 0.6882246732711792, "learning_rate": 9.673037996587333e-05, "loss": 0.0268, "step": 47990 }, { "epoch": 30.592734225621413, "grad_norm": 0.4233173727989197, "learning_rate": 9.672841932526315e-05, "loss": 0.0426, "step": 48000 }, { "epoch": 30.59910771191842, "grad_norm": 0.8273032903671265, "learning_rate": 9.672645811685857e-05, "loss": 0.0248, "step": 48010 }, { "epoch": 30.605481198215422, "grad_norm": 0.617384672164917, "learning_rate": 9.672449634068339e-05, "loss": 0.0301, "step": 48020 }, { "epoch": 30.61185468451243, "grad_norm": 0.6358820199966431, "learning_rate": 9.672253399676148e-05, "loss": 0.0378, "step": 48030 }, { "epoch": 30.618228170809434, "grad_norm": 0.41155993938446045, "learning_rate": 9.672057108511665e-05, "loss": 0.0319, "step": 48040 }, { "epoch": 30.62460165710644, "grad_norm": 0.6127799153327942, "learning_rate": 9.671860760577279e-05, "loss": 0.0313, "step": 48050 }, { "epoch": 30.630975143403443, "grad_norm": 0.5372704267501831, "learning_rate": 9.671664355875373e-05, "loss": 0.0345, "step": 48060 }, { "epoch": 30.637348629700448, "grad_norm": 0.6143667697906494, "learning_rate": 9.671467894408335e-05, "loss": 0.0288, "step": 48070 }, { "epoch": 30.643722115997452, "grad_norm": 0.6638042330741882, "learning_rate": 9.671271376178551e-05, "loss": 0.0431, "step": 48080 }, { "epoch": 30.650095602294456, "grad_norm": 0.566862940788269, "learning_rate": 9.671074801188408e-05, "loss": 0.0333, "step": 48090 }, { "epoch": 30.65646908859146, "grad_norm": 0.7406484484672546, "learning_rate": 9.670878169440297e-05, "loss": 0.0404, "step": 48100 }, { "epoch": 30.662842574888465, "grad_norm": 0.38298800587654114, "learning_rate": 9.670681480936606e-05, "loss": 0.0245, "step": 48110 }, { "epoch": 30.66921606118547, "grad_norm": 0.6056684255599976, "learning_rate": 9.670484735679725e-05, "loss": 0.0394, "step": 48120 }, { "epoch": 30.675589547482474, "grad_norm": 0.6433191895484924, "learning_rate": 9.670287933672046e-05, "loss": 0.0307, "step": 48130 }, { "epoch": 30.68196303377948, "grad_norm": 0.41848137974739075, "learning_rate": 9.670091074915959e-05, "loss": 0.0196, "step": 48140 }, { "epoch": 30.688336520076483, "grad_norm": 0.430083692073822, "learning_rate": 9.669894159413854e-05, "loss": 0.022, "step": 48150 }, { "epoch": 30.694710006373487, "grad_norm": 0.4374866187572479, "learning_rate": 9.669697187168127e-05, "loss": 0.0303, "step": 48160 }, { "epoch": 30.70108349267049, "grad_norm": 0.47613564133644104, "learning_rate": 9.669500158181172e-05, "loss": 0.0347, "step": 48170 }, { "epoch": 30.707456978967496, "grad_norm": 0.7548603415489197, "learning_rate": 9.66930307245538e-05, "loss": 0.0363, "step": 48180 }, { "epoch": 30.7138304652645, "grad_norm": 0.7053880095481873, "learning_rate": 9.669105929993147e-05, "loss": 0.0312, "step": 48190 }, { "epoch": 30.720203951561505, "grad_norm": 0.4214024841785431, "learning_rate": 9.668908730796868e-05, "loss": 0.0274, "step": 48200 }, { "epoch": 30.72657743785851, "grad_norm": 0.5566748976707458, "learning_rate": 9.668711474868941e-05, "loss": 0.0334, "step": 48210 }, { "epoch": 30.732950924155514, "grad_norm": 0.8187327980995178, "learning_rate": 9.668514162211761e-05, "loss": 0.0335, "step": 48220 }, { "epoch": 30.739324410452518, "grad_norm": 0.3835656940937042, "learning_rate": 9.668316792827725e-05, "loss": 0.0172, "step": 48230 }, { "epoch": 30.745697896749522, "grad_norm": 0.4533531367778778, "learning_rate": 9.668119366719235e-05, "loss": 0.0358, "step": 48240 }, { "epoch": 30.752071383046527, "grad_norm": 1.1595898866653442, "learning_rate": 9.667921883888685e-05, "loss": 0.0411, "step": 48250 }, { "epoch": 30.75844486934353, "grad_norm": 0.4471672773361206, "learning_rate": 9.667724344338478e-05, "loss": 0.039, "step": 48260 }, { "epoch": 30.764818355640536, "grad_norm": 0.4533998966217041, "learning_rate": 9.667526748071011e-05, "loss": 0.0374, "step": 48270 }, { "epoch": 30.77119184193754, "grad_norm": 0.4383496046066284, "learning_rate": 9.66732909508869e-05, "loss": 0.0223, "step": 48280 }, { "epoch": 30.777565328234544, "grad_norm": 0.4665463864803314, "learning_rate": 9.667131385393913e-05, "loss": 0.0256, "step": 48290 }, { "epoch": 30.78393881453155, "grad_norm": 0.48245003819465637, "learning_rate": 9.666933618989082e-05, "loss": 0.0339, "step": 48300 }, { "epoch": 30.790312300828553, "grad_norm": 0.5799371600151062, "learning_rate": 9.666735795876602e-05, "loss": 0.0197, "step": 48310 }, { "epoch": 30.796685787125558, "grad_norm": 0.4700395464897156, "learning_rate": 9.666537916058876e-05, "loss": 0.0271, "step": 48320 }, { "epoch": 30.803059273422562, "grad_norm": 0.769071102142334, "learning_rate": 9.666339979538308e-05, "loss": 0.0263, "step": 48330 }, { "epoch": 30.809432759719567, "grad_norm": 0.5273532271385193, "learning_rate": 9.666141986317305e-05, "loss": 0.0193, "step": 48340 }, { "epoch": 30.81580624601657, "grad_norm": 0.43907466530799866, "learning_rate": 9.66594393639827e-05, "loss": 0.0274, "step": 48350 }, { "epoch": 30.822179732313575, "grad_norm": 0.43947866559028625, "learning_rate": 9.66574582978361e-05, "loss": 0.0273, "step": 48360 }, { "epoch": 30.82855321861058, "grad_norm": 0.7551102638244629, "learning_rate": 9.665547666475734e-05, "loss": 0.0238, "step": 48370 }, { "epoch": 30.834926704907584, "grad_norm": 0.6669016480445862, "learning_rate": 9.66534944647705e-05, "loss": 0.0353, "step": 48380 }, { "epoch": 30.84130019120459, "grad_norm": 0.659733235836029, "learning_rate": 9.665151169789964e-05, "loss": 0.0287, "step": 48390 }, { "epoch": 30.847673677501593, "grad_norm": 0.34366127848625183, "learning_rate": 9.664952836416887e-05, "loss": 0.0295, "step": 48400 }, { "epoch": 30.854047163798597, "grad_norm": 0.5763676166534424, "learning_rate": 9.664754446360227e-05, "loss": 0.0306, "step": 48410 }, { "epoch": 30.8604206500956, "grad_norm": 0.45096004009246826, "learning_rate": 9.664555999622397e-05, "loss": 0.0205, "step": 48420 }, { "epoch": 30.866794136392606, "grad_norm": 0.5856727361679077, "learning_rate": 9.66435749620581e-05, "loss": 0.0285, "step": 48430 }, { "epoch": 30.87316762268961, "grad_norm": 0.6432901620864868, "learning_rate": 9.664158936112873e-05, "loss": 0.0362, "step": 48440 }, { "epoch": 30.879541108986615, "grad_norm": 0.34328076243400574, "learning_rate": 9.663960319346002e-05, "loss": 0.0254, "step": 48450 }, { "epoch": 30.88591459528362, "grad_norm": 0.4030042290687561, "learning_rate": 9.663761645907609e-05, "loss": 0.0302, "step": 48460 }, { "epoch": 30.892288081580624, "grad_norm": 0.41659975051879883, "learning_rate": 9.66356291580011e-05, "loss": 0.0387, "step": 48470 }, { "epoch": 30.898661567877628, "grad_norm": 0.5034846663475037, "learning_rate": 9.663364129025919e-05, "loss": 0.0307, "step": 48480 }, { "epoch": 30.905035054174633, "grad_norm": 0.5030421614646912, "learning_rate": 9.663165285587448e-05, "loss": 0.035, "step": 48490 }, { "epoch": 30.911408540471637, "grad_norm": 0.4905811548233032, "learning_rate": 9.662966385487118e-05, "loss": 0.0312, "step": 48500 }, { "epoch": 30.91778202676864, "grad_norm": 0.37994247674942017, "learning_rate": 9.662767428727343e-05, "loss": 0.0302, "step": 48510 }, { "epoch": 30.924155513065646, "grad_norm": 0.5185877084732056, "learning_rate": 9.662568415310543e-05, "loss": 0.0338, "step": 48520 }, { "epoch": 30.93052899936265, "grad_norm": 0.5717117190361023, "learning_rate": 9.662369345239132e-05, "loss": 0.0225, "step": 48530 }, { "epoch": 30.936902485659655, "grad_norm": 0.7184933423995972, "learning_rate": 9.662170218515533e-05, "loss": 0.0377, "step": 48540 }, { "epoch": 30.94327597195666, "grad_norm": 0.5517489314079285, "learning_rate": 9.661971035142164e-05, "loss": 0.0238, "step": 48550 }, { "epoch": 30.949649458253663, "grad_norm": 0.3884011209011078, "learning_rate": 9.661771795121444e-05, "loss": 0.0413, "step": 48560 }, { "epoch": 30.956022944550668, "grad_norm": 0.2566922605037689, "learning_rate": 9.661572498455798e-05, "loss": 0.0261, "step": 48570 }, { "epoch": 30.962396430847672, "grad_norm": 0.35354575514793396, "learning_rate": 9.661373145147643e-05, "loss": 0.0553, "step": 48580 }, { "epoch": 30.968769917144677, "grad_norm": 0.8680574297904968, "learning_rate": 9.661173735199402e-05, "loss": 0.0394, "step": 48590 }, { "epoch": 30.97514340344168, "grad_norm": 0.34516721963882446, "learning_rate": 9.660974268613501e-05, "loss": 0.025, "step": 48600 }, { "epoch": 30.981516889738685, "grad_norm": 0.3806225657463074, "learning_rate": 9.660774745392363e-05, "loss": 0.0274, "step": 48610 }, { "epoch": 30.98789037603569, "grad_norm": 0.884701132774353, "learning_rate": 9.66057516553841e-05, "loss": 0.0649, "step": 48620 }, { "epoch": 30.994263862332694, "grad_norm": 0.5785786509513855, "learning_rate": 9.660375529054068e-05, "loss": 0.0295, "step": 48630 }, { "epoch": 31.000637348629702, "grad_norm": 0.4000411033630371, "learning_rate": 9.660175835941763e-05, "loss": 0.0262, "step": 48640 }, { "epoch": 31.007010834926707, "grad_norm": 0.5221824049949646, "learning_rate": 9.659976086203923e-05, "loss": 0.0242, "step": 48650 }, { "epoch": 31.01338432122371, "grad_norm": 0.6969224810600281, "learning_rate": 9.659776279842972e-05, "loss": 0.0306, "step": 48660 }, { "epoch": 31.019757807520715, "grad_norm": 0.32996734976768494, "learning_rate": 9.659576416861342e-05, "loss": 0.0218, "step": 48670 }, { "epoch": 31.02613129381772, "grad_norm": 0.5652149319648743, "learning_rate": 9.659376497261457e-05, "loss": 0.0307, "step": 48680 }, { "epoch": 31.032504780114724, "grad_norm": 0.6288396716117859, "learning_rate": 9.65917652104575e-05, "loss": 0.0291, "step": 48690 }, { "epoch": 31.03887826641173, "grad_norm": 0.9513260722160339, "learning_rate": 9.658976488216648e-05, "loss": 0.036, "step": 48700 }, { "epoch": 31.045251752708733, "grad_norm": 0.40537938475608826, "learning_rate": 9.658776398776583e-05, "loss": 0.0283, "step": 48710 }, { "epoch": 31.051625239005737, "grad_norm": 0.5784873366355896, "learning_rate": 9.658576252727987e-05, "loss": 0.0272, "step": 48720 }, { "epoch": 31.057998725302742, "grad_norm": 0.425845742225647, "learning_rate": 9.65837605007329e-05, "loss": 0.0255, "step": 48730 }, { "epoch": 31.064372211599746, "grad_norm": 0.6609985828399658, "learning_rate": 9.658175790814927e-05, "loss": 0.0416, "step": 48740 }, { "epoch": 31.07074569789675, "grad_norm": 0.5502713322639465, "learning_rate": 9.65797547495533e-05, "loss": 0.0394, "step": 48750 }, { "epoch": 31.077119184193755, "grad_norm": 0.43580299615859985, "learning_rate": 9.657775102496933e-05, "loss": 0.0243, "step": 48760 }, { "epoch": 31.08349267049076, "grad_norm": 0.4713056683540344, "learning_rate": 9.65757467344217e-05, "loss": 0.028, "step": 48770 }, { "epoch": 31.089866156787764, "grad_norm": 0.5350761413574219, "learning_rate": 9.657374187793477e-05, "loss": 0.0302, "step": 48780 }, { "epoch": 31.09623964308477, "grad_norm": 0.6443381309509277, "learning_rate": 9.65717364555329e-05, "loss": 0.0203, "step": 48790 }, { "epoch": 31.102613129381773, "grad_norm": 0.6345484256744385, "learning_rate": 9.656973046724048e-05, "loss": 0.0217, "step": 48800 }, { "epoch": 31.108986615678777, "grad_norm": 0.6308213472366333, "learning_rate": 9.656772391308186e-05, "loss": 0.028, "step": 48810 }, { "epoch": 31.11536010197578, "grad_norm": 0.39703309535980225, "learning_rate": 9.656571679308141e-05, "loss": 0.0262, "step": 48820 }, { "epoch": 31.121733588272786, "grad_norm": 0.3152744770050049, "learning_rate": 9.656370910726355e-05, "loss": 0.0406, "step": 48830 }, { "epoch": 31.12810707456979, "grad_norm": 0.5273758172988892, "learning_rate": 9.656170085565269e-05, "loss": 0.037, "step": 48840 }, { "epoch": 31.134480560866795, "grad_norm": 0.47815024852752686, "learning_rate": 9.655969203827316e-05, "loss": 0.0305, "step": 48850 }, { "epoch": 31.1408540471638, "grad_norm": 0.7427622675895691, "learning_rate": 9.655768265514941e-05, "loss": 0.0353, "step": 48860 }, { "epoch": 31.147227533460804, "grad_norm": 0.32691600918769836, "learning_rate": 9.655567270630589e-05, "loss": 0.0314, "step": 48870 }, { "epoch": 31.153601019757808, "grad_norm": 0.7550122737884521, "learning_rate": 9.655366219176696e-05, "loss": 0.0277, "step": 48880 }, { "epoch": 31.159974506054812, "grad_norm": 0.4847235083580017, "learning_rate": 9.65516511115571e-05, "loss": 0.0366, "step": 48890 }, { "epoch": 31.166347992351817, "grad_norm": 0.5646427869796753, "learning_rate": 9.654963946570073e-05, "loss": 0.0253, "step": 48900 }, { "epoch": 31.17272147864882, "grad_norm": 0.41712501645088196, "learning_rate": 9.654762725422229e-05, "loss": 0.0399, "step": 48910 }, { "epoch": 31.179094964945826, "grad_norm": 0.462252676486969, "learning_rate": 9.654561447714621e-05, "loss": 0.0357, "step": 48920 }, { "epoch": 31.18546845124283, "grad_norm": 0.36987707018852234, "learning_rate": 9.654360113449698e-05, "loss": 0.0295, "step": 48930 }, { "epoch": 31.191841937539834, "grad_norm": 0.4034726619720459, "learning_rate": 9.654158722629906e-05, "loss": 0.0362, "step": 48940 }, { "epoch": 31.19821542383684, "grad_norm": 0.506635844707489, "learning_rate": 9.65395727525769e-05, "loss": 0.0268, "step": 48950 }, { "epoch": 31.204588910133843, "grad_norm": 0.7329914569854736, "learning_rate": 9.653755771335499e-05, "loss": 0.032, "step": 48960 }, { "epoch": 31.210962396430848, "grad_norm": 0.6020264029502869, "learning_rate": 9.653554210865783e-05, "loss": 0.0394, "step": 48970 }, { "epoch": 31.217335882727852, "grad_norm": 0.6978441476821899, "learning_rate": 9.653352593850988e-05, "loss": 0.0305, "step": 48980 }, { "epoch": 31.223709369024856, "grad_norm": 0.4542914032936096, "learning_rate": 9.653150920293568e-05, "loss": 0.0296, "step": 48990 }, { "epoch": 31.23008285532186, "grad_norm": 0.47818824648857117, "learning_rate": 9.652949190195967e-05, "loss": 0.0289, "step": 49000 }, { "epoch": 31.236456341618865, "grad_norm": 0.6142327785491943, "learning_rate": 9.652747403560644e-05, "loss": 0.0319, "step": 49010 }, { "epoch": 31.24282982791587, "grad_norm": 0.42436355352401733, "learning_rate": 9.652545560390045e-05, "loss": 0.0377, "step": 49020 }, { "epoch": 31.249203314212874, "grad_norm": 0.33607783913612366, "learning_rate": 9.652343660686626e-05, "loss": 0.0272, "step": 49030 }, { "epoch": 31.25557680050988, "grad_norm": 0.7621344923973083, "learning_rate": 9.652141704452837e-05, "loss": 0.0289, "step": 49040 }, { "epoch": 31.261950286806883, "grad_norm": 0.6974853873252869, "learning_rate": 9.651939691691137e-05, "loss": 0.0288, "step": 49050 }, { "epoch": 31.268323773103887, "grad_norm": 0.3906075954437256, "learning_rate": 9.651737622403974e-05, "loss": 0.0339, "step": 49060 }, { "epoch": 31.27469725940089, "grad_norm": 0.47291985154151917, "learning_rate": 9.651535496593809e-05, "loss": 0.0233, "step": 49070 }, { "epoch": 31.281070745697896, "grad_norm": 0.2513379454612732, "learning_rate": 9.651333314263096e-05, "loss": 0.0318, "step": 49080 }, { "epoch": 31.2874442319949, "grad_norm": 0.8028595447540283, "learning_rate": 9.651131075414292e-05, "loss": 0.0311, "step": 49090 }, { "epoch": 31.293817718291905, "grad_norm": 0.4388154447078705, "learning_rate": 9.650928780049853e-05, "loss": 0.0339, "step": 49100 }, { "epoch": 31.30019120458891, "grad_norm": 0.41119831800460815, "learning_rate": 9.650726428172239e-05, "loss": 0.0297, "step": 49110 }, { "epoch": 31.306564690885914, "grad_norm": 0.3754115700721741, "learning_rate": 9.650524019783906e-05, "loss": 0.0333, "step": 49120 }, { "epoch": 31.312938177182918, "grad_norm": 0.3825179636478424, "learning_rate": 9.650321554887318e-05, "loss": 0.0289, "step": 49130 }, { "epoch": 31.319311663479922, "grad_norm": 0.6352419257164001, "learning_rate": 9.65011903348493e-05, "loss": 0.0363, "step": 49140 }, { "epoch": 31.325685149776927, "grad_norm": 0.8752667903900146, "learning_rate": 9.649916455579206e-05, "loss": 0.0317, "step": 49150 }, { "epoch": 31.33205863607393, "grad_norm": 0.575877845287323, "learning_rate": 9.649713821172608e-05, "loss": 0.0234, "step": 49160 }, { "epoch": 31.338432122370936, "grad_norm": 0.4051464796066284, "learning_rate": 9.649511130267597e-05, "loss": 0.0307, "step": 49170 }, { "epoch": 31.34480560866794, "grad_norm": 0.32039958238601685, "learning_rate": 9.649308382866635e-05, "loss": 0.031, "step": 49180 }, { "epoch": 31.351179094964944, "grad_norm": 0.4514513313770294, "learning_rate": 9.649105578972186e-05, "loss": 0.0222, "step": 49190 }, { "epoch": 31.35755258126195, "grad_norm": 0.5966686010360718, "learning_rate": 9.648902718586717e-05, "loss": 0.0307, "step": 49200 }, { "epoch": 31.363926067558953, "grad_norm": 0.40468645095825195, "learning_rate": 9.648699801712688e-05, "loss": 0.0198, "step": 49210 }, { "epoch": 31.370299553855958, "grad_norm": 0.7375828623771667, "learning_rate": 9.648496828352569e-05, "loss": 0.0335, "step": 49220 }, { "epoch": 31.376673040152962, "grad_norm": 0.6003609299659729, "learning_rate": 9.648293798508823e-05, "loss": 0.0257, "step": 49230 }, { "epoch": 31.383046526449967, "grad_norm": 0.508283257484436, "learning_rate": 9.648090712183919e-05, "loss": 0.0287, "step": 49240 }, { "epoch": 31.38942001274697, "grad_norm": 0.3484508693218231, "learning_rate": 9.647887569380326e-05, "loss": 0.0257, "step": 49250 }, { "epoch": 31.395793499043975, "grad_norm": 0.5096694827079773, "learning_rate": 9.64768437010051e-05, "loss": 0.0362, "step": 49260 }, { "epoch": 31.402166985340983, "grad_norm": 0.3822021186351776, "learning_rate": 9.647481114346941e-05, "loss": 0.0237, "step": 49270 }, { "epoch": 31.408540471637988, "grad_norm": 0.6328350305557251, "learning_rate": 9.647277802122087e-05, "loss": 0.0289, "step": 49280 }, { "epoch": 31.414913957934992, "grad_norm": 0.3835970163345337, "learning_rate": 9.647074433428421e-05, "loss": 0.0291, "step": 49290 }, { "epoch": 31.421287444231996, "grad_norm": 0.6144000291824341, "learning_rate": 9.646871008268412e-05, "loss": 0.0254, "step": 49300 }, { "epoch": 31.427660930529, "grad_norm": 0.45338183641433716, "learning_rate": 9.646667526644535e-05, "loss": 0.0207, "step": 49310 }, { "epoch": 31.434034416826005, "grad_norm": 0.5927661657333374, "learning_rate": 9.64646398855926e-05, "loss": 0.0236, "step": 49320 }, { "epoch": 31.44040790312301, "grad_norm": 0.8147389888763428, "learning_rate": 9.646260394015059e-05, "loss": 0.0296, "step": 49330 }, { "epoch": 31.446781389420014, "grad_norm": 0.7373713254928589, "learning_rate": 9.646056743014407e-05, "loss": 0.0262, "step": 49340 }, { "epoch": 31.45315487571702, "grad_norm": 0.3558279275894165, "learning_rate": 9.645853035559781e-05, "loss": 0.0354, "step": 49350 }, { "epoch": 31.459528362014023, "grad_norm": 0.5918961763381958, "learning_rate": 9.645649271653654e-05, "loss": 0.0371, "step": 49360 }, { "epoch": 31.465901848311027, "grad_norm": 0.32327085733413696, "learning_rate": 9.645445451298501e-05, "loss": 0.0251, "step": 49370 }, { "epoch": 31.47227533460803, "grad_norm": 0.26392844319343567, "learning_rate": 9.6452415744968e-05, "loss": 0.0222, "step": 49380 }, { "epoch": 31.478648820905036, "grad_norm": 0.5202863216400146, "learning_rate": 9.645037641251029e-05, "loss": 0.0385, "step": 49390 }, { "epoch": 31.48502230720204, "grad_norm": 1.6697717905044556, "learning_rate": 9.644833651563664e-05, "loss": 0.0412, "step": 49400 }, { "epoch": 31.491395793499045, "grad_norm": 0.4944114089012146, "learning_rate": 9.644629605437186e-05, "loss": 0.0289, "step": 49410 }, { "epoch": 31.49776927979605, "grad_norm": 0.5295724868774414, "learning_rate": 9.64442550287407e-05, "loss": 0.0276, "step": 49420 }, { "epoch": 31.504142766093054, "grad_norm": 0.4605574607849121, "learning_rate": 9.644221343876801e-05, "loss": 0.0268, "step": 49430 }, { "epoch": 31.510516252390058, "grad_norm": 0.35252270102500916, "learning_rate": 9.644017128447859e-05, "loss": 0.0241, "step": 49440 }, { "epoch": 31.516889738687063, "grad_norm": 0.6916472315788269, "learning_rate": 9.643812856589723e-05, "loss": 0.0319, "step": 49450 }, { "epoch": 31.523263224984067, "grad_norm": 0.5201977491378784, "learning_rate": 9.643608528304877e-05, "loss": 0.0324, "step": 49460 }, { "epoch": 31.52963671128107, "grad_norm": 0.34854909777641296, "learning_rate": 9.643404143595801e-05, "loss": 0.0199, "step": 49470 }, { "epoch": 31.536010197578076, "grad_norm": 0.40513482689857483, "learning_rate": 9.643199702464982e-05, "loss": 0.0243, "step": 49480 }, { "epoch": 31.54238368387508, "grad_norm": 0.4340430796146393, "learning_rate": 9.642995204914902e-05, "loss": 0.0225, "step": 49490 }, { "epoch": 31.548757170172085, "grad_norm": 1.0601682662963867, "learning_rate": 9.642790650948048e-05, "loss": 0.0359, "step": 49500 }, { "epoch": 31.55513065646909, "grad_norm": 0.6526304483413696, "learning_rate": 9.642586040566903e-05, "loss": 0.0414, "step": 49510 }, { "epoch": 31.561504142766093, "grad_norm": 0.7227376103401184, "learning_rate": 9.642381373773955e-05, "loss": 0.0393, "step": 49520 }, { "epoch": 31.567877629063098, "grad_norm": 0.40774422883987427, "learning_rate": 9.642176650571691e-05, "loss": 0.0333, "step": 49530 }, { "epoch": 31.574251115360102, "grad_norm": 0.48242467641830444, "learning_rate": 9.641971870962597e-05, "loss": 0.0397, "step": 49540 }, { "epoch": 31.580624601657107, "grad_norm": 0.6240224242210388, "learning_rate": 9.641767034949161e-05, "loss": 0.0298, "step": 49550 }, { "epoch": 31.58699808795411, "grad_norm": 0.4099906086921692, "learning_rate": 9.641562142533874e-05, "loss": 0.0357, "step": 49560 }, { "epoch": 31.593371574251115, "grad_norm": 0.23491650819778442, "learning_rate": 9.641357193719223e-05, "loss": 0.0283, "step": 49570 }, { "epoch": 31.59974506054812, "grad_norm": 0.35146304965019226, "learning_rate": 9.6411521885077e-05, "loss": 0.0317, "step": 49580 }, { "epoch": 31.606118546845124, "grad_norm": 0.5646169185638428, "learning_rate": 9.640947126901798e-05, "loss": 0.0372, "step": 49590 }, { "epoch": 31.61249203314213, "grad_norm": 0.3802034556865692, "learning_rate": 9.640742008904005e-05, "loss": 0.041, "step": 49600 }, { "epoch": 31.618865519439133, "grad_norm": 0.3893008530139923, "learning_rate": 9.640536834516815e-05, "loss": 0.0363, "step": 49610 }, { "epoch": 31.625239005736137, "grad_norm": 0.6261109709739685, "learning_rate": 9.640331603742723e-05, "loss": 0.0341, "step": 49620 }, { "epoch": 31.631612492033142, "grad_norm": 0.5088911056518555, "learning_rate": 9.640126316584219e-05, "loss": 0.0392, "step": 49630 }, { "epoch": 31.637985978330146, "grad_norm": 0.9107030034065247, "learning_rate": 9.639920973043801e-05, "loss": 0.0324, "step": 49640 }, { "epoch": 31.64435946462715, "grad_norm": 0.9197406768798828, "learning_rate": 9.639715573123961e-05, "loss": 0.0286, "step": 49650 }, { "epoch": 31.650732950924155, "grad_norm": 0.39591729640960693, "learning_rate": 9.639510116827196e-05, "loss": 0.0288, "step": 49660 }, { "epoch": 31.65710643722116, "grad_norm": 0.3422752320766449, "learning_rate": 9.639304604156003e-05, "loss": 0.0285, "step": 49670 }, { "epoch": 31.663479923518164, "grad_norm": 0.4514029026031494, "learning_rate": 9.63909903511288e-05, "loss": 0.024, "step": 49680 }, { "epoch": 31.66985340981517, "grad_norm": 0.5566706657409668, "learning_rate": 9.638893409700321e-05, "loss": 0.0432, "step": 49690 }, { "epoch": 31.676226896112173, "grad_norm": 0.6060683131217957, "learning_rate": 9.638687727920828e-05, "loss": 0.0281, "step": 49700 }, { "epoch": 31.682600382409177, "grad_norm": 0.6065380573272705, "learning_rate": 9.638481989776901e-05, "loss": 0.0321, "step": 49710 }, { "epoch": 31.68897386870618, "grad_norm": 0.5228275656700134, "learning_rate": 9.638276195271037e-05, "loss": 0.0246, "step": 49720 }, { "epoch": 31.695347355003186, "grad_norm": 0.442179411649704, "learning_rate": 9.638070344405738e-05, "loss": 0.025, "step": 49730 }, { "epoch": 31.70172084130019, "grad_norm": 0.4860314130783081, "learning_rate": 9.637864437183507e-05, "loss": 0.0332, "step": 49740 }, { "epoch": 31.708094327597195, "grad_norm": 0.5204603672027588, "learning_rate": 9.637658473606843e-05, "loss": 0.0288, "step": 49750 }, { "epoch": 31.7144678138942, "grad_norm": 0.4228981137275696, "learning_rate": 9.637452453678251e-05, "loss": 0.0262, "step": 49760 }, { "epoch": 31.720841300191204, "grad_norm": 0.43562695384025574, "learning_rate": 9.63724637740023e-05, "loss": 0.0276, "step": 49770 }, { "epoch": 31.727214786488208, "grad_norm": 0.6446609497070312, "learning_rate": 9.637040244775291e-05, "loss": 0.0247, "step": 49780 }, { "epoch": 31.733588272785212, "grad_norm": 0.5088351964950562, "learning_rate": 9.636834055805933e-05, "loss": 0.0243, "step": 49790 }, { "epoch": 31.739961759082217, "grad_norm": 0.6405149102210999, "learning_rate": 9.636627810494663e-05, "loss": 0.0227, "step": 49800 }, { "epoch": 31.74633524537922, "grad_norm": 0.6196313500404358, "learning_rate": 9.636421508843988e-05, "loss": 0.0231, "step": 49810 }, { "epoch": 31.752708731676226, "grad_norm": 0.8192743062973022, "learning_rate": 9.636215150856414e-05, "loss": 0.0289, "step": 49820 }, { "epoch": 31.75908221797323, "grad_norm": 0.43991485238075256, "learning_rate": 9.636008736534448e-05, "loss": 0.0329, "step": 49830 }, { "epoch": 31.765455704270234, "grad_norm": 0.7172949910163879, "learning_rate": 9.6358022658806e-05, "loss": 0.0247, "step": 49840 }, { "epoch": 31.77182919056724, "grad_norm": 0.4867013394832611, "learning_rate": 9.635595738897377e-05, "loss": 0.0273, "step": 49850 }, { "epoch": 31.778202676864243, "grad_norm": 0.5021018981933594, "learning_rate": 9.635389155587289e-05, "loss": 0.0366, "step": 49860 }, { "epoch": 31.784576163161248, "grad_norm": 0.42545363306999207, "learning_rate": 9.635182515952847e-05, "loss": 0.0263, "step": 49870 }, { "epoch": 31.790949649458252, "grad_norm": 0.2706247568130493, "learning_rate": 9.63497581999656e-05, "loss": 0.0219, "step": 49880 }, { "epoch": 31.797323135755256, "grad_norm": 0.5228594541549683, "learning_rate": 9.634769067720941e-05, "loss": 0.032, "step": 49890 }, { "epoch": 31.803696622052264, "grad_norm": 0.42217695713043213, "learning_rate": 9.634562259128502e-05, "loss": 0.04, "step": 49900 }, { "epoch": 31.81007010834927, "grad_norm": 0.4393623173236847, "learning_rate": 9.634355394221756e-05, "loss": 0.0306, "step": 49910 }, { "epoch": 31.816443594646273, "grad_norm": 0.3546094596385956, "learning_rate": 9.634148473003216e-05, "loss": 0.0201, "step": 49920 }, { "epoch": 31.822817080943278, "grad_norm": 0.3944132328033447, "learning_rate": 9.633941495475396e-05, "loss": 0.0263, "step": 49930 }, { "epoch": 31.829190567240282, "grad_norm": 0.441276490688324, "learning_rate": 9.633734461640813e-05, "loss": 0.0178, "step": 49940 }, { "epoch": 31.835564053537286, "grad_norm": 0.3799179494380951, "learning_rate": 9.63352737150198e-05, "loss": 0.0268, "step": 49950 }, { "epoch": 31.84193753983429, "grad_norm": 0.46664437651634216, "learning_rate": 9.633320225061417e-05, "loss": 0.0354, "step": 49960 }, { "epoch": 31.848311026131295, "grad_norm": 0.2994579076766968, "learning_rate": 9.633113022321636e-05, "loss": 0.031, "step": 49970 }, { "epoch": 31.8546845124283, "grad_norm": 0.5496512651443481, "learning_rate": 9.632905763285158e-05, "loss": 0.0224, "step": 49980 }, { "epoch": 31.861057998725304, "grad_norm": 0.2695121467113495, "learning_rate": 9.632698447954503e-05, "loss": 0.022, "step": 49990 }, { "epoch": 31.86743148502231, "grad_norm": 0.4434918761253357, "learning_rate": 9.632491076332185e-05, "loss": 0.0284, "step": 50000 }, { "epoch": 31.873804971319313, "grad_norm": 0.5567672848701477, "learning_rate": 9.632283648420728e-05, "loss": 0.0339, "step": 50010 }, { "epoch": 31.880178457616317, "grad_norm": 0.5594049096107483, "learning_rate": 9.632076164222651e-05, "loss": 0.0381, "step": 50020 }, { "epoch": 31.88655194391332, "grad_norm": 0.32747602462768555, "learning_rate": 9.631868623740475e-05, "loss": 0.0192, "step": 50030 }, { "epoch": 31.892925430210326, "grad_norm": 0.41065672039985657, "learning_rate": 9.631661026976722e-05, "loss": 0.0282, "step": 50040 }, { "epoch": 31.89929891650733, "grad_norm": 0.44742387533187866, "learning_rate": 9.631453373933914e-05, "loss": 0.0371, "step": 50050 }, { "epoch": 31.905672402804335, "grad_norm": 0.5311201810836792, "learning_rate": 9.631245664614576e-05, "loss": 0.0387, "step": 50060 }, { "epoch": 31.91204588910134, "grad_norm": 0.6947133541107178, "learning_rate": 9.631037899021229e-05, "loss": 0.0325, "step": 50070 }, { "epoch": 31.918419375398344, "grad_norm": 0.526229977607727, "learning_rate": 9.630830077156401e-05, "loss": 0.0218, "step": 50080 }, { "epoch": 31.924792861695348, "grad_norm": 0.6932626962661743, "learning_rate": 9.630622199022614e-05, "loss": 0.0275, "step": 50090 }, { "epoch": 31.931166347992352, "grad_norm": 0.41859298944473267, "learning_rate": 9.630414264622397e-05, "loss": 0.0208, "step": 50100 }, { "epoch": 31.937539834289357, "grad_norm": 0.3434845507144928, "learning_rate": 9.630206273958272e-05, "loss": 0.0305, "step": 50110 }, { "epoch": 31.94391332058636, "grad_norm": 0.47382214665412903, "learning_rate": 9.62999822703277e-05, "loss": 0.0268, "step": 50120 }, { "epoch": 31.950286806883366, "grad_norm": 0.637179970741272, "learning_rate": 9.62979012384842e-05, "loss": 0.0285, "step": 50130 }, { "epoch": 31.95666029318037, "grad_norm": 0.5213580131530762, "learning_rate": 9.629581964407746e-05, "loss": 0.0279, "step": 50140 }, { "epoch": 31.963033779477374, "grad_norm": 1.0581862926483154, "learning_rate": 9.629373748713281e-05, "loss": 0.024, "step": 50150 }, { "epoch": 31.96940726577438, "grad_norm": 0.5908228754997253, "learning_rate": 9.629165476767555e-05, "loss": 0.0258, "step": 50160 }, { "epoch": 31.975780752071383, "grad_norm": 0.6178209185600281, "learning_rate": 9.628957148573098e-05, "loss": 0.0234, "step": 50170 }, { "epoch": 31.982154238368388, "grad_norm": 0.44477248191833496, "learning_rate": 9.628748764132441e-05, "loss": 0.0359, "step": 50180 }, { "epoch": 31.988527724665392, "grad_norm": 0.509045422077179, "learning_rate": 9.628540323448114e-05, "loss": 0.0273, "step": 50190 }, { "epoch": 31.994901210962396, "grad_norm": 0.404618501663208, "learning_rate": 9.628331826522654e-05, "loss": 0.0324, "step": 50200 }, { "epoch": 32.001274697259404, "grad_norm": 0.3674522638320923, "learning_rate": 9.628123273358594e-05, "loss": 0.0415, "step": 50210 }, { "epoch": 32.007648183556405, "grad_norm": 0.4915797710418701, "learning_rate": 9.627914663958463e-05, "loss": 0.0279, "step": 50220 }, { "epoch": 32.01402166985341, "grad_norm": 0.5052920579910278, "learning_rate": 9.627705998324802e-05, "loss": 0.0204, "step": 50230 }, { "epoch": 32.020395156150414, "grad_norm": 0.3605380058288574, "learning_rate": 9.627497276460142e-05, "loss": 0.0294, "step": 50240 }, { "epoch": 32.02676864244742, "grad_norm": 0.31230372190475464, "learning_rate": 9.627288498367022e-05, "loss": 0.0274, "step": 50250 }, { "epoch": 32.03314212874442, "grad_norm": 0.3686187267303467, "learning_rate": 9.627079664047979e-05, "loss": 0.02, "step": 50260 }, { "epoch": 32.03951561504143, "grad_norm": 0.511961042881012, "learning_rate": 9.626870773505548e-05, "loss": 0.0274, "step": 50270 }, { "epoch": 32.04588910133843, "grad_norm": 0.6962223052978516, "learning_rate": 9.62666182674227e-05, "loss": 0.0363, "step": 50280 }, { "epoch": 32.05226258763544, "grad_norm": 0.4706366956233978, "learning_rate": 9.626452823760681e-05, "loss": 0.0371, "step": 50290 }, { "epoch": 32.05863607393244, "grad_norm": 0.6609902381896973, "learning_rate": 9.626243764563321e-05, "loss": 0.0296, "step": 50300 }, { "epoch": 32.06500956022945, "grad_norm": 0.4340435564517975, "learning_rate": 9.626034649152733e-05, "loss": 0.0263, "step": 50310 }, { "epoch": 32.07138304652645, "grad_norm": 0.632668137550354, "learning_rate": 9.625825477531456e-05, "loss": 0.0258, "step": 50320 }, { "epoch": 32.07775653282346, "grad_norm": 0.5861436128616333, "learning_rate": 9.625616249702032e-05, "loss": 0.0415, "step": 50330 }, { "epoch": 32.08413001912046, "grad_norm": 0.40851935744285583, "learning_rate": 9.625406965667003e-05, "loss": 0.0309, "step": 50340 }, { "epoch": 32.090503505417466, "grad_norm": 0.5697430372238159, "learning_rate": 9.625197625428913e-05, "loss": 0.0297, "step": 50350 }, { "epoch": 32.09687699171447, "grad_norm": 0.4416266679763794, "learning_rate": 9.624988228990305e-05, "loss": 0.0264, "step": 50360 }, { "epoch": 32.103250478011475, "grad_norm": 0.8499746322631836, "learning_rate": 9.624778776353722e-05, "loss": 0.0295, "step": 50370 }, { "epoch": 32.109623964308476, "grad_norm": 0.3335327208042145, "learning_rate": 9.624569267521712e-05, "loss": 0.0236, "step": 50380 }, { "epoch": 32.115997450605484, "grad_norm": 0.3305918872356415, "learning_rate": 9.624359702496819e-05, "loss": 0.0269, "step": 50390 }, { "epoch": 32.122370936902485, "grad_norm": 0.46877411007881165, "learning_rate": 9.624150081281589e-05, "loss": 0.0357, "step": 50400 }, { "epoch": 32.12874442319949, "grad_norm": 0.4583114981651306, "learning_rate": 9.623940403878569e-05, "loss": 0.0304, "step": 50410 }, { "epoch": 32.13511790949649, "grad_norm": 0.4105294644832611, "learning_rate": 9.623730670290307e-05, "loss": 0.0304, "step": 50420 }, { "epoch": 32.1414913957935, "grad_norm": 0.596894383430481, "learning_rate": 9.623520880519352e-05, "loss": 0.0278, "step": 50430 }, { "epoch": 32.1478648820905, "grad_norm": 0.6973862648010254, "learning_rate": 9.623311034568254e-05, "loss": 0.0235, "step": 50440 }, { "epoch": 32.15423836838751, "grad_norm": 0.38368988037109375, "learning_rate": 9.623101132439561e-05, "loss": 0.0222, "step": 50450 }, { "epoch": 32.16061185468451, "grad_norm": 0.3807137906551361, "learning_rate": 9.622891174135825e-05, "loss": 0.0261, "step": 50460 }, { "epoch": 32.16698534098152, "grad_norm": 0.6037473678588867, "learning_rate": 9.622681159659597e-05, "loss": 0.0314, "step": 50470 }, { "epoch": 32.17335882727852, "grad_norm": 0.45340391993522644, "learning_rate": 9.622471089013427e-05, "loss": 0.0205, "step": 50480 }, { "epoch": 32.17973231357553, "grad_norm": 0.3425014019012451, "learning_rate": 9.622260962199869e-05, "loss": 0.0234, "step": 50490 }, { "epoch": 32.18610579987253, "grad_norm": 0.4424443244934082, "learning_rate": 9.622050779221477e-05, "loss": 0.024, "step": 50500 }, { "epoch": 32.19247928616954, "grad_norm": 0.5424277782440186, "learning_rate": 9.621840540080805e-05, "loss": 0.0309, "step": 50510 }, { "epoch": 32.19885277246654, "grad_norm": 0.7074220180511475, "learning_rate": 9.621630244780404e-05, "loss": 0.0236, "step": 50520 }, { "epoch": 32.205226258763545, "grad_norm": 0.9161022901535034, "learning_rate": 9.621419893322834e-05, "loss": 0.0443, "step": 50530 }, { "epoch": 32.211599745060546, "grad_norm": 0.48749053478240967, "learning_rate": 9.621209485710647e-05, "loss": 0.03, "step": 50540 }, { "epoch": 32.217973231357554, "grad_norm": 0.8491482734680176, "learning_rate": 9.620999021946404e-05, "loss": 0.0275, "step": 50550 }, { "epoch": 32.224346717654555, "grad_norm": 0.2147703766822815, "learning_rate": 9.620788502032658e-05, "loss": 0.0201, "step": 50560 }, { "epoch": 32.23072020395156, "grad_norm": 0.4647914469242096, "learning_rate": 9.62057792597197e-05, "loss": 0.0328, "step": 50570 }, { "epoch": 32.237093690248564, "grad_norm": 0.9474709630012512, "learning_rate": 9.620367293766898e-05, "loss": 0.0281, "step": 50580 }, { "epoch": 32.24346717654557, "grad_norm": 0.49123984575271606, "learning_rate": 9.620156605420001e-05, "loss": 0.028, "step": 50590 }, { "epoch": 32.24984066284257, "grad_norm": 0.3277948200702667, "learning_rate": 9.619945860933839e-05, "loss": 0.035, "step": 50600 }, { "epoch": 32.25621414913958, "grad_norm": 0.6156045198440552, "learning_rate": 9.619735060310972e-05, "loss": 0.0205, "step": 50610 }, { "epoch": 32.26258763543658, "grad_norm": 0.44093379378318787, "learning_rate": 9.619524203553962e-05, "loss": 0.0226, "step": 50620 }, { "epoch": 32.26896112173359, "grad_norm": 0.3858446478843689, "learning_rate": 9.619313290665373e-05, "loss": 0.0361, "step": 50630 }, { "epoch": 32.27533460803059, "grad_norm": 0.6209191083908081, "learning_rate": 9.619102321647765e-05, "loss": 0.0273, "step": 50640 }, { "epoch": 32.2817080943276, "grad_norm": 0.7387955784797668, "learning_rate": 9.618891296503702e-05, "loss": 0.0465, "step": 50650 }, { "epoch": 32.2880815806246, "grad_norm": 0.6510863304138184, "learning_rate": 9.618680215235751e-05, "loss": 0.0319, "step": 50660 }, { "epoch": 32.29445506692161, "grad_norm": 0.5081324577331543, "learning_rate": 9.618469077846473e-05, "loss": 0.0232, "step": 50670 }, { "epoch": 32.30082855321861, "grad_norm": 0.5236969590187073, "learning_rate": 9.618257884338435e-05, "loss": 0.0409, "step": 50680 }, { "epoch": 32.307202039515616, "grad_norm": 0.7517256736755371, "learning_rate": 9.618046634714205e-05, "loss": 0.0314, "step": 50690 }, { "epoch": 32.31357552581262, "grad_norm": 0.3475639522075653, "learning_rate": 9.617835328976345e-05, "loss": 0.0273, "step": 50700 }, { "epoch": 32.319949012109625, "grad_norm": 0.38038545846939087, "learning_rate": 9.617623967127429e-05, "loss": 0.0228, "step": 50710 }, { "epoch": 32.326322498406626, "grad_norm": 0.3361956477165222, "learning_rate": 9.61741254917002e-05, "loss": 0.0277, "step": 50720 }, { "epoch": 32.33269598470363, "grad_norm": 0.5007058382034302, "learning_rate": 9.61720107510669e-05, "loss": 0.0289, "step": 50730 }, { "epoch": 32.339069471000634, "grad_norm": 0.45574307441711426, "learning_rate": 9.616989544940009e-05, "loss": 0.0255, "step": 50740 }, { "epoch": 32.34544295729764, "grad_norm": 0.6038351058959961, "learning_rate": 9.616777958672543e-05, "loss": 0.0381, "step": 50750 }, { "epoch": 32.35181644359464, "grad_norm": 0.5222229957580566, "learning_rate": 9.616566316306867e-05, "loss": 0.0316, "step": 50760 }, { "epoch": 32.35818992989165, "grad_norm": 0.3345721662044525, "learning_rate": 9.616354617845552e-05, "loss": 0.0377, "step": 50770 }, { "epoch": 32.36456341618865, "grad_norm": 0.40095987915992737, "learning_rate": 9.616142863291169e-05, "loss": 0.0253, "step": 50780 }, { "epoch": 32.37093690248566, "grad_norm": 0.9229083061218262, "learning_rate": 9.615931052646293e-05, "loss": 0.0407, "step": 50790 }, { "epoch": 32.37731038878266, "grad_norm": 0.5151830315589905, "learning_rate": 9.615719185913496e-05, "loss": 0.0293, "step": 50800 }, { "epoch": 32.38368387507967, "grad_norm": 0.2776039242744446, "learning_rate": 9.615507263095353e-05, "loss": 0.0349, "step": 50810 }, { "epoch": 32.39005736137667, "grad_norm": 0.5033051371574402, "learning_rate": 9.61529528419444e-05, "loss": 0.0267, "step": 50820 }, { "epoch": 32.39643084767368, "grad_norm": 0.544008731842041, "learning_rate": 9.615083249213329e-05, "loss": 0.029, "step": 50830 }, { "epoch": 32.40280433397068, "grad_norm": 0.5722498297691345, "learning_rate": 9.614871158154601e-05, "loss": 0.026, "step": 50840 }, { "epoch": 32.409177820267686, "grad_norm": 0.46081364154815674, "learning_rate": 9.61465901102083e-05, "loss": 0.0224, "step": 50850 }, { "epoch": 32.415551306564694, "grad_norm": 0.3459358513355255, "learning_rate": 9.614446807814597e-05, "loss": 0.0283, "step": 50860 }, { "epoch": 32.421924792861695, "grad_norm": 0.33563676476478577, "learning_rate": 9.614234548538477e-05, "loss": 0.0316, "step": 50870 }, { "epoch": 32.4282982791587, "grad_norm": 0.3282836973667145, "learning_rate": 9.614022233195051e-05, "loss": 0.0232, "step": 50880 }, { "epoch": 32.434671765455704, "grad_norm": 0.44209009408950806, "learning_rate": 9.613809861786899e-05, "loss": 0.0357, "step": 50890 }, { "epoch": 32.44104525175271, "grad_norm": 0.6703401207923889, "learning_rate": 9.613597434316602e-05, "loss": 0.0256, "step": 50900 }, { "epoch": 32.44741873804971, "grad_norm": 0.4466400742530823, "learning_rate": 9.613384950786738e-05, "loss": 0.0292, "step": 50910 }, { "epoch": 32.45379222434672, "grad_norm": 0.49755531549453735, "learning_rate": 9.613172411199891e-05, "loss": 0.0291, "step": 50920 }, { "epoch": 32.46016571064372, "grad_norm": 0.354056179523468, "learning_rate": 9.612959815558645e-05, "loss": 0.0261, "step": 50930 }, { "epoch": 32.46653919694073, "grad_norm": 0.43406248092651367, "learning_rate": 9.612747163865582e-05, "loss": 0.025, "step": 50940 }, { "epoch": 32.47291268323773, "grad_norm": 0.32500842213630676, "learning_rate": 9.612534456123285e-05, "loss": 0.0256, "step": 50950 }, { "epoch": 32.47928616953474, "grad_norm": 0.2974274456501007, "learning_rate": 9.612321692334341e-05, "loss": 0.0306, "step": 50960 }, { "epoch": 32.48565965583174, "grad_norm": 0.5435872673988342, "learning_rate": 9.612108872501331e-05, "loss": 0.0419, "step": 50970 }, { "epoch": 32.49203314212875, "grad_norm": 0.2850504517555237, "learning_rate": 9.611895996626845e-05, "loss": 0.0225, "step": 50980 }, { "epoch": 32.49840662842575, "grad_norm": 0.564734160900116, "learning_rate": 9.611683064713469e-05, "loss": 0.0297, "step": 50990 }, { "epoch": 32.504780114722756, "grad_norm": 0.4351910650730133, "learning_rate": 9.611470076763788e-05, "loss": 0.0272, "step": 51000 }, { "epoch": 32.51115360101976, "grad_norm": 0.4046146869659424, "learning_rate": 9.611257032780394e-05, "loss": 0.035, "step": 51010 }, { "epoch": 32.517527087316765, "grad_norm": 0.4468485116958618, "learning_rate": 9.611043932765872e-05, "loss": 0.0279, "step": 51020 }, { "epoch": 32.523900573613766, "grad_norm": 0.40404176712036133, "learning_rate": 9.610830776722812e-05, "loss": 0.0288, "step": 51030 }, { "epoch": 32.530274059910774, "grad_norm": 0.21042655408382416, "learning_rate": 9.610617564653805e-05, "loss": 0.0247, "step": 51040 }, { "epoch": 32.536647546207774, "grad_norm": 0.29448527097702026, "learning_rate": 9.610404296561441e-05, "loss": 0.0273, "step": 51050 }, { "epoch": 32.54302103250478, "grad_norm": 0.7024926543235779, "learning_rate": 9.610190972448314e-05, "loss": 0.0216, "step": 51060 }, { "epoch": 32.54939451880178, "grad_norm": 0.39891892671585083, "learning_rate": 9.609977592317012e-05, "loss": 0.0323, "step": 51070 }, { "epoch": 32.55576800509879, "grad_norm": 0.34664294123649597, "learning_rate": 9.609764156170132e-05, "loss": 0.0197, "step": 51080 }, { "epoch": 32.56214149139579, "grad_norm": 0.6028919219970703, "learning_rate": 9.609550664010263e-05, "loss": 0.0254, "step": 51090 }, { "epoch": 32.5685149776928, "grad_norm": 0.5032874941825867, "learning_rate": 9.60933711584e-05, "loss": 0.0275, "step": 51100 }, { "epoch": 32.5748884639898, "grad_norm": 0.800987720489502, "learning_rate": 9.609123511661941e-05, "loss": 0.0366, "step": 51110 }, { "epoch": 32.58126195028681, "grad_norm": 0.4672933518886566, "learning_rate": 9.60890985147868e-05, "loss": 0.0417, "step": 51120 }, { "epoch": 32.58763543658381, "grad_norm": 0.5694705843925476, "learning_rate": 9.608696135292812e-05, "loss": 0.0278, "step": 51130 }, { "epoch": 32.59400892288082, "grad_norm": 0.47093504667282104, "learning_rate": 9.608482363106935e-05, "loss": 0.0255, "step": 51140 }, { "epoch": 32.60038240917782, "grad_norm": 0.5619297623634338, "learning_rate": 9.608268534923646e-05, "loss": 0.0222, "step": 51150 }, { "epoch": 32.606755895474826, "grad_norm": 0.38592392206192017, "learning_rate": 9.608054650745543e-05, "loss": 0.029, "step": 51160 }, { "epoch": 32.61312938177183, "grad_norm": 0.39945876598358154, "learning_rate": 9.607840710575226e-05, "loss": 0.0571, "step": 51170 }, { "epoch": 32.619502868068835, "grad_norm": 0.37257975339889526, "learning_rate": 9.607626714415294e-05, "loss": 0.0308, "step": 51180 }, { "epoch": 32.625876354365836, "grad_norm": 0.30795133113861084, "learning_rate": 9.607412662268346e-05, "loss": 0.0288, "step": 51190 }, { "epoch": 32.632249840662844, "grad_norm": 0.404982328414917, "learning_rate": 9.607198554136985e-05, "loss": 0.0286, "step": 51200 }, { "epoch": 32.638623326959845, "grad_norm": 0.5144482851028442, "learning_rate": 9.606984390023812e-05, "loss": 0.0297, "step": 51210 }, { "epoch": 32.64499681325685, "grad_norm": 0.49482566118240356, "learning_rate": 9.606770169931427e-05, "loss": 0.0469, "step": 51220 }, { "epoch": 32.651370299553854, "grad_norm": 1.3468883037567139, "learning_rate": 9.606555893862437e-05, "loss": 0.0313, "step": 51230 }, { "epoch": 32.65774378585086, "grad_norm": 0.5524077415466309, "learning_rate": 9.606341561819443e-05, "loss": 0.0287, "step": 51240 }, { "epoch": 32.66411727214786, "grad_norm": 0.5672935843467712, "learning_rate": 9.606127173805048e-05, "loss": 0.0275, "step": 51250 }, { "epoch": 32.67049075844487, "grad_norm": 0.3613284230232239, "learning_rate": 9.605912729821862e-05, "loss": 0.0292, "step": 51260 }, { "epoch": 32.67686424474187, "grad_norm": 0.7732415199279785, "learning_rate": 9.605698229872487e-05, "loss": 0.026, "step": 51270 }, { "epoch": 32.68323773103888, "grad_norm": 0.4910607635974884, "learning_rate": 9.605483673959528e-05, "loss": 0.0364, "step": 51280 }, { "epoch": 32.68961121733588, "grad_norm": 0.7407321929931641, "learning_rate": 9.605269062085596e-05, "loss": 0.0261, "step": 51290 }, { "epoch": 32.69598470363289, "grad_norm": 0.4703577756881714, "learning_rate": 9.605054394253295e-05, "loss": 0.0273, "step": 51300 }, { "epoch": 32.70235818992989, "grad_norm": 0.8633629679679871, "learning_rate": 9.604839670465236e-05, "loss": 0.0311, "step": 51310 }, { "epoch": 32.7087316762269, "grad_norm": 0.5776593685150146, "learning_rate": 9.604624890724027e-05, "loss": 0.0277, "step": 51320 }, { "epoch": 32.7151051625239, "grad_norm": 0.5495419502258301, "learning_rate": 9.60441005503228e-05, "loss": 0.0302, "step": 51330 }, { "epoch": 32.721478648820906, "grad_norm": 0.43936818838119507, "learning_rate": 9.604195163392602e-05, "loss": 0.026, "step": 51340 }, { "epoch": 32.72785213511791, "grad_norm": 0.41285133361816406, "learning_rate": 9.603980215807607e-05, "loss": 0.0353, "step": 51350 }, { "epoch": 32.734225621414915, "grad_norm": 0.45568233728408813, "learning_rate": 9.603765212279904e-05, "loss": 0.04, "step": 51360 }, { "epoch": 32.740599107711915, "grad_norm": 0.4877910912036896, "learning_rate": 9.603550152812106e-05, "loss": 0.0269, "step": 51370 }, { "epoch": 32.74697259400892, "grad_norm": 0.6525394916534424, "learning_rate": 9.60333503740683e-05, "loss": 0.0307, "step": 51380 }, { "epoch": 32.753346080305924, "grad_norm": 0.40296053886413574, "learning_rate": 9.603119866066686e-05, "loss": 0.0172, "step": 51390 }, { "epoch": 32.75971956660293, "grad_norm": 0.47951367497444153, "learning_rate": 9.602904638794288e-05, "loss": 0.0274, "step": 51400 }, { "epoch": 32.76609305289993, "grad_norm": 0.6736000180244446, "learning_rate": 9.602689355592254e-05, "loss": 0.0244, "step": 51410 }, { "epoch": 32.77246653919694, "grad_norm": 0.4851997196674347, "learning_rate": 9.602474016463199e-05, "loss": 0.0264, "step": 51420 }, { "epoch": 32.77884002549394, "grad_norm": 0.3178149163722992, "learning_rate": 9.602258621409739e-05, "loss": 0.0248, "step": 51430 }, { "epoch": 32.78521351179095, "grad_norm": 0.4149443507194519, "learning_rate": 9.602043170434491e-05, "loss": 0.0338, "step": 51440 }, { "epoch": 32.79158699808795, "grad_norm": 0.3450547754764557, "learning_rate": 9.601827663540075e-05, "loss": 0.0273, "step": 51450 }, { "epoch": 32.79796048438496, "grad_norm": 0.4411133825778961, "learning_rate": 9.601612100729106e-05, "loss": 0.0273, "step": 51460 }, { "epoch": 32.80433397068197, "grad_norm": 0.4255078136920929, "learning_rate": 9.601396482004206e-05, "loss": 0.0304, "step": 51470 }, { "epoch": 32.81070745697897, "grad_norm": 0.9307108521461487, "learning_rate": 9.601180807367995e-05, "loss": 0.0317, "step": 51480 }, { "epoch": 32.817080943275975, "grad_norm": 0.6013964414596558, "learning_rate": 9.600965076823092e-05, "loss": 0.0237, "step": 51490 }, { "epoch": 32.823454429572976, "grad_norm": 0.43137577176094055, "learning_rate": 9.600749290372119e-05, "loss": 0.029, "step": 51500 }, { "epoch": 32.829827915869984, "grad_norm": 0.47421103715896606, "learning_rate": 9.6005334480177e-05, "loss": 0.0301, "step": 51510 }, { "epoch": 32.836201402166985, "grad_norm": 0.4701686501502991, "learning_rate": 9.600317549762454e-05, "loss": 0.0259, "step": 51520 }, { "epoch": 32.84257488846399, "grad_norm": 0.32566794753074646, "learning_rate": 9.600101595609008e-05, "loss": 0.0296, "step": 51530 }, { "epoch": 32.848948374760994, "grad_norm": 0.6722625494003296, "learning_rate": 9.599885585559985e-05, "loss": 0.0319, "step": 51540 }, { "epoch": 32.855321861058, "grad_norm": 0.6065362095832825, "learning_rate": 9.599669519618006e-05, "loss": 0.0322, "step": 51550 }, { "epoch": 32.861695347355, "grad_norm": 0.7035350203514099, "learning_rate": 9.599453397785702e-05, "loss": 0.0371, "step": 51560 }, { "epoch": 32.86806883365201, "grad_norm": 0.6945169568061829, "learning_rate": 9.599237220065696e-05, "loss": 0.0346, "step": 51570 }, { "epoch": 32.87444231994901, "grad_norm": 0.895824134349823, "learning_rate": 9.599020986460616e-05, "loss": 0.0311, "step": 51580 }, { "epoch": 32.88081580624602, "grad_norm": 0.6051656603813171, "learning_rate": 9.598804696973087e-05, "loss": 0.0267, "step": 51590 }, { "epoch": 32.88718929254302, "grad_norm": 0.5784096717834473, "learning_rate": 9.598588351605742e-05, "loss": 0.0208, "step": 51600 }, { "epoch": 32.89356277884003, "grad_norm": 0.39280831813812256, "learning_rate": 9.598371950361205e-05, "loss": 0.0237, "step": 51610 }, { "epoch": 32.89993626513703, "grad_norm": 0.42418429255485535, "learning_rate": 9.598155493242106e-05, "loss": 0.0223, "step": 51620 }, { "epoch": 32.90630975143404, "grad_norm": 0.6621795296669006, "learning_rate": 9.597938980251078e-05, "loss": 0.028, "step": 51630 }, { "epoch": 32.91268323773104, "grad_norm": 0.3239007592201233, "learning_rate": 9.597722411390749e-05, "loss": 0.0254, "step": 51640 }, { "epoch": 32.919056724028046, "grad_norm": 0.45672088861465454, "learning_rate": 9.597505786663754e-05, "loss": 0.0223, "step": 51650 }, { "epoch": 32.92543021032505, "grad_norm": 0.7999601364135742, "learning_rate": 9.597289106072719e-05, "loss": 0.0321, "step": 51660 }, { "epoch": 32.931803696622055, "grad_norm": 0.9373602271080017, "learning_rate": 9.597072369620284e-05, "loss": 0.0302, "step": 51670 }, { "epoch": 32.938177182919055, "grad_norm": 0.5340915322303772, "learning_rate": 9.596855577309078e-05, "loss": 0.0271, "step": 51680 }, { "epoch": 32.94455066921606, "grad_norm": 0.5211122632026672, "learning_rate": 9.596638729141738e-05, "loss": 0.0466, "step": 51690 }, { "epoch": 32.950924155513064, "grad_norm": 0.38493096828460693, "learning_rate": 9.596421825120895e-05, "loss": 0.0293, "step": 51700 }, { "epoch": 32.95729764181007, "grad_norm": 0.6660537123680115, "learning_rate": 9.59620486524919e-05, "loss": 0.0285, "step": 51710 }, { "epoch": 32.96367112810707, "grad_norm": 0.5970039963722229, "learning_rate": 9.595987849529255e-05, "loss": 0.0295, "step": 51720 }, { "epoch": 32.97004461440408, "grad_norm": 0.401458740234375, "learning_rate": 9.595770777963728e-05, "loss": 0.0253, "step": 51730 }, { "epoch": 32.97641810070108, "grad_norm": 0.4435962438583374, "learning_rate": 9.595553650555248e-05, "loss": 0.0304, "step": 51740 }, { "epoch": 32.98279158699809, "grad_norm": 0.4422203302383423, "learning_rate": 9.595336467306451e-05, "loss": 0.0217, "step": 51750 }, { "epoch": 32.98916507329509, "grad_norm": 0.3956891596317291, "learning_rate": 9.595119228219978e-05, "loss": 0.0242, "step": 51760 }, { "epoch": 32.9955385595921, "grad_norm": 0.5664805769920349, "learning_rate": 9.594901933298468e-05, "loss": 0.0253, "step": 51770 }, { "epoch": 33.0019120458891, "grad_norm": 0.39192429184913635, "learning_rate": 9.594684582544559e-05, "loss": 0.031, "step": 51780 }, { "epoch": 33.00828553218611, "grad_norm": 0.6645326018333435, "learning_rate": 9.594467175960897e-05, "loss": 0.0294, "step": 51790 }, { "epoch": 33.01465901848311, "grad_norm": 0.8251082301139832, "learning_rate": 9.59424971355012e-05, "loss": 0.0323, "step": 51800 }, { "epoch": 33.021032504780116, "grad_norm": 0.5013106465339661, "learning_rate": 9.594032195314872e-05, "loss": 0.0222, "step": 51810 }, { "epoch": 33.02740599107712, "grad_norm": 0.4512283504009247, "learning_rate": 9.593814621257794e-05, "loss": 0.0299, "step": 51820 }, { "epoch": 33.033779477374125, "grad_norm": 0.4122163951396942, "learning_rate": 9.593596991381531e-05, "loss": 0.0271, "step": 51830 }, { "epoch": 33.040152963671126, "grad_norm": 0.5083367228507996, "learning_rate": 9.593379305688728e-05, "loss": 0.0203, "step": 51840 }, { "epoch": 33.046526449968134, "grad_norm": 0.6539205312728882, "learning_rate": 9.593161564182029e-05, "loss": 0.0352, "step": 51850 }, { "epoch": 33.052899936265135, "grad_norm": 0.5143155455589294, "learning_rate": 9.592943766864083e-05, "loss": 0.0223, "step": 51860 }, { "epoch": 33.05927342256214, "grad_norm": 0.6443670988082886, "learning_rate": 9.59272591373753e-05, "loss": 0.0339, "step": 51870 }, { "epoch": 33.065646908859144, "grad_norm": 0.5199339985847473, "learning_rate": 9.592508004805023e-05, "loss": 0.0371, "step": 51880 }, { "epoch": 33.07202039515615, "grad_norm": 0.6474002599716187, "learning_rate": 9.592290040069207e-05, "loss": 0.034, "step": 51890 }, { "epoch": 33.07839388145315, "grad_norm": 0.6562907695770264, "learning_rate": 9.592072019532731e-05, "loss": 0.0352, "step": 51900 }, { "epoch": 33.08476736775016, "grad_norm": 0.39504918456077576, "learning_rate": 9.591853943198245e-05, "loss": 0.0229, "step": 51910 }, { "epoch": 33.09114085404716, "grad_norm": 0.7300223708152771, "learning_rate": 9.5916358110684e-05, "loss": 0.0282, "step": 51920 }, { "epoch": 33.09751434034417, "grad_norm": 0.7987372875213623, "learning_rate": 9.591417623145843e-05, "loss": 0.0337, "step": 51930 }, { "epoch": 33.10388782664117, "grad_norm": 0.524081826210022, "learning_rate": 9.591199379433225e-05, "loss": 0.0297, "step": 51940 }, { "epoch": 33.11026131293818, "grad_norm": 0.28298643231391907, "learning_rate": 9.590981079933202e-05, "loss": 0.0201, "step": 51950 }, { "epoch": 33.11663479923518, "grad_norm": 0.4383884072303772, "learning_rate": 9.590762724648423e-05, "loss": 0.0306, "step": 51960 }, { "epoch": 33.12300828553219, "grad_norm": 0.42452484369277954, "learning_rate": 9.590544313581545e-05, "loss": 0.0325, "step": 51970 }, { "epoch": 33.12938177182919, "grad_norm": 0.7442895770072937, "learning_rate": 9.590325846735216e-05, "loss": 0.0312, "step": 51980 }, { "epoch": 33.135755258126196, "grad_norm": 0.42221128940582275, "learning_rate": 9.590107324112095e-05, "loss": 0.0249, "step": 51990 }, { "epoch": 33.1421287444232, "grad_norm": 0.3364163041114807, "learning_rate": 9.589888745714837e-05, "loss": 0.0246, "step": 52000 }, { "epoch": 33.148502230720204, "grad_norm": 0.759209930896759, "learning_rate": 9.589670111546097e-05, "loss": 0.0368, "step": 52010 }, { "epoch": 33.154875717017205, "grad_norm": 0.3822955787181854, "learning_rate": 9.589451421608531e-05, "loss": 0.0434, "step": 52020 }, { "epoch": 33.16124920331421, "grad_norm": 0.6044666767120361, "learning_rate": 9.589232675904798e-05, "loss": 0.0298, "step": 52030 }, { "epoch": 33.167622689611214, "grad_norm": 1.056251883506775, "learning_rate": 9.589013874437555e-05, "loss": 0.0308, "step": 52040 }, { "epoch": 33.17399617590822, "grad_norm": 0.48414215445518494, "learning_rate": 9.58879501720946e-05, "loss": 0.0446, "step": 52050 }, { "epoch": 33.18036966220522, "grad_norm": 0.29902100563049316, "learning_rate": 9.588576104223174e-05, "loss": 0.0269, "step": 52060 }, { "epoch": 33.18674314850223, "grad_norm": 0.5396019220352173, "learning_rate": 9.588357135481355e-05, "loss": 0.0387, "step": 52070 }, { "epoch": 33.19311663479923, "grad_norm": 0.43449875712394714, "learning_rate": 9.588138110986665e-05, "loss": 0.0289, "step": 52080 }, { "epoch": 33.19949012109624, "grad_norm": 0.4832167625427246, "learning_rate": 9.587919030741766e-05, "loss": 0.026, "step": 52090 }, { "epoch": 33.20586360739325, "grad_norm": 0.5578222870826721, "learning_rate": 9.587699894749317e-05, "loss": 0.0305, "step": 52100 }, { "epoch": 33.21223709369025, "grad_norm": 0.6065574884414673, "learning_rate": 9.587480703011984e-05, "loss": 0.0389, "step": 52110 }, { "epoch": 33.218610579987256, "grad_norm": 0.5374851822853088, "learning_rate": 9.587261455532429e-05, "loss": 0.0267, "step": 52120 }, { "epoch": 33.22498406628426, "grad_norm": 0.5225486755371094, "learning_rate": 9.587042152313317e-05, "loss": 0.0269, "step": 52130 }, { "epoch": 33.231357552581265, "grad_norm": 0.47357842326164246, "learning_rate": 9.586822793357311e-05, "loss": 0.0308, "step": 52140 }, { "epoch": 33.237731038878266, "grad_norm": 0.25090691447257996, "learning_rate": 9.586603378667077e-05, "loss": 0.0246, "step": 52150 }, { "epoch": 33.244104525175274, "grad_norm": 0.4330727159976959, "learning_rate": 9.586383908245283e-05, "loss": 0.0239, "step": 52160 }, { "epoch": 33.250478011472275, "grad_norm": 0.6424544453620911, "learning_rate": 9.586164382094593e-05, "loss": 0.0254, "step": 52170 }, { "epoch": 33.25685149776928, "grad_norm": 0.44416797161102295, "learning_rate": 9.585944800217677e-05, "loss": 0.0176, "step": 52180 }, { "epoch": 33.263224984066284, "grad_norm": 0.42545875906944275, "learning_rate": 9.5857251626172e-05, "loss": 0.037, "step": 52190 }, { "epoch": 33.26959847036329, "grad_norm": 0.4206511676311493, "learning_rate": 9.585505469295833e-05, "loss": 0.029, "step": 52200 }, { "epoch": 33.27597195666029, "grad_norm": 0.6629734635353088, "learning_rate": 9.585285720256246e-05, "loss": 0.0288, "step": 52210 }, { "epoch": 33.2823454429573, "grad_norm": 0.37238195538520813, "learning_rate": 9.585065915501108e-05, "loss": 0.0237, "step": 52220 }, { "epoch": 33.2887189292543, "grad_norm": 0.4782288372516632, "learning_rate": 9.58484605503309e-05, "loss": 0.0252, "step": 52230 }, { "epoch": 33.29509241555131, "grad_norm": 0.5243447422981262, "learning_rate": 9.584626138854864e-05, "loss": 0.0316, "step": 52240 }, { "epoch": 33.30146590184831, "grad_norm": 0.4381398558616638, "learning_rate": 9.584406166969101e-05, "loss": 0.0307, "step": 52250 }, { "epoch": 33.30783938814532, "grad_norm": 0.4061771333217621, "learning_rate": 9.584186139378474e-05, "loss": 0.0173, "step": 52260 }, { "epoch": 33.31421287444232, "grad_norm": 0.32514333724975586, "learning_rate": 9.58396605608566e-05, "loss": 0.0248, "step": 52270 }, { "epoch": 33.32058636073933, "grad_norm": 0.4628959596157074, "learning_rate": 9.583745917093327e-05, "loss": 0.0204, "step": 52280 }, { "epoch": 33.32695984703633, "grad_norm": 0.4943941533565521, "learning_rate": 9.583525722404157e-05, "loss": 0.019, "step": 52290 }, { "epoch": 33.333333333333336, "grad_norm": 0.5929449796676636, "learning_rate": 9.58330547202082e-05, "loss": 0.0221, "step": 52300 }, { "epoch": 33.33970681963034, "grad_norm": 0.6916621327400208, "learning_rate": 9.583085165945992e-05, "loss": 0.0221, "step": 52310 }, { "epoch": 33.346080305927345, "grad_norm": 0.4577091336250305, "learning_rate": 9.582864804182356e-05, "loss": 0.0222, "step": 52320 }, { "epoch": 33.352453792224345, "grad_norm": 0.6032233834266663, "learning_rate": 9.582644386732584e-05, "loss": 0.0299, "step": 52330 }, { "epoch": 33.35882727852135, "grad_norm": 0.48720675706863403, "learning_rate": 9.582423913599357e-05, "loss": 0.0351, "step": 52340 }, { "epoch": 33.365200764818354, "grad_norm": 0.5394951105117798, "learning_rate": 9.582203384785351e-05, "loss": 0.0269, "step": 52350 }, { "epoch": 33.37157425111536, "grad_norm": 0.528304934501648, "learning_rate": 9.581982800293248e-05, "loss": 0.031, "step": 52360 }, { "epoch": 33.37794773741236, "grad_norm": 0.5044497847557068, "learning_rate": 9.581762160125729e-05, "loss": 0.0252, "step": 52370 }, { "epoch": 33.38432122370937, "grad_norm": 0.3245835602283478, "learning_rate": 9.581541464285472e-05, "loss": 0.0241, "step": 52380 }, { "epoch": 33.39069471000637, "grad_norm": 0.5321206450462341, "learning_rate": 9.581320712775161e-05, "loss": 0.033, "step": 52390 }, { "epoch": 33.39706819630338, "grad_norm": 0.5315662622451782, "learning_rate": 9.58109990559748e-05, "loss": 0.0302, "step": 52400 }, { "epoch": 33.40344168260038, "grad_norm": 0.7033866047859192, "learning_rate": 9.580879042755109e-05, "loss": 0.0304, "step": 52410 }, { "epoch": 33.40981516889739, "grad_norm": 0.5001955032348633, "learning_rate": 9.580658124250731e-05, "loss": 0.0289, "step": 52420 }, { "epoch": 33.41618865519439, "grad_norm": 0.5968204736709595, "learning_rate": 9.580437150087033e-05, "loss": 0.0286, "step": 52430 }, { "epoch": 33.4225621414914, "grad_norm": 0.32420679926872253, "learning_rate": 9.580216120266698e-05, "loss": 0.019, "step": 52440 }, { "epoch": 33.4289356277884, "grad_norm": 0.6301494836807251, "learning_rate": 9.579995034792414e-05, "loss": 0.0248, "step": 52450 }, { "epoch": 33.435309114085406, "grad_norm": 0.44887086749076843, "learning_rate": 9.579773893666865e-05, "loss": 0.0211, "step": 52460 }, { "epoch": 33.44168260038241, "grad_norm": 0.32745301723480225, "learning_rate": 9.579552696892741e-05, "loss": 0.0333, "step": 52470 }, { "epoch": 33.448056086679415, "grad_norm": 0.44621390104293823, "learning_rate": 9.579331444472725e-05, "loss": 0.0257, "step": 52480 }, { "epoch": 33.454429572976416, "grad_norm": 0.4460764229297638, "learning_rate": 9.579110136409511e-05, "loss": 0.0329, "step": 52490 }, { "epoch": 33.460803059273424, "grad_norm": 0.6321269273757935, "learning_rate": 9.578888772705785e-05, "loss": 0.045, "step": 52500 }, { "epoch": 33.467176545570425, "grad_norm": 0.41663962602615356, "learning_rate": 9.578667353364237e-05, "loss": 0.0252, "step": 52510 }, { "epoch": 33.47355003186743, "grad_norm": 0.3875863254070282, "learning_rate": 9.578445878387556e-05, "loss": 0.0253, "step": 52520 }, { "epoch": 33.47992351816443, "grad_norm": 0.4657875895500183, "learning_rate": 9.578224347778437e-05, "loss": 0.026, "step": 52530 }, { "epoch": 33.48629700446144, "grad_norm": 0.6377483606338501, "learning_rate": 9.578002761539568e-05, "loss": 0.0245, "step": 52540 }, { "epoch": 33.49267049075844, "grad_norm": 0.6499089598655701, "learning_rate": 9.577781119673645e-05, "loss": 0.0355, "step": 52550 }, { "epoch": 33.49904397705545, "grad_norm": 0.45889222621917725, "learning_rate": 9.577559422183357e-05, "loss": 0.0239, "step": 52560 }, { "epoch": 33.50541746335245, "grad_norm": 0.5011894702911377, "learning_rate": 9.577337669071403e-05, "loss": 0.0304, "step": 52570 }, { "epoch": 33.51179094964946, "grad_norm": 0.37935492396354675, "learning_rate": 9.577115860340473e-05, "loss": 0.0266, "step": 52580 }, { "epoch": 33.51816443594646, "grad_norm": 0.3869948983192444, "learning_rate": 9.576893995993264e-05, "loss": 0.0344, "step": 52590 }, { "epoch": 33.52453792224347, "grad_norm": 0.5484520792961121, "learning_rate": 9.57667207603247e-05, "loss": 0.0314, "step": 52600 }, { "epoch": 33.53091140854047, "grad_norm": 0.3637152910232544, "learning_rate": 9.57645010046079e-05, "loss": 0.0284, "step": 52610 }, { "epoch": 33.53728489483748, "grad_norm": 0.3845587372779846, "learning_rate": 9.576228069280921e-05, "loss": 0.0223, "step": 52620 }, { "epoch": 33.54365838113448, "grad_norm": 0.46730637550354004, "learning_rate": 9.576005982495562e-05, "loss": 0.0221, "step": 52630 }, { "epoch": 33.550031867431485, "grad_norm": 0.44513535499572754, "learning_rate": 9.575783840107407e-05, "loss": 0.0245, "step": 52640 }, { "epoch": 33.556405353728486, "grad_norm": 0.40078699588775635, "learning_rate": 9.575561642119159e-05, "loss": 0.025, "step": 52650 }, { "epoch": 33.562778840025494, "grad_norm": 0.3006082773208618, "learning_rate": 9.575339388533517e-05, "loss": 0.025, "step": 52660 }, { "epoch": 33.569152326322495, "grad_norm": 0.3805831968784332, "learning_rate": 9.575117079353182e-05, "loss": 0.0305, "step": 52670 }, { "epoch": 33.5755258126195, "grad_norm": 0.4649515151977539, "learning_rate": 9.574894714580854e-05, "loss": 0.0316, "step": 52680 }, { "epoch": 33.581899298916504, "grad_norm": 0.446331650018692, "learning_rate": 9.574672294219237e-05, "loss": 0.0342, "step": 52690 }, { "epoch": 33.58827278521351, "grad_norm": 0.5373594760894775, "learning_rate": 9.574449818271032e-05, "loss": 0.0227, "step": 52700 }, { "epoch": 33.59464627151051, "grad_norm": 0.5900424718856812, "learning_rate": 9.574227286738942e-05, "loss": 0.0319, "step": 52710 }, { "epoch": 33.60101975780752, "grad_norm": 0.9526788592338562, "learning_rate": 9.574004699625671e-05, "loss": 0.0295, "step": 52720 }, { "epoch": 33.60739324410453, "grad_norm": 0.6568556427955627, "learning_rate": 9.573782056933927e-05, "loss": 0.0283, "step": 52730 }, { "epoch": 33.61376673040153, "grad_norm": 0.27466288208961487, "learning_rate": 9.57355935866641e-05, "loss": 0.0214, "step": 52740 }, { "epoch": 33.62014021669854, "grad_norm": 0.32533931732177734, "learning_rate": 9.573336604825831e-05, "loss": 0.0281, "step": 52750 }, { "epoch": 33.62651370299554, "grad_norm": 0.2134103775024414, "learning_rate": 9.573113795414891e-05, "loss": 0.0266, "step": 52760 }, { "epoch": 33.632887189292546, "grad_norm": 0.39872175455093384, "learning_rate": 9.572890930436305e-05, "loss": 0.026, "step": 52770 }, { "epoch": 33.63926067558955, "grad_norm": 0.6040008068084717, "learning_rate": 9.572668009892773e-05, "loss": 0.0437, "step": 52780 }, { "epoch": 33.645634161886555, "grad_norm": 0.49851861596107483, "learning_rate": 9.57244503378701e-05, "loss": 0.0274, "step": 52790 }, { "epoch": 33.652007648183556, "grad_norm": 0.34538549184799194, "learning_rate": 9.572222002121722e-05, "loss": 0.0386, "step": 52800 }, { "epoch": 33.658381134480564, "grad_norm": 0.28822439908981323, "learning_rate": 9.571998914899619e-05, "loss": 0.0329, "step": 52810 }, { "epoch": 33.664754620777565, "grad_norm": 0.5181159377098083, "learning_rate": 9.571775772123412e-05, "loss": 0.0314, "step": 52820 }, { "epoch": 33.67112810707457, "grad_norm": 0.4672962427139282, "learning_rate": 9.571552573795815e-05, "loss": 0.0314, "step": 52830 }, { "epoch": 33.677501593371574, "grad_norm": 0.509365439414978, "learning_rate": 9.571329319919535e-05, "loss": 0.0278, "step": 52840 }, { "epoch": 33.68387507966858, "grad_norm": 0.3330395519733429, "learning_rate": 9.571106010497291e-05, "loss": 0.0302, "step": 52850 }, { "epoch": 33.69024856596558, "grad_norm": 0.4655858874320984, "learning_rate": 9.57088264553179e-05, "loss": 0.0402, "step": 52860 }, { "epoch": 33.69662205226259, "grad_norm": 0.3230804204940796, "learning_rate": 9.570659225025751e-05, "loss": 0.0281, "step": 52870 }, { "epoch": 33.70299553855959, "grad_norm": 0.23053815960884094, "learning_rate": 9.570435748981888e-05, "loss": 0.0264, "step": 52880 }, { "epoch": 33.7093690248566, "grad_norm": 0.4922768175601959, "learning_rate": 9.570212217402913e-05, "loss": 0.0332, "step": 52890 }, { "epoch": 33.7157425111536, "grad_norm": 0.4251912534236908, "learning_rate": 9.569988630291545e-05, "loss": 0.0373, "step": 52900 }, { "epoch": 33.72211599745061, "grad_norm": 0.6231818795204163, "learning_rate": 9.569764987650502e-05, "loss": 0.0281, "step": 52910 }, { "epoch": 33.72848948374761, "grad_norm": 0.5457112193107605, "learning_rate": 9.569541289482498e-05, "loss": 0.0365, "step": 52920 }, { "epoch": 33.73486297004462, "grad_norm": 0.5576350092887878, "learning_rate": 9.569317535790255e-05, "loss": 0.0275, "step": 52930 }, { "epoch": 33.74123645634162, "grad_norm": 0.43591880798339844, "learning_rate": 9.569093726576488e-05, "loss": 0.0318, "step": 52940 }, { "epoch": 33.747609942638626, "grad_norm": 0.40061303973197937, "learning_rate": 9.568869861843919e-05, "loss": 0.0393, "step": 52950 }, { "epoch": 33.753983428935626, "grad_norm": 0.7621817588806152, "learning_rate": 9.568645941595267e-05, "loss": 0.0419, "step": 52960 }, { "epoch": 33.760356915232634, "grad_norm": 0.5062728524208069, "learning_rate": 9.568421965833254e-05, "loss": 0.0245, "step": 52970 }, { "epoch": 33.766730401529635, "grad_norm": 0.3432307243347168, "learning_rate": 9.568197934560602e-05, "loss": 0.0331, "step": 52980 }, { "epoch": 33.77310388782664, "grad_norm": 0.4563245475292206, "learning_rate": 9.567973847780029e-05, "loss": 0.0379, "step": 52990 }, { "epoch": 33.779477374123644, "grad_norm": 0.3369235098361969, "learning_rate": 9.567749705494263e-05, "loss": 0.0222, "step": 53000 }, { "epoch": 33.78585086042065, "grad_norm": 0.65120530128479, "learning_rate": 9.567525507706024e-05, "loss": 0.0305, "step": 53010 }, { "epoch": 33.79222434671765, "grad_norm": 0.572632372379303, "learning_rate": 9.567301254418038e-05, "loss": 0.0249, "step": 53020 }, { "epoch": 33.79859783301466, "grad_norm": 0.6785462498664856, "learning_rate": 9.567076945633031e-05, "loss": 0.0331, "step": 53030 }, { "epoch": 33.80497131931166, "grad_norm": 0.4580390155315399, "learning_rate": 9.566852581353726e-05, "loss": 0.0363, "step": 53040 }, { "epoch": 33.81134480560867, "grad_norm": 0.27569130063056946, "learning_rate": 9.56662816158285e-05, "loss": 0.0338, "step": 53050 }, { "epoch": 33.81771829190567, "grad_norm": 0.46244460344314575, "learning_rate": 9.566403686323131e-05, "loss": 0.0356, "step": 53060 }, { "epoch": 33.82409177820268, "grad_norm": 0.3413628935813904, "learning_rate": 9.566179155577294e-05, "loss": 0.0303, "step": 53070 }, { "epoch": 33.83046526449968, "grad_norm": 0.34665679931640625, "learning_rate": 9.565954569348072e-05, "loss": 0.0355, "step": 53080 }, { "epoch": 33.83683875079669, "grad_norm": 0.4611565172672272, "learning_rate": 9.565729927638188e-05, "loss": 0.051, "step": 53090 }, { "epoch": 33.84321223709369, "grad_norm": 0.6459419131278992, "learning_rate": 9.565505230450376e-05, "loss": 0.0478, "step": 53100 }, { "epoch": 33.849585723390696, "grad_norm": 0.39568012952804565, "learning_rate": 9.565280477787367e-05, "loss": 0.0346, "step": 53110 }, { "epoch": 33.8559592096877, "grad_norm": 0.47443461418151855, "learning_rate": 9.565055669651886e-05, "loss": 0.0283, "step": 53120 }, { "epoch": 33.862332695984705, "grad_norm": 0.6358253359794617, "learning_rate": 9.56483080604667e-05, "loss": 0.0392, "step": 53130 }, { "epoch": 33.868706182281706, "grad_norm": 0.4592743217945099, "learning_rate": 9.564605886974451e-05, "loss": 0.0295, "step": 53140 }, { "epoch": 33.875079668578714, "grad_norm": 0.42860621213912964, "learning_rate": 9.56438091243796e-05, "loss": 0.0276, "step": 53150 }, { "epoch": 33.881453154875715, "grad_norm": 0.5611061453819275, "learning_rate": 9.564155882439931e-05, "loss": 0.0396, "step": 53160 }, { "epoch": 33.88782664117272, "grad_norm": 0.48615095019340515, "learning_rate": 9.563930796983098e-05, "loss": 0.0377, "step": 53170 }, { "epoch": 33.89420012746972, "grad_norm": 0.42084577679634094, "learning_rate": 9.563705656070198e-05, "loss": 0.0301, "step": 53180 }, { "epoch": 33.90057361376673, "grad_norm": 0.567521333694458, "learning_rate": 9.563480459703966e-05, "loss": 0.0427, "step": 53190 }, { "epoch": 33.90694710006373, "grad_norm": 0.5445899963378906, "learning_rate": 9.563255207887136e-05, "loss": 0.0334, "step": 53200 }, { "epoch": 33.91332058636074, "grad_norm": 0.4061892330646515, "learning_rate": 9.563029900622449e-05, "loss": 0.0293, "step": 53210 }, { "epoch": 33.91969407265774, "grad_norm": 0.2565276026725769, "learning_rate": 9.562804537912639e-05, "loss": 0.0281, "step": 53220 }, { "epoch": 33.92606755895475, "grad_norm": 0.5534552335739136, "learning_rate": 9.562579119760447e-05, "loss": 0.0274, "step": 53230 }, { "epoch": 33.93244104525175, "grad_norm": 0.3412909209728241, "learning_rate": 9.56235364616861e-05, "loss": 0.0186, "step": 53240 }, { "epoch": 33.93881453154876, "grad_norm": 0.31539711356163025, "learning_rate": 9.562128117139868e-05, "loss": 0.036, "step": 53250 }, { "epoch": 33.94518801784576, "grad_norm": 0.3508473038673401, "learning_rate": 9.561902532676963e-05, "loss": 0.0246, "step": 53260 }, { "epoch": 33.95156150414277, "grad_norm": 0.48333466053009033, "learning_rate": 9.561676892782633e-05, "loss": 0.0296, "step": 53270 }, { "epoch": 33.95793499043977, "grad_norm": 0.5551915764808655, "learning_rate": 9.561451197459625e-05, "loss": 0.0302, "step": 53280 }, { "epoch": 33.964308476736775, "grad_norm": 0.33949464559555054, "learning_rate": 9.561225446710676e-05, "loss": 0.0323, "step": 53290 }, { "epoch": 33.970681963033776, "grad_norm": 0.32581931352615356, "learning_rate": 9.560999640538532e-05, "loss": 0.0208, "step": 53300 }, { "epoch": 33.977055449330784, "grad_norm": 0.39329954981803894, "learning_rate": 9.560773778945938e-05, "loss": 0.038, "step": 53310 }, { "epoch": 33.983428935627785, "grad_norm": 0.6731166243553162, "learning_rate": 9.560547861935634e-05, "loss": 0.0339, "step": 53320 }, { "epoch": 33.98980242192479, "grad_norm": 0.3781405985355377, "learning_rate": 9.560321889510368e-05, "loss": 0.0355, "step": 53330 }, { "epoch": 33.9961759082218, "grad_norm": 0.47978052496910095, "learning_rate": 9.560095861672885e-05, "loss": 0.0261, "step": 53340 }, { "epoch": 34.0025493945188, "grad_norm": 0.4785864055156708, "learning_rate": 9.559869778425933e-05, "loss": 0.0305, "step": 53350 }, { "epoch": 34.00892288081581, "grad_norm": 0.36711546778678894, "learning_rate": 9.559643639772258e-05, "loss": 0.0305, "step": 53360 }, { "epoch": 34.01529636711281, "grad_norm": 0.4253818094730377, "learning_rate": 9.559417445714607e-05, "loss": 0.0293, "step": 53370 }, { "epoch": 34.02166985340982, "grad_norm": 0.6237525343894958, "learning_rate": 9.55919119625573e-05, "loss": 0.0236, "step": 53380 }, { "epoch": 34.02804333970682, "grad_norm": 0.34273648262023926, "learning_rate": 9.558964891398374e-05, "loss": 0.0277, "step": 53390 }, { "epoch": 34.03441682600383, "grad_norm": 0.4689711332321167, "learning_rate": 9.55873853114529e-05, "loss": 0.0246, "step": 53400 }, { "epoch": 34.04079031230083, "grad_norm": 0.24570055305957794, "learning_rate": 9.55851211549923e-05, "loss": 0.0204, "step": 53410 }, { "epoch": 34.047163798597836, "grad_norm": 0.49715811014175415, "learning_rate": 9.558285644462943e-05, "loss": 0.0334, "step": 53420 }, { "epoch": 34.05353728489484, "grad_norm": 0.5334717631340027, "learning_rate": 9.558059118039183e-05, "loss": 0.0255, "step": 53430 }, { "epoch": 34.059910771191845, "grad_norm": 0.546329915523529, "learning_rate": 9.557832536230701e-05, "loss": 0.0222, "step": 53440 }, { "epoch": 34.066284257488846, "grad_norm": 1.4655948877334595, "learning_rate": 9.55760589904025e-05, "loss": 0.0274, "step": 53450 }, { "epoch": 34.072657743785854, "grad_norm": 0.3237227499485016, "learning_rate": 9.557379206470583e-05, "loss": 0.0282, "step": 53460 }, { "epoch": 34.079031230082855, "grad_norm": 0.6544840335845947, "learning_rate": 9.557152458524457e-05, "loss": 0.0323, "step": 53470 }, { "epoch": 34.08540471637986, "grad_norm": 0.2570575475692749, "learning_rate": 9.556925655204626e-05, "loss": 0.0191, "step": 53480 }, { "epoch": 34.09177820267686, "grad_norm": 0.3816926181316376, "learning_rate": 9.556698796513845e-05, "loss": 0.0226, "step": 53490 }, { "epoch": 34.09815168897387, "grad_norm": 0.49775460362434387, "learning_rate": 9.556471882454872e-05, "loss": 0.0448, "step": 53500 }, { "epoch": 34.10452517527087, "grad_norm": 0.4969298243522644, "learning_rate": 9.556244913030465e-05, "loss": 0.0225, "step": 53510 }, { "epoch": 34.11089866156788, "grad_norm": 0.4886188209056854, "learning_rate": 9.556017888243378e-05, "loss": 0.0261, "step": 53520 }, { "epoch": 34.11727214786488, "grad_norm": 0.5399004817008972, "learning_rate": 9.555790808096374e-05, "loss": 0.0265, "step": 53530 }, { "epoch": 34.12364563416189, "grad_norm": 0.28880321979522705, "learning_rate": 9.55556367259221e-05, "loss": 0.019, "step": 53540 }, { "epoch": 34.13001912045889, "grad_norm": 0.4671781361103058, "learning_rate": 9.555336481733644e-05, "loss": 0.0374, "step": 53550 }, { "epoch": 34.1363926067559, "grad_norm": 0.2675691246986389, "learning_rate": 9.55510923552344e-05, "loss": 0.0214, "step": 53560 }, { "epoch": 34.1427660930529, "grad_norm": 0.2760460674762726, "learning_rate": 9.554881933964359e-05, "loss": 0.029, "step": 53570 }, { "epoch": 34.14913957934991, "grad_norm": 0.5502656102180481, "learning_rate": 9.554654577059163e-05, "loss": 0.0203, "step": 53580 }, { "epoch": 34.15551306564691, "grad_norm": 0.5731120109558105, "learning_rate": 9.554427164810611e-05, "loss": 0.0248, "step": 53590 }, { "epoch": 34.161886551943915, "grad_norm": 0.5067721605300903, "learning_rate": 9.55419969722147e-05, "loss": 0.0273, "step": 53600 }, { "epoch": 34.168260038240916, "grad_norm": 0.3770888149738312, "learning_rate": 9.553972174294503e-05, "loss": 0.0276, "step": 53610 }, { "epoch": 34.174633524537924, "grad_norm": 0.45629173517227173, "learning_rate": 9.553744596032474e-05, "loss": 0.0292, "step": 53620 }, { "epoch": 34.181007010834925, "grad_norm": 0.6713749170303345, "learning_rate": 9.553516962438149e-05, "loss": 0.0247, "step": 53630 }, { "epoch": 34.18738049713193, "grad_norm": 0.6958839297294617, "learning_rate": 9.553289273514294e-05, "loss": 0.037, "step": 53640 }, { "epoch": 34.193753983428934, "grad_norm": 0.5203536152839661, "learning_rate": 9.553061529263673e-05, "loss": 0.0298, "step": 53650 }, { "epoch": 34.20012746972594, "grad_norm": 0.5386590957641602, "learning_rate": 9.55283372968906e-05, "loss": 0.0307, "step": 53660 }, { "epoch": 34.20650095602294, "grad_norm": 0.5252214074134827, "learning_rate": 9.552605874793214e-05, "loss": 0.0223, "step": 53670 }, { "epoch": 34.21287444231995, "grad_norm": 0.31366869807243347, "learning_rate": 9.552377964578912e-05, "loss": 0.0292, "step": 53680 }, { "epoch": 34.21924792861695, "grad_norm": 0.39047500491142273, "learning_rate": 9.552149999048917e-05, "loss": 0.0267, "step": 53690 }, { "epoch": 34.22562141491396, "grad_norm": 0.6531780362129211, "learning_rate": 9.551921978206001e-05, "loss": 0.0413, "step": 53700 }, { "epoch": 34.23199490121096, "grad_norm": 0.5505098700523376, "learning_rate": 9.551693902052938e-05, "loss": 0.0258, "step": 53710 }, { "epoch": 34.23836838750797, "grad_norm": 0.575623631477356, "learning_rate": 9.551465770592495e-05, "loss": 0.0319, "step": 53720 }, { "epoch": 34.24474187380497, "grad_norm": 0.4888959228992462, "learning_rate": 9.551237583827446e-05, "loss": 0.0325, "step": 53730 }, { "epoch": 34.25111536010198, "grad_norm": 0.2585799992084503, "learning_rate": 9.551009341760564e-05, "loss": 0.0349, "step": 53740 }, { "epoch": 34.25748884639898, "grad_norm": 0.5387828946113586, "learning_rate": 9.550781044394621e-05, "loss": 0.0234, "step": 53750 }, { "epoch": 34.263862332695986, "grad_norm": 0.5920343995094299, "learning_rate": 9.55055269173239e-05, "loss": 0.031, "step": 53760 }, { "epoch": 34.27023581899299, "grad_norm": 0.4863366186618805, "learning_rate": 9.55032428377665e-05, "loss": 0.0467, "step": 53770 }, { "epoch": 34.276609305289995, "grad_norm": 0.3027154505252838, "learning_rate": 9.550095820530172e-05, "loss": 0.0272, "step": 53780 }, { "epoch": 34.282982791586996, "grad_norm": 0.7009127140045166, "learning_rate": 9.549867301995734e-05, "loss": 0.0244, "step": 53790 }, { "epoch": 34.289356277884, "grad_norm": 0.816160261631012, "learning_rate": 9.549638728176114e-05, "loss": 0.0365, "step": 53800 }, { "epoch": 34.295729764181004, "grad_norm": 0.48131388425827026, "learning_rate": 9.549410099074085e-05, "loss": 0.0234, "step": 53810 }, { "epoch": 34.30210325047801, "grad_norm": 0.49172544479370117, "learning_rate": 9.549181414692431e-05, "loss": 0.033, "step": 53820 }, { "epoch": 34.30847673677501, "grad_norm": 0.35445553064346313, "learning_rate": 9.548952675033926e-05, "loss": 0.0307, "step": 53830 }, { "epoch": 34.31485022307202, "grad_norm": 0.24412131309509277, "learning_rate": 9.548723880101351e-05, "loss": 0.0259, "step": 53840 }, { "epoch": 34.32122370936902, "grad_norm": 0.627521276473999, "learning_rate": 9.548495029897486e-05, "loss": 0.0314, "step": 53850 }, { "epoch": 34.32759719566603, "grad_norm": 0.31412380933761597, "learning_rate": 9.548266124425113e-05, "loss": 0.0297, "step": 53860 }, { "epoch": 34.33397068196303, "grad_norm": 0.2314710021018982, "learning_rate": 9.548037163687011e-05, "loss": 0.02, "step": 53870 }, { "epoch": 34.34034416826004, "grad_norm": 0.7032288312911987, "learning_rate": 9.547808147685965e-05, "loss": 0.0356, "step": 53880 }, { "epoch": 34.34671765455704, "grad_norm": 0.43201225996017456, "learning_rate": 9.547579076424756e-05, "loss": 0.0228, "step": 53890 }, { "epoch": 34.35309114085405, "grad_norm": 0.18556854128837585, "learning_rate": 9.547349949906166e-05, "loss": 0.0156, "step": 53900 }, { "epoch": 34.35946462715105, "grad_norm": 0.32275819778442383, "learning_rate": 9.547120768132982e-05, "loss": 0.0193, "step": 53910 }, { "epoch": 34.365838113448056, "grad_norm": 0.7590206265449524, "learning_rate": 9.546891531107986e-05, "loss": 0.0311, "step": 53920 }, { "epoch": 34.37221159974506, "grad_norm": 0.5695422291755676, "learning_rate": 9.546662238833967e-05, "loss": 0.0305, "step": 53930 }, { "epoch": 34.378585086042065, "grad_norm": 0.5140156745910645, "learning_rate": 9.546432891313708e-05, "loss": 0.0426, "step": 53940 }, { "epoch": 34.384958572339066, "grad_norm": 0.3449597656726837, "learning_rate": 9.546203488549997e-05, "loss": 0.0425, "step": 53950 }, { "epoch": 34.391332058636074, "grad_norm": 0.44223153591156006, "learning_rate": 9.545974030545621e-05, "loss": 0.0258, "step": 53960 }, { "epoch": 34.397705544933075, "grad_norm": 0.780102014541626, "learning_rate": 9.545744517303368e-05, "loss": 0.0477, "step": 53970 }, { "epoch": 34.40407903123008, "grad_norm": 0.7015387415885925, "learning_rate": 9.545514948826028e-05, "loss": 0.0457, "step": 53980 }, { "epoch": 34.41045251752709, "grad_norm": 0.6713115572929382, "learning_rate": 9.54528532511639e-05, "loss": 0.0209, "step": 53990 }, { "epoch": 34.41682600382409, "grad_norm": 0.9521574974060059, "learning_rate": 9.545055646177243e-05, "loss": 0.0448, "step": 54000 }, { "epoch": 34.4231994901211, "grad_norm": 0.4362484812736511, "learning_rate": 9.544825912011379e-05, "loss": 0.022, "step": 54010 }, { "epoch": 34.4295729764181, "grad_norm": 0.46086394786834717, "learning_rate": 9.54459612262159e-05, "loss": 0.0295, "step": 54020 }, { "epoch": 34.43594646271511, "grad_norm": 0.4730875790119171, "learning_rate": 9.544366278010666e-05, "loss": 0.0276, "step": 54030 }, { "epoch": 34.44231994901211, "grad_norm": 0.5375385284423828, "learning_rate": 9.544136378181401e-05, "loss": 0.0375, "step": 54040 }, { "epoch": 34.44869343530912, "grad_norm": 0.4185850918292999, "learning_rate": 9.543906423136589e-05, "loss": 0.0377, "step": 54050 }, { "epoch": 34.45506692160612, "grad_norm": 0.4055817425251007, "learning_rate": 9.543676412879024e-05, "loss": 0.0207, "step": 54060 }, { "epoch": 34.461440407903126, "grad_norm": 0.27319684624671936, "learning_rate": 9.543446347411501e-05, "loss": 0.0205, "step": 54070 }, { "epoch": 34.46781389420013, "grad_norm": 0.43446555733680725, "learning_rate": 9.543216226736815e-05, "loss": 0.0282, "step": 54080 }, { "epoch": 34.474187380497135, "grad_norm": 0.3387250304222107, "learning_rate": 9.542986050857761e-05, "loss": 0.02, "step": 54090 }, { "epoch": 34.480560866794136, "grad_norm": 0.3333646357059479, "learning_rate": 9.542755819777139e-05, "loss": 0.0295, "step": 54100 }, { "epoch": 34.486934353091144, "grad_norm": 0.4125802516937256, "learning_rate": 9.542525533497743e-05, "loss": 0.0177, "step": 54110 }, { "epoch": 34.493307839388144, "grad_norm": 0.4044288694858551, "learning_rate": 9.542295192022374e-05, "loss": 0.0353, "step": 54120 }, { "epoch": 34.49968132568515, "grad_norm": 0.3926030695438385, "learning_rate": 9.542064795353831e-05, "loss": 0.0182, "step": 54130 }, { "epoch": 34.50605481198215, "grad_norm": 0.3920355439186096, "learning_rate": 9.541834343494911e-05, "loss": 0.0328, "step": 54140 }, { "epoch": 34.51242829827916, "grad_norm": 0.44078636169433594, "learning_rate": 9.541603836448417e-05, "loss": 0.02, "step": 54150 }, { "epoch": 34.51880178457616, "grad_norm": 0.31956973671913147, "learning_rate": 9.541373274217145e-05, "loss": 0.0377, "step": 54160 }, { "epoch": 34.52517527087317, "grad_norm": 0.3992490768432617, "learning_rate": 9.541142656803903e-05, "loss": 0.0325, "step": 54170 }, { "epoch": 34.53154875717017, "grad_norm": 0.8118324279785156, "learning_rate": 9.540911984211489e-05, "loss": 0.0326, "step": 54180 }, { "epoch": 34.53792224346718, "grad_norm": 0.4829874336719513, "learning_rate": 9.540681256442707e-05, "loss": 0.026, "step": 54190 }, { "epoch": 34.54429572976418, "grad_norm": 0.31661680340766907, "learning_rate": 9.54045047350036e-05, "loss": 0.0285, "step": 54200 }, { "epoch": 34.55066921606119, "grad_norm": 0.28740260004997253, "learning_rate": 9.540219635387253e-05, "loss": 0.0189, "step": 54210 }, { "epoch": 34.55704270235819, "grad_norm": 0.45540308952331543, "learning_rate": 9.539988742106191e-05, "loss": 0.0361, "step": 54220 }, { "epoch": 34.5634161886552, "grad_norm": 0.7181097865104675, "learning_rate": 9.53975779365998e-05, "loss": 0.0308, "step": 54230 }, { "epoch": 34.5697896749522, "grad_norm": 0.449716717004776, "learning_rate": 9.539526790051426e-05, "loss": 0.0214, "step": 54240 }, { "epoch": 34.576163161249205, "grad_norm": 0.5261098742485046, "learning_rate": 9.539295731283336e-05, "loss": 0.0302, "step": 54250 }, { "epoch": 34.582536647546206, "grad_norm": 0.5469397306442261, "learning_rate": 9.539064617358515e-05, "loss": 0.0325, "step": 54260 }, { "epoch": 34.588910133843214, "grad_norm": 0.5277042984962463, "learning_rate": 9.538833448279773e-05, "loss": 0.0227, "step": 54270 }, { "epoch": 34.595283620140215, "grad_norm": 0.4310486912727356, "learning_rate": 9.538602224049922e-05, "loss": 0.0258, "step": 54280 }, { "epoch": 34.60165710643722, "grad_norm": 0.5732216238975525, "learning_rate": 9.538370944671766e-05, "loss": 0.029, "step": 54290 }, { "epoch": 34.608030592734224, "grad_norm": 0.6305785775184631, "learning_rate": 9.53813961014812e-05, "loss": 0.0409, "step": 54300 }, { "epoch": 34.61440407903123, "grad_norm": 0.5343277454376221, "learning_rate": 9.537908220481791e-05, "loss": 0.0303, "step": 54310 }, { "epoch": 34.62077756532823, "grad_norm": 4.150912761688232, "learning_rate": 9.537676775675595e-05, "loss": 0.028, "step": 54320 }, { "epoch": 34.62715105162524, "grad_norm": 0.5228842496871948, "learning_rate": 9.537445275732341e-05, "loss": 0.0355, "step": 54330 }, { "epoch": 34.63352453792224, "grad_norm": 0.48362424969673157, "learning_rate": 9.537213720654843e-05, "loss": 0.035, "step": 54340 }, { "epoch": 34.63989802421925, "grad_norm": 0.5665836334228516, "learning_rate": 9.536982110445914e-05, "loss": 0.0269, "step": 54350 }, { "epoch": 34.64627151051625, "grad_norm": 0.64130699634552, "learning_rate": 9.536750445108367e-05, "loss": 0.0272, "step": 54360 }, { "epoch": 34.65264499681326, "grad_norm": 0.38070589303970337, "learning_rate": 9.536518724645022e-05, "loss": 0.0248, "step": 54370 }, { "epoch": 34.65901848311026, "grad_norm": 0.5269302129745483, "learning_rate": 9.53628694905869e-05, "loss": 0.0277, "step": 54380 }, { "epoch": 34.66539196940727, "grad_norm": 0.65503990650177, "learning_rate": 9.536055118352187e-05, "loss": 0.0479, "step": 54390 }, { "epoch": 34.67176545570427, "grad_norm": 0.45364704728126526, "learning_rate": 9.535823232528334e-05, "loss": 0.0215, "step": 54400 }, { "epoch": 34.678138942001276, "grad_norm": 0.4646311402320862, "learning_rate": 9.535591291589945e-05, "loss": 0.0329, "step": 54410 }, { "epoch": 34.68451242829828, "grad_norm": 0.9024292230606079, "learning_rate": 9.53535929553984e-05, "loss": 0.0274, "step": 54420 }, { "epoch": 34.690885914595285, "grad_norm": 0.7289599776268005, "learning_rate": 9.535127244380837e-05, "loss": 0.0274, "step": 54430 }, { "epoch": 34.697259400892285, "grad_norm": 0.42252904176712036, "learning_rate": 9.534895138115755e-05, "loss": 0.0196, "step": 54440 }, { "epoch": 34.70363288718929, "grad_norm": 0.6122227907180786, "learning_rate": 9.534662976747418e-05, "loss": 0.0498, "step": 54450 }, { "epoch": 34.710006373486294, "grad_norm": 0.48873403668403625, "learning_rate": 9.534430760278644e-05, "loss": 0.027, "step": 54460 }, { "epoch": 34.7163798597833, "grad_norm": 0.4227951169013977, "learning_rate": 9.534198488712253e-05, "loss": 0.0221, "step": 54470 }, { "epoch": 34.7227533460803, "grad_norm": 0.512873113155365, "learning_rate": 9.533966162051071e-05, "loss": 0.032, "step": 54480 }, { "epoch": 34.72912683237731, "grad_norm": 0.6129639744758606, "learning_rate": 9.533733780297919e-05, "loss": 0.0273, "step": 54490 }, { "epoch": 34.73550031867431, "grad_norm": 0.4442170262336731, "learning_rate": 9.53350134345562e-05, "loss": 0.0362, "step": 54500 }, { "epoch": 34.74187380497132, "grad_norm": 0.2548813819885254, "learning_rate": 9.533268851527001e-05, "loss": 0.0289, "step": 54510 }, { "epoch": 34.74824729126832, "grad_norm": 0.27000072598457336, "learning_rate": 9.533036304514884e-05, "loss": 0.0314, "step": 54520 }, { "epoch": 34.75462077756533, "grad_norm": 0.4531651437282562, "learning_rate": 9.532803702422098e-05, "loss": 0.0336, "step": 54530 }, { "epoch": 34.76099426386233, "grad_norm": 0.4736124277114868, "learning_rate": 9.532571045251465e-05, "loss": 0.0235, "step": 54540 }, { "epoch": 34.76736775015934, "grad_norm": 0.3686412274837494, "learning_rate": 9.532338333005815e-05, "loss": 0.0238, "step": 54550 }, { "epoch": 34.77374123645634, "grad_norm": 0.5054984092712402, "learning_rate": 9.532105565687976e-05, "loss": 0.0394, "step": 54560 }, { "epoch": 34.780114722753346, "grad_norm": 0.49800899624824524, "learning_rate": 9.531872743300774e-05, "loss": 0.021, "step": 54570 }, { "epoch": 34.78648820905035, "grad_norm": 0.3700025975704193, "learning_rate": 9.531639865847039e-05, "loss": 0.0244, "step": 54580 }, { "epoch": 34.792861695347355, "grad_norm": 0.39722928404808044, "learning_rate": 9.531406933329602e-05, "loss": 0.0259, "step": 54590 }, { "epoch": 34.79923518164436, "grad_norm": 0.5059988498687744, "learning_rate": 9.531173945751293e-05, "loss": 0.0236, "step": 54600 }, { "epoch": 34.805608667941364, "grad_norm": 0.48618635535240173, "learning_rate": 9.530940903114942e-05, "loss": 0.0219, "step": 54610 }, { "epoch": 34.81198215423837, "grad_norm": 0.49334630370140076, "learning_rate": 9.530707805423381e-05, "loss": 0.0348, "step": 54620 }, { "epoch": 34.81835564053537, "grad_norm": 0.4374242126941681, "learning_rate": 9.530474652679444e-05, "loss": 0.0409, "step": 54630 }, { "epoch": 34.82472912683238, "grad_norm": 0.661401093006134, "learning_rate": 9.530241444885961e-05, "loss": 0.0306, "step": 54640 }, { "epoch": 34.83110261312938, "grad_norm": 0.5589414834976196, "learning_rate": 9.530008182045766e-05, "loss": 0.0268, "step": 54650 }, { "epoch": 34.83747609942639, "grad_norm": 0.626943051815033, "learning_rate": 9.529774864161695e-05, "loss": 0.0254, "step": 54660 }, { "epoch": 34.84384958572339, "grad_norm": 0.5908910632133484, "learning_rate": 9.529541491236585e-05, "loss": 0.0273, "step": 54670 }, { "epoch": 34.8502230720204, "grad_norm": 0.3634604513645172, "learning_rate": 9.529308063273265e-05, "loss": 0.0218, "step": 54680 }, { "epoch": 34.8565965583174, "grad_norm": 0.5779263377189636, "learning_rate": 9.529074580274579e-05, "loss": 0.0247, "step": 54690 }, { "epoch": 34.86297004461441, "grad_norm": 0.4129598140716553, "learning_rate": 9.52884104224336e-05, "loss": 0.0397, "step": 54700 }, { "epoch": 34.86934353091141, "grad_norm": 0.5561365485191345, "learning_rate": 9.528607449182447e-05, "loss": 0.0272, "step": 54710 }, { "epoch": 34.875717017208416, "grad_norm": 0.7617877125740051, "learning_rate": 9.528373801094676e-05, "loss": 0.02, "step": 54720 }, { "epoch": 34.88209050350542, "grad_norm": 0.5030522346496582, "learning_rate": 9.528140097982889e-05, "loss": 0.035, "step": 54730 }, { "epoch": 34.888463989802425, "grad_norm": 0.4373551309108734, "learning_rate": 9.527906339849925e-05, "loss": 0.037, "step": 54740 }, { "epoch": 34.894837476099426, "grad_norm": 0.5599916577339172, "learning_rate": 9.527672526698624e-05, "loss": 0.0296, "step": 54750 }, { "epoch": 34.90121096239643, "grad_norm": 0.9060662984848022, "learning_rate": 9.527438658531826e-05, "loss": 0.0271, "step": 54760 }, { "epoch": 34.907584448693434, "grad_norm": 0.28670892119407654, "learning_rate": 9.527204735352374e-05, "loss": 0.026, "step": 54770 }, { "epoch": 34.91395793499044, "grad_norm": 0.49428582191467285, "learning_rate": 9.52697075716311e-05, "loss": 0.0417, "step": 54780 }, { "epoch": 34.92033142128744, "grad_norm": 0.6374714970588684, "learning_rate": 9.526736723966878e-05, "loss": 0.0271, "step": 54790 }, { "epoch": 34.92670490758445, "grad_norm": 0.5660468935966492, "learning_rate": 9.52650263576652e-05, "loss": 0.0277, "step": 54800 }, { "epoch": 34.93307839388145, "grad_norm": 0.3216094970703125, "learning_rate": 9.52626849256488e-05, "loss": 0.032, "step": 54810 }, { "epoch": 34.93945188017846, "grad_norm": 0.5788775682449341, "learning_rate": 9.526034294364807e-05, "loss": 0.0273, "step": 54820 }, { "epoch": 34.94582536647546, "grad_norm": 0.5020880699157715, "learning_rate": 9.525800041169142e-05, "loss": 0.0226, "step": 54830 }, { "epoch": 34.95219885277247, "grad_norm": 0.542186975479126, "learning_rate": 9.525565732980735e-05, "loss": 0.0286, "step": 54840 }, { "epoch": 34.95857233906947, "grad_norm": 0.2930257320404053, "learning_rate": 9.525331369802431e-05, "loss": 0.0336, "step": 54850 }, { "epoch": 34.96494582536648, "grad_norm": 0.3854008913040161, "learning_rate": 9.525096951637078e-05, "loss": 0.0238, "step": 54860 }, { "epoch": 34.97131931166348, "grad_norm": 0.4955374598503113, "learning_rate": 9.524862478487524e-05, "loss": 0.0243, "step": 54870 }, { "epoch": 34.977692797960486, "grad_norm": 0.4580172300338745, "learning_rate": 9.52462795035662e-05, "loss": 0.0214, "step": 54880 }, { "epoch": 34.98406628425749, "grad_norm": 0.5698228478431702, "learning_rate": 9.524393367247212e-05, "loss": 0.0247, "step": 54890 }, { "epoch": 34.990439770554495, "grad_norm": 0.955376386642456, "learning_rate": 9.524158729162155e-05, "loss": 0.0398, "step": 54900 }, { "epoch": 34.996813256851496, "grad_norm": 0.3933910131454468, "learning_rate": 9.523924036104297e-05, "loss": 0.0243, "step": 54910 }, { "epoch": 35.003186743148504, "grad_norm": 0.5582911372184753, "learning_rate": 9.523689288076491e-05, "loss": 0.0293, "step": 54920 }, { "epoch": 35.009560229445505, "grad_norm": 0.6082283854484558, "learning_rate": 9.52345448508159e-05, "loss": 0.0252, "step": 54930 }, { "epoch": 35.01593371574251, "grad_norm": 0.5585810542106628, "learning_rate": 9.523219627122445e-05, "loss": 0.0357, "step": 54940 }, { "epoch": 35.022307202039514, "grad_norm": 0.4258825480937958, "learning_rate": 9.522984714201911e-05, "loss": 0.022, "step": 54950 }, { "epoch": 35.02868068833652, "grad_norm": 0.522302508354187, "learning_rate": 9.522749746322844e-05, "loss": 0.0243, "step": 54960 }, { "epoch": 35.03505417463352, "grad_norm": 0.3039967715740204, "learning_rate": 9.522514723488095e-05, "loss": 0.0359, "step": 54970 }, { "epoch": 35.04142766093053, "grad_norm": 0.3911643624305725, "learning_rate": 9.522279645700522e-05, "loss": 0.0358, "step": 54980 }, { "epoch": 35.04780114722753, "grad_norm": 0.35110360383987427, "learning_rate": 9.522044512962984e-05, "loss": 0.0293, "step": 54990 }, { "epoch": 35.05417463352454, "grad_norm": 0.5617950558662415, "learning_rate": 9.521809325278335e-05, "loss": 0.0383, "step": 55000 }, { "epoch": 35.06054811982154, "grad_norm": 0.4283028841018677, "learning_rate": 9.521574082649431e-05, "loss": 0.0341, "step": 55010 }, { "epoch": 35.06692160611855, "grad_norm": 0.5666895508766174, "learning_rate": 9.521338785079134e-05, "loss": 0.0288, "step": 55020 }, { "epoch": 35.07329509241555, "grad_norm": 0.381862074136734, "learning_rate": 9.521103432570304e-05, "loss": 0.0352, "step": 55030 }, { "epoch": 35.07966857871256, "grad_norm": 1.3599945306777954, "learning_rate": 9.520868025125798e-05, "loss": 0.0284, "step": 55040 }, { "epoch": 35.08604206500956, "grad_norm": 0.6772026419639587, "learning_rate": 9.520632562748474e-05, "loss": 0.0346, "step": 55050 }, { "epoch": 35.092415551306566, "grad_norm": 0.4685651361942291, "learning_rate": 9.5203970454412e-05, "loss": 0.0257, "step": 55060 }, { "epoch": 35.09878903760357, "grad_norm": 0.5068112015724182, "learning_rate": 9.520161473206832e-05, "loss": 0.0318, "step": 55070 }, { "epoch": 35.105162523900574, "grad_norm": 0.38628146052360535, "learning_rate": 9.519925846048234e-05, "loss": 0.0299, "step": 55080 }, { "epoch": 35.111536010197575, "grad_norm": 0.4938599169254303, "learning_rate": 9.519690163968272e-05, "loss": 0.0234, "step": 55090 }, { "epoch": 35.11790949649458, "grad_norm": 0.4402371048927307, "learning_rate": 9.519454426969803e-05, "loss": 0.0356, "step": 55100 }, { "epoch": 35.124282982791584, "grad_norm": 1.0453944206237793, "learning_rate": 9.5192186350557e-05, "loss": 0.0309, "step": 55110 }, { "epoch": 35.13065646908859, "grad_norm": 0.3025345206260681, "learning_rate": 9.518982788228821e-05, "loss": 0.0282, "step": 55120 }, { "epoch": 35.13702995538559, "grad_norm": 0.4899325966835022, "learning_rate": 9.518746886492035e-05, "loss": 0.0252, "step": 55130 }, { "epoch": 35.1434034416826, "grad_norm": 0.4341687262058258, "learning_rate": 9.518510929848208e-05, "loss": 0.0301, "step": 55140 }, { "epoch": 35.1497769279796, "grad_norm": 0.5626493096351624, "learning_rate": 9.518274918300208e-05, "loss": 0.0349, "step": 55150 }, { "epoch": 35.15615041427661, "grad_norm": 0.3852962851524353, "learning_rate": 9.5180388518509e-05, "loss": 0.0283, "step": 55160 }, { "epoch": 35.16252390057361, "grad_norm": 0.5742796659469604, "learning_rate": 9.517802730503156e-05, "loss": 0.0266, "step": 55170 }, { "epoch": 35.16889738687062, "grad_norm": 0.35183802247047424, "learning_rate": 9.517566554259842e-05, "loss": 0.0202, "step": 55180 }, { "epoch": 35.17527087316762, "grad_norm": 0.7214723825454712, "learning_rate": 9.51733032312383e-05, "loss": 0.0347, "step": 55190 }, { "epoch": 35.18164435946463, "grad_norm": 0.5835719108581543, "learning_rate": 9.51709403709799e-05, "loss": 0.0312, "step": 55200 }, { "epoch": 35.18801784576163, "grad_norm": 0.3250637650489807, "learning_rate": 9.51685769618519e-05, "loss": 0.0226, "step": 55210 }, { "epoch": 35.194391332058636, "grad_norm": 0.37022122740745544, "learning_rate": 9.516621300388306e-05, "loss": 0.0436, "step": 55220 }, { "epoch": 35.20076481835564, "grad_norm": 0.469418466091156, "learning_rate": 9.51638484971021e-05, "loss": 0.0297, "step": 55230 }, { "epoch": 35.207138304652645, "grad_norm": 0.5068747401237488, "learning_rate": 9.516148344153773e-05, "loss": 0.0325, "step": 55240 }, { "epoch": 35.21351179094965, "grad_norm": 0.596798300743103, "learning_rate": 9.515911783721868e-05, "loss": 0.0405, "step": 55250 }, { "epoch": 35.219885277246654, "grad_norm": 0.4990018904209137, "learning_rate": 9.515675168417372e-05, "loss": 0.0284, "step": 55260 }, { "epoch": 35.22625876354366, "grad_norm": 0.44395509362220764, "learning_rate": 9.515438498243161e-05, "loss": 0.0233, "step": 55270 }, { "epoch": 35.23263224984066, "grad_norm": 0.22774653136730194, "learning_rate": 9.515201773202106e-05, "loss": 0.0316, "step": 55280 }, { "epoch": 35.23900573613767, "grad_norm": 0.5739774107933044, "learning_rate": 9.514964993297089e-05, "loss": 0.04, "step": 55290 }, { "epoch": 35.24537922243467, "grad_norm": 0.3615114390850067, "learning_rate": 9.514728158530983e-05, "loss": 0.0371, "step": 55300 }, { "epoch": 35.25175270873168, "grad_norm": 0.41419804096221924, "learning_rate": 9.51449126890667e-05, "loss": 0.0224, "step": 55310 }, { "epoch": 35.25812619502868, "grad_norm": 0.5439110994338989, "learning_rate": 9.514254324427022e-05, "loss": 0.0207, "step": 55320 }, { "epoch": 35.26449968132569, "grad_norm": 0.5592920184135437, "learning_rate": 9.514017325094923e-05, "loss": 0.0262, "step": 55330 }, { "epoch": 35.27087316762269, "grad_norm": 0.5115230083465576, "learning_rate": 9.513780270913253e-05, "loss": 0.0319, "step": 55340 }, { "epoch": 35.2772466539197, "grad_norm": 0.635106086730957, "learning_rate": 9.51354316188489e-05, "loss": 0.0295, "step": 55350 }, { "epoch": 35.2836201402167, "grad_norm": 0.3669269382953644, "learning_rate": 9.513305998012716e-05, "loss": 0.0231, "step": 55360 }, { "epoch": 35.289993626513706, "grad_norm": 0.6878235936164856, "learning_rate": 9.513068779299612e-05, "loss": 0.029, "step": 55370 }, { "epoch": 35.29636711281071, "grad_norm": 0.4709893763065338, "learning_rate": 9.512831505748463e-05, "loss": 0.0277, "step": 55380 }, { "epoch": 35.302740599107715, "grad_norm": 0.8442989587783813, "learning_rate": 9.512594177362148e-05, "loss": 0.0354, "step": 55390 }, { "epoch": 35.309114085404715, "grad_norm": 0.24564409255981445, "learning_rate": 9.512356794143556e-05, "loss": 0.0259, "step": 55400 }, { "epoch": 35.31548757170172, "grad_norm": 0.21884040534496307, "learning_rate": 9.512119356095567e-05, "loss": 0.0216, "step": 55410 }, { "epoch": 35.321861057998724, "grad_norm": 0.42159944772720337, "learning_rate": 9.511881863221068e-05, "loss": 0.0327, "step": 55420 }, { "epoch": 35.32823454429573, "grad_norm": 0.4503278434276581, "learning_rate": 9.511644315522945e-05, "loss": 0.0338, "step": 55430 }, { "epoch": 35.33460803059273, "grad_norm": 0.5629932284355164, "learning_rate": 9.511406713004084e-05, "loss": 0.0256, "step": 55440 }, { "epoch": 35.34098151688974, "grad_norm": 0.3847126364707947, "learning_rate": 9.511169055667371e-05, "loss": 0.0284, "step": 55450 }, { "epoch": 35.34735500318674, "grad_norm": 0.3572671413421631, "learning_rate": 9.510931343515694e-05, "loss": 0.0189, "step": 55460 }, { "epoch": 35.35372848948375, "grad_norm": 0.5695170760154724, "learning_rate": 9.510693576551943e-05, "loss": 0.0322, "step": 55470 }, { "epoch": 35.36010197578075, "grad_norm": 0.5786896347999573, "learning_rate": 9.510455754779006e-05, "loss": 0.0253, "step": 55480 }, { "epoch": 35.36647546207776, "grad_norm": 0.644467294216156, "learning_rate": 9.510217878199774e-05, "loss": 0.0231, "step": 55490 }, { "epoch": 35.37284894837476, "grad_norm": 0.47169211506843567, "learning_rate": 9.509979946817136e-05, "loss": 0.029, "step": 55500 }, { "epoch": 35.37922243467177, "grad_norm": 0.49006518721580505, "learning_rate": 9.509741960633981e-05, "loss": 0.0265, "step": 55510 }, { "epoch": 35.38559592096877, "grad_norm": 0.5282463431358337, "learning_rate": 9.509503919653207e-05, "loss": 0.0198, "step": 55520 }, { "epoch": 35.391969407265776, "grad_norm": 1.186556339263916, "learning_rate": 9.5092658238777e-05, "loss": 0.0295, "step": 55530 }, { "epoch": 35.39834289356278, "grad_norm": 0.3059556186199188, "learning_rate": 9.509027673310358e-05, "loss": 0.0318, "step": 55540 }, { "epoch": 35.404716379859785, "grad_norm": 0.41900452971458435, "learning_rate": 9.508789467954072e-05, "loss": 0.0278, "step": 55550 }, { "epoch": 35.411089866156786, "grad_norm": 0.7526875734329224, "learning_rate": 9.508551207811735e-05, "loss": 0.0279, "step": 55560 }, { "epoch": 35.417463352453794, "grad_norm": 0.5941905975341797, "learning_rate": 9.508312892886247e-05, "loss": 0.0369, "step": 55570 }, { "epoch": 35.423836838750795, "grad_norm": 0.37143465876579285, "learning_rate": 9.5080745231805e-05, "loss": 0.0272, "step": 55580 }, { "epoch": 35.4302103250478, "grad_norm": 0.45726409554481506, "learning_rate": 9.507836098697389e-05, "loss": 0.0271, "step": 55590 }, { "epoch": 35.4365838113448, "grad_norm": 0.38343968987464905, "learning_rate": 9.507597619439816e-05, "loss": 0.0245, "step": 55600 }, { "epoch": 35.44295729764181, "grad_norm": 0.298199862241745, "learning_rate": 9.507359085410675e-05, "loss": 0.027, "step": 55610 }, { "epoch": 35.44933078393881, "grad_norm": 0.6016449332237244, "learning_rate": 9.507120496612865e-05, "loss": 0.0313, "step": 55620 }, { "epoch": 35.45570427023582, "grad_norm": 0.34862378239631653, "learning_rate": 9.506881853049285e-05, "loss": 0.0174, "step": 55630 }, { "epoch": 35.46207775653282, "grad_norm": 0.29296115040779114, "learning_rate": 9.506643154722837e-05, "loss": 0.0238, "step": 55640 }, { "epoch": 35.46845124282983, "grad_norm": 0.4980362057685852, "learning_rate": 9.50640440163642e-05, "loss": 0.0333, "step": 55650 }, { "epoch": 35.47482472912683, "grad_norm": 0.40781158208847046, "learning_rate": 9.506165593792933e-05, "loss": 0.0246, "step": 55660 }, { "epoch": 35.48119821542384, "grad_norm": 0.6440739035606384, "learning_rate": 9.50592673119528e-05, "loss": 0.032, "step": 55670 }, { "epoch": 35.48757170172084, "grad_norm": 0.31374576687812805, "learning_rate": 9.505687813846363e-05, "loss": 0.0252, "step": 55680 }, { "epoch": 35.49394518801785, "grad_norm": 0.5982542037963867, "learning_rate": 9.505448841749085e-05, "loss": 0.0381, "step": 55690 }, { "epoch": 35.50031867431485, "grad_norm": 0.5191935896873474, "learning_rate": 9.505209814906351e-05, "loss": 0.0378, "step": 55700 }, { "epoch": 35.506692160611856, "grad_norm": 0.6178092956542969, "learning_rate": 9.504970733321062e-05, "loss": 0.027, "step": 55710 }, { "epoch": 35.513065646908856, "grad_norm": 0.3382032513618469, "learning_rate": 9.504731596996125e-05, "loss": 0.0292, "step": 55720 }, { "epoch": 35.519439133205864, "grad_norm": 0.4072098135948181, "learning_rate": 9.504492405934448e-05, "loss": 0.0207, "step": 55730 }, { "epoch": 35.525812619502865, "grad_norm": 0.5382463932037354, "learning_rate": 9.504253160138934e-05, "loss": 0.0239, "step": 55740 }, { "epoch": 35.53218610579987, "grad_norm": 0.3895913064479828, "learning_rate": 9.504013859612493e-05, "loss": 0.028, "step": 55750 }, { "epoch": 35.538559592096874, "grad_norm": 0.5287757515907288, "learning_rate": 9.503774504358029e-05, "loss": 0.0402, "step": 55760 }, { "epoch": 35.54493307839388, "grad_norm": 0.28353437781333923, "learning_rate": 9.503535094378453e-05, "loss": 0.021, "step": 55770 }, { "epoch": 35.55130656469088, "grad_norm": 0.2986859381198883, "learning_rate": 9.503295629676674e-05, "loss": 0.04, "step": 55780 }, { "epoch": 35.55768005098789, "grad_norm": 0.5009360909461975, "learning_rate": 9.503056110255601e-05, "loss": 0.022, "step": 55790 }, { "epoch": 35.56405353728489, "grad_norm": 0.41437825560569763, "learning_rate": 9.502816536118147e-05, "loss": 0.0378, "step": 55800 }, { "epoch": 35.5704270235819, "grad_norm": 0.384902685880661, "learning_rate": 9.502576907267218e-05, "loss": 0.0293, "step": 55810 }, { "epoch": 35.5768005098789, "grad_norm": 0.46506819128990173, "learning_rate": 9.502337223705729e-05, "loss": 0.0308, "step": 55820 }, { "epoch": 35.58317399617591, "grad_norm": 0.5410088896751404, "learning_rate": 9.502097485436591e-05, "loss": 0.0236, "step": 55830 }, { "epoch": 35.58954748247291, "grad_norm": 0.3385242819786072, "learning_rate": 9.501857692462719e-05, "loss": 0.0343, "step": 55840 }, { "epoch": 35.59592096876992, "grad_norm": 0.46601220965385437, "learning_rate": 9.501617844787025e-05, "loss": 0.0268, "step": 55850 }, { "epoch": 35.602294455066925, "grad_norm": 0.804152250289917, "learning_rate": 9.501377942412425e-05, "loss": 0.0273, "step": 55860 }, { "epoch": 35.608667941363926, "grad_norm": 0.40002262592315674, "learning_rate": 9.501137985341832e-05, "loss": 0.0167, "step": 55870 }, { "epoch": 35.615041427660934, "grad_norm": 0.5190397500991821, "learning_rate": 9.500897973578162e-05, "loss": 0.0314, "step": 55880 }, { "epoch": 35.621414913957935, "grad_norm": 0.2816859185695648, "learning_rate": 9.500657907124334e-05, "loss": 0.0268, "step": 55890 }, { "epoch": 35.62778840025494, "grad_norm": 0.33525821566581726, "learning_rate": 9.500417785983262e-05, "loss": 0.0267, "step": 55900 }, { "epoch": 35.634161886551944, "grad_norm": 0.5705966353416443, "learning_rate": 9.500177610157865e-05, "loss": 0.0236, "step": 55910 }, { "epoch": 35.64053537284895, "grad_norm": 0.7921620607376099, "learning_rate": 9.499937379651062e-05, "loss": 0.0314, "step": 55920 }, { "epoch": 35.64690885914595, "grad_norm": 0.3798553943634033, "learning_rate": 9.499697094465769e-05, "loss": 0.0266, "step": 55930 }, { "epoch": 35.65328234544296, "grad_norm": 0.580742359161377, "learning_rate": 9.499456754604909e-05, "loss": 0.0261, "step": 55940 }, { "epoch": 35.65965583173996, "grad_norm": 0.38662663102149963, "learning_rate": 9.499216360071402e-05, "loss": 0.0338, "step": 55950 }, { "epoch": 35.66602931803697, "grad_norm": 0.3803071677684784, "learning_rate": 9.498975910868168e-05, "loss": 0.0221, "step": 55960 }, { "epoch": 35.67240280433397, "grad_norm": 0.4569070041179657, "learning_rate": 9.498735406998128e-05, "loss": 0.0245, "step": 55970 }, { "epoch": 35.67877629063098, "grad_norm": 0.2580690085887909, "learning_rate": 9.498494848464206e-05, "loss": 0.0173, "step": 55980 }, { "epoch": 35.68514977692798, "grad_norm": 0.5265932083129883, "learning_rate": 9.498254235269325e-05, "loss": 0.0279, "step": 55990 }, { "epoch": 35.69152326322499, "grad_norm": 0.2920020520687103, "learning_rate": 9.498013567416407e-05, "loss": 0.03, "step": 56000 }, { "epoch": 35.69789674952199, "grad_norm": 0.3040582239627838, "learning_rate": 9.497772844908376e-05, "loss": 0.0193, "step": 56010 }, { "epoch": 35.704270235818996, "grad_norm": 0.4648805558681488, "learning_rate": 9.497532067748162e-05, "loss": 0.0209, "step": 56020 }, { "epoch": 35.710643722116, "grad_norm": 0.5383450388908386, "learning_rate": 9.497291235938685e-05, "loss": 0.0336, "step": 56030 }, { "epoch": 35.717017208413004, "grad_norm": 0.5212481617927551, "learning_rate": 9.497050349482872e-05, "loss": 0.0258, "step": 56040 }, { "epoch": 35.723390694710005, "grad_norm": 0.5252798199653625, "learning_rate": 9.496809408383654e-05, "loss": 0.032, "step": 56050 }, { "epoch": 35.72976418100701, "grad_norm": 0.2740311920642853, "learning_rate": 9.496568412643954e-05, "loss": 0.0213, "step": 56060 }, { "epoch": 35.736137667304014, "grad_norm": 0.718986988067627, "learning_rate": 9.496327362266702e-05, "loss": 0.0344, "step": 56070 }, { "epoch": 35.74251115360102, "grad_norm": 0.5655689239501953, "learning_rate": 9.496086257254829e-05, "loss": 0.023, "step": 56080 }, { "epoch": 35.74888463989802, "grad_norm": 0.34078076481819153, "learning_rate": 9.495845097611262e-05, "loss": 0.0358, "step": 56090 }, { "epoch": 35.75525812619503, "grad_norm": 0.6018723249435425, "learning_rate": 9.495603883338932e-05, "loss": 0.0304, "step": 56100 }, { "epoch": 35.76163161249203, "grad_norm": 0.6271097660064697, "learning_rate": 9.495362614440771e-05, "loss": 0.0243, "step": 56110 }, { "epoch": 35.76800509878904, "grad_norm": 0.5130691528320312, "learning_rate": 9.49512129091971e-05, "loss": 0.0262, "step": 56120 }, { "epoch": 35.77437858508604, "grad_norm": 0.5798001289367676, "learning_rate": 9.49487991277868e-05, "loss": 0.0295, "step": 56130 }, { "epoch": 35.78075207138305, "grad_norm": 0.4628540575504303, "learning_rate": 9.494638480020617e-05, "loss": 0.0202, "step": 56140 }, { "epoch": 35.78712555768005, "grad_norm": 0.2786974012851715, "learning_rate": 9.494396992648451e-05, "loss": 0.0324, "step": 56150 }, { "epoch": 35.79349904397706, "grad_norm": 0.8057183027267456, "learning_rate": 9.494155450665119e-05, "loss": 0.0324, "step": 56160 }, { "epoch": 35.79987253027406, "grad_norm": 0.6558979749679565, "learning_rate": 9.493913854073554e-05, "loss": 0.0263, "step": 56170 }, { "epoch": 35.806246016571066, "grad_norm": 0.44564953446388245, "learning_rate": 9.493672202876696e-05, "loss": 0.0217, "step": 56180 }, { "epoch": 35.81261950286807, "grad_norm": 0.37708356976509094, "learning_rate": 9.493430497077476e-05, "loss": 0.0222, "step": 56190 }, { "epoch": 35.818992989165075, "grad_norm": 0.461723268032074, "learning_rate": 9.493188736678831e-05, "loss": 0.0257, "step": 56200 }, { "epoch": 35.825366475462076, "grad_norm": 0.28750985860824585, "learning_rate": 9.492946921683703e-05, "loss": 0.0257, "step": 56210 }, { "epoch": 35.831739961759084, "grad_norm": 0.2941114008426666, "learning_rate": 9.492705052095028e-05, "loss": 0.0272, "step": 56220 }, { "epoch": 35.838113448056085, "grad_norm": 0.5967600345611572, "learning_rate": 9.492463127915742e-05, "loss": 0.0326, "step": 56230 }, { "epoch": 35.84448693435309, "grad_norm": 0.3645625710487366, "learning_rate": 9.49222114914879e-05, "loss": 0.0237, "step": 56240 }, { "epoch": 35.85086042065009, "grad_norm": 0.7042148113250732, "learning_rate": 9.491979115797109e-05, "loss": 0.03, "step": 56250 }, { "epoch": 35.8572339069471, "grad_norm": 0.33184170722961426, "learning_rate": 9.49173702786364e-05, "loss": 0.0271, "step": 56260 }, { "epoch": 35.8636073932441, "grad_norm": 0.4953864812850952, "learning_rate": 9.491494885351328e-05, "loss": 0.0255, "step": 56270 }, { "epoch": 35.86998087954111, "grad_norm": 0.40998318791389465, "learning_rate": 9.49125268826311e-05, "loss": 0.0429, "step": 56280 }, { "epoch": 35.87635436583811, "grad_norm": 0.5258964896202087, "learning_rate": 9.491010436601931e-05, "loss": 0.0268, "step": 56290 }, { "epoch": 35.88272785213512, "grad_norm": 0.47657454013824463, "learning_rate": 9.490768130370736e-05, "loss": 0.0398, "step": 56300 }, { "epoch": 35.88910133843212, "grad_norm": 0.5450493693351746, "learning_rate": 9.490525769572468e-05, "loss": 0.0208, "step": 56310 }, { "epoch": 35.89547482472913, "grad_norm": 0.4829743802547455, "learning_rate": 9.490283354210072e-05, "loss": 0.0261, "step": 56320 }, { "epoch": 35.90184831102613, "grad_norm": 0.7247356176376343, "learning_rate": 9.490040884286493e-05, "loss": 0.0234, "step": 56330 }, { "epoch": 35.90822179732314, "grad_norm": 0.3666912317276001, "learning_rate": 9.48979835980468e-05, "loss": 0.0202, "step": 56340 }, { "epoch": 35.91459528362014, "grad_norm": 0.3548961579799652, "learning_rate": 9.489555780767576e-05, "loss": 0.0198, "step": 56350 }, { "epoch": 35.920968769917145, "grad_norm": 0.6006926894187927, "learning_rate": 9.489313147178131e-05, "loss": 0.03, "step": 56360 }, { "epoch": 35.927342256214146, "grad_norm": 0.5787987112998962, "learning_rate": 9.489070459039293e-05, "loss": 0.0269, "step": 56370 }, { "epoch": 35.933715742511154, "grad_norm": 0.5361904501914978, "learning_rate": 9.48882771635401e-05, "loss": 0.0237, "step": 56380 }, { "epoch": 35.940089228808155, "grad_norm": 0.4886249303817749, "learning_rate": 9.488584919125232e-05, "loss": 0.0254, "step": 56390 }, { "epoch": 35.94646271510516, "grad_norm": 0.530318558216095, "learning_rate": 9.48834206735591e-05, "loss": 0.023, "step": 56400 }, { "epoch": 35.952836201402164, "grad_norm": 0.3418777883052826, "learning_rate": 9.488099161048994e-05, "loss": 0.0315, "step": 56410 }, { "epoch": 35.95920968769917, "grad_norm": 0.3609558939933777, "learning_rate": 9.487856200207436e-05, "loss": 0.0345, "step": 56420 }, { "epoch": 35.96558317399617, "grad_norm": 0.424207478761673, "learning_rate": 9.487613184834187e-05, "loss": 0.0256, "step": 56430 }, { "epoch": 35.97195666029318, "grad_norm": 0.26957613229751587, "learning_rate": 9.487370114932202e-05, "loss": 0.0173, "step": 56440 }, { "epoch": 35.97833014659018, "grad_norm": 0.2965833842754364, "learning_rate": 9.487126990504431e-05, "loss": 0.0249, "step": 56450 }, { "epoch": 35.98470363288719, "grad_norm": 0.32169368863105774, "learning_rate": 9.486883811553833e-05, "loss": 0.0302, "step": 56460 }, { "epoch": 35.99107711918419, "grad_norm": 0.7866684794425964, "learning_rate": 9.486640578083359e-05, "loss": 0.0257, "step": 56470 }, { "epoch": 35.9974506054812, "grad_norm": 0.5600200891494751, "learning_rate": 9.486397290095968e-05, "loss": 0.0239, "step": 56480 }, { "epoch": 36.003824091778206, "grad_norm": 0.2826383411884308, "learning_rate": 9.486153947594612e-05, "loss": 0.0227, "step": 56490 }, { "epoch": 36.01019757807521, "grad_norm": 0.49822118878364563, "learning_rate": 9.48591055058225e-05, "loss": 0.0229, "step": 56500 }, { "epoch": 36.016571064372215, "grad_norm": 0.599690318107605, "learning_rate": 9.485667099061839e-05, "loss": 0.0368, "step": 56510 }, { "epoch": 36.022944550669216, "grad_norm": 0.5329281091690063, "learning_rate": 9.485423593036339e-05, "loss": 0.0273, "step": 56520 }, { "epoch": 36.029318036966224, "grad_norm": 0.9028069376945496, "learning_rate": 9.485180032508708e-05, "loss": 0.025, "step": 56530 }, { "epoch": 36.035691523263225, "grad_norm": 0.3643564283847809, "learning_rate": 9.484936417481902e-05, "loss": 0.0332, "step": 56540 }, { "epoch": 36.04206500956023, "grad_norm": 0.5311803817749023, "learning_rate": 9.484692747958885e-05, "loss": 0.026, "step": 56550 }, { "epoch": 36.04843849585723, "grad_norm": 0.9790709018707275, "learning_rate": 9.484449023942619e-05, "loss": 0.0483, "step": 56560 }, { "epoch": 36.05481198215424, "grad_norm": 0.4993636906147003, "learning_rate": 9.484205245436062e-05, "loss": 0.0312, "step": 56570 }, { "epoch": 36.06118546845124, "grad_norm": 0.6763119697570801, "learning_rate": 9.483961412442177e-05, "loss": 0.0218, "step": 56580 }, { "epoch": 36.06755895474825, "grad_norm": 0.6167675256729126, "learning_rate": 9.483717524963926e-05, "loss": 0.0295, "step": 56590 }, { "epoch": 36.07393244104525, "grad_norm": 0.614103376865387, "learning_rate": 9.483473583004276e-05, "loss": 0.017, "step": 56600 }, { "epoch": 36.08030592734226, "grad_norm": 0.41841888427734375, "learning_rate": 9.483229586566189e-05, "loss": 0.0365, "step": 56610 }, { "epoch": 36.08667941363926, "grad_norm": 0.4662361145019531, "learning_rate": 9.482985535652629e-05, "loss": 0.0342, "step": 56620 }, { "epoch": 36.09305289993627, "grad_norm": 0.48006853461265564, "learning_rate": 9.482741430266561e-05, "loss": 0.0246, "step": 56630 }, { "epoch": 36.09942638623327, "grad_norm": 0.4246199429035187, "learning_rate": 9.482497270410953e-05, "loss": 0.0198, "step": 56640 }, { "epoch": 36.10579987253028, "grad_norm": 0.4166896939277649, "learning_rate": 9.482253056088773e-05, "loss": 0.0372, "step": 56650 }, { "epoch": 36.11217335882728, "grad_norm": 0.7195228338241577, "learning_rate": 9.482008787302983e-05, "loss": 0.0218, "step": 56660 }, { "epoch": 36.118546845124285, "grad_norm": 0.4906434118747711, "learning_rate": 9.481764464056556e-05, "loss": 0.0374, "step": 56670 }, { "epoch": 36.124920331421286, "grad_norm": 0.448713481426239, "learning_rate": 9.48152008635246e-05, "loss": 0.0178, "step": 56680 }, { "epoch": 36.131293817718294, "grad_norm": 0.29854315519332886, "learning_rate": 9.481275654193661e-05, "loss": 0.0338, "step": 56690 }, { "epoch": 36.137667304015295, "grad_norm": 0.4892030954360962, "learning_rate": 9.481031167583135e-05, "loss": 0.0234, "step": 56700 }, { "epoch": 36.1440407903123, "grad_norm": 0.7320384979248047, "learning_rate": 9.480786626523849e-05, "loss": 0.0369, "step": 56710 }, { "epoch": 36.150414276609304, "grad_norm": 0.4274139106273651, "learning_rate": 9.480542031018775e-05, "loss": 0.0186, "step": 56720 }, { "epoch": 36.15678776290631, "grad_norm": 0.34931451082229614, "learning_rate": 9.480297381070883e-05, "loss": 0.0262, "step": 56730 }, { "epoch": 36.16316124920331, "grad_norm": 0.4003312289714813, "learning_rate": 9.48005267668315e-05, "loss": 0.0196, "step": 56740 }, { "epoch": 36.16953473550032, "grad_norm": 0.5395488739013672, "learning_rate": 9.479807917858545e-05, "loss": 0.0301, "step": 56750 }, { "epoch": 36.17590822179732, "grad_norm": 0.457661509513855, "learning_rate": 9.479563104600046e-05, "loss": 0.0394, "step": 56760 }, { "epoch": 36.18228170809433, "grad_norm": 0.2268311232328415, "learning_rate": 9.479318236910625e-05, "loss": 0.0164, "step": 56770 }, { "epoch": 36.18865519439133, "grad_norm": 0.4309782385826111, "learning_rate": 9.479073314793259e-05, "loss": 0.0263, "step": 56780 }, { "epoch": 36.19502868068834, "grad_norm": 0.7037017941474915, "learning_rate": 9.478828338250922e-05, "loss": 0.0285, "step": 56790 }, { "epoch": 36.20140216698534, "grad_norm": 0.6677271127700806, "learning_rate": 9.478583307286592e-05, "loss": 0.0382, "step": 56800 }, { "epoch": 36.20777565328235, "grad_norm": 0.332619309425354, "learning_rate": 9.478338221903248e-05, "loss": 0.0167, "step": 56810 }, { "epoch": 36.21414913957935, "grad_norm": 0.4014318287372589, "learning_rate": 9.478093082103865e-05, "loss": 0.037, "step": 56820 }, { "epoch": 36.220522625876356, "grad_norm": 0.663598358631134, "learning_rate": 9.477847887891423e-05, "loss": 0.0286, "step": 56830 }, { "epoch": 36.22689611217336, "grad_norm": 0.7025123834609985, "learning_rate": 9.477602639268902e-05, "loss": 0.0284, "step": 56840 }, { "epoch": 36.233269598470365, "grad_norm": 0.5149930715560913, "learning_rate": 9.477357336239281e-05, "loss": 0.0205, "step": 56850 }, { "epoch": 36.239643084767366, "grad_norm": 0.47478237748146057, "learning_rate": 9.477111978805542e-05, "loss": 0.0344, "step": 56860 }, { "epoch": 36.246016571064374, "grad_norm": 0.32208940386772156, "learning_rate": 9.476866566970664e-05, "loss": 0.0451, "step": 56870 }, { "epoch": 36.252390057361374, "grad_norm": 0.3925299048423767, "learning_rate": 9.47662110073763e-05, "loss": 0.0354, "step": 56880 }, { "epoch": 36.25876354365838, "grad_norm": 0.2112465500831604, "learning_rate": 9.476375580109425e-05, "loss": 0.0187, "step": 56890 }, { "epoch": 36.26513702995538, "grad_norm": 0.5465281009674072, "learning_rate": 9.476130005089029e-05, "loss": 0.0351, "step": 56900 }, { "epoch": 36.27151051625239, "grad_norm": 0.46439221501350403, "learning_rate": 9.475884375679427e-05, "loss": 0.0317, "step": 56910 }, { "epoch": 36.27788400254939, "grad_norm": 0.4916335642337799, "learning_rate": 9.475638691883604e-05, "loss": 0.0352, "step": 56920 }, { "epoch": 36.2842574888464, "grad_norm": 0.4705188572406769, "learning_rate": 9.475392953704546e-05, "loss": 0.0265, "step": 56930 }, { "epoch": 36.2906309751434, "grad_norm": 0.5308962464332581, "learning_rate": 9.475147161145237e-05, "loss": 0.0277, "step": 56940 }, { "epoch": 36.29700446144041, "grad_norm": 0.4140111207962036, "learning_rate": 9.474901314208666e-05, "loss": 0.021, "step": 56950 }, { "epoch": 36.30337794773741, "grad_norm": 0.257232129573822, "learning_rate": 9.474655412897818e-05, "loss": 0.0204, "step": 56960 }, { "epoch": 36.30975143403442, "grad_norm": 0.4791267216205597, "learning_rate": 9.474409457215683e-05, "loss": 0.0245, "step": 56970 }, { "epoch": 36.31612492033142, "grad_norm": 0.47617340087890625, "learning_rate": 9.474163447165247e-05, "loss": 0.0294, "step": 56980 }, { "epoch": 36.322498406628426, "grad_norm": 0.4410161077976227, "learning_rate": 9.473917382749502e-05, "loss": 0.0263, "step": 56990 }, { "epoch": 36.32887189292543, "grad_norm": 0.5325472950935364, "learning_rate": 9.473671263971438e-05, "loss": 0.0277, "step": 57000 }, { "epoch": 36.335245379222435, "grad_norm": 0.4779888391494751, "learning_rate": 9.473425090834041e-05, "loss": 0.0218, "step": 57010 }, { "epoch": 36.341618865519436, "grad_norm": 0.8897963166236877, "learning_rate": 9.473178863340309e-05, "loss": 0.0502, "step": 57020 }, { "epoch": 36.347992351816444, "grad_norm": 0.45764678716659546, "learning_rate": 9.472932581493228e-05, "loss": 0.0277, "step": 57030 }, { "epoch": 36.354365838113445, "grad_norm": 0.5744030475616455, "learning_rate": 9.472686245295793e-05, "loss": 0.0233, "step": 57040 }, { "epoch": 36.36073932441045, "grad_norm": 0.2924349904060364, "learning_rate": 9.472439854750999e-05, "loss": 0.0251, "step": 57050 }, { "epoch": 36.367112810707454, "grad_norm": 0.43266424536705017, "learning_rate": 9.472193409861836e-05, "loss": 0.0391, "step": 57060 }, { "epoch": 36.37348629700446, "grad_norm": 0.6007348895072937, "learning_rate": 9.471946910631302e-05, "loss": 0.0259, "step": 57070 }, { "epoch": 36.37985978330146, "grad_norm": 0.4023219645023346, "learning_rate": 9.47170035706239e-05, "loss": 0.0148, "step": 57080 }, { "epoch": 36.38623326959847, "grad_norm": 0.4193379580974579, "learning_rate": 9.471453749158096e-05, "loss": 0.0303, "step": 57090 }, { "epoch": 36.39260675589547, "grad_norm": 0.5523746013641357, "learning_rate": 9.47120708692142e-05, "loss": 0.0215, "step": 57100 }, { "epoch": 36.39898024219248, "grad_norm": 0.5907568335533142, "learning_rate": 9.470960370355356e-05, "loss": 0.0269, "step": 57110 }, { "epoch": 36.40535372848949, "grad_norm": 0.43363630771636963, "learning_rate": 9.4707135994629e-05, "loss": 0.0294, "step": 57120 }, { "epoch": 36.41172721478649, "grad_norm": 0.6485674977302551, "learning_rate": 9.470466774247054e-05, "loss": 0.0273, "step": 57130 }, { "epoch": 36.418100701083496, "grad_norm": 0.4767267107963562, "learning_rate": 9.470219894710814e-05, "loss": 0.0184, "step": 57140 }, { "epoch": 36.4244741873805, "grad_norm": 0.5569348931312561, "learning_rate": 9.469972960857182e-05, "loss": 0.0222, "step": 57150 }, { "epoch": 36.430847673677505, "grad_norm": 0.35149672627449036, "learning_rate": 9.469725972689161e-05, "loss": 0.0355, "step": 57160 }, { "epoch": 36.437221159974506, "grad_norm": 0.3694823384284973, "learning_rate": 9.469478930209748e-05, "loss": 0.0258, "step": 57170 }, { "epoch": 36.443594646271514, "grad_norm": 0.3965859115123749, "learning_rate": 9.469231833421945e-05, "loss": 0.0226, "step": 57180 }, { "epoch": 36.449968132568515, "grad_norm": 0.5353180170059204, "learning_rate": 9.468984682328756e-05, "loss": 0.0275, "step": 57190 }, { "epoch": 36.45634161886552, "grad_norm": 0.5391090512275696, "learning_rate": 9.468737476933184e-05, "loss": 0.0326, "step": 57200 }, { "epoch": 36.46271510516252, "grad_norm": 0.36551183462142944, "learning_rate": 9.468490217238232e-05, "loss": 0.0306, "step": 57210 }, { "epoch": 36.46908859145953, "grad_norm": 0.3970765769481659, "learning_rate": 9.468242903246907e-05, "loss": 0.0218, "step": 57220 }, { "epoch": 36.47546207775653, "grad_norm": 0.3243199586868286, "learning_rate": 9.46799553496221e-05, "loss": 0.0205, "step": 57230 }, { "epoch": 36.48183556405354, "grad_norm": 0.44766804575920105, "learning_rate": 9.46774811238715e-05, "loss": 0.0241, "step": 57240 }, { "epoch": 36.48820905035054, "grad_norm": 0.5407687425613403, "learning_rate": 9.467500635524731e-05, "loss": 0.0214, "step": 57250 }, { "epoch": 36.49458253664755, "grad_norm": 0.5734948515892029, "learning_rate": 9.46725310437796e-05, "loss": 0.0256, "step": 57260 }, { "epoch": 36.50095602294455, "grad_norm": 0.5349106192588806, "learning_rate": 9.467005518949848e-05, "loss": 0.0358, "step": 57270 }, { "epoch": 36.50732950924156, "grad_norm": 0.646597146987915, "learning_rate": 9.466757879243402e-05, "loss": 0.0354, "step": 57280 }, { "epoch": 36.51370299553856, "grad_norm": 0.3602616488933563, "learning_rate": 9.46651018526163e-05, "loss": 0.0351, "step": 57290 }, { "epoch": 36.52007648183557, "grad_norm": 0.4239743947982788, "learning_rate": 9.46626243700754e-05, "loss": 0.029, "step": 57300 }, { "epoch": 36.52644996813257, "grad_norm": 0.2851369380950928, "learning_rate": 9.466014634484147e-05, "loss": 0.0335, "step": 57310 }, { "epoch": 36.532823454429575, "grad_norm": 0.5277607440948486, "learning_rate": 9.465766777694458e-05, "loss": 0.0334, "step": 57320 }, { "epoch": 36.539196940726576, "grad_norm": 0.6391661763191223, "learning_rate": 9.465518866641487e-05, "loss": 0.0339, "step": 57330 }, { "epoch": 36.545570427023584, "grad_norm": 0.6888791918754578, "learning_rate": 9.465270901328247e-05, "loss": 0.0287, "step": 57340 }, { "epoch": 36.551943913320585, "grad_norm": 0.3791259527206421, "learning_rate": 9.465022881757749e-05, "loss": 0.0259, "step": 57350 }, { "epoch": 36.55831739961759, "grad_norm": 0.5581915974617004, "learning_rate": 9.464774807933005e-05, "loss": 0.027, "step": 57360 }, { "epoch": 36.564690885914594, "grad_norm": 0.6255170106887817, "learning_rate": 9.464526679857034e-05, "loss": 0.0469, "step": 57370 }, { "epoch": 36.5710643722116, "grad_norm": 0.5116237998008728, "learning_rate": 9.464278497532848e-05, "loss": 0.0294, "step": 57380 }, { "epoch": 36.5774378585086, "grad_norm": 0.5179383158683777, "learning_rate": 9.464030260963462e-05, "loss": 0.0261, "step": 57390 }, { "epoch": 36.58381134480561, "grad_norm": 0.32617050409317017, "learning_rate": 9.463781970151895e-05, "loss": 0.0283, "step": 57400 }, { "epoch": 36.59018483110261, "grad_norm": 0.30168455839157104, "learning_rate": 9.463533625101162e-05, "loss": 0.0248, "step": 57410 }, { "epoch": 36.59655831739962, "grad_norm": 0.3762986958026886, "learning_rate": 9.463285225814283e-05, "loss": 0.0201, "step": 57420 }, { "epoch": 36.60293180369662, "grad_norm": 0.42218831181526184, "learning_rate": 9.463036772294272e-05, "loss": 0.0302, "step": 57430 }, { "epoch": 36.60930528999363, "grad_norm": 0.5094795227050781, "learning_rate": 9.462788264544152e-05, "loss": 0.0262, "step": 57440 }, { "epoch": 36.61567877629063, "grad_norm": 0.8828285932540894, "learning_rate": 9.462539702566939e-05, "loss": 0.0315, "step": 57450 }, { "epoch": 36.62205226258764, "grad_norm": 0.26420992612838745, "learning_rate": 9.462291086365657e-05, "loss": 0.0425, "step": 57460 }, { "epoch": 36.62842574888464, "grad_norm": 0.5642569065093994, "learning_rate": 9.462042415943324e-05, "loss": 0.022, "step": 57470 }, { "epoch": 36.634799235181646, "grad_norm": 0.5613830089569092, "learning_rate": 9.461793691302965e-05, "loss": 0.0286, "step": 57480 }, { "epoch": 36.64117272147865, "grad_norm": 0.38027775287628174, "learning_rate": 9.461544912447597e-05, "loss": 0.0301, "step": 57490 }, { "epoch": 36.647546207775655, "grad_norm": 0.36848384141921997, "learning_rate": 9.461296079380249e-05, "loss": 0.0292, "step": 57500 }, { "epoch": 36.653919694072655, "grad_norm": 0.3908831775188446, "learning_rate": 9.461047192103939e-05, "loss": 0.0274, "step": 57510 }, { "epoch": 36.66029318036966, "grad_norm": 0.8432488441467285, "learning_rate": 9.460798250621694e-05, "loss": 0.0366, "step": 57520 }, { "epoch": 36.666666666666664, "grad_norm": 0.8290928602218628, "learning_rate": 9.46054925493654e-05, "loss": 0.0282, "step": 57530 }, { "epoch": 36.67304015296367, "grad_norm": 0.5668666362762451, "learning_rate": 9.4603002050515e-05, "loss": 0.0316, "step": 57540 }, { "epoch": 36.67941363926067, "grad_norm": 0.5342413187026978, "learning_rate": 9.460051100969602e-05, "loss": 0.0345, "step": 57550 }, { "epoch": 36.68578712555768, "grad_norm": 0.3098469078540802, "learning_rate": 9.459801942693871e-05, "loss": 0.0329, "step": 57560 }, { "epoch": 36.69216061185468, "grad_norm": 0.3256721794605255, "learning_rate": 9.459552730227336e-05, "loss": 0.0156, "step": 57570 }, { "epoch": 36.69853409815169, "grad_norm": 0.6516679525375366, "learning_rate": 9.459303463573026e-05, "loss": 0.0268, "step": 57580 }, { "epoch": 36.70490758444869, "grad_norm": 0.33256176114082336, "learning_rate": 9.459054142733968e-05, "loss": 0.022, "step": 57590 }, { "epoch": 36.7112810707457, "grad_norm": 0.34782370924949646, "learning_rate": 9.458804767713192e-05, "loss": 0.0282, "step": 57600 }, { "epoch": 36.7176545570427, "grad_norm": 0.32653582096099854, "learning_rate": 9.458555338513728e-05, "loss": 0.0356, "step": 57610 }, { "epoch": 36.72402804333971, "grad_norm": 0.6436353325843811, "learning_rate": 9.458305855138607e-05, "loss": 0.0221, "step": 57620 }, { "epoch": 36.73040152963671, "grad_norm": 0.5634167790412903, "learning_rate": 9.458056317590862e-05, "loss": 0.0333, "step": 57630 }, { "epoch": 36.736775015933716, "grad_norm": 0.4234979450702667, "learning_rate": 9.457806725873522e-05, "loss": 0.0418, "step": 57640 }, { "epoch": 36.74314850223072, "grad_norm": 0.48141583800315857, "learning_rate": 9.457557079989622e-05, "loss": 0.0262, "step": 57650 }, { "epoch": 36.749521988527725, "grad_norm": 0.40675675868988037, "learning_rate": 9.457307379942195e-05, "loss": 0.0298, "step": 57660 }, { "epoch": 36.755895474824726, "grad_norm": 0.6099407076835632, "learning_rate": 9.457057625734273e-05, "loss": 0.0338, "step": 57670 }, { "epoch": 36.762268961121734, "grad_norm": 0.5035131573677063, "learning_rate": 9.456807817368895e-05, "loss": 0.0253, "step": 57680 }, { "epoch": 36.768642447418735, "grad_norm": 0.5587304830551147, "learning_rate": 9.456557954849094e-05, "loss": 0.0233, "step": 57690 }, { "epoch": 36.77501593371574, "grad_norm": 0.2693192958831787, "learning_rate": 9.456308038177906e-05, "loss": 0.0208, "step": 57700 }, { "epoch": 36.781389420012744, "grad_norm": 0.4463195204734802, "learning_rate": 9.456058067358367e-05, "loss": 0.0228, "step": 57710 }, { "epoch": 36.78776290630975, "grad_norm": 0.3880036175251007, "learning_rate": 9.455808042393516e-05, "loss": 0.0223, "step": 57720 }, { "epoch": 36.79413639260676, "grad_norm": 0.4854073226451874, "learning_rate": 9.45555796328639e-05, "loss": 0.0204, "step": 57730 }, { "epoch": 36.80050987890376, "grad_norm": 0.46327298879623413, "learning_rate": 9.455307830040028e-05, "loss": 0.0255, "step": 57740 }, { "epoch": 36.80688336520077, "grad_norm": 0.43582096695899963, "learning_rate": 9.455057642657469e-05, "loss": 0.0221, "step": 57750 }, { "epoch": 36.81325685149777, "grad_norm": 0.3683633804321289, "learning_rate": 9.454807401141754e-05, "loss": 0.03, "step": 57760 }, { "epoch": 36.81963033779478, "grad_norm": 0.38868263363838196, "learning_rate": 9.454557105495923e-05, "loss": 0.0179, "step": 57770 }, { "epoch": 36.82600382409178, "grad_norm": 0.47523003816604614, "learning_rate": 9.454306755723017e-05, "loss": 0.0314, "step": 57780 }, { "epoch": 36.832377310388786, "grad_norm": 0.6168904900550842, "learning_rate": 9.45405635182608e-05, "loss": 0.0303, "step": 57790 }, { "epoch": 36.83875079668579, "grad_norm": 0.4319334328174591, "learning_rate": 9.45380589380815e-05, "loss": 0.0189, "step": 57800 }, { "epoch": 36.845124282982795, "grad_norm": 0.7553130388259888, "learning_rate": 9.453555381672275e-05, "loss": 0.0532, "step": 57810 }, { "epoch": 36.851497769279796, "grad_norm": 0.3500904440879822, "learning_rate": 9.453304815421497e-05, "loss": 0.0298, "step": 57820 }, { "epoch": 36.8578712555768, "grad_norm": 0.40126967430114746, "learning_rate": 9.453054195058862e-05, "loss": 0.0204, "step": 57830 }, { "epoch": 36.864244741873804, "grad_norm": 0.5441057682037354, "learning_rate": 9.452803520587412e-05, "loss": 0.0256, "step": 57840 }, { "epoch": 36.87061822817081, "grad_norm": 0.4571659564971924, "learning_rate": 9.452552792010197e-05, "loss": 0.0304, "step": 57850 }, { "epoch": 36.87699171446781, "grad_norm": 0.442874014377594, "learning_rate": 9.452302009330261e-05, "loss": 0.0252, "step": 57860 }, { "epoch": 36.88336520076482, "grad_norm": 0.5466029047966003, "learning_rate": 9.452051172550652e-05, "loss": 0.0381, "step": 57870 }, { "epoch": 36.88973868706182, "grad_norm": 0.8083130717277527, "learning_rate": 9.451800281674417e-05, "loss": 0.0271, "step": 57880 }, { "epoch": 36.89611217335883, "grad_norm": 0.3117665648460388, "learning_rate": 9.451549336704606e-05, "loss": 0.0347, "step": 57890 }, { "epoch": 36.90248565965583, "grad_norm": 0.28044503927230835, "learning_rate": 9.451298337644267e-05, "loss": 0.035, "step": 57900 }, { "epoch": 36.90885914595284, "grad_norm": 0.5083986520767212, "learning_rate": 9.451047284496452e-05, "loss": 0.0298, "step": 57910 }, { "epoch": 36.91523263224984, "grad_norm": 0.5868697762489319, "learning_rate": 9.450796177264209e-05, "loss": 0.0222, "step": 57920 }, { "epoch": 36.92160611854685, "grad_norm": 0.8156291246414185, "learning_rate": 9.450545015950591e-05, "loss": 0.0455, "step": 57930 }, { "epoch": 36.92797960484385, "grad_norm": 0.8249725699424744, "learning_rate": 9.450293800558648e-05, "loss": 0.0342, "step": 57940 }, { "epoch": 36.934353091140856, "grad_norm": 0.3781333267688751, "learning_rate": 9.450042531091433e-05, "loss": 0.0297, "step": 57950 }, { "epoch": 36.94072657743786, "grad_norm": 0.40890398621559143, "learning_rate": 9.449791207552001e-05, "loss": 0.0294, "step": 57960 }, { "epoch": 36.947100063734865, "grad_norm": 0.33422401547431946, "learning_rate": 9.449539829943404e-05, "loss": 0.0239, "step": 57970 }, { "epoch": 36.953473550031866, "grad_norm": 0.4204844534397125, "learning_rate": 9.449288398268697e-05, "loss": 0.0289, "step": 57980 }, { "epoch": 36.959847036328874, "grad_norm": 0.5481045246124268, "learning_rate": 9.449036912530935e-05, "loss": 0.0292, "step": 57990 }, { "epoch": 36.966220522625875, "grad_norm": 0.2501024007797241, "learning_rate": 9.448785372733174e-05, "loss": 0.0182, "step": 58000 }, { "epoch": 36.97259400892288, "grad_norm": 0.43681061267852783, "learning_rate": 9.448533778878471e-05, "loss": 0.0218, "step": 58010 }, { "epoch": 36.978967495219884, "grad_norm": 0.6796426177024841, "learning_rate": 9.448282130969882e-05, "loss": 0.0228, "step": 58020 }, { "epoch": 36.98534098151689, "grad_norm": 0.40712788701057434, "learning_rate": 9.448030429010466e-05, "loss": 0.0411, "step": 58030 }, { "epoch": 36.99171446781389, "grad_norm": 0.6452886462211609, "learning_rate": 9.44777867300328e-05, "loss": 0.0242, "step": 58040 }, { "epoch": 36.9980879541109, "grad_norm": 0.4238446354866028, "learning_rate": 9.447526862951384e-05, "loss": 0.0361, "step": 58050 }, { "epoch": 37.0044614404079, "grad_norm": 0.5474826693534851, "learning_rate": 9.447274998857836e-05, "loss": 0.0163, "step": 58060 }, { "epoch": 37.01083492670491, "grad_norm": 0.4672966003417969, "learning_rate": 9.447023080725698e-05, "loss": 0.0234, "step": 58070 }, { "epoch": 37.01720841300191, "grad_norm": 0.6289032101631165, "learning_rate": 9.446771108558033e-05, "loss": 0.0282, "step": 58080 }, { "epoch": 37.02358189929892, "grad_norm": 0.8783146739006042, "learning_rate": 9.446519082357898e-05, "loss": 0.0209, "step": 58090 }, { "epoch": 37.02995538559592, "grad_norm": 0.3712984025478363, "learning_rate": 9.44626700212836e-05, "loss": 0.0352, "step": 58100 }, { "epoch": 37.03632887189293, "grad_norm": 0.48685890436172485, "learning_rate": 9.44601486787248e-05, "loss": 0.0257, "step": 58110 }, { "epoch": 37.04270235818993, "grad_norm": 0.3748556673526764, "learning_rate": 9.445762679593321e-05, "loss": 0.0211, "step": 58120 }, { "epoch": 37.049075844486936, "grad_norm": 0.3443540036678314, "learning_rate": 9.445510437293948e-05, "loss": 0.023, "step": 58130 }, { "epoch": 37.05544933078394, "grad_norm": 0.3426295816898346, "learning_rate": 9.445258140977426e-05, "loss": 0.0327, "step": 58140 }, { "epoch": 37.061822817080945, "grad_norm": 0.5828007459640503, "learning_rate": 9.445005790646819e-05, "loss": 0.0364, "step": 58150 }, { "epoch": 37.068196303377945, "grad_norm": 0.4944082796573639, "learning_rate": 9.444753386305196e-05, "loss": 0.0278, "step": 58160 }, { "epoch": 37.07456978967495, "grad_norm": 0.2439521700143814, "learning_rate": 9.444500927955624e-05, "loss": 0.0222, "step": 58170 }, { "epoch": 37.080943275971954, "grad_norm": 0.6633405089378357, "learning_rate": 9.444248415601168e-05, "loss": 0.0231, "step": 58180 }, { "epoch": 37.08731676226896, "grad_norm": 0.7136917114257812, "learning_rate": 9.443995849244899e-05, "loss": 0.0209, "step": 58190 }, { "epoch": 37.09369024856596, "grad_norm": 0.5466375350952148, "learning_rate": 9.443743228889883e-05, "loss": 0.0216, "step": 58200 }, { "epoch": 37.10006373486297, "grad_norm": 0.8081802725791931, "learning_rate": 9.443490554539191e-05, "loss": 0.0434, "step": 58210 }, { "epoch": 37.10643722115997, "grad_norm": 0.3328581750392914, "learning_rate": 9.443237826195897e-05, "loss": 0.0218, "step": 58220 }, { "epoch": 37.11281070745698, "grad_norm": 0.45257583260536194, "learning_rate": 9.442985043863064e-05, "loss": 0.031, "step": 58230 }, { "epoch": 37.11918419375398, "grad_norm": 0.5781674385070801, "learning_rate": 9.44273220754377e-05, "loss": 0.0232, "step": 58240 }, { "epoch": 37.12555768005099, "grad_norm": 0.7794846296310425, "learning_rate": 9.442479317241084e-05, "loss": 0.0332, "step": 58250 }, { "epoch": 37.13193116634799, "grad_norm": 0.43148553371429443, "learning_rate": 9.44222637295808e-05, "loss": 0.019, "step": 58260 }, { "epoch": 37.138304652645, "grad_norm": 0.45630818605422974, "learning_rate": 9.441973374697833e-05, "loss": 0.0263, "step": 58270 }, { "epoch": 37.144678138942, "grad_norm": 1.0317178964614868, "learning_rate": 9.441720322463414e-05, "loss": 0.0304, "step": 58280 }, { "epoch": 37.151051625239006, "grad_norm": 0.6345905065536499, "learning_rate": 9.4414672162579e-05, "loss": 0.0165, "step": 58290 }, { "epoch": 37.15742511153601, "grad_norm": 0.4350857138633728, "learning_rate": 9.441214056084365e-05, "loss": 0.0307, "step": 58300 }, { "epoch": 37.163798597833015, "grad_norm": 0.431850403547287, "learning_rate": 9.440960841945887e-05, "loss": 0.0392, "step": 58310 }, { "epoch": 37.170172084130016, "grad_norm": 0.6173940300941467, "learning_rate": 9.440707573845542e-05, "loss": 0.0203, "step": 58320 }, { "epoch": 37.176545570427024, "grad_norm": 0.6035770773887634, "learning_rate": 9.440454251786406e-05, "loss": 0.0261, "step": 58330 }, { "epoch": 37.182919056724025, "grad_norm": 0.46005141735076904, "learning_rate": 9.44020087577156e-05, "loss": 0.0243, "step": 58340 }, { "epoch": 37.18929254302103, "grad_norm": 1.101370930671692, "learning_rate": 9.439947445804081e-05, "loss": 0.0362, "step": 58350 }, { "epoch": 37.19566602931803, "grad_norm": 0.4578830301761627, "learning_rate": 9.439693961887046e-05, "loss": 0.0262, "step": 58360 }, { "epoch": 37.20203951561504, "grad_norm": 0.3971228003501892, "learning_rate": 9.43944042402354e-05, "loss": 0.0178, "step": 58370 }, { "epoch": 37.20841300191205, "grad_norm": 0.4319360852241516, "learning_rate": 9.43918683221664e-05, "loss": 0.0264, "step": 58380 }, { "epoch": 37.21478648820905, "grad_norm": 0.7146967053413391, "learning_rate": 9.43893318646943e-05, "loss": 0.0286, "step": 58390 }, { "epoch": 37.22115997450606, "grad_norm": 0.6462353467941284, "learning_rate": 9.43867948678499e-05, "loss": 0.03, "step": 58400 }, { "epoch": 37.22753346080306, "grad_norm": 0.876550555229187, "learning_rate": 9.438425733166404e-05, "loss": 0.0225, "step": 58410 }, { "epoch": 37.23390694710007, "grad_norm": 0.40169620513916016, "learning_rate": 9.438171925616754e-05, "loss": 0.0297, "step": 58420 }, { "epoch": 37.24028043339707, "grad_norm": 0.39487162232398987, "learning_rate": 9.437918064139126e-05, "loss": 0.0228, "step": 58430 }, { "epoch": 37.246653919694076, "grad_norm": 0.2838194966316223, "learning_rate": 9.437664148736603e-05, "loss": 0.0334, "step": 58440 }, { "epoch": 37.25302740599108, "grad_norm": 0.4718831777572632, "learning_rate": 9.437410179412269e-05, "loss": 0.0348, "step": 58450 }, { "epoch": 37.259400892288085, "grad_norm": 0.7466816902160645, "learning_rate": 9.437156156169213e-05, "loss": 0.0373, "step": 58460 }, { "epoch": 37.265774378585085, "grad_norm": 0.5572332143783569, "learning_rate": 9.436902079010521e-05, "loss": 0.0225, "step": 58470 }, { "epoch": 37.27214786488209, "grad_norm": 0.3881951570510864, "learning_rate": 9.436647947939279e-05, "loss": 0.0255, "step": 58480 }, { "epoch": 37.278521351179094, "grad_norm": 0.38517072796821594, "learning_rate": 9.436393762958575e-05, "loss": 0.0266, "step": 58490 }, { "epoch": 37.2848948374761, "grad_norm": 0.36928829550743103, "learning_rate": 9.4361395240715e-05, "loss": 0.0354, "step": 58500 }, { "epoch": 37.2912683237731, "grad_norm": 0.5534390807151794, "learning_rate": 9.435885231281139e-05, "loss": 0.0387, "step": 58510 }, { "epoch": 37.29764181007011, "grad_norm": 0.4806787073612213, "learning_rate": 9.435630884590586e-05, "loss": 0.0294, "step": 58520 }, { "epoch": 37.30401529636711, "grad_norm": 0.5615505576133728, "learning_rate": 9.435376484002928e-05, "loss": 0.031, "step": 58530 }, { "epoch": 37.31038878266412, "grad_norm": 0.5432212948799133, "learning_rate": 9.435122029521259e-05, "loss": 0.0266, "step": 58540 }, { "epoch": 37.31676226896112, "grad_norm": 0.3455148935317993, "learning_rate": 9.43486752114867e-05, "loss": 0.0337, "step": 58550 }, { "epoch": 37.32313575525813, "grad_norm": 0.9200009703636169, "learning_rate": 9.434612958888253e-05, "loss": 0.022, "step": 58560 }, { "epoch": 37.32950924155513, "grad_norm": 0.3465256094932556, "learning_rate": 9.434358342743103e-05, "loss": 0.0222, "step": 58570 }, { "epoch": 37.33588272785214, "grad_norm": 0.5053335428237915, "learning_rate": 9.434103672716311e-05, "loss": 0.0283, "step": 58580 }, { "epoch": 37.34225621414914, "grad_norm": 0.3810831308364868, "learning_rate": 9.433848948810973e-05, "loss": 0.0257, "step": 58590 }, { "epoch": 37.348629700446146, "grad_norm": 0.5449327230453491, "learning_rate": 9.433594171030184e-05, "loss": 0.0189, "step": 58600 }, { "epoch": 37.35500318674315, "grad_norm": 0.5504589676856995, "learning_rate": 9.43333933937704e-05, "loss": 0.0224, "step": 58610 }, { "epoch": 37.361376673040155, "grad_norm": 0.6452327966690063, "learning_rate": 9.433084453854638e-05, "loss": 0.0262, "step": 58620 }, { "epoch": 37.367750159337156, "grad_norm": 0.505751371383667, "learning_rate": 9.432829514466074e-05, "loss": 0.0241, "step": 58630 }, { "epoch": 37.374123645634164, "grad_norm": 0.727211594581604, "learning_rate": 9.432574521214445e-05, "loss": 0.0328, "step": 58640 }, { "epoch": 37.380497131931165, "grad_norm": 0.5360155701637268, "learning_rate": 9.432319474102852e-05, "loss": 0.0205, "step": 58650 }, { "epoch": 37.38687061822817, "grad_norm": 0.5676209926605225, "learning_rate": 9.432064373134392e-05, "loss": 0.0271, "step": 58660 }, { "epoch": 37.393244104525174, "grad_norm": 0.3800312578678131, "learning_rate": 9.431809218312166e-05, "loss": 0.023, "step": 58670 }, { "epoch": 37.39961759082218, "grad_norm": 0.4834483861923218, "learning_rate": 9.431554009639272e-05, "loss": 0.0268, "step": 58680 }, { "epoch": 37.40599107711918, "grad_norm": 0.35569557547569275, "learning_rate": 9.431298747118814e-05, "loss": 0.0273, "step": 58690 }, { "epoch": 37.41236456341619, "grad_norm": 0.2903531789779663, "learning_rate": 9.431043430753892e-05, "loss": 0.0249, "step": 58700 }, { "epoch": 37.41873804971319, "grad_norm": 0.5252476334571838, "learning_rate": 9.430788060547609e-05, "loss": 0.0279, "step": 58710 }, { "epoch": 37.4251115360102, "grad_norm": 0.6529276967048645, "learning_rate": 9.430532636503068e-05, "loss": 0.0355, "step": 58720 }, { "epoch": 37.4314850223072, "grad_norm": 0.3895089328289032, "learning_rate": 9.43027715862337e-05, "loss": 0.0277, "step": 58730 }, { "epoch": 37.43785850860421, "grad_norm": 0.36202213168144226, "learning_rate": 9.430021626911625e-05, "loss": 0.0335, "step": 58740 }, { "epoch": 37.44423199490121, "grad_norm": 0.5615801811218262, "learning_rate": 9.429766041370934e-05, "loss": 0.0183, "step": 58750 }, { "epoch": 37.45060548119822, "grad_norm": 0.3760916590690613, "learning_rate": 9.429510402004401e-05, "loss": 0.0257, "step": 58760 }, { "epoch": 37.45697896749522, "grad_norm": 0.622738778591156, "learning_rate": 9.429254708815136e-05, "loss": 0.0256, "step": 58770 }, { "epoch": 37.463352453792226, "grad_norm": 0.4987794756889343, "learning_rate": 9.428998961806244e-05, "loss": 0.0268, "step": 58780 }, { "epoch": 37.469725940089226, "grad_norm": 0.5066813230514526, "learning_rate": 9.428743160980833e-05, "loss": 0.0271, "step": 58790 }, { "epoch": 37.476099426386234, "grad_norm": 0.34712347388267517, "learning_rate": 9.428487306342011e-05, "loss": 0.0235, "step": 58800 }, { "epoch": 37.482472912683235, "grad_norm": 0.49838587641716003, "learning_rate": 9.428231397892888e-05, "loss": 0.0259, "step": 58810 }, { "epoch": 37.48884639898024, "grad_norm": 0.8514861464500427, "learning_rate": 9.427975435636571e-05, "loss": 0.044, "step": 58820 }, { "epoch": 37.495219885277244, "grad_norm": 0.3483809232711792, "learning_rate": 9.427719419576173e-05, "loss": 0.0397, "step": 58830 }, { "epoch": 37.50159337157425, "grad_norm": 0.5594362020492554, "learning_rate": 9.427463349714804e-05, "loss": 0.0352, "step": 58840 }, { "epoch": 37.50796685787125, "grad_norm": 0.41040951013565063, "learning_rate": 9.427207226055574e-05, "loss": 0.0331, "step": 58850 }, { "epoch": 37.51434034416826, "grad_norm": 0.37323230504989624, "learning_rate": 9.426951048601595e-05, "loss": 0.02, "step": 58860 }, { "epoch": 37.52071383046526, "grad_norm": 0.8023868799209595, "learning_rate": 9.426694817355983e-05, "loss": 0.028, "step": 58870 }, { "epoch": 37.52708731676227, "grad_norm": 0.5003507733345032, "learning_rate": 9.42643853232185e-05, "loss": 0.0366, "step": 58880 }, { "epoch": 37.53346080305927, "grad_norm": 0.5807331800460815, "learning_rate": 9.426182193502309e-05, "loss": 0.0409, "step": 58890 }, { "epoch": 37.53983428935628, "grad_norm": 0.4494272470474243, "learning_rate": 9.425925800900475e-05, "loss": 0.0311, "step": 58900 }, { "epoch": 37.54620777565328, "grad_norm": 0.3875611126422882, "learning_rate": 9.425669354519464e-05, "loss": 0.0264, "step": 58910 }, { "epoch": 37.55258126195029, "grad_norm": 0.4444333612918854, "learning_rate": 9.42541285436239e-05, "loss": 0.033, "step": 58920 }, { "epoch": 37.55895474824729, "grad_norm": 0.3254602551460266, "learning_rate": 9.425156300432373e-05, "loss": 0.0381, "step": 58930 }, { "epoch": 37.565328234544296, "grad_norm": 0.3972413241863251, "learning_rate": 9.424899692732529e-05, "loss": 0.0331, "step": 58940 }, { "epoch": 37.5717017208413, "grad_norm": 0.4516952335834503, "learning_rate": 9.424643031265976e-05, "loss": 0.0243, "step": 58950 }, { "epoch": 37.578075207138305, "grad_norm": 0.8825690150260925, "learning_rate": 9.424386316035833e-05, "loss": 0.0451, "step": 58960 }, { "epoch": 37.584448693435306, "grad_norm": 0.3919873535633087, "learning_rate": 9.424129547045218e-05, "loss": 0.0238, "step": 58970 }, { "epoch": 37.590822179732314, "grad_norm": 0.2833164930343628, "learning_rate": 9.423872724297252e-05, "loss": 0.0222, "step": 58980 }, { "epoch": 37.59719566602932, "grad_norm": 0.33000099658966064, "learning_rate": 9.423615847795057e-05, "loss": 0.0276, "step": 58990 }, { "epoch": 37.60356915232632, "grad_norm": 0.5424593687057495, "learning_rate": 9.423358917541751e-05, "loss": 0.0191, "step": 59000 }, { "epoch": 37.60994263862333, "grad_norm": 0.4948881268501282, "learning_rate": 9.423101933540459e-05, "loss": 0.0403, "step": 59010 }, { "epoch": 37.61631612492033, "grad_norm": 0.42425987124443054, "learning_rate": 9.422844895794301e-05, "loss": 0.0305, "step": 59020 }, { "epoch": 37.62268961121734, "grad_norm": 0.5077105164527893, "learning_rate": 9.422587804306404e-05, "loss": 0.0202, "step": 59030 }, { "epoch": 37.62906309751434, "grad_norm": 0.2441534847021103, "learning_rate": 9.422330659079887e-05, "loss": 0.0359, "step": 59040 }, { "epoch": 37.63543658381135, "grad_norm": 0.3992559313774109, "learning_rate": 9.42207346011788e-05, "loss": 0.0368, "step": 59050 }, { "epoch": 37.64181007010835, "grad_norm": 0.5145223736763, "learning_rate": 9.421816207423503e-05, "loss": 0.0318, "step": 59060 }, { "epoch": 37.64818355640536, "grad_norm": 0.4393545687198639, "learning_rate": 9.421558900999885e-05, "loss": 0.0318, "step": 59070 }, { "epoch": 37.65455704270236, "grad_norm": 0.6667615175247192, "learning_rate": 9.421301540850152e-05, "loss": 0.0298, "step": 59080 }, { "epoch": 37.660930528999366, "grad_norm": 0.5049098134040833, "learning_rate": 9.42104412697743e-05, "loss": 0.0382, "step": 59090 }, { "epoch": 37.66730401529637, "grad_norm": 0.5856661200523376, "learning_rate": 9.420786659384848e-05, "loss": 0.0408, "step": 59100 }, { "epoch": 37.673677501593374, "grad_norm": 0.4838673770427704, "learning_rate": 9.420529138075535e-05, "loss": 0.0333, "step": 59110 }, { "epoch": 37.680050987890375, "grad_norm": 0.6913846731185913, "learning_rate": 9.420271563052618e-05, "loss": 0.0326, "step": 59120 }, { "epoch": 37.68642447418738, "grad_norm": 0.5608755350112915, "learning_rate": 9.42001393431923e-05, "loss": 0.0266, "step": 59130 }, { "epoch": 37.692797960484384, "grad_norm": 0.617890477180481, "learning_rate": 9.419756251878498e-05, "loss": 0.0309, "step": 59140 }, { "epoch": 37.69917144678139, "grad_norm": 0.5727347731590271, "learning_rate": 9.419498515733554e-05, "loss": 0.0233, "step": 59150 }, { "epoch": 37.70554493307839, "grad_norm": 0.7098250389099121, "learning_rate": 9.41924072588753e-05, "loss": 0.0283, "step": 59160 }, { "epoch": 37.7119184193754, "grad_norm": 0.7657014727592468, "learning_rate": 9.418982882343561e-05, "loss": 0.0366, "step": 59170 }, { "epoch": 37.7182919056724, "grad_norm": 0.45679304003715515, "learning_rate": 9.418724985104777e-05, "loss": 0.0257, "step": 59180 }, { "epoch": 37.72466539196941, "grad_norm": 0.5330288410186768, "learning_rate": 9.418467034174312e-05, "loss": 0.0202, "step": 59190 }, { "epoch": 37.73103887826641, "grad_norm": 0.3801381587982178, "learning_rate": 9.4182090295553e-05, "loss": 0.0258, "step": 59200 }, { "epoch": 37.73741236456342, "grad_norm": 0.673989474773407, "learning_rate": 9.417950971250878e-05, "loss": 0.0173, "step": 59210 }, { "epoch": 37.74378585086042, "grad_norm": 0.6289236545562744, "learning_rate": 9.417692859264181e-05, "loss": 0.0275, "step": 59220 }, { "epoch": 37.75015933715743, "grad_norm": 0.6513596773147583, "learning_rate": 9.417434693598344e-05, "loss": 0.0222, "step": 59230 }, { "epoch": 37.75653282345443, "grad_norm": 0.5019389986991882, "learning_rate": 9.417176474256504e-05, "loss": 0.0275, "step": 59240 }, { "epoch": 37.762906309751436, "grad_norm": 0.39428678154945374, "learning_rate": 9.4169182012418e-05, "loss": 0.0198, "step": 59250 }, { "epoch": 37.76927979604844, "grad_norm": 0.3980458676815033, "learning_rate": 9.41665987455737e-05, "loss": 0.0275, "step": 59260 }, { "epoch": 37.775653282345445, "grad_norm": 0.3053942024707794, "learning_rate": 9.416401494206353e-05, "loss": 0.0244, "step": 59270 }, { "epoch": 37.782026768642446, "grad_norm": 0.3392542004585266, "learning_rate": 9.416143060191886e-05, "loss": 0.0207, "step": 59280 }, { "epoch": 37.788400254939454, "grad_norm": 0.43882572650909424, "learning_rate": 9.415884572517113e-05, "loss": 0.0204, "step": 59290 }, { "epoch": 37.794773741236455, "grad_norm": 0.40536171197891235, "learning_rate": 9.415626031185172e-05, "loss": 0.0239, "step": 59300 }, { "epoch": 37.80114722753346, "grad_norm": 0.42050808668136597, "learning_rate": 9.415367436199207e-05, "loss": 0.0255, "step": 59310 }, { "epoch": 37.80752071383046, "grad_norm": 0.7898723483085632, "learning_rate": 9.41510878756236e-05, "loss": 0.0336, "step": 59320 }, { "epoch": 37.81389420012747, "grad_norm": 0.2493704855442047, "learning_rate": 9.414850085277769e-05, "loss": 0.0296, "step": 59330 }, { "epoch": 37.82026768642447, "grad_norm": 0.4026285409927368, "learning_rate": 9.414591329348583e-05, "loss": 0.03, "step": 59340 }, { "epoch": 37.82664117272148, "grad_norm": 0.3279447853565216, "learning_rate": 9.414332519777944e-05, "loss": 0.0212, "step": 59350 }, { "epoch": 37.83301465901848, "grad_norm": 0.4612616300582886, "learning_rate": 9.414073656568998e-05, "loss": 0.0225, "step": 59360 }, { "epoch": 37.83938814531549, "grad_norm": 0.6776186227798462, "learning_rate": 9.413814739724889e-05, "loss": 0.0313, "step": 59370 }, { "epoch": 37.84576163161249, "grad_norm": 0.5331281423568726, "learning_rate": 9.413555769248763e-05, "loss": 0.0328, "step": 59380 }, { "epoch": 37.8521351179095, "grad_norm": 0.3487010896205902, "learning_rate": 9.413296745143769e-05, "loss": 0.0201, "step": 59390 }, { "epoch": 37.8585086042065, "grad_norm": 0.37419599294662476, "learning_rate": 9.413037667413051e-05, "loss": 0.0275, "step": 59400 }, { "epoch": 37.86488209050351, "grad_norm": 0.8000113368034363, "learning_rate": 9.412778536059759e-05, "loss": 0.0241, "step": 59410 }, { "epoch": 37.87125557680051, "grad_norm": 0.47382041811943054, "learning_rate": 9.412519351087041e-05, "loss": 0.0276, "step": 59420 }, { "epoch": 37.877629063097515, "grad_norm": 0.3366382420063019, "learning_rate": 9.412260112498049e-05, "loss": 0.0289, "step": 59430 }, { "epoch": 37.884002549394516, "grad_norm": 0.4883997142314911, "learning_rate": 9.412000820295928e-05, "loss": 0.0278, "step": 59440 }, { "epoch": 37.890376035691524, "grad_norm": 0.4499445855617523, "learning_rate": 9.411741474483833e-05, "loss": 0.0293, "step": 59450 }, { "epoch": 37.896749521988525, "grad_norm": 0.3948924243450165, "learning_rate": 9.411482075064913e-05, "loss": 0.0219, "step": 59460 }, { "epoch": 37.90312300828553, "grad_norm": 0.42939478158950806, "learning_rate": 9.411222622042322e-05, "loss": 0.0294, "step": 59470 }, { "epoch": 37.909496494582534, "grad_norm": 0.4224923849105835, "learning_rate": 9.410963115419209e-05, "loss": 0.0215, "step": 59480 }, { "epoch": 37.91586998087954, "grad_norm": 0.36642277240753174, "learning_rate": 9.410703555198731e-05, "loss": 0.0258, "step": 59490 }, { "epoch": 37.92224346717654, "grad_norm": 0.33882126212120056, "learning_rate": 9.410443941384039e-05, "loss": 0.0294, "step": 59500 }, { "epoch": 37.92861695347355, "grad_norm": 0.42448851466178894, "learning_rate": 9.41018427397829e-05, "loss": 0.0342, "step": 59510 }, { "epoch": 37.93499043977055, "grad_norm": 0.45555663108825684, "learning_rate": 9.40992455298464e-05, "loss": 0.0208, "step": 59520 }, { "epoch": 37.94136392606756, "grad_norm": 0.43489423394203186, "learning_rate": 9.40966477840624e-05, "loss": 0.023, "step": 59530 }, { "epoch": 37.94773741236456, "grad_norm": 0.6702353954315186, "learning_rate": 9.409404950246251e-05, "loss": 0.0325, "step": 59540 }, { "epoch": 37.95411089866157, "grad_norm": 0.41264376044273376, "learning_rate": 9.409145068507829e-05, "loss": 0.0263, "step": 59550 }, { "epoch": 37.96048438495857, "grad_norm": 0.48045140504837036, "learning_rate": 9.408885133194131e-05, "loss": 0.0297, "step": 59560 }, { "epoch": 37.96685787125558, "grad_norm": 0.37375152111053467, "learning_rate": 9.408625144308316e-05, "loss": 0.0389, "step": 59570 }, { "epoch": 37.97323135755258, "grad_norm": 0.4405043125152588, "learning_rate": 9.408365101853543e-05, "loss": 0.039, "step": 59580 }, { "epoch": 37.979604843849586, "grad_norm": 0.33477380871772766, "learning_rate": 9.408105005832972e-05, "loss": 0.0233, "step": 59590 }, { "epoch": 37.98597833014659, "grad_norm": 0.43110743165016174, "learning_rate": 9.407844856249763e-05, "loss": 0.0156, "step": 59600 }, { "epoch": 37.992351816443595, "grad_norm": 0.20999041199684143, "learning_rate": 9.407584653107078e-05, "loss": 0.0209, "step": 59610 }, { "epoch": 37.998725302740596, "grad_norm": 0.5987453460693359, "learning_rate": 9.407324396408077e-05, "loss": 0.0204, "step": 59620 }, { "epoch": 38.0050987890376, "grad_norm": 0.5122214555740356, "learning_rate": 9.407064086155924e-05, "loss": 0.0291, "step": 59630 }, { "epoch": 38.01147227533461, "grad_norm": 0.3935222029685974, "learning_rate": 9.40680372235378e-05, "loss": 0.0243, "step": 59640 }, { "epoch": 38.01784576163161, "grad_norm": 0.3285004198551178, "learning_rate": 9.406543305004812e-05, "loss": 0.0182, "step": 59650 }, { "epoch": 38.02421924792862, "grad_norm": 0.5359097719192505, "learning_rate": 9.406282834112182e-05, "loss": 0.031, "step": 59660 }, { "epoch": 38.03059273422562, "grad_norm": 0.25976693630218506, "learning_rate": 9.406022309679054e-05, "loss": 0.0389, "step": 59670 }, { "epoch": 38.03696622052263, "grad_norm": 0.4073411226272583, "learning_rate": 9.405761731708597e-05, "loss": 0.0399, "step": 59680 }, { "epoch": 38.04333970681963, "grad_norm": 0.43172791600227356, "learning_rate": 9.405501100203974e-05, "loss": 0.0283, "step": 59690 }, { "epoch": 38.04971319311664, "grad_norm": 0.6576762795448303, "learning_rate": 9.405240415168353e-05, "loss": 0.0442, "step": 59700 }, { "epoch": 38.05608667941364, "grad_norm": 0.5279595255851746, "learning_rate": 9.404979676604902e-05, "loss": 0.0363, "step": 59710 }, { "epoch": 38.06246016571065, "grad_norm": 0.33692070841789246, "learning_rate": 9.40471888451679e-05, "loss": 0.0295, "step": 59720 }, { "epoch": 38.06883365200765, "grad_norm": 0.3834635019302368, "learning_rate": 9.404458038907183e-05, "loss": 0.0277, "step": 59730 }, { "epoch": 38.075207138304656, "grad_norm": 0.444149374961853, "learning_rate": 9.404197139779253e-05, "loss": 0.0245, "step": 59740 }, { "epoch": 38.081580624601656, "grad_norm": 0.4036004841327667, "learning_rate": 9.403936187136169e-05, "loss": 0.0371, "step": 59750 }, { "epoch": 38.087954110898664, "grad_norm": 0.7655401229858398, "learning_rate": 9.403675180981104e-05, "loss": 0.0373, "step": 59760 }, { "epoch": 38.094327597195665, "grad_norm": 0.7658129334449768, "learning_rate": 9.403414121317226e-05, "loss": 0.0259, "step": 59770 }, { "epoch": 38.10070108349267, "grad_norm": 0.6385941505432129, "learning_rate": 9.403153008147709e-05, "loss": 0.0339, "step": 59780 }, { "epoch": 38.107074569789674, "grad_norm": 2.1804392337799072, "learning_rate": 9.402891841475728e-05, "loss": 0.0239, "step": 59790 }, { "epoch": 38.11344805608668, "grad_norm": 0.42595407366752625, "learning_rate": 9.402630621304452e-05, "loss": 0.0332, "step": 59800 }, { "epoch": 38.11982154238368, "grad_norm": 0.6152677536010742, "learning_rate": 9.402369347637057e-05, "loss": 0.0247, "step": 59810 }, { "epoch": 38.12619502868069, "grad_norm": 0.7131217122077942, "learning_rate": 9.402108020476718e-05, "loss": 0.0243, "step": 59820 }, { "epoch": 38.13256851497769, "grad_norm": 0.39470013976097107, "learning_rate": 9.401846639826612e-05, "loss": 0.0207, "step": 59830 }, { "epoch": 38.1389420012747, "grad_norm": 0.5159879326820374, "learning_rate": 9.401585205689912e-05, "loss": 0.0237, "step": 59840 }, { "epoch": 38.1453154875717, "grad_norm": 0.2219592183828354, "learning_rate": 9.401323718069796e-05, "loss": 0.0218, "step": 59850 }, { "epoch": 38.15168897386871, "grad_norm": 0.43782442808151245, "learning_rate": 9.401062176969442e-05, "loss": 0.0196, "step": 59860 }, { "epoch": 38.15806246016571, "grad_norm": 0.44834721088409424, "learning_rate": 9.400800582392027e-05, "loss": 0.0181, "step": 59870 }, { "epoch": 38.16443594646272, "grad_norm": 0.3449564576148987, "learning_rate": 9.400538934340728e-05, "loss": 0.0263, "step": 59880 }, { "epoch": 38.17080943275972, "grad_norm": 0.5342633724212646, "learning_rate": 9.400277232818728e-05, "loss": 0.0207, "step": 59890 }, { "epoch": 38.177182919056726, "grad_norm": 0.4120038151741028, "learning_rate": 9.400015477829205e-05, "loss": 0.0221, "step": 59900 }, { "epoch": 38.18355640535373, "grad_norm": 0.4158679246902466, "learning_rate": 9.399753669375339e-05, "loss": 0.0331, "step": 59910 }, { "epoch": 38.189929891650735, "grad_norm": 0.43505528569221497, "learning_rate": 9.399491807460312e-05, "loss": 0.0321, "step": 59920 }, { "epoch": 38.196303377947736, "grad_norm": 0.40071165561676025, "learning_rate": 9.399229892087306e-05, "loss": 0.0217, "step": 59930 }, { "epoch": 38.202676864244744, "grad_norm": 0.3870788514614105, "learning_rate": 9.398967923259503e-05, "loss": 0.0288, "step": 59940 }, { "epoch": 38.209050350541744, "grad_norm": 0.7408939003944397, "learning_rate": 9.398705900980088e-05, "loss": 0.0251, "step": 59950 }, { "epoch": 38.21542383683875, "grad_norm": 0.7131556868553162, "learning_rate": 9.398443825252241e-05, "loss": 0.0347, "step": 59960 }, { "epoch": 38.22179732313575, "grad_norm": 0.5342478156089783, "learning_rate": 9.398181696079149e-05, "loss": 0.028, "step": 59970 }, { "epoch": 38.22817080943276, "grad_norm": 0.5344670414924622, "learning_rate": 9.397919513463998e-05, "loss": 0.0263, "step": 59980 }, { "epoch": 38.23454429572976, "grad_norm": 0.33856165409088135, "learning_rate": 9.397657277409971e-05, "loss": 0.0215, "step": 59990 }, { "epoch": 38.24091778202677, "grad_norm": 0.5994165539741516, "learning_rate": 9.397394987920257e-05, "loss": 0.0366, "step": 60000 }, { "epoch": 38.24729126832377, "grad_norm": 0.491438090801239, "learning_rate": 9.397132644998042e-05, "loss": 0.0234, "step": 60010 }, { "epoch": 38.25366475462078, "grad_norm": 0.5649241209030151, "learning_rate": 9.396870248646515e-05, "loss": 0.0301, "step": 60020 }, { "epoch": 38.26003824091778, "grad_norm": 0.449146032333374, "learning_rate": 9.396607798868861e-05, "loss": 0.0196, "step": 60030 }, { "epoch": 38.26641172721479, "grad_norm": 0.3796714246273041, "learning_rate": 9.396345295668272e-05, "loss": 0.0219, "step": 60040 }, { "epoch": 38.27278521351179, "grad_norm": 0.8207438588142395, "learning_rate": 9.396082739047937e-05, "loss": 0.0242, "step": 60050 }, { "epoch": 38.2791586998088, "grad_norm": 0.3021965026855469, "learning_rate": 9.395820129011046e-05, "loss": 0.032, "step": 60060 }, { "epoch": 38.2855321861058, "grad_norm": 0.37927284836769104, "learning_rate": 9.39555746556079e-05, "loss": 0.0194, "step": 60070 }, { "epoch": 38.291905672402805, "grad_norm": 0.6051521301269531, "learning_rate": 9.395294748700361e-05, "loss": 0.0215, "step": 60080 }, { "epoch": 38.298279158699806, "grad_norm": 1.012482762336731, "learning_rate": 9.395031978432952e-05, "loss": 0.0348, "step": 60090 }, { "epoch": 38.304652644996814, "grad_norm": 0.5415579080581665, "learning_rate": 9.394769154761754e-05, "loss": 0.0177, "step": 60100 }, { "epoch": 38.311026131293815, "grad_norm": 0.5627006888389587, "learning_rate": 9.39450627768996e-05, "loss": 0.0371, "step": 60110 }, { "epoch": 38.31739961759082, "grad_norm": 0.6162716746330261, "learning_rate": 9.394243347220766e-05, "loss": 0.026, "step": 60120 }, { "epoch": 38.323773103887824, "grad_norm": 0.6367555260658264, "learning_rate": 9.393980363357368e-05, "loss": 0.031, "step": 60130 }, { "epoch": 38.33014659018483, "grad_norm": 0.520797848701477, "learning_rate": 9.393717326102959e-05, "loss": 0.0227, "step": 60140 }, { "epoch": 38.33652007648183, "grad_norm": 0.7482666969299316, "learning_rate": 9.393454235460736e-05, "loss": 0.0232, "step": 60150 }, { "epoch": 38.34289356277884, "grad_norm": 0.4850766956806183, "learning_rate": 9.393191091433896e-05, "loss": 0.0287, "step": 60160 }, { "epoch": 38.34926704907584, "grad_norm": 0.30845627188682556, "learning_rate": 9.392927894025637e-05, "loss": 0.0306, "step": 60170 }, { "epoch": 38.35564053537285, "grad_norm": 0.41747283935546875, "learning_rate": 9.392664643239155e-05, "loss": 0.0295, "step": 60180 }, { "epoch": 38.36201402166985, "grad_norm": 0.31774675846099854, "learning_rate": 9.392401339077652e-05, "loss": 0.0226, "step": 60190 }, { "epoch": 38.36838750796686, "grad_norm": 0.3680015504360199, "learning_rate": 9.392137981544325e-05, "loss": 0.0335, "step": 60200 }, { "epoch": 38.37476099426386, "grad_norm": 0.4650542438030243, "learning_rate": 9.391874570642375e-05, "loss": 0.0257, "step": 60210 }, { "epoch": 38.38113448056087, "grad_norm": 0.5423802137374878, "learning_rate": 9.391611106375002e-05, "loss": 0.0198, "step": 60220 }, { "epoch": 38.38750796685787, "grad_norm": 0.41541942954063416, "learning_rate": 9.391347588745407e-05, "loss": 0.0176, "step": 60230 }, { "epoch": 38.393881453154876, "grad_norm": 0.6082161068916321, "learning_rate": 9.391084017756794e-05, "loss": 0.0274, "step": 60240 }, { "epoch": 38.400254939451884, "grad_norm": 0.5188995599746704, "learning_rate": 9.390820393412363e-05, "loss": 0.0244, "step": 60250 }, { "epoch": 38.406628425748885, "grad_norm": 0.4696905016899109, "learning_rate": 9.39055671571532e-05, "loss": 0.031, "step": 60260 }, { "epoch": 38.41300191204589, "grad_norm": 0.4666711091995239, "learning_rate": 9.390292984668867e-05, "loss": 0.0278, "step": 60270 }, { "epoch": 38.41937539834289, "grad_norm": 0.3953295350074768, "learning_rate": 9.390029200276208e-05, "loss": 0.0273, "step": 60280 }, { "epoch": 38.4257488846399, "grad_norm": 0.4609638452529907, "learning_rate": 9.389765362540551e-05, "loss": 0.0218, "step": 60290 }, { "epoch": 38.4321223709369, "grad_norm": 0.6821250319480896, "learning_rate": 9.389501471465099e-05, "loss": 0.0269, "step": 60300 }, { "epoch": 38.43849585723391, "grad_norm": 0.46804070472717285, "learning_rate": 9.38923752705306e-05, "loss": 0.0241, "step": 60310 }, { "epoch": 38.44486934353091, "grad_norm": 0.46511170268058777, "learning_rate": 9.388973529307641e-05, "loss": 0.0194, "step": 60320 }, { "epoch": 38.45124282982792, "grad_norm": 0.35595694184303284, "learning_rate": 9.388709478232049e-05, "loss": 0.0253, "step": 60330 }, { "epoch": 38.45761631612492, "grad_norm": 0.5269393920898438, "learning_rate": 9.388445373829495e-05, "loss": 0.0221, "step": 60340 }, { "epoch": 38.46398980242193, "grad_norm": 0.7659209370613098, "learning_rate": 9.388181216103185e-05, "loss": 0.0344, "step": 60350 }, { "epoch": 38.47036328871893, "grad_norm": 0.709048330783844, "learning_rate": 9.38791700505633e-05, "loss": 0.0238, "step": 60360 }, { "epoch": 38.47673677501594, "grad_norm": 0.32942870259284973, "learning_rate": 9.38765274069214e-05, "loss": 0.0244, "step": 60370 }, { "epoch": 38.48311026131294, "grad_norm": 0.3576360046863556, "learning_rate": 9.38738842301383e-05, "loss": 0.0159, "step": 60380 }, { "epoch": 38.489483747609945, "grad_norm": 0.4606381356716156, "learning_rate": 9.387124052024605e-05, "loss": 0.0311, "step": 60390 }, { "epoch": 38.495857233906946, "grad_norm": 0.21920731663703918, "learning_rate": 9.386859627727682e-05, "loss": 0.0226, "step": 60400 }, { "epoch": 38.502230720203954, "grad_norm": 0.695374608039856, "learning_rate": 9.386595150126272e-05, "loss": 0.0239, "step": 60410 }, { "epoch": 38.508604206500955, "grad_norm": 0.5035502314567566, "learning_rate": 9.386330619223589e-05, "loss": 0.0264, "step": 60420 }, { "epoch": 38.51497769279796, "grad_norm": 0.3692714273929596, "learning_rate": 9.386066035022848e-05, "loss": 0.0173, "step": 60430 }, { "epoch": 38.521351179094964, "grad_norm": 0.698500394821167, "learning_rate": 9.385801397527263e-05, "loss": 0.0348, "step": 60440 }, { "epoch": 38.52772466539197, "grad_norm": 0.3472888171672821, "learning_rate": 9.385536706740051e-05, "loss": 0.0321, "step": 60450 }, { "epoch": 38.53409815168897, "grad_norm": 0.6455478668212891, "learning_rate": 9.385271962664427e-05, "loss": 0.0395, "step": 60460 }, { "epoch": 38.54047163798598, "grad_norm": 0.3985987603664398, "learning_rate": 9.385007165303609e-05, "loss": 0.0424, "step": 60470 }, { "epoch": 38.54684512428298, "grad_norm": 0.5377212762832642, "learning_rate": 9.384742314660812e-05, "loss": 0.0283, "step": 60480 }, { "epoch": 38.55321861057999, "grad_norm": 0.47219470143318176, "learning_rate": 9.384477410739257e-05, "loss": 0.0201, "step": 60490 }, { "epoch": 38.55959209687699, "grad_norm": 0.6544003486633301, "learning_rate": 9.384212453542162e-05, "loss": 0.0245, "step": 60500 }, { "epoch": 38.565965583174, "grad_norm": 0.36290815472602844, "learning_rate": 9.383947443072747e-05, "loss": 0.0366, "step": 60510 }, { "epoch": 38.572339069471, "grad_norm": 0.5753461122512817, "learning_rate": 9.383682379334231e-05, "loss": 0.0228, "step": 60520 }, { "epoch": 38.57871255576801, "grad_norm": 0.3562580645084381, "learning_rate": 9.383417262329834e-05, "loss": 0.0233, "step": 60530 }, { "epoch": 38.58508604206501, "grad_norm": 0.41233503818511963, "learning_rate": 9.38315209206278e-05, "loss": 0.0199, "step": 60540 }, { "epoch": 38.591459528362016, "grad_norm": 0.4829746186733246, "learning_rate": 9.38288686853629e-05, "loss": 0.0235, "step": 60550 }, { "epoch": 38.59783301465902, "grad_norm": 0.4323595464229584, "learning_rate": 9.382621591753584e-05, "loss": 0.0303, "step": 60560 }, { "epoch": 38.604206500956025, "grad_norm": 0.6684671640396118, "learning_rate": 9.38235626171789e-05, "loss": 0.0229, "step": 60570 }, { "epoch": 38.610579987253026, "grad_norm": 0.40282732248306274, "learning_rate": 9.38209087843243e-05, "loss": 0.0232, "step": 60580 }, { "epoch": 38.61695347355003, "grad_norm": 1.5885354280471802, "learning_rate": 9.381825441900427e-05, "loss": 0.0417, "step": 60590 }, { "epoch": 38.623326959847034, "grad_norm": 0.8120958805084229, "learning_rate": 9.381559952125108e-05, "loss": 0.0293, "step": 60600 }, { "epoch": 38.62970044614404, "grad_norm": 0.593139111995697, "learning_rate": 9.3812944091097e-05, "loss": 0.0214, "step": 60610 }, { "epoch": 38.63607393244104, "grad_norm": 0.32957184314727783, "learning_rate": 9.381028812857426e-05, "loss": 0.0215, "step": 60620 }, { "epoch": 38.64244741873805, "grad_norm": 0.5119672417640686, "learning_rate": 9.380763163371517e-05, "loss": 0.0216, "step": 60630 }, { "epoch": 38.64882090503505, "grad_norm": 0.42426764965057373, "learning_rate": 9.3804974606552e-05, "loss": 0.0211, "step": 60640 }, { "epoch": 38.65519439133206, "grad_norm": 0.48537662625312805, "learning_rate": 9.380231704711703e-05, "loss": 0.0238, "step": 60650 }, { "epoch": 38.66156787762906, "grad_norm": 0.4036189317703247, "learning_rate": 9.379965895544254e-05, "loss": 0.021, "step": 60660 }, { "epoch": 38.66794136392607, "grad_norm": 0.618143618106842, "learning_rate": 9.379700033156085e-05, "loss": 0.0205, "step": 60670 }, { "epoch": 38.67431485022307, "grad_norm": 0.675474226474762, "learning_rate": 9.379434117550425e-05, "loss": 0.0268, "step": 60680 }, { "epoch": 38.68068833652008, "grad_norm": 0.47942447662353516, "learning_rate": 9.379168148730506e-05, "loss": 0.0253, "step": 60690 }, { "epoch": 38.68706182281708, "grad_norm": 0.4218752384185791, "learning_rate": 9.378902126699558e-05, "loss": 0.0269, "step": 60700 }, { "epoch": 38.693435309114086, "grad_norm": 0.39426541328430176, "learning_rate": 9.378636051460816e-05, "loss": 0.0206, "step": 60710 }, { "epoch": 38.69980879541109, "grad_norm": 0.2343386709690094, "learning_rate": 9.378369923017511e-05, "loss": 0.0249, "step": 60720 }, { "epoch": 38.706182281708095, "grad_norm": 0.32120057940483093, "learning_rate": 9.378103741372879e-05, "loss": 0.0275, "step": 60730 }, { "epoch": 38.712555768005096, "grad_norm": 0.5255295038223267, "learning_rate": 9.377837506530151e-05, "loss": 0.0299, "step": 60740 }, { "epoch": 38.718929254302104, "grad_norm": 0.6949279308319092, "learning_rate": 9.377571218492567e-05, "loss": 0.0282, "step": 60750 }, { "epoch": 38.725302740599105, "grad_norm": 0.395681232213974, "learning_rate": 9.377304877263358e-05, "loss": 0.0191, "step": 60760 }, { "epoch": 38.73167622689611, "grad_norm": 0.48582664132118225, "learning_rate": 9.37703848284576e-05, "loss": 0.0327, "step": 60770 }, { "epoch": 38.738049713193114, "grad_norm": 1.597266674041748, "learning_rate": 9.376772035243013e-05, "loss": 0.0235, "step": 60780 }, { "epoch": 38.74442319949012, "grad_norm": 0.4729820489883423, "learning_rate": 9.376505534458355e-05, "loss": 0.0448, "step": 60790 }, { "epoch": 38.75079668578712, "grad_norm": 0.4519246220588684, "learning_rate": 9.37623898049502e-05, "loss": 0.0284, "step": 60800 }, { "epoch": 38.75717017208413, "grad_norm": 0.4270004630088806, "learning_rate": 9.375972373356252e-05, "loss": 0.0229, "step": 60810 }, { "epoch": 38.76354365838113, "grad_norm": 0.4639846682548523, "learning_rate": 9.375705713045287e-05, "loss": 0.0263, "step": 60820 }, { "epoch": 38.76991714467814, "grad_norm": 0.5524167418479919, "learning_rate": 9.375438999565366e-05, "loss": 0.0277, "step": 60830 }, { "epoch": 38.77629063097514, "grad_norm": 0.2680951654911041, "learning_rate": 9.375172232919731e-05, "loss": 0.0237, "step": 60840 }, { "epoch": 38.78266411727215, "grad_norm": 0.3372533321380615, "learning_rate": 9.374905413111622e-05, "loss": 0.026, "step": 60850 }, { "epoch": 38.78903760356915, "grad_norm": 0.5349087119102478, "learning_rate": 9.374638540144281e-05, "loss": 0.0304, "step": 60860 }, { "epoch": 38.79541108986616, "grad_norm": 0.5707732439041138, "learning_rate": 9.374371614020951e-05, "loss": 0.0275, "step": 60870 }, { "epoch": 38.80178457616316, "grad_norm": 0.4641343951225281, "learning_rate": 9.374104634744878e-05, "loss": 0.0258, "step": 60880 }, { "epoch": 38.808158062460166, "grad_norm": 0.44850394129753113, "learning_rate": 9.373837602319301e-05, "loss": 0.0194, "step": 60890 }, { "epoch": 38.814531548757174, "grad_norm": 0.4024102985858917, "learning_rate": 9.37357051674747e-05, "loss": 0.0221, "step": 60900 }, { "epoch": 38.820905035054174, "grad_norm": 0.5499832630157471, "learning_rate": 9.373303378032628e-05, "loss": 0.023, "step": 60910 }, { "epoch": 38.82727852135118, "grad_norm": 0.6478255987167358, "learning_rate": 9.373036186178019e-05, "loss": 0.0289, "step": 60920 }, { "epoch": 38.83365200764818, "grad_norm": 0.5420502424240112, "learning_rate": 9.372768941186891e-05, "loss": 0.0341, "step": 60930 }, { "epoch": 38.84002549394519, "grad_norm": 0.31860363483428955, "learning_rate": 9.372501643062494e-05, "loss": 0.0334, "step": 60940 }, { "epoch": 38.84639898024219, "grad_norm": 0.37389856576919556, "learning_rate": 9.372234291808073e-05, "loss": 0.0247, "step": 60950 }, { "epoch": 38.8527724665392, "grad_norm": 0.4862537682056427, "learning_rate": 9.371966887426877e-05, "loss": 0.0415, "step": 60960 }, { "epoch": 38.8591459528362, "grad_norm": 0.29928916692733765, "learning_rate": 9.371699429922157e-05, "loss": 0.0353, "step": 60970 }, { "epoch": 38.86551943913321, "grad_norm": 0.5489560961723328, "learning_rate": 9.371431919297158e-05, "loss": 0.0311, "step": 60980 }, { "epoch": 38.87189292543021, "grad_norm": 0.7020895481109619, "learning_rate": 9.371164355555135e-05, "loss": 0.0352, "step": 60990 }, { "epoch": 38.87826641172722, "grad_norm": 0.379985511302948, "learning_rate": 9.370896738699339e-05, "loss": 0.0264, "step": 61000 }, { "epoch": 38.88463989802422, "grad_norm": 0.36817467212677, "learning_rate": 9.370629068733022e-05, "loss": 0.0302, "step": 61010 }, { "epoch": 38.89101338432123, "grad_norm": 0.34796997904777527, "learning_rate": 9.370361345659433e-05, "loss": 0.0289, "step": 61020 }, { "epoch": 38.89738687061823, "grad_norm": 0.3631622791290283, "learning_rate": 9.370093569481829e-05, "loss": 0.0238, "step": 61030 }, { "epoch": 38.903760356915235, "grad_norm": 0.809110701084137, "learning_rate": 9.369825740203462e-05, "loss": 0.0251, "step": 61040 }, { "epoch": 38.910133843212236, "grad_norm": 0.5102279186248779, "learning_rate": 9.369557857827584e-05, "loss": 0.0356, "step": 61050 }, { "epoch": 38.916507329509244, "grad_norm": 0.6785749197006226, "learning_rate": 9.369289922357455e-05, "loss": 0.0321, "step": 61060 }, { "epoch": 38.922880815806245, "grad_norm": 0.5038653612136841, "learning_rate": 9.369021933796327e-05, "loss": 0.0249, "step": 61070 }, { "epoch": 38.92925430210325, "grad_norm": 0.5456666946411133, "learning_rate": 9.368753892147458e-05, "loss": 0.0359, "step": 61080 }, { "epoch": 38.935627788400254, "grad_norm": 0.6632387638092041, "learning_rate": 9.368485797414105e-05, "loss": 0.0303, "step": 61090 }, { "epoch": 38.94200127469726, "grad_norm": 0.3330095708370209, "learning_rate": 9.368217649599524e-05, "loss": 0.0225, "step": 61100 }, { "epoch": 38.94837476099426, "grad_norm": 0.32467252016067505, "learning_rate": 9.367949448706974e-05, "loss": 0.0284, "step": 61110 }, { "epoch": 38.95474824729127, "grad_norm": 0.4425029456615448, "learning_rate": 9.367681194739715e-05, "loss": 0.0212, "step": 61120 }, { "epoch": 38.96112173358827, "grad_norm": 0.5201866030693054, "learning_rate": 9.367412887701005e-05, "loss": 0.0283, "step": 61130 }, { "epoch": 38.96749521988528, "grad_norm": 0.45859822630882263, "learning_rate": 9.367144527594106e-05, "loss": 0.0276, "step": 61140 }, { "epoch": 38.97386870618228, "grad_norm": 0.4274231195449829, "learning_rate": 9.366876114422278e-05, "loss": 0.0224, "step": 61150 }, { "epoch": 38.98024219247929, "grad_norm": 0.5036986470222473, "learning_rate": 9.36660764818878e-05, "loss": 0.0315, "step": 61160 }, { "epoch": 38.98661567877629, "grad_norm": 0.3120930790901184, "learning_rate": 9.366339128896878e-05, "loss": 0.0251, "step": 61170 }, { "epoch": 38.9929891650733, "grad_norm": 0.25410112738609314, "learning_rate": 9.366070556549833e-05, "loss": 0.0384, "step": 61180 }, { "epoch": 38.9993626513703, "grad_norm": 0.8482842445373535, "learning_rate": 9.365801931150909e-05, "loss": 0.024, "step": 61190 }, { "epoch": 39.005736137667306, "grad_norm": 0.5147600173950195, "learning_rate": 9.365533252703368e-05, "loss": 0.0246, "step": 61200 }, { "epoch": 39.01210962396431, "grad_norm": 0.3764215111732483, "learning_rate": 9.365264521210478e-05, "loss": 0.0249, "step": 61210 }, { "epoch": 39.018483110261315, "grad_norm": 0.4297805726528168, "learning_rate": 9.3649957366755e-05, "loss": 0.0231, "step": 61220 }, { "epoch": 39.024856596558315, "grad_norm": 0.30081427097320557, "learning_rate": 9.364726899101703e-05, "loss": 0.0224, "step": 61230 }, { "epoch": 39.03123008285532, "grad_norm": 0.49720096588134766, "learning_rate": 9.364458008492356e-05, "loss": 0.0265, "step": 61240 }, { "epoch": 39.037603569152324, "grad_norm": 0.4793117046356201, "learning_rate": 9.364189064850721e-05, "loss": 0.0381, "step": 61250 }, { "epoch": 39.04397705544933, "grad_norm": 0.5174895524978638, "learning_rate": 9.36392006818007e-05, "loss": 0.0303, "step": 61260 }, { "epoch": 39.05035054174633, "grad_norm": 0.43084970116615295, "learning_rate": 9.363651018483667e-05, "loss": 0.0379, "step": 61270 }, { "epoch": 39.05672402804334, "grad_norm": 0.3522895574569702, "learning_rate": 9.363381915764788e-05, "loss": 0.0193, "step": 61280 }, { "epoch": 39.06309751434034, "grad_norm": 0.614811897277832, "learning_rate": 9.363112760026698e-05, "loss": 0.0396, "step": 61290 }, { "epoch": 39.06947100063735, "grad_norm": 0.6192757487297058, "learning_rate": 9.362843551272666e-05, "loss": 0.0377, "step": 61300 }, { "epoch": 39.07584448693435, "grad_norm": 0.7819889187812805, "learning_rate": 9.362574289505968e-05, "loss": 0.0377, "step": 61310 }, { "epoch": 39.08221797323136, "grad_norm": 0.4721345007419586, "learning_rate": 9.362304974729874e-05, "loss": 0.0328, "step": 61320 }, { "epoch": 39.08859145952836, "grad_norm": 0.3917187452316284, "learning_rate": 9.362035606947654e-05, "loss": 0.0292, "step": 61330 }, { "epoch": 39.09496494582537, "grad_norm": 0.3970746099948883, "learning_rate": 9.361766186162584e-05, "loss": 0.0322, "step": 61340 }, { "epoch": 39.10133843212237, "grad_norm": 0.7976200580596924, "learning_rate": 9.361496712377935e-05, "loss": 0.0241, "step": 61350 }, { "epoch": 39.107711918419376, "grad_norm": 0.5349223613739014, "learning_rate": 9.361227185596985e-05, "loss": 0.0224, "step": 61360 }, { "epoch": 39.11408540471638, "grad_norm": 0.4572047293186188, "learning_rate": 9.360957605823007e-05, "loss": 0.0232, "step": 61370 }, { "epoch": 39.120458891013385, "grad_norm": 0.4388844966888428, "learning_rate": 9.360687973059275e-05, "loss": 0.03, "step": 61380 }, { "epoch": 39.126832377310386, "grad_norm": 0.4839017391204834, "learning_rate": 9.360418287309067e-05, "loss": 0.0343, "step": 61390 }, { "epoch": 39.133205863607394, "grad_norm": 0.6416274309158325, "learning_rate": 9.360148548575662e-05, "loss": 0.031, "step": 61400 }, { "epoch": 39.139579349904395, "grad_norm": 0.624721884727478, "learning_rate": 9.359878756862333e-05, "loss": 0.0264, "step": 61410 }, { "epoch": 39.1459528362014, "grad_norm": 0.5643642544746399, "learning_rate": 9.359608912172362e-05, "loss": 0.0265, "step": 61420 }, { "epoch": 39.1523263224984, "grad_norm": 0.42867419123649597, "learning_rate": 9.359339014509028e-05, "loss": 0.0211, "step": 61430 }, { "epoch": 39.15869980879541, "grad_norm": 0.6147358417510986, "learning_rate": 9.359069063875607e-05, "loss": 0.0261, "step": 61440 }, { "epoch": 39.16507329509241, "grad_norm": 0.6237074732780457, "learning_rate": 9.358799060275382e-05, "loss": 0.0267, "step": 61450 }, { "epoch": 39.17144678138942, "grad_norm": 0.6123011112213135, "learning_rate": 9.358529003711632e-05, "loss": 0.0243, "step": 61460 }, { "epoch": 39.17782026768642, "grad_norm": 0.4978525638580322, "learning_rate": 9.35825889418764e-05, "loss": 0.0235, "step": 61470 }, { "epoch": 39.18419375398343, "grad_norm": 0.5927453637123108, "learning_rate": 9.357988731706689e-05, "loss": 0.0256, "step": 61480 }, { "epoch": 39.19056724028043, "grad_norm": 0.42637959122657776, "learning_rate": 9.357718516272058e-05, "loss": 0.0255, "step": 61490 }, { "epoch": 39.19694072657744, "grad_norm": 0.4307953715324402, "learning_rate": 9.357448247887036e-05, "loss": 0.0252, "step": 61500 }, { "epoch": 39.203314212874446, "grad_norm": 0.5068081617355347, "learning_rate": 9.3571779265549e-05, "loss": 0.0281, "step": 61510 }, { "epoch": 39.20968769917145, "grad_norm": 0.5475598573684692, "learning_rate": 9.35690755227894e-05, "loss": 0.0289, "step": 61520 }, { "epoch": 39.216061185468455, "grad_norm": 0.6064635515213013, "learning_rate": 9.356637125062439e-05, "loss": 0.025, "step": 61530 }, { "epoch": 39.222434671765456, "grad_norm": 0.3752914071083069, "learning_rate": 9.356366644908685e-05, "loss": 0.018, "step": 61540 }, { "epoch": 39.22880815806246, "grad_norm": 0.7047579884529114, "learning_rate": 9.356096111820963e-05, "loss": 0.0272, "step": 61550 }, { "epoch": 39.235181644359464, "grad_norm": 0.44205716252326965, "learning_rate": 9.355825525802559e-05, "loss": 0.0332, "step": 61560 }, { "epoch": 39.24155513065647, "grad_norm": 0.46331724524497986, "learning_rate": 9.355554886856762e-05, "loss": 0.0462, "step": 61570 }, { "epoch": 39.24792861695347, "grad_norm": 0.48321545124053955, "learning_rate": 9.355284194986862e-05, "loss": 0.0254, "step": 61580 }, { "epoch": 39.25430210325048, "grad_norm": 0.4457411766052246, "learning_rate": 9.355013450196146e-05, "loss": 0.0292, "step": 61590 }, { "epoch": 39.26067558954748, "grad_norm": 0.4657999575138092, "learning_rate": 9.354742652487907e-05, "loss": 0.0184, "step": 61600 }, { "epoch": 39.26704907584449, "grad_norm": 0.40327826142311096, "learning_rate": 9.35447180186543e-05, "loss": 0.0374, "step": 61610 }, { "epoch": 39.27342256214149, "grad_norm": 0.7968728542327881, "learning_rate": 9.354200898332012e-05, "loss": 0.0404, "step": 61620 }, { "epoch": 39.2797960484385, "grad_norm": 0.7786111235618591, "learning_rate": 9.35392994189094e-05, "loss": 0.0233, "step": 61630 }, { "epoch": 39.2861695347355, "grad_norm": 0.5833590626716614, "learning_rate": 9.35365893254551e-05, "loss": 0.0303, "step": 61640 }, { "epoch": 39.29254302103251, "grad_norm": 0.3441862165927887, "learning_rate": 9.353387870299013e-05, "loss": 0.0207, "step": 61650 }, { "epoch": 39.29891650732951, "grad_norm": 0.6298415660858154, "learning_rate": 9.353116755154742e-05, "loss": 0.0343, "step": 61660 }, { "epoch": 39.305289993626516, "grad_norm": 0.32209378480911255, "learning_rate": 9.352845587115994e-05, "loss": 0.047, "step": 61670 }, { "epoch": 39.31166347992352, "grad_norm": 0.44098782539367676, "learning_rate": 9.352574366186062e-05, "loss": 0.0229, "step": 61680 }, { "epoch": 39.318036966220525, "grad_norm": 0.4994329810142517, "learning_rate": 9.352303092368242e-05, "loss": 0.023, "step": 61690 }, { "epoch": 39.324410452517526, "grad_norm": 0.6623849868774414, "learning_rate": 9.35203176566583e-05, "loss": 0.0243, "step": 61700 }, { "epoch": 39.330783938814534, "grad_norm": 0.3024030029773712, "learning_rate": 9.351760386082123e-05, "loss": 0.0214, "step": 61710 }, { "epoch": 39.337157425111535, "grad_norm": 0.3235529065132141, "learning_rate": 9.351488953620418e-05, "loss": 0.0237, "step": 61720 }, { "epoch": 39.34353091140854, "grad_norm": 0.31255900859832764, "learning_rate": 9.351217468284014e-05, "loss": 0.0285, "step": 61730 }, { "epoch": 39.349904397705544, "grad_norm": 0.3639899790287018, "learning_rate": 9.35094593007621e-05, "loss": 0.0216, "step": 61740 }, { "epoch": 39.35627788400255, "grad_norm": 0.4521467387676239, "learning_rate": 9.350674339000303e-05, "loss": 0.0293, "step": 61750 }, { "epoch": 39.36265137029955, "grad_norm": 0.2472415268421173, "learning_rate": 9.350402695059598e-05, "loss": 0.0246, "step": 61760 }, { "epoch": 39.36902485659656, "grad_norm": 0.3704747259616852, "learning_rate": 9.350130998257391e-05, "loss": 0.0275, "step": 61770 }, { "epoch": 39.37539834289356, "grad_norm": 0.37372714281082153, "learning_rate": 9.349859248596986e-05, "loss": 0.0323, "step": 61780 }, { "epoch": 39.38177182919057, "grad_norm": 0.4915224015712738, "learning_rate": 9.349587446081684e-05, "loss": 0.0353, "step": 61790 }, { "epoch": 39.38814531548757, "grad_norm": 0.41226059198379517, "learning_rate": 9.349315590714786e-05, "loss": 0.0253, "step": 61800 }, { "epoch": 39.39451880178458, "grad_norm": 0.6475129723548889, "learning_rate": 9.3490436824996e-05, "loss": 0.0223, "step": 61810 }, { "epoch": 39.40089228808158, "grad_norm": 0.8030534982681274, "learning_rate": 9.348771721439426e-05, "loss": 0.0256, "step": 61820 }, { "epoch": 39.40726577437859, "grad_norm": 0.3132134974002838, "learning_rate": 9.34849970753757e-05, "loss": 0.0311, "step": 61830 }, { "epoch": 39.41363926067559, "grad_norm": 0.8907923102378845, "learning_rate": 9.348227640797336e-05, "loss": 0.0234, "step": 61840 }, { "epoch": 39.420012746972596, "grad_norm": 0.6327041387557983, "learning_rate": 9.347955521222032e-05, "loss": 0.0312, "step": 61850 }, { "epoch": 39.4263862332696, "grad_norm": 0.6635070443153381, "learning_rate": 9.347683348814962e-05, "loss": 0.0183, "step": 61860 }, { "epoch": 39.432759719566604, "grad_norm": 0.3856315612792969, "learning_rate": 9.347411123579435e-05, "loss": 0.0207, "step": 61870 }, { "epoch": 39.439133205863605, "grad_norm": 0.7494692802429199, "learning_rate": 9.347138845518757e-05, "loss": 0.0342, "step": 61880 }, { "epoch": 39.44550669216061, "grad_norm": 0.47494277358055115, "learning_rate": 9.34686651463624e-05, "loss": 0.03, "step": 61890 }, { "epoch": 39.451880178457614, "grad_norm": 0.602017343044281, "learning_rate": 9.346594130935189e-05, "loss": 0.0263, "step": 61900 }, { "epoch": 39.45825366475462, "grad_norm": 0.588264524936676, "learning_rate": 9.346321694418916e-05, "loss": 0.0213, "step": 61910 }, { "epoch": 39.46462715105162, "grad_norm": 0.7017486095428467, "learning_rate": 9.34604920509073e-05, "loss": 0.0276, "step": 61920 }, { "epoch": 39.47100063734863, "grad_norm": 0.6558878421783447, "learning_rate": 9.345776662953944e-05, "loss": 0.0236, "step": 61930 }, { "epoch": 39.47737412364563, "grad_norm": 0.6183075308799744, "learning_rate": 9.345504068011868e-05, "loss": 0.041, "step": 61940 }, { "epoch": 39.48374760994264, "grad_norm": 0.8397691249847412, "learning_rate": 9.345231420267814e-05, "loss": 0.0338, "step": 61950 }, { "epoch": 39.49012109623964, "grad_norm": 0.7653963565826416, "learning_rate": 9.344958719725098e-05, "loss": 0.0409, "step": 61960 }, { "epoch": 39.49649458253665, "grad_norm": 0.7158079743385315, "learning_rate": 9.34468596638703e-05, "loss": 0.03, "step": 61970 }, { "epoch": 39.50286806883365, "grad_norm": 0.3591374456882477, "learning_rate": 9.344413160256924e-05, "loss": 0.0279, "step": 61980 }, { "epoch": 39.50924155513066, "grad_norm": 0.45469602942466736, "learning_rate": 9.344140301338098e-05, "loss": 0.0186, "step": 61990 }, { "epoch": 39.51561504142766, "grad_norm": 0.44513916969299316, "learning_rate": 9.343867389633864e-05, "loss": 0.0313, "step": 62000 }, { "epoch": 39.521988527724666, "grad_norm": 0.5575897693634033, "learning_rate": 9.343594425147542e-05, "loss": 0.0249, "step": 62010 }, { "epoch": 39.52836201402167, "grad_norm": 0.4167698919773102, "learning_rate": 9.343321407882446e-05, "loss": 0.024, "step": 62020 }, { "epoch": 39.534735500318675, "grad_norm": 0.5615195035934448, "learning_rate": 9.343048337841893e-05, "loss": 0.024, "step": 62030 }, { "epoch": 39.541108986615676, "grad_norm": 0.34862884879112244, "learning_rate": 9.342775215029204e-05, "loss": 0.0166, "step": 62040 }, { "epoch": 39.547482472912684, "grad_norm": 0.3746243715286255, "learning_rate": 9.342502039447695e-05, "loss": 0.0244, "step": 62050 }, { "epoch": 39.553855959209685, "grad_norm": 0.649435818195343, "learning_rate": 9.342228811100685e-05, "loss": 0.0301, "step": 62060 }, { "epoch": 39.56022944550669, "grad_norm": 0.42625269293785095, "learning_rate": 9.341955529991497e-05, "loss": 0.0267, "step": 62070 }, { "epoch": 39.56660293180369, "grad_norm": 0.6124297976493835, "learning_rate": 9.34168219612345e-05, "loss": 0.0156, "step": 62080 }, { "epoch": 39.5729764181007, "grad_norm": 0.42768728733062744, "learning_rate": 9.341408809499863e-05, "loss": 0.0269, "step": 62090 }, { "epoch": 39.5793499043977, "grad_norm": 0.7170265316963196, "learning_rate": 9.341135370124062e-05, "loss": 0.0242, "step": 62100 }, { "epoch": 39.58572339069471, "grad_norm": 0.4428454041481018, "learning_rate": 9.340861877999367e-05, "loss": 0.0412, "step": 62110 }, { "epoch": 39.59209687699171, "grad_norm": 0.5291985869407654, "learning_rate": 9.340588333129101e-05, "loss": 0.0288, "step": 62120 }, { "epoch": 39.59847036328872, "grad_norm": 0.3868374526500702, "learning_rate": 9.340314735516588e-05, "loss": 0.0236, "step": 62130 }, { "epoch": 39.60484384958572, "grad_norm": 0.46146371960639954, "learning_rate": 9.340041085165156e-05, "loss": 0.0369, "step": 62140 }, { "epoch": 39.61121733588273, "grad_norm": 0.28697240352630615, "learning_rate": 9.339767382078125e-05, "loss": 0.0215, "step": 62150 }, { "epoch": 39.617590822179736, "grad_norm": 0.8732864856719971, "learning_rate": 9.339493626258824e-05, "loss": 0.0292, "step": 62160 }, { "epoch": 39.62396430847674, "grad_norm": 0.5142582654953003, "learning_rate": 9.339219817710576e-05, "loss": 0.0264, "step": 62170 }, { "epoch": 39.630337794773745, "grad_norm": 0.5248042345046997, "learning_rate": 9.338945956436712e-05, "loss": 0.0263, "step": 62180 }, { "epoch": 39.636711281070745, "grad_norm": 0.3663114905357361, "learning_rate": 9.338672042440559e-05, "loss": 0.0266, "step": 62190 }, { "epoch": 39.64308476736775, "grad_norm": 0.5371996760368347, "learning_rate": 9.338398075725442e-05, "loss": 0.0279, "step": 62200 }, { "epoch": 39.649458253664754, "grad_norm": 0.5203249454498291, "learning_rate": 9.338124056294694e-05, "loss": 0.0181, "step": 62210 }, { "epoch": 39.65583173996176, "grad_norm": 0.5507537722587585, "learning_rate": 9.337849984151642e-05, "loss": 0.0297, "step": 62220 }, { "epoch": 39.66220522625876, "grad_norm": 0.4123222231864929, "learning_rate": 9.337575859299617e-05, "loss": 0.0215, "step": 62230 }, { "epoch": 39.66857871255577, "grad_norm": 0.49891847372055054, "learning_rate": 9.33730168174195e-05, "loss": 0.0176, "step": 62240 }, { "epoch": 39.67495219885277, "grad_norm": 0.22477129101753235, "learning_rate": 9.337027451481973e-05, "loss": 0.0235, "step": 62250 }, { "epoch": 39.68132568514978, "grad_norm": 0.4633645713329315, "learning_rate": 9.336753168523017e-05, "loss": 0.0229, "step": 62260 }, { "epoch": 39.68769917144678, "grad_norm": 0.41032442450523376, "learning_rate": 9.336478832868416e-05, "loss": 0.0252, "step": 62270 }, { "epoch": 39.69407265774379, "grad_norm": 0.2654260993003845, "learning_rate": 9.336204444521502e-05, "loss": 0.0131, "step": 62280 }, { "epoch": 39.70044614404079, "grad_norm": 0.5611236095428467, "learning_rate": 9.335930003485612e-05, "loss": 0.0242, "step": 62290 }, { "epoch": 39.7068196303378, "grad_norm": 0.6043351888656616, "learning_rate": 9.335655509764076e-05, "loss": 0.0215, "step": 62300 }, { "epoch": 39.7131931166348, "grad_norm": 0.3117479085922241, "learning_rate": 9.335380963360233e-05, "loss": 0.0382, "step": 62310 }, { "epoch": 39.719566602931806, "grad_norm": 0.5316615700721741, "learning_rate": 9.335106364277419e-05, "loss": 0.0232, "step": 62320 }, { "epoch": 39.72594008922881, "grad_norm": 0.6081122159957886, "learning_rate": 9.334831712518969e-05, "loss": 0.0265, "step": 62330 }, { "epoch": 39.732313575525815, "grad_norm": 0.46377450227737427, "learning_rate": 9.334557008088221e-05, "loss": 0.0288, "step": 62340 }, { "epoch": 39.738687061822816, "grad_norm": 0.2957437336444855, "learning_rate": 9.334282250988511e-05, "loss": 0.0243, "step": 62350 }, { "epoch": 39.745060548119824, "grad_norm": 0.37826284766197205, "learning_rate": 9.334007441223183e-05, "loss": 0.0311, "step": 62360 }, { "epoch": 39.751434034416825, "grad_norm": 0.49150553345680237, "learning_rate": 9.333732578795571e-05, "loss": 0.0341, "step": 62370 }, { "epoch": 39.75780752071383, "grad_norm": 0.34751540422439575, "learning_rate": 9.333457663709015e-05, "loss": 0.0277, "step": 62380 }, { "epoch": 39.76418100701083, "grad_norm": 0.36657750606536865, "learning_rate": 9.333182695966858e-05, "loss": 0.0228, "step": 62390 }, { "epoch": 39.77055449330784, "grad_norm": 0.41618847846984863, "learning_rate": 9.33290767557244e-05, "loss": 0.0234, "step": 62400 }, { "epoch": 39.77692797960484, "grad_norm": 0.496868759393692, "learning_rate": 9.332632602529103e-05, "loss": 0.0253, "step": 62410 }, { "epoch": 39.78330146590185, "grad_norm": 0.40036311745643616, "learning_rate": 9.33235747684019e-05, "loss": 0.0339, "step": 62420 }, { "epoch": 39.78967495219885, "grad_norm": 0.5525063872337341, "learning_rate": 9.332082298509042e-05, "loss": 0.022, "step": 62430 }, { "epoch": 39.79604843849586, "grad_norm": 0.6169453263282776, "learning_rate": 9.331807067539005e-05, "loss": 0.0211, "step": 62440 }, { "epoch": 39.80242192479286, "grad_norm": 0.6020433306694031, "learning_rate": 9.331531783933421e-05, "loss": 0.0328, "step": 62450 }, { "epoch": 39.80879541108987, "grad_norm": 0.9072852730751038, "learning_rate": 9.331256447695635e-05, "loss": 0.0337, "step": 62460 }, { "epoch": 39.81516889738687, "grad_norm": 0.5409811735153198, "learning_rate": 9.330981058828996e-05, "loss": 0.0292, "step": 62470 }, { "epoch": 39.82154238368388, "grad_norm": 0.4983304440975189, "learning_rate": 9.330705617336848e-05, "loss": 0.0353, "step": 62480 }, { "epoch": 39.82791586998088, "grad_norm": 0.48281770944595337, "learning_rate": 9.330430123222537e-05, "loss": 0.0302, "step": 62490 }, { "epoch": 39.834289356277885, "grad_norm": 0.526093602180481, "learning_rate": 9.330154576489411e-05, "loss": 0.0225, "step": 62500 }, { "epoch": 39.840662842574886, "grad_norm": 0.568004310131073, "learning_rate": 9.32987897714082e-05, "loss": 0.0196, "step": 62510 }, { "epoch": 39.847036328871894, "grad_norm": 0.6009873747825623, "learning_rate": 9.32960332518011e-05, "loss": 0.0218, "step": 62520 }, { "epoch": 39.853409815168895, "grad_norm": 0.35112130641937256, "learning_rate": 9.329327620610633e-05, "loss": 0.0212, "step": 62530 }, { "epoch": 39.8597833014659, "grad_norm": 0.49526679515838623, "learning_rate": 9.329051863435738e-05, "loss": 0.0168, "step": 62540 }, { "epoch": 39.866156787762904, "grad_norm": 0.6273272037506104, "learning_rate": 9.328776053658774e-05, "loss": 0.0238, "step": 62550 }, { "epoch": 39.87253027405991, "grad_norm": 0.5736474394798279, "learning_rate": 9.328500191283096e-05, "loss": 0.0288, "step": 62560 }, { "epoch": 39.87890376035691, "grad_norm": 0.8673955798149109, "learning_rate": 9.328224276312052e-05, "loss": 0.0211, "step": 62570 }, { "epoch": 39.88527724665392, "grad_norm": 0.3162112534046173, "learning_rate": 9.327948308748997e-05, "loss": 0.0346, "step": 62580 }, { "epoch": 39.89165073295092, "grad_norm": 0.24026086926460266, "learning_rate": 9.327672288597285e-05, "loss": 0.0206, "step": 62590 }, { "epoch": 39.89802421924793, "grad_norm": 0.8014615774154663, "learning_rate": 9.327396215860267e-05, "loss": 0.044, "step": 62600 }, { "epoch": 39.90439770554493, "grad_norm": 0.38810208439826965, "learning_rate": 9.327120090541299e-05, "loss": 0.0269, "step": 62610 }, { "epoch": 39.91077119184194, "grad_norm": 0.34893250465393066, "learning_rate": 9.326843912643738e-05, "loss": 0.021, "step": 62620 }, { "epoch": 39.91714467813894, "grad_norm": 0.6214011311531067, "learning_rate": 9.326567682170936e-05, "loss": 0.0289, "step": 62630 }, { "epoch": 39.92351816443595, "grad_norm": 0.47269535064697266, "learning_rate": 9.326291399126255e-05, "loss": 0.016, "step": 62640 }, { "epoch": 39.92989165073295, "grad_norm": 0.5334193706512451, "learning_rate": 9.326015063513047e-05, "loss": 0.0355, "step": 62650 }, { "epoch": 39.936265137029956, "grad_norm": 0.6120707988739014, "learning_rate": 9.32573867533467e-05, "loss": 0.018, "step": 62660 }, { "epoch": 39.94263862332696, "grad_norm": 0.33862704038619995, "learning_rate": 9.325462234594486e-05, "loss": 0.0223, "step": 62670 }, { "epoch": 39.949012109623965, "grad_norm": 0.5869870185852051, "learning_rate": 9.325185741295852e-05, "loss": 0.0285, "step": 62680 }, { "epoch": 39.955385595920966, "grad_norm": 0.636843204498291, "learning_rate": 9.324909195442126e-05, "loss": 0.0246, "step": 62690 }, { "epoch": 39.961759082217974, "grad_norm": 0.5684895515441895, "learning_rate": 9.324632597036671e-05, "loss": 0.0296, "step": 62700 }, { "epoch": 39.968132568514974, "grad_norm": 0.2403639554977417, "learning_rate": 9.324355946082848e-05, "loss": 0.0166, "step": 62710 }, { "epoch": 39.97450605481198, "grad_norm": 0.2785743772983551, "learning_rate": 9.324079242584016e-05, "loss": 0.0314, "step": 62720 }, { "epoch": 39.98087954110898, "grad_norm": 0.3789971172809601, "learning_rate": 9.323802486543538e-05, "loss": 0.0148, "step": 62730 }, { "epoch": 39.98725302740599, "grad_norm": 0.5234839916229248, "learning_rate": 9.323525677964778e-05, "loss": 0.0301, "step": 62740 }, { "epoch": 39.99362651370299, "grad_norm": 0.4090384542942047, "learning_rate": 9.3232488168511e-05, "loss": 0.0157, "step": 62750 }, { "epoch": 40.0, "grad_norm": 0.5289058685302734, "learning_rate": 9.322971903205866e-05, "loss": 0.0295, "step": 62760 }, { "epoch": 40.00637348629701, "grad_norm": 0.3625575304031372, "learning_rate": 9.322694937032441e-05, "loss": 0.0412, "step": 62770 }, { "epoch": 40.01274697259401, "grad_norm": 0.3322247266769409, "learning_rate": 9.322417918334192e-05, "loss": 0.0196, "step": 62780 }, { "epoch": 40.01912045889102, "grad_norm": 0.33960163593292236, "learning_rate": 9.322140847114486e-05, "loss": 0.0202, "step": 62790 }, { "epoch": 40.02549394518802, "grad_norm": 0.8471006751060486, "learning_rate": 9.321863723376685e-05, "loss": 0.0252, "step": 62800 }, { "epoch": 40.031867431485026, "grad_norm": 0.37825411558151245, "learning_rate": 9.321586547124161e-05, "loss": 0.0203, "step": 62810 }, { "epoch": 40.038240917782026, "grad_norm": 0.36321723461151123, "learning_rate": 9.32130931836028e-05, "loss": 0.0276, "step": 62820 }, { "epoch": 40.044614404079034, "grad_norm": 0.4991523325443268, "learning_rate": 9.32103203708841e-05, "loss": 0.0347, "step": 62830 }, { "epoch": 40.050987890376035, "grad_norm": 0.6861948370933533, "learning_rate": 9.320754703311922e-05, "loss": 0.0229, "step": 62840 }, { "epoch": 40.05736137667304, "grad_norm": 0.6041613817214966, "learning_rate": 9.320477317034185e-05, "loss": 0.0253, "step": 62850 }, { "epoch": 40.063734862970044, "grad_norm": 0.3739321827888489, "learning_rate": 9.320199878258568e-05, "loss": 0.0238, "step": 62860 }, { "epoch": 40.07010834926705, "grad_norm": 0.40952154994010925, "learning_rate": 9.319922386988445e-05, "loss": 0.0262, "step": 62870 }, { "epoch": 40.07648183556405, "grad_norm": 0.31995755434036255, "learning_rate": 9.319644843227186e-05, "loss": 0.0266, "step": 62880 }, { "epoch": 40.08285532186106, "grad_norm": 0.5926610827445984, "learning_rate": 9.319367246978163e-05, "loss": 0.0265, "step": 62890 }, { "epoch": 40.08922880815806, "grad_norm": 0.3666624426841736, "learning_rate": 9.31908959824475e-05, "loss": 0.0244, "step": 62900 }, { "epoch": 40.09560229445507, "grad_norm": 0.4483577311038971, "learning_rate": 9.318811897030322e-05, "loss": 0.018, "step": 62910 }, { "epoch": 40.10197578075207, "grad_norm": 0.28789639472961426, "learning_rate": 9.318534143338249e-05, "loss": 0.0268, "step": 62920 }, { "epoch": 40.10834926704908, "grad_norm": 0.3110656142234802, "learning_rate": 9.318256337171911e-05, "loss": 0.0207, "step": 62930 }, { "epoch": 40.11472275334608, "grad_norm": 0.44846904277801514, "learning_rate": 9.317978478534681e-05, "loss": 0.0281, "step": 62940 }, { "epoch": 40.12109623964309, "grad_norm": 0.7120790481567383, "learning_rate": 9.317700567429935e-05, "loss": 0.0265, "step": 62950 }, { "epoch": 40.12746972594009, "grad_norm": 0.4908425807952881, "learning_rate": 9.31742260386105e-05, "loss": 0.0279, "step": 62960 }, { "epoch": 40.133843212237096, "grad_norm": 0.3664989173412323, "learning_rate": 9.317144587831405e-05, "loss": 0.0358, "step": 62970 }, { "epoch": 40.1402166985341, "grad_norm": 0.4467090368270874, "learning_rate": 9.316866519344378e-05, "loss": 0.0269, "step": 62980 }, { "epoch": 40.146590184831105, "grad_norm": 0.2641473114490509, "learning_rate": 9.316588398403345e-05, "loss": 0.024, "step": 62990 }, { "epoch": 40.152963671128106, "grad_norm": 0.712022066116333, "learning_rate": 9.316310225011689e-05, "loss": 0.0275, "step": 63000 }, { "epoch": 40.159337157425114, "grad_norm": 0.7244455218315125, "learning_rate": 9.316031999172788e-05, "loss": 0.0192, "step": 63010 }, { "epoch": 40.165710643722115, "grad_norm": 0.42831501364707947, "learning_rate": 9.315753720890024e-05, "loss": 0.0253, "step": 63020 }, { "epoch": 40.17208413001912, "grad_norm": 0.3438226282596588, "learning_rate": 9.315475390166777e-05, "loss": 0.0222, "step": 63030 }, { "epoch": 40.17845761631612, "grad_norm": 0.4604009985923767, "learning_rate": 9.315197007006429e-05, "loss": 0.0223, "step": 63040 }, { "epoch": 40.18483110261313, "grad_norm": 0.5738576650619507, "learning_rate": 9.314918571412363e-05, "loss": 0.0316, "step": 63050 }, { "epoch": 40.19120458891013, "grad_norm": 0.35562795400619507, "learning_rate": 9.314640083387963e-05, "loss": 0.0266, "step": 63060 }, { "epoch": 40.19757807520714, "grad_norm": 0.7249412536621094, "learning_rate": 9.314361542936613e-05, "loss": 0.0283, "step": 63070 }, { "epoch": 40.20395156150414, "grad_norm": 0.34809356927871704, "learning_rate": 9.314082950061695e-05, "loss": 0.0313, "step": 63080 }, { "epoch": 40.21032504780115, "grad_norm": 0.8902215361595154, "learning_rate": 9.313804304766598e-05, "loss": 0.0258, "step": 63090 }, { "epoch": 40.21669853409815, "grad_norm": 0.46959972381591797, "learning_rate": 9.313525607054705e-05, "loss": 0.0305, "step": 63100 }, { "epoch": 40.22307202039516, "grad_norm": 0.486623615026474, "learning_rate": 9.313246856929403e-05, "loss": 0.0303, "step": 63110 }, { "epoch": 40.22944550669216, "grad_norm": 0.4028576612472534, "learning_rate": 9.312968054394078e-05, "loss": 0.0285, "step": 63120 }, { "epoch": 40.23581899298917, "grad_norm": 0.4004812240600586, "learning_rate": 9.31268919945212e-05, "loss": 0.0261, "step": 63130 }, { "epoch": 40.24219247928617, "grad_norm": 0.5265500545501709, "learning_rate": 9.312410292106917e-05, "loss": 0.0301, "step": 63140 }, { "epoch": 40.248565965583175, "grad_norm": 0.9219235181808472, "learning_rate": 9.312131332361857e-05, "loss": 0.0224, "step": 63150 }, { "epoch": 40.254939451880176, "grad_norm": 0.5000272989273071, "learning_rate": 9.311852320220331e-05, "loss": 0.0261, "step": 63160 }, { "epoch": 40.261312938177184, "grad_norm": 0.5270541906356812, "learning_rate": 9.311573255685726e-05, "loss": 0.0336, "step": 63170 }, { "epoch": 40.267686424474185, "grad_norm": 0.4113840162754059, "learning_rate": 9.311294138761437e-05, "loss": 0.02, "step": 63180 }, { "epoch": 40.27405991077119, "grad_norm": 0.4303252696990967, "learning_rate": 9.311014969450853e-05, "loss": 0.026, "step": 63190 }, { "epoch": 40.280433397068194, "grad_norm": 0.40886190533638, "learning_rate": 9.310735747757367e-05, "loss": 0.022, "step": 63200 }, { "epoch": 40.2868068833652, "grad_norm": 0.6992054581642151, "learning_rate": 9.310456473684371e-05, "loss": 0.0365, "step": 63210 }, { "epoch": 40.2931803696622, "grad_norm": 0.5420033931732178, "learning_rate": 9.31017714723526e-05, "loss": 0.0274, "step": 63220 }, { "epoch": 40.29955385595921, "grad_norm": 0.48933863639831543, "learning_rate": 9.309897768413428e-05, "loss": 0.0308, "step": 63230 }, { "epoch": 40.30592734225621, "grad_norm": 0.3360384702682495, "learning_rate": 9.309618337222267e-05, "loss": 0.025, "step": 63240 }, { "epoch": 40.31230082855322, "grad_norm": 0.49482646584510803, "learning_rate": 9.309338853665174e-05, "loss": 0.0274, "step": 63250 }, { "epoch": 40.31867431485022, "grad_norm": 0.6703147292137146, "learning_rate": 9.309059317745547e-05, "loss": 0.0279, "step": 63260 }, { "epoch": 40.32504780114723, "grad_norm": 0.7898859977722168, "learning_rate": 9.308779729466781e-05, "loss": 0.0402, "step": 63270 }, { "epoch": 40.33142128744423, "grad_norm": 0.21609419584274292, "learning_rate": 9.308500088832272e-05, "loss": 0.0327, "step": 63280 }, { "epoch": 40.33779477374124, "grad_norm": 0.6299142241477966, "learning_rate": 9.30822039584542e-05, "loss": 0.0479, "step": 63290 }, { "epoch": 40.34416826003824, "grad_norm": 0.2556610703468323, "learning_rate": 9.307940650509621e-05, "loss": 0.0214, "step": 63300 }, { "epoch": 40.350541746335246, "grad_norm": 0.44743067026138306, "learning_rate": 9.307660852828276e-05, "loss": 0.0268, "step": 63310 }, { "epoch": 40.35691523263225, "grad_norm": 0.2864423394203186, "learning_rate": 9.307381002804785e-05, "loss": 0.0192, "step": 63320 }, { "epoch": 40.363288718929255, "grad_norm": 0.22436396777629852, "learning_rate": 9.307101100442548e-05, "loss": 0.0312, "step": 63330 }, { "epoch": 40.369662205226255, "grad_norm": 0.4444640874862671, "learning_rate": 9.306821145744967e-05, "loss": 0.025, "step": 63340 }, { "epoch": 40.37603569152326, "grad_norm": 0.3553168773651123, "learning_rate": 9.306541138715441e-05, "loss": 0.0197, "step": 63350 }, { "epoch": 40.382409177820264, "grad_norm": 0.2972046136856079, "learning_rate": 9.306261079357374e-05, "loss": 0.0299, "step": 63360 }, { "epoch": 40.38878266411727, "grad_norm": 0.4508771300315857, "learning_rate": 9.305980967674171e-05, "loss": 0.0203, "step": 63370 }, { "epoch": 40.39515615041427, "grad_norm": 0.5381671786308289, "learning_rate": 9.305700803669232e-05, "loss": 0.0399, "step": 63380 }, { "epoch": 40.40152963671128, "grad_norm": 0.5347617864608765, "learning_rate": 9.305420587345966e-05, "loss": 0.0258, "step": 63390 }, { "epoch": 40.40790312300829, "grad_norm": 0.45342910289764404, "learning_rate": 9.305140318707772e-05, "loss": 0.0238, "step": 63400 }, { "epoch": 40.41427660930529, "grad_norm": 0.5812029242515564, "learning_rate": 9.304859997758059e-05, "loss": 0.0274, "step": 63410 }, { "epoch": 40.4206500956023, "grad_norm": 0.3914230763912201, "learning_rate": 9.304579624500232e-05, "loss": 0.0181, "step": 63420 }, { "epoch": 40.4270235818993, "grad_norm": 0.6828516125679016, "learning_rate": 9.304299198937699e-05, "loss": 0.0254, "step": 63430 }, { "epoch": 40.43339706819631, "grad_norm": 0.36527177691459656, "learning_rate": 9.304018721073867e-05, "loss": 0.0268, "step": 63440 }, { "epoch": 40.43977055449331, "grad_norm": 0.6079586148262024, "learning_rate": 9.303738190912143e-05, "loss": 0.0351, "step": 63450 }, { "epoch": 40.446144040790315, "grad_norm": 0.23152729868888855, "learning_rate": 9.303457608455939e-05, "loss": 0.021, "step": 63460 }, { "epoch": 40.452517527087316, "grad_norm": 0.6136364936828613, "learning_rate": 9.30317697370866e-05, "loss": 0.0256, "step": 63470 }, { "epoch": 40.458891013384324, "grad_norm": 0.641066312789917, "learning_rate": 9.302896286673717e-05, "loss": 0.0306, "step": 63480 }, { "epoch": 40.465264499681325, "grad_norm": 0.48491403460502625, "learning_rate": 9.302615547354523e-05, "loss": 0.0219, "step": 63490 }, { "epoch": 40.47163798597833, "grad_norm": 0.8864009380340576, "learning_rate": 9.302334755754487e-05, "loss": 0.0379, "step": 63500 }, { "epoch": 40.478011472275334, "grad_norm": 0.3024265766143799, "learning_rate": 9.302053911877022e-05, "loss": 0.0302, "step": 63510 }, { "epoch": 40.48438495857234, "grad_norm": 0.3069208860397339, "learning_rate": 9.301773015725541e-05, "loss": 0.0301, "step": 63520 }, { "epoch": 40.49075844486934, "grad_norm": 0.3112746477127075, "learning_rate": 9.301492067303454e-05, "loss": 0.0377, "step": 63530 }, { "epoch": 40.49713193116635, "grad_norm": 0.3950579762458801, "learning_rate": 9.301211066614179e-05, "loss": 0.0183, "step": 63540 }, { "epoch": 40.50350541746335, "grad_norm": 0.5443411469459534, "learning_rate": 9.300930013661129e-05, "loss": 0.0257, "step": 63550 }, { "epoch": 40.50987890376036, "grad_norm": 0.3092659115791321, "learning_rate": 9.300648908447718e-05, "loss": 0.0205, "step": 63560 }, { "epoch": 40.51625239005736, "grad_norm": 0.33626043796539307, "learning_rate": 9.300367750977362e-05, "loss": 0.0238, "step": 63570 }, { "epoch": 40.52262587635437, "grad_norm": 0.6261965036392212, "learning_rate": 9.300086541253476e-05, "loss": 0.0394, "step": 63580 }, { "epoch": 40.52899936265137, "grad_norm": 0.8300643563270569, "learning_rate": 9.299805279279481e-05, "loss": 0.0298, "step": 63590 }, { "epoch": 40.53537284894838, "grad_norm": 0.40425753593444824, "learning_rate": 9.299523965058791e-05, "loss": 0.0343, "step": 63600 }, { "epoch": 40.54174633524538, "grad_norm": 0.4221994876861572, "learning_rate": 9.299242598594826e-05, "loss": 0.0225, "step": 63610 }, { "epoch": 40.548119821542386, "grad_norm": 0.4299912750720978, "learning_rate": 9.298961179891004e-05, "loss": 0.0285, "step": 63620 }, { "epoch": 40.55449330783939, "grad_norm": 0.6174660325050354, "learning_rate": 9.298679708950744e-05, "loss": 0.0374, "step": 63630 }, { "epoch": 40.560866794136395, "grad_norm": 0.5485066175460815, "learning_rate": 9.298398185777468e-05, "loss": 0.0274, "step": 63640 }, { "epoch": 40.567240280433396, "grad_norm": 0.36455947160720825, "learning_rate": 9.298116610374595e-05, "loss": 0.0238, "step": 63650 }, { "epoch": 40.5736137667304, "grad_norm": 0.7693423628807068, "learning_rate": 9.297834982745549e-05, "loss": 0.0261, "step": 63660 }, { "epoch": 40.579987253027404, "grad_norm": 0.4184761643409729, "learning_rate": 9.297553302893747e-05, "loss": 0.0276, "step": 63670 }, { "epoch": 40.58636073932441, "grad_norm": 0.3175077736377716, "learning_rate": 9.297271570822617e-05, "loss": 0.0182, "step": 63680 }, { "epoch": 40.59273422562141, "grad_norm": 0.6506162881851196, "learning_rate": 9.296989786535579e-05, "loss": 0.0291, "step": 63690 }, { "epoch": 40.59910771191842, "grad_norm": 0.515789806842804, "learning_rate": 9.296707950036059e-05, "loss": 0.0257, "step": 63700 }, { "epoch": 40.60548119821542, "grad_norm": 0.6423799395561218, "learning_rate": 9.296426061327479e-05, "loss": 0.0499, "step": 63710 }, { "epoch": 40.61185468451243, "grad_norm": 0.4790751039981842, "learning_rate": 9.296144120413267e-05, "loss": 0.0448, "step": 63720 }, { "epoch": 40.61822817080943, "grad_norm": 0.723077654838562, "learning_rate": 9.295862127296848e-05, "loss": 0.0266, "step": 63730 }, { "epoch": 40.62460165710644, "grad_norm": 0.33511605858802795, "learning_rate": 9.295580081981647e-05, "loss": 0.0229, "step": 63740 }, { "epoch": 40.63097514340344, "grad_norm": 0.3703979551792145, "learning_rate": 9.295297984471094e-05, "loss": 0.0208, "step": 63750 }, { "epoch": 40.63734862970045, "grad_norm": 0.31570249795913696, "learning_rate": 9.295015834768614e-05, "loss": 0.0165, "step": 63760 }, { "epoch": 40.64372211599745, "grad_norm": 0.3661767244338989, "learning_rate": 9.294733632877637e-05, "loss": 0.0211, "step": 63770 }, { "epoch": 40.650095602294456, "grad_norm": 0.2925328314304352, "learning_rate": 9.294451378801591e-05, "loss": 0.0198, "step": 63780 }, { "epoch": 40.65646908859146, "grad_norm": 0.35155346989631653, "learning_rate": 9.294169072543906e-05, "loss": 0.0263, "step": 63790 }, { "epoch": 40.662842574888465, "grad_norm": 0.4355350434780121, "learning_rate": 9.293886714108013e-05, "loss": 0.0298, "step": 63800 }, { "epoch": 40.669216061185466, "grad_norm": 0.5581322908401489, "learning_rate": 9.293604303497343e-05, "loss": 0.0266, "step": 63810 }, { "epoch": 40.675589547482474, "grad_norm": 0.5159093141555786, "learning_rate": 9.293321840715326e-05, "loss": 0.0361, "step": 63820 }, { "epoch": 40.681963033779475, "grad_norm": 0.3896491825580597, "learning_rate": 9.293039325765393e-05, "loss": 0.032, "step": 63830 }, { "epoch": 40.68833652007648, "grad_norm": 0.3469632863998413, "learning_rate": 9.292756758650983e-05, "loss": 0.037, "step": 63840 }, { "epoch": 40.694710006373484, "grad_norm": 0.8771584033966064, "learning_rate": 9.292474139375522e-05, "loss": 0.0369, "step": 63850 }, { "epoch": 40.70108349267049, "grad_norm": 0.9623972177505493, "learning_rate": 9.292191467942449e-05, "loss": 0.0325, "step": 63860 }, { "epoch": 40.70745697896749, "grad_norm": 0.5989234447479248, "learning_rate": 9.291908744355196e-05, "loss": 0.0237, "step": 63870 }, { "epoch": 40.7138304652645, "grad_norm": 0.48549893498420715, "learning_rate": 9.291625968617201e-05, "loss": 0.0212, "step": 63880 }, { "epoch": 40.7202039515615, "grad_norm": 0.7818636894226074, "learning_rate": 9.291343140731899e-05, "loss": 0.0239, "step": 63890 }, { "epoch": 40.72657743785851, "grad_norm": 0.6163133382797241, "learning_rate": 9.291060260702724e-05, "loss": 0.0365, "step": 63900 }, { "epoch": 40.73295092415551, "grad_norm": 0.32581809163093567, "learning_rate": 9.290777328533115e-05, "loss": 0.0213, "step": 63910 }, { "epoch": 40.73932441045252, "grad_norm": 0.33073508739471436, "learning_rate": 9.290494344226512e-05, "loss": 0.0241, "step": 63920 }, { "epoch": 40.74569789674952, "grad_norm": 0.41424766182899475, "learning_rate": 9.290211307786352e-05, "loss": 0.0278, "step": 63930 }, { "epoch": 40.75207138304653, "grad_norm": 0.4511762261390686, "learning_rate": 9.289928219216073e-05, "loss": 0.0339, "step": 63940 }, { "epoch": 40.75844486934353, "grad_norm": 0.2600332498550415, "learning_rate": 9.289645078519117e-05, "loss": 0.0246, "step": 63950 }, { "epoch": 40.764818355640536, "grad_norm": 0.44549131393432617, "learning_rate": 9.289361885698922e-05, "loss": 0.0228, "step": 63960 }, { "epoch": 40.77119184193754, "grad_norm": 0.3620595335960388, "learning_rate": 9.28907864075893e-05, "loss": 0.0245, "step": 63970 }, { "epoch": 40.777565328234544, "grad_norm": 0.6010586619377136, "learning_rate": 9.288795343702584e-05, "loss": 0.025, "step": 63980 }, { "epoch": 40.783938814531545, "grad_norm": 0.5105922222137451, "learning_rate": 9.288511994533325e-05, "loss": 0.0216, "step": 63990 }, { "epoch": 40.79031230082855, "grad_norm": 0.5473626255989075, "learning_rate": 9.288228593254597e-05, "loss": 0.0403, "step": 64000 }, { "epoch": 40.796685787125554, "grad_norm": 0.4037788212299347, "learning_rate": 9.287945139869843e-05, "loss": 0.0236, "step": 64010 }, { "epoch": 40.80305927342256, "grad_norm": 0.6704573631286621, "learning_rate": 9.287661634382505e-05, "loss": 0.0282, "step": 64020 }, { "epoch": 40.80943275971957, "grad_norm": 0.2981036603450775, "learning_rate": 9.287378076796031e-05, "loss": 0.0254, "step": 64030 }, { "epoch": 40.81580624601657, "grad_norm": 0.4433252215385437, "learning_rate": 9.287094467113865e-05, "loss": 0.02, "step": 64040 }, { "epoch": 40.82217973231358, "grad_norm": 0.4513382017612457, "learning_rate": 9.286810805339454e-05, "loss": 0.0245, "step": 64050 }, { "epoch": 40.82855321861058, "grad_norm": 0.4106238782405853, "learning_rate": 9.286527091476245e-05, "loss": 0.0233, "step": 64060 }, { "epoch": 40.83492670490759, "grad_norm": 0.36267948150634766, "learning_rate": 9.286243325527684e-05, "loss": 0.0181, "step": 64070 }, { "epoch": 40.84130019120459, "grad_norm": 0.3590107262134552, "learning_rate": 9.28595950749722e-05, "loss": 0.0201, "step": 64080 }, { "epoch": 40.8476736775016, "grad_norm": 0.4199380874633789, "learning_rate": 9.285675637388301e-05, "loss": 0.0232, "step": 64090 }, { "epoch": 40.8540471637986, "grad_norm": 0.24521631002426147, "learning_rate": 9.285391715204375e-05, "loss": 0.0213, "step": 64100 }, { "epoch": 40.860420650095605, "grad_norm": 0.5479651093482971, "learning_rate": 9.285107740948896e-05, "loss": 0.0272, "step": 64110 }, { "epoch": 40.866794136392606, "grad_norm": 0.6227468252182007, "learning_rate": 9.28482371462531e-05, "loss": 0.0263, "step": 64120 }, { "epoch": 40.873167622689614, "grad_norm": 0.5131120681762695, "learning_rate": 9.284539636237072e-05, "loss": 0.0241, "step": 64130 }, { "epoch": 40.879541108986615, "grad_norm": 0.693159818649292, "learning_rate": 9.284255505787632e-05, "loss": 0.0306, "step": 64140 }, { "epoch": 40.88591459528362, "grad_norm": 0.30935630202293396, "learning_rate": 9.283971323280442e-05, "loss": 0.0179, "step": 64150 }, { "epoch": 40.892288081580624, "grad_norm": 0.40742743015289307, "learning_rate": 9.283687088718957e-05, "loss": 0.0274, "step": 64160 }, { "epoch": 40.89866156787763, "grad_norm": 0.716603696346283, "learning_rate": 9.283402802106629e-05, "loss": 0.0334, "step": 64170 }, { "epoch": 40.90503505417463, "grad_norm": 0.4814022183418274, "learning_rate": 9.283118463446911e-05, "loss": 0.0216, "step": 64180 }, { "epoch": 40.91140854047164, "grad_norm": 0.6144996881484985, "learning_rate": 9.282834072743262e-05, "loss": 0.0347, "step": 64190 }, { "epoch": 40.91778202676864, "grad_norm": 0.48591917753219604, "learning_rate": 9.282549629999134e-05, "loss": 0.0242, "step": 64200 }, { "epoch": 40.92415551306565, "grad_norm": 0.43578240275382996, "learning_rate": 9.282265135217985e-05, "loss": 0.0348, "step": 64210 }, { "epoch": 40.93052899936265, "grad_norm": 0.8711801767349243, "learning_rate": 9.281980588403272e-05, "loss": 0.0238, "step": 64220 }, { "epoch": 40.93690248565966, "grad_norm": 0.53843754529953, "learning_rate": 9.281695989558451e-05, "loss": 0.0216, "step": 64230 }, { "epoch": 40.94327597195666, "grad_norm": 0.36928272247314453, "learning_rate": 9.281411338686982e-05, "loss": 0.0242, "step": 64240 }, { "epoch": 40.94964945825367, "grad_norm": 0.3841590881347656, "learning_rate": 9.281126635792322e-05, "loss": 0.0267, "step": 64250 }, { "epoch": 40.95602294455067, "grad_norm": 0.557015597820282, "learning_rate": 9.280841880877932e-05, "loss": 0.0183, "step": 64260 }, { "epoch": 40.962396430847676, "grad_norm": 0.6608373522758484, "learning_rate": 9.280557073947271e-05, "loss": 0.0234, "step": 64270 }, { "epoch": 40.96876991714468, "grad_norm": 0.3394624888896942, "learning_rate": 9.280272215003801e-05, "loss": 0.0179, "step": 64280 }, { "epoch": 40.975143403441685, "grad_norm": 0.41730642318725586, "learning_rate": 9.279987304050982e-05, "loss": 0.0308, "step": 64290 }, { "epoch": 40.981516889738685, "grad_norm": 0.7683445811271667, "learning_rate": 9.279702341092274e-05, "loss": 0.0417, "step": 64300 }, { "epoch": 40.98789037603569, "grad_norm": 0.3285045325756073, "learning_rate": 9.279417326131143e-05, "loss": 0.0354, "step": 64310 }, { "epoch": 40.994263862332694, "grad_norm": 0.5637664198875427, "learning_rate": 9.279132259171053e-05, "loss": 0.0273, "step": 64320 }, { "epoch": 41.0006373486297, "grad_norm": 0.5384771823883057, "learning_rate": 9.278847140215464e-05, "loss": 0.0303, "step": 64330 }, { "epoch": 41.0070108349267, "grad_norm": 0.3501203954219818, "learning_rate": 9.278561969267842e-05, "loss": 0.0258, "step": 64340 }, { "epoch": 41.01338432122371, "grad_norm": 0.37752172350883484, "learning_rate": 9.278276746331654e-05, "loss": 0.0229, "step": 64350 }, { "epoch": 41.01975780752071, "grad_norm": 0.2993519604206085, "learning_rate": 9.277991471410363e-05, "loss": 0.0384, "step": 64360 }, { "epoch": 41.02613129381772, "grad_norm": 0.3055190145969391, "learning_rate": 9.27770614450744e-05, "loss": 0.0166, "step": 64370 }, { "epoch": 41.03250478011472, "grad_norm": 0.4883407652378082, "learning_rate": 9.277420765626343e-05, "loss": 0.0233, "step": 64380 }, { "epoch": 41.03887826641173, "grad_norm": 0.7747842073440552, "learning_rate": 9.277135334770548e-05, "loss": 0.0321, "step": 64390 }, { "epoch": 41.04525175270873, "grad_norm": 0.42877596616744995, "learning_rate": 9.276849851943521e-05, "loss": 0.0291, "step": 64400 }, { "epoch": 41.05162523900574, "grad_norm": 0.41594821214675903, "learning_rate": 9.27656431714873e-05, "loss": 0.0218, "step": 64410 }, { "epoch": 41.05799872530274, "grad_norm": 0.41545432806015015, "learning_rate": 9.276278730389642e-05, "loss": 0.0182, "step": 64420 }, { "epoch": 41.064372211599746, "grad_norm": 0.6110652685165405, "learning_rate": 9.275993091669734e-05, "loss": 0.0335, "step": 64430 }, { "epoch": 41.07074569789675, "grad_norm": 0.3596121668815613, "learning_rate": 9.27570740099247e-05, "loss": 0.0159, "step": 64440 }, { "epoch": 41.077119184193755, "grad_norm": 0.6247264742851257, "learning_rate": 9.275421658361325e-05, "loss": 0.0231, "step": 64450 }, { "epoch": 41.083492670490756, "grad_norm": 0.8417141437530518, "learning_rate": 9.27513586377977e-05, "loss": 0.0276, "step": 64460 }, { "epoch": 41.089866156787764, "grad_norm": 0.3102891445159912, "learning_rate": 9.274850017251278e-05, "loss": 0.0234, "step": 64470 }, { "epoch": 41.096239643084765, "grad_norm": 0.4532087743282318, "learning_rate": 9.274564118779323e-05, "loss": 0.0283, "step": 64480 }, { "epoch": 41.10261312938177, "grad_norm": 0.7715798616409302, "learning_rate": 9.274278168367378e-05, "loss": 0.0369, "step": 64490 }, { "epoch": 41.10898661567877, "grad_norm": 0.5587839484214783, "learning_rate": 9.273992166018916e-05, "loss": 0.0241, "step": 64500 }, { "epoch": 41.11536010197578, "grad_norm": 0.5093590021133423, "learning_rate": 9.273706111737415e-05, "loss": 0.0239, "step": 64510 }, { "epoch": 41.12173358827278, "grad_norm": 0.6072325706481934, "learning_rate": 9.27342000552635e-05, "loss": 0.0251, "step": 64520 }, { "epoch": 41.12810707456979, "grad_norm": 0.21390171349048615, "learning_rate": 9.273133847389197e-05, "loss": 0.023, "step": 64530 }, { "epoch": 41.13448056086679, "grad_norm": 0.38769009709358215, "learning_rate": 9.272847637329431e-05, "loss": 0.0226, "step": 64540 }, { "epoch": 41.1408540471638, "grad_norm": 0.1946212202310562, "learning_rate": 9.272561375350535e-05, "loss": 0.026, "step": 64550 }, { "epoch": 41.1472275334608, "grad_norm": 0.4732208549976349, "learning_rate": 9.272275061455984e-05, "loss": 0.0364, "step": 64560 }, { "epoch": 41.15360101975781, "grad_norm": 0.5318770408630371, "learning_rate": 9.271988695649256e-05, "loss": 0.0289, "step": 64570 }, { "epoch": 41.15997450605481, "grad_norm": 0.4918212592601776, "learning_rate": 9.271702277933833e-05, "loss": 0.0257, "step": 64580 }, { "epoch": 41.16634799235182, "grad_norm": 0.5507063269615173, "learning_rate": 9.271415808313193e-05, "loss": 0.0422, "step": 64590 }, { "epoch": 41.17272147864882, "grad_norm": 0.4971425533294678, "learning_rate": 9.271129286790819e-05, "loss": 0.0175, "step": 64600 }, { "epoch": 41.179094964945826, "grad_norm": 0.4122069478034973, "learning_rate": 9.270842713370191e-05, "loss": 0.0222, "step": 64610 }, { "epoch": 41.185468451242826, "grad_norm": 0.5299763679504395, "learning_rate": 9.270556088054791e-05, "loss": 0.0318, "step": 64620 }, { "epoch": 41.191841937539834, "grad_norm": 0.2601962983608246, "learning_rate": 9.270269410848105e-05, "loss": 0.023, "step": 64630 }, { "epoch": 41.198215423836835, "grad_norm": 0.5905089378356934, "learning_rate": 9.269982681753612e-05, "loss": 0.0346, "step": 64640 }, { "epoch": 41.20458891013384, "grad_norm": 0.3155181407928467, "learning_rate": 9.269695900774798e-05, "loss": 0.0219, "step": 64650 }, { "epoch": 41.21096239643085, "grad_norm": 0.37132367491722107, "learning_rate": 9.269409067915147e-05, "loss": 0.0273, "step": 64660 }, { "epoch": 41.21733588272785, "grad_norm": 0.3521001636981964, "learning_rate": 9.269122183178146e-05, "loss": 0.023, "step": 64670 }, { "epoch": 41.22370936902486, "grad_norm": 0.3338051736354828, "learning_rate": 9.26883524656728e-05, "loss": 0.0226, "step": 64680 }, { "epoch": 41.23008285532186, "grad_norm": 0.3368802070617676, "learning_rate": 9.268548258086033e-05, "loss": 0.02, "step": 64690 }, { "epoch": 41.23645634161887, "grad_norm": 0.5896616578102112, "learning_rate": 9.268261217737897e-05, "loss": 0.0262, "step": 64700 }, { "epoch": 41.24282982791587, "grad_norm": 0.669954776763916, "learning_rate": 9.267974125526356e-05, "loss": 0.0475, "step": 64710 }, { "epoch": 41.24920331421288, "grad_norm": 0.32509857416152954, "learning_rate": 9.2676869814549e-05, "loss": 0.0276, "step": 64720 }, { "epoch": 41.25557680050988, "grad_norm": 0.3558478057384491, "learning_rate": 9.267399785527017e-05, "loss": 0.0192, "step": 64730 }, { "epoch": 41.261950286806886, "grad_norm": 0.6734727621078491, "learning_rate": 9.2671125377462e-05, "loss": 0.0325, "step": 64740 }, { "epoch": 41.26832377310389, "grad_norm": 0.6073101758956909, "learning_rate": 9.266825238115935e-05, "loss": 0.0324, "step": 64750 }, { "epoch": 41.274697259400895, "grad_norm": 0.4673481583595276, "learning_rate": 9.266537886639714e-05, "loss": 0.0277, "step": 64760 }, { "epoch": 41.281070745697896, "grad_norm": 0.39584973454475403, "learning_rate": 9.266250483321031e-05, "loss": 0.0234, "step": 64770 }, { "epoch": 41.287444231994904, "grad_norm": 0.5112573504447937, "learning_rate": 9.265963028163376e-05, "loss": 0.0215, "step": 64780 }, { "epoch": 41.293817718291905, "grad_norm": 0.5653212070465088, "learning_rate": 9.265675521170243e-05, "loss": 0.0209, "step": 64790 }, { "epoch": 41.30019120458891, "grad_norm": 0.5304442644119263, "learning_rate": 9.265387962345125e-05, "loss": 0.034, "step": 64800 }, { "epoch": 41.306564690885914, "grad_norm": 0.24975214898586273, "learning_rate": 9.265100351691516e-05, "loss": 0.0349, "step": 64810 }, { "epoch": 41.31293817718292, "grad_norm": 0.33544230461120605, "learning_rate": 9.26481268921291e-05, "loss": 0.0205, "step": 64820 }, { "epoch": 41.31931166347992, "grad_norm": 0.4971437454223633, "learning_rate": 9.264524974912803e-05, "loss": 0.0274, "step": 64830 }, { "epoch": 41.32568514977693, "grad_norm": 0.44031015038490295, "learning_rate": 9.264237208794692e-05, "loss": 0.0177, "step": 64840 }, { "epoch": 41.33205863607393, "grad_norm": 0.4225476086139679, "learning_rate": 9.263949390862073e-05, "loss": 0.0391, "step": 64850 }, { "epoch": 41.33843212237094, "grad_norm": 0.592457115650177, "learning_rate": 9.263661521118442e-05, "loss": 0.0203, "step": 64860 }, { "epoch": 41.34480560866794, "grad_norm": 0.7723691463470459, "learning_rate": 9.263373599567298e-05, "loss": 0.0218, "step": 64870 }, { "epoch": 41.35117909496495, "grad_norm": 0.6781921982765198, "learning_rate": 9.26308562621214e-05, "loss": 0.0384, "step": 64880 }, { "epoch": 41.35755258126195, "grad_norm": 0.3004898130893707, "learning_rate": 9.262797601056468e-05, "loss": 0.0176, "step": 64890 }, { "epoch": 41.36392606755896, "grad_norm": 0.902631402015686, "learning_rate": 9.262509524103778e-05, "loss": 0.0276, "step": 64900 }, { "epoch": 41.37029955385596, "grad_norm": 0.5635977387428284, "learning_rate": 9.262221395357574e-05, "loss": 0.0182, "step": 64910 }, { "epoch": 41.376673040152966, "grad_norm": 0.7531383633613586, "learning_rate": 9.261933214821357e-05, "loss": 0.0336, "step": 64920 }, { "epoch": 41.38304652644997, "grad_norm": 0.32179540395736694, "learning_rate": 9.261644982498626e-05, "loss": 0.0317, "step": 64930 }, { "epoch": 41.389420012746974, "grad_norm": 0.3562071919441223, "learning_rate": 9.261356698392884e-05, "loss": 0.0225, "step": 64940 }, { "epoch": 41.395793499043975, "grad_norm": 0.5225604772567749, "learning_rate": 9.261068362507636e-05, "loss": 0.0227, "step": 64950 }, { "epoch": 41.40216698534098, "grad_norm": 0.5845931172370911, "learning_rate": 9.260779974846383e-05, "loss": 0.03, "step": 64960 }, { "epoch": 41.408540471637984, "grad_norm": 0.5382035374641418, "learning_rate": 9.260491535412631e-05, "loss": 0.0305, "step": 64970 }, { "epoch": 41.41491395793499, "grad_norm": 0.3906625807285309, "learning_rate": 9.260203044209885e-05, "loss": 0.0401, "step": 64980 }, { "epoch": 41.42128744423199, "grad_norm": 0.4308006763458252, "learning_rate": 9.259914501241652e-05, "loss": 0.0248, "step": 64990 }, { "epoch": 41.427660930529, "grad_norm": 0.5970350503921509, "learning_rate": 9.259625906511431e-05, "loss": 0.0293, "step": 65000 }, { "epoch": 41.434034416826, "grad_norm": 0.3551969528198242, "learning_rate": 9.259337260022736e-05, "loss": 0.0158, "step": 65010 }, { "epoch": 41.44040790312301, "grad_norm": 0.4995523691177368, "learning_rate": 9.259048561779071e-05, "loss": 0.0362, "step": 65020 }, { "epoch": 41.44678138942001, "grad_norm": 0.6094891428947449, "learning_rate": 9.258759811783946e-05, "loss": 0.0245, "step": 65030 }, { "epoch": 41.45315487571702, "grad_norm": 0.5610607862472534, "learning_rate": 9.258471010040868e-05, "loss": 0.0286, "step": 65040 }, { "epoch": 41.45952836201402, "grad_norm": 0.6460846066474915, "learning_rate": 9.258182156553346e-05, "loss": 0.0372, "step": 65050 }, { "epoch": 41.46590184831103, "grad_norm": 0.46597814559936523, "learning_rate": 9.257893251324892e-05, "loss": 0.0313, "step": 65060 }, { "epoch": 41.47227533460803, "grad_norm": 0.5409643650054932, "learning_rate": 9.257604294359012e-05, "loss": 0.0208, "step": 65070 }, { "epoch": 41.478648820905036, "grad_norm": 0.7999165058135986, "learning_rate": 9.257315285659222e-05, "loss": 0.0296, "step": 65080 }, { "epoch": 41.48502230720204, "grad_norm": 0.9311299324035645, "learning_rate": 9.257026225229031e-05, "loss": 0.0169, "step": 65090 }, { "epoch": 41.491395793499045, "grad_norm": 0.5974019169807434, "learning_rate": 9.256737113071951e-05, "loss": 0.0233, "step": 65100 }, { "epoch": 41.497769279796046, "grad_norm": 0.5159413814544678, "learning_rate": 9.256447949191498e-05, "loss": 0.0229, "step": 65110 }, { "epoch": 41.504142766093054, "grad_norm": 0.3097835183143616, "learning_rate": 9.256158733591184e-05, "loss": 0.0214, "step": 65120 }, { "epoch": 41.510516252390055, "grad_norm": 0.31960171461105347, "learning_rate": 9.255869466274522e-05, "loss": 0.0238, "step": 65130 }, { "epoch": 41.51688973868706, "grad_norm": 0.3472208082675934, "learning_rate": 9.255580147245027e-05, "loss": 0.0298, "step": 65140 }, { "epoch": 41.52326322498406, "grad_norm": 0.5966345071792603, "learning_rate": 9.255290776506215e-05, "loss": 0.0417, "step": 65150 }, { "epoch": 41.52963671128107, "grad_norm": 0.3870909810066223, "learning_rate": 9.255001354061604e-05, "loss": 0.0218, "step": 65160 }, { "epoch": 41.53601019757807, "grad_norm": 0.3882139027118683, "learning_rate": 9.25471187991471e-05, "loss": 0.0184, "step": 65170 }, { "epoch": 41.54238368387508, "grad_norm": 0.38385850191116333, "learning_rate": 9.254422354069048e-05, "loss": 0.0157, "step": 65180 }, { "epoch": 41.54875717017208, "grad_norm": 0.6244174242019653, "learning_rate": 9.254132776528137e-05, "loss": 0.0237, "step": 65190 }, { "epoch": 41.55513065646909, "grad_norm": 0.5376093983650208, "learning_rate": 9.253843147295497e-05, "loss": 0.0245, "step": 65200 }, { "epoch": 41.56150414276609, "grad_norm": 0.7808029651641846, "learning_rate": 9.253553466374646e-05, "loss": 0.0274, "step": 65210 }, { "epoch": 41.5678776290631, "grad_norm": 0.45590806007385254, "learning_rate": 9.253263733769105e-05, "loss": 0.0239, "step": 65220 }, { "epoch": 41.5742511153601, "grad_norm": 0.46953079104423523, "learning_rate": 9.252973949482393e-05, "loss": 0.0236, "step": 65230 }, { "epoch": 41.58062460165711, "grad_norm": 0.35137003660202026, "learning_rate": 9.252684113518033e-05, "loss": 0.0287, "step": 65240 }, { "epoch": 41.58699808795411, "grad_norm": 0.32690155506134033, "learning_rate": 9.252394225879545e-05, "loss": 0.0216, "step": 65250 }, { "epoch": 41.593371574251115, "grad_norm": 0.40403011441230774, "learning_rate": 9.252104286570453e-05, "loss": 0.0235, "step": 65260 }, { "epoch": 41.599745060548116, "grad_norm": 0.3692602217197418, "learning_rate": 9.251814295594279e-05, "loss": 0.0266, "step": 65270 }, { "epoch": 41.606118546845124, "grad_norm": 0.35846632719039917, "learning_rate": 9.251524252954547e-05, "loss": 0.0305, "step": 65280 }, { "epoch": 41.61249203314213, "grad_norm": 0.4267905652523041, "learning_rate": 9.25123415865478e-05, "loss": 0.0196, "step": 65290 }, { "epoch": 41.61886551943913, "grad_norm": 0.4155419170856476, "learning_rate": 9.250944012698506e-05, "loss": 0.0282, "step": 65300 }, { "epoch": 41.62523900573614, "grad_norm": 0.6177881956100464, "learning_rate": 9.250653815089247e-05, "loss": 0.0211, "step": 65310 }, { "epoch": 41.63161249203314, "grad_norm": 0.6805815696716309, "learning_rate": 9.250363565830532e-05, "loss": 0.0293, "step": 65320 }, { "epoch": 41.63798597833015, "grad_norm": 0.2599300444126129, "learning_rate": 9.250073264925887e-05, "loss": 0.0296, "step": 65330 }, { "epoch": 41.64435946462715, "grad_norm": 0.6139668822288513, "learning_rate": 9.249782912378838e-05, "loss": 0.03, "step": 65340 }, { "epoch": 41.65073295092416, "grad_norm": 0.4658694863319397, "learning_rate": 9.249492508192914e-05, "loss": 0.0316, "step": 65350 }, { "epoch": 41.65710643722116, "grad_norm": 0.3623603582382202, "learning_rate": 9.249202052371646e-05, "loss": 0.0196, "step": 65360 }, { "epoch": 41.66347992351817, "grad_norm": 0.5396273136138916, "learning_rate": 9.248911544918559e-05, "loss": 0.0211, "step": 65370 }, { "epoch": 41.66985340981517, "grad_norm": 0.37033963203430176, "learning_rate": 9.248620985837186e-05, "loss": 0.017, "step": 65380 }, { "epoch": 41.676226896112176, "grad_norm": 0.6679165959358215, "learning_rate": 9.248330375131055e-05, "loss": 0.0181, "step": 65390 }, { "epoch": 41.68260038240918, "grad_norm": 0.26984667778015137, "learning_rate": 9.248039712803701e-05, "loss": 0.0303, "step": 65400 }, { "epoch": 41.688973868706185, "grad_norm": 0.7891584634780884, "learning_rate": 9.247748998858653e-05, "loss": 0.0221, "step": 65410 }, { "epoch": 41.695347355003186, "grad_norm": 0.5628881454467773, "learning_rate": 9.247458233299444e-05, "loss": 0.0239, "step": 65420 }, { "epoch": 41.701720841300194, "grad_norm": 0.3488200008869171, "learning_rate": 9.247167416129607e-05, "loss": 0.0327, "step": 65430 }, { "epoch": 41.708094327597195, "grad_norm": 0.6590137481689453, "learning_rate": 9.246876547352676e-05, "loss": 0.027, "step": 65440 }, { "epoch": 41.7144678138942, "grad_norm": 0.6593069434165955, "learning_rate": 9.246585626972186e-05, "loss": 0.027, "step": 65450 }, { "epoch": 41.7208413001912, "grad_norm": 0.4028383195400238, "learning_rate": 9.24629465499167e-05, "loss": 0.0227, "step": 65460 }, { "epoch": 41.72721478648821, "grad_norm": 0.39276933670043945, "learning_rate": 9.246003631414666e-05, "loss": 0.0203, "step": 65470 }, { "epoch": 41.73358827278521, "grad_norm": 0.7703338265419006, "learning_rate": 9.245712556244708e-05, "loss": 0.0277, "step": 65480 }, { "epoch": 41.73996175908222, "grad_norm": 0.5041046738624573, "learning_rate": 9.245421429485335e-05, "loss": 0.0189, "step": 65490 }, { "epoch": 41.74633524537922, "grad_norm": 0.6143819093704224, "learning_rate": 9.245130251140082e-05, "loss": 0.0292, "step": 65500 }, { "epoch": 41.75270873167623, "grad_norm": 0.6391715407371521, "learning_rate": 9.244839021212488e-05, "loss": 0.023, "step": 65510 }, { "epoch": 41.75908221797323, "grad_norm": 0.32556939125061035, "learning_rate": 9.244547739706093e-05, "loss": 0.0311, "step": 65520 }, { "epoch": 41.76545570427024, "grad_norm": 0.3450542092323303, "learning_rate": 9.244256406624436e-05, "loss": 0.042, "step": 65530 }, { "epoch": 41.77182919056724, "grad_norm": 0.8907830119132996, "learning_rate": 9.243965021971057e-05, "loss": 0.0268, "step": 65540 }, { "epoch": 41.77820267686425, "grad_norm": 0.33248305320739746, "learning_rate": 9.243673585749494e-05, "loss": 0.0231, "step": 65550 }, { "epoch": 41.78457616316125, "grad_norm": 0.3961670994758606, "learning_rate": 9.243382097963291e-05, "loss": 0.0179, "step": 65560 }, { "epoch": 41.790949649458256, "grad_norm": 0.17833109200000763, "learning_rate": 9.243090558615989e-05, "loss": 0.0283, "step": 65570 }, { "epoch": 41.797323135755256, "grad_norm": 0.35153061151504517, "learning_rate": 9.24279896771113e-05, "loss": 0.0232, "step": 65580 }, { "epoch": 41.803696622052264, "grad_norm": 0.3453674614429474, "learning_rate": 9.242507325252259e-05, "loss": 0.0363, "step": 65590 }, { "epoch": 41.810070108349265, "grad_norm": 0.5939841866493225, "learning_rate": 9.242215631242919e-05, "loss": 0.0253, "step": 65600 }, { "epoch": 41.81644359464627, "grad_norm": 0.3945414423942566, "learning_rate": 9.241923885686652e-05, "loss": 0.0222, "step": 65610 }, { "epoch": 41.822817080943274, "grad_norm": 0.6010554432868958, "learning_rate": 9.241632088587004e-05, "loss": 0.0347, "step": 65620 }, { "epoch": 41.82919056724028, "grad_norm": 0.5998820662498474, "learning_rate": 9.241340239947523e-05, "loss": 0.0249, "step": 65630 }, { "epoch": 41.83556405353728, "grad_norm": 0.4723265469074249, "learning_rate": 9.241048339771754e-05, "loss": 0.047, "step": 65640 }, { "epoch": 41.84193753983429, "grad_norm": 0.5665018558502197, "learning_rate": 9.240756388063242e-05, "loss": 0.0292, "step": 65650 }, { "epoch": 41.84831102613129, "grad_norm": 0.8665750622749329, "learning_rate": 9.240464384825538e-05, "loss": 0.0233, "step": 65660 }, { "epoch": 41.8546845124283, "grad_norm": 0.642570972442627, "learning_rate": 9.240172330062186e-05, "loss": 0.0242, "step": 65670 }, { "epoch": 41.8610579987253, "grad_norm": 0.43970707058906555, "learning_rate": 9.239880223776738e-05, "loss": 0.0239, "step": 65680 }, { "epoch": 41.86743148502231, "grad_norm": 0.6904773116111755, "learning_rate": 9.239588065972742e-05, "loss": 0.0288, "step": 65690 }, { "epoch": 41.87380497131931, "grad_norm": 0.3054540455341339, "learning_rate": 9.239295856653747e-05, "loss": 0.0205, "step": 65700 }, { "epoch": 41.88017845761632, "grad_norm": 0.45055291056632996, "learning_rate": 9.239003595823306e-05, "loss": 0.0181, "step": 65710 }, { "epoch": 41.88655194391332, "grad_norm": 1.852647304534912, "learning_rate": 9.238711283484968e-05, "loss": 0.0327, "step": 65720 }, { "epoch": 41.892925430210326, "grad_norm": 0.48864656686782837, "learning_rate": 9.238418919642287e-05, "loss": 0.0176, "step": 65730 }, { "epoch": 41.89929891650733, "grad_norm": 0.3609170913696289, "learning_rate": 9.238126504298814e-05, "loss": 0.0157, "step": 65740 }, { "epoch": 41.905672402804335, "grad_norm": 0.6786608099937439, "learning_rate": 9.237834037458102e-05, "loss": 0.0294, "step": 65750 }, { "epoch": 41.912045889101336, "grad_norm": 0.4320574104785919, "learning_rate": 9.237541519123707e-05, "loss": 0.0246, "step": 65760 }, { "epoch": 41.918419375398344, "grad_norm": 0.4893729090690613, "learning_rate": 9.23724894929918e-05, "loss": 0.0262, "step": 65770 }, { "epoch": 41.924792861695344, "grad_norm": 0.5465174913406372, "learning_rate": 9.236956327988078e-05, "loss": 0.0182, "step": 65780 }, { "epoch": 41.93116634799235, "grad_norm": 0.3505243957042694, "learning_rate": 9.236663655193955e-05, "loss": 0.0195, "step": 65790 }, { "epoch": 41.93753983428935, "grad_norm": 1.6846966743469238, "learning_rate": 9.236370930920371e-05, "loss": 0.033, "step": 65800 }, { "epoch": 41.94391332058636, "grad_norm": 0.51729816198349, "learning_rate": 9.23607815517088e-05, "loss": 0.0147, "step": 65810 }, { "epoch": 41.95028680688336, "grad_norm": 0.4280257225036621, "learning_rate": 9.235785327949038e-05, "loss": 0.0203, "step": 65820 }, { "epoch": 41.95666029318037, "grad_norm": 0.2931711673736572, "learning_rate": 9.235492449258407e-05, "loss": 0.0199, "step": 65830 }, { "epoch": 41.96303377947737, "grad_norm": 0.32455819845199585, "learning_rate": 9.235199519102542e-05, "loss": 0.0178, "step": 65840 }, { "epoch": 41.96940726577438, "grad_norm": 0.44798725843429565, "learning_rate": 9.234906537485006e-05, "loss": 0.0244, "step": 65850 }, { "epoch": 41.97578075207138, "grad_norm": 0.5284509062767029, "learning_rate": 9.234613504409356e-05, "loss": 0.0328, "step": 65860 }, { "epoch": 41.98215423836839, "grad_norm": 0.3808888792991638, "learning_rate": 9.234320419879154e-05, "loss": 0.0226, "step": 65870 }, { "epoch": 41.98852772466539, "grad_norm": 0.5114511847496033, "learning_rate": 9.234027283897961e-05, "loss": 0.0318, "step": 65880 }, { "epoch": 41.9949012109624, "grad_norm": 0.433860719203949, "learning_rate": 9.233734096469339e-05, "loss": 0.0238, "step": 65890 }, { "epoch": 42.001274697259404, "grad_norm": 0.49082469940185547, "learning_rate": 9.23344085759685e-05, "loss": 0.0243, "step": 65900 }, { "epoch": 42.007648183556405, "grad_norm": 0.6280624270439148, "learning_rate": 9.233147567284057e-05, "loss": 0.0252, "step": 65910 }, { "epoch": 42.01402166985341, "grad_norm": 0.47607970237731934, "learning_rate": 9.232854225534527e-05, "loss": 0.0285, "step": 65920 }, { "epoch": 42.020395156150414, "grad_norm": 0.35698407888412476, "learning_rate": 9.23256083235182e-05, "loss": 0.0231, "step": 65930 }, { "epoch": 42.02676864244742, "grad_norm": 0.4576590061187744, "learning_rate": 9.232267387739502e-05, "loss": 0.0315, "step": 65940 }, { "epoch": 42.03314212874442, "grad_norm": 0.3582839071750641, "learning_rate": 9.23197389170114e-05, "loss": 0.0185, "step": 65950 }, { "epoch": 42.03951561504143, "grad_norm": 0.4310932159423828, "learning_rate": 9.231680344240298e-05, "loss": 0.0345, "step": 65960 }, { "epoch": 42.04588910133843, "grad_norm": 0.5213659405708313, "learning_rate": 9.231386745360546e-05, "loss": 0.0269, "step": 65970 }, { "epoch": 42.05226258763544, "grad_norm": 0.414416640996933, "learning_rate": 9.231093095065451e-05, "loss": 0.029, "step": 65980 }, { "epoch": 42.05863607393244, "grad_norm": 0.4381742477416992, "learning_rate": 9.23079939335858e-05, "loss": 0.0244, "step": 65990 }, { "epoch": 42.06500956022945, "grad_norm": 0.5713891983032227, "learning_rate": 9.2305056402435e-05, "loss": 0.0272, "step": 66000 }, { "epoch": 42.07138304652645, "grad_norm": 0.6084326505661011, "learning_rate": 9.230211835723783e-05, "loss": 0.0204, "step": 66010 }, { "epoch": 42.07775653282346, "grad_norm": 0.6460614800453186, "learning_rate": 9.229917979802997e-05, "loss": 0.025, "step": 66020 }, { "epoch": 42.08413001912046, "grad_norm": 0.31622523069381714, "learning_rate": 9.229624072484716e-05, "loss": 0.022, "step": 66030 }, { "epoch": 42.090503505417466, "grad_norm": 0.355973482131958, "learning_rate": 9.229330113772507e-05, "loss": 0.0181, "step": 66040 }, { "epoch": 42.09687699171447, "grad_norm": 0.4074205458164215, "learning_rate": 9.229036103669946e-05, "loss": 0.0253, "step": 66050 }, { "epoch": 42.103250478011475, "grad_norm": 0.5366103053092957, "learning_rate": 9.228742042180603e-05, "loss": 0.0278, "step": 66060 }, { "epoch": 42.109623964308476, "grad_norm": 0.39441221952438354, "learning_rate": 9.228447929308051e-05, "loss": 0.0268, "step": 66070 }, { "epoch": 42.115997450605484, "grad_norm": 0.5194296836853027, "learning_rate": 9.228153765055864e-05, "loss": 0.0229, "step": 66080 }, { "epoch": 42.122370936902485, "grad_norm": 0.43909651041030884, "learning_rate": 9.227859549427618e-05, "loss": 0.0237, "step": 66090 }, { "epoch": 42.12874442319949, "grad_norm": 0.40062376856803894, "learning_rate": 9.227565282426887e-05, "loss": 0.0266, "step": 66100 }, { "epoch": 42.13511790949649, "grad_norm": 0.5049161314964294, "learning_rate": 9.227270964057245e-05, "loss": 0.0309, "step": 66110 }, { "epoch": 42.1414913957935, "grad_norm": 0.29995375871658325, "learning_rate": 9.22697659432227e-05, "loss": 0.0167, "step": 66120 }, { "epoch": 42.1478648820905, "grad_norm": 0.5678418874740601, "learning_rate": 9.226682173225537e-05, "loss": 0.0207, "step": 66130 }, { "epoch": 42.15423836838751, "grad_norm": 0.3566882312297821, "learning_rate": 9.226387700770626e-05, "loss": 0.0143, "step": 66140 }, { "epoch": 42.16061185468451, "grad_norm": 0.5927200317382812, "learning_rate": 9.226093176961115e-05, "loss": 0.0182, "step": 66150 }, { "epoch": 42.16698534098152, "grad_norm": 0.33880048990249634, "learning_rate": 9.225798601800582e-05, "loss": 0.0249, "step": 66160 }, { "epoch": 42.17335882727852, "grad_norm": 0.44698598980903625, "learning_rate": 9.225503975292604e-05, "loss": 0.0232, "step": 66170 }, { "epoch": 42.17973231357553, "grad_norm": 0.4112945795059204, "learning_rate": 9.225209297440764e-05, "loss": 0.0248, "step": 66180 }, { "epoch": 42.18610579987253, "grad_norm": 0.46955057978630066, "learning_rate": 9.224914568248644e-05, "loss": 0.0387, "step": 66190 }, { "epoch": 42.19247928616954, "grad_norm": 0.6163523197174072, "learning_rate": 9.22461978771982e-05, "loss": 0.0165, "step": 66200 }, { "epoch": 42.19885277246654, "grad_norm": 0.40762218832969666, "learning_rate": 9.224324955857879e-05, "loss": 0.0206, "step": 66210 }, { "epoch": 42.205226258763545, "grad_norm": 0.2340753972530365, "learning_rate": 9.224030072666401e-05, "loss": 0.018, "step": 66220 }, { "epoch": 42.211599745060546, "grad_norm": 0.25114962458610535, "learning_rate": 9.223735138148968e-05, "loss": 0.0246, "step": 66230 }, { "epoch": 42.217973231357554, "grad_norm": 0.4387468993663788, "learning_rate": 9.223440152309166e-05, "loss": 0.0235, "step": 66240 }, { "epoch": 42.224346717654555, "grad_norm": 0.37101712822914124, "learning_rate": 9.22314511515058e-05, "loss": 0.021, "step": 66250 }, { "epoch": 42.23072020395156, "grad_norm": 0.3785618543624878, "learning_rate": 9.222850026676791e-05, "loss": 0.0275, "step": 66260 }, { "epoch": 42.237093690248564, "grad_norm": 0.4702027440071106, "learning_rate": 9.222554886891389e-05, "loss": 0.0241, "step": 66270 }, { "epoch": 42.24346717654557, "grad_norm": 0.33600151538848877, "learning_rate": 9.222259695797958e-05, "loss": 0.0224, "step": 66280 }, { "epoch": 42.24984066284257, "grad_norm": 0.47428736090660095, "learning_rate": 9.221964453400084e-05, "loss": 0.0267, "step": 66290 }, { "epoch": 42.25621414913958, "grad_norm": 0.660992443561554, "learning_rate": 9.221669159701358e-05, "loss": 0.0322, "step": 66300 }, { "epoch": 42.26258763543658, "grad_norm": 0.336344450712204, "learning_rate": 9.221373814705365e-05, "loss": 0.0254, "step": 66310 }, { "epoch": 42.26896112173359, "grad_norm": 0.47817614674568176, "learning_rate": 9.221078418415693e-05, "loss": 0.0269, "step": 66320 }, { "epoch": 42.27533460803059, "grad_norm": 0.6063815355300903, "learning_rate": 9.220782970835934e-05, "loss": 0.0244, "step": 66330 }, { "epoch": 42.2817080943276, "grad_norm": 0.44965827465057373, "learning_rate": 9.220487471969677e-05, "loss": 0.0296, "step": 66340 }, { "epoch": 42.2880815806246, "grad_norm": 0.455697238445282, "learning_rate": 9.220191921820511e-05, "loss": 0.0261, "step": 66350 }, { "epoch": 42.29445506692161, "grad_norm": 0.4224787652492523, "learning_rate": 9.219896320392031e-05, "loss": 0.0188, "step": 66360 }, { "epoch": 42.30082855321861, "grad_norm": 0.42280644178390503, "learning_rate": 9.219600667687824e-05, "loss": 0.0191, "step": 66370 }, { "epoch": 42.307202039515616, "grad_norm": 0.6243449449539185, "learning_rate": 9.219304963711487e-05, "loss": 0.0292, "step": 66380 }, { "epoch": 42.31357552581262, "grad_norm": 0.5976722836494446, "learning_rate": 9.21900920846661e-05, "loss": 0.0279, "step": 66390 }, { "epoch": 42.319949012109625, "grad_norm": 0.6923816204071045, "learning_rate": 9.218713401956788e-05, "loss": 0.0312, "step": 66400 }, { "epoch": 42.326322498406626, "grad_norm": 0.37685689330101013, "learning_rate": 9.218417544185615e-05, "loss": 0.0238, "step": 66410 }, { "epoch": 42.33269598470363, "grad_norm": 0.31143316626548767, "learning_rate": 9.218121635156686e-05, "loss": 0.021, "step": 66420 }, { "epoch": 42.339069471000634, "grad_norm": 0.7185904383659363, "learning_rate": 9.217825674873597e-05, "loss": 0.0335, "step": 66430 }, { "epoch": 42.34544295729764, "grad_norm": 0.30323293805122375, "learning_rate": 9.217529663339942e-05, "loss": 0.0205, "step": 66440 }, { "epoch": 42.35181644359464, "grad_norm": 0.4969540238380432, "learning_rate": 9.21723360055932e-05, "loss": 0.0245, "step": 66450 }, { "epoch": 42.35818992989165, "grad_norm": 0.48340144753456116, "learning_rate": 9.216937486535331e-05, "loss": 0.0251, "step": 66460 }, { "epoch": 42.36456341618865, "grad_norm": 0.3917814791202545, "learning_rate": 9.216641321271567e-05, "loss": 0.0205, "step": 66470 }, { "epoch": 42.37093690248566, "grad_norm": 0.6298745274543762, "learning_rate": 9.216345104771631e-05, "loss": 0.0307, "step": 66480 }, { "epoch": 42.37731038878266, "grad_norm": 0.4075070321559906, "learning_rate": 9.21604883703912e-05, "loss": 0.0218, "step": 66490 }, { "epoch": 42.38368387507967, "grad_norm": 0.44286903738975525, "learning_rate": 9.215752518077635e-05, "loss": 0.0298, "step": 66500 }, { "epoch": 42.39005736137667, "grad_norm": 0.4755772352218628, "learning_rate": 9.215456147890779e-05, "loss": 0.0311, "step": 66510 }, { "epoch": 42.39643084767368, "grad_norm": 0.5091001391410828, "learning_rate": 9.215159726482148e-05, "loss": 0.0169, "step": 66520 }, { "epoch": 42.40280433397068, "grad_norm": 0.4735628068447113, "learning_rate": 9.214863253855348e-05, "loss": 0.0214, "step": 66530 }, { "epoch": 42.409177820267686, "grad_norm": 0.42597267031669617, "learning_rate": 9.214566730013981e-05, "loss": 0.0291, "step": 66540 }, { "epoch": 42.415551306564694, "grad_norm": 0.8328556418418884, "learning_rate": 9.214270154961647e-05, "loss": 0.0214, "step": 66550 }, { "epoch": 42.421924792861695, "grad_norm": 0.4224740266799927, "learning_rate": 9.213973528701952e-05, "loss": 0.0208, "step": 66560 }, { "epoch": 42.4282982791587, "grad_norm": 0.5940009355545044, "learning_rate": 9.213676851238501e-05, "loss": 0.0201, "step": 66570 }, { "epoch": 42.434671765455704, "grad_norm": 0.3207134008407593, "learning_rate": 9.213380122574898e-05, "loss": 0.0259, "step": 66580 }, { "epoch": 42.44104525175271, "grad_norm": 0.1958329826593399, "learning_rate": 9.213083342714747e-05, "loss": 0.025, "step": 66590 }, { "epoch": 42.44741873804971, "grad_norm": 0.668378472328186, "learning_rate": 9.212786511661656e-05, "loss": 0.0269, "step": 66600 }, { "epoch": 42.45379222434672, "grad_norm": 0.4601955711841583, "learning_rate": 9.212489629419232e-05, "loss": 0.0323, "step": 66610 }, { "epoch": 42.46016571064372, "grad_norm": 0.3183055520057678, "learning_rate": 9.212192695991082e-05, "loss": 0.028, "step": 66620 }, { "epoch": 42.46653919694073, "grad_norm": 0.23550714552402496, "learning_rate": 9.211895711380812e-05, "loss": 0.0236, "step": 66630 }, { "epoch": 42.47291268323773, "grad_norm": 0.6197898983955383, "learning_rate": 9.211598675592034e-05, "loss": 0.0316, "step": 66640 }, { "epoch": 42.47928616953474, "grad_norm": 0.3372287154197693, "learning_rate": 9.211301588628354e-05, "loss": 0.016, "step": 66650 }, { "epoch": 42.48565965583174, "grad_norm": 0.438781201839447, "learning_rate": 9.211004450493386e-05, "loss": 0.0186, "step": 66660 }, { "epoch": 42.49203314212875, "grad_norm": 0.5597378611564636, "learning_rate": 9.210707261190736e-05, "loss": 0.0255, "step": 66670 }, { "epoch": 42.49840662842575, "grad_norm": 0.4239688515663147, "learning_rate": 9.210410020724018e-05, "loss": 0.0225, "step": 66680 }, { "epoch": 42.504780114722756, "grad_norm": 0.3857112228870392, "learning_rate": 9.210112729096843e-05, "loss": 0.0245, "step": 66690 }, { "epoch": 42.51115360101976, "grad_norm": 0.5746787190437317, "learning_rate": 9.209815386312824e-05, "loss": 0.0335, "step": 66700 }, { "epoch": 42.517527087316765, "grad_norm": 0.5966720581054688, "learning_rate": 9.209517992375572e-05, "loss": 0.0208, "step": 66710 }, { "epoch": 42.523900573613766, "grad_norm": 0.5027617812156677, "learning_rate": 9.209220547288704e-05, "loss": 0.0249, "step": 66720 }, { "epoch": 42.530274059910774, "grad_norm": 0.8104925751686096, "learning_rate": 9.20892305105583e-05, "loss": 0.0224, "step": 66730 }, { "epoch": 42.536647546207774, "grad_norm": 1.0852450132369995, "learning_rate": 9.208625503680568e-05, "loss": 0.0225, "step": 66740 }, { "epoch": 42.54302103250478, "grad_norm": 0.3320201635360718, "learning_rate": 9.208327905166533e-05, "loss": 0.0253, "step": 66750 }, { "epoch": 42.54939451880178, "grad_norm": 0.3694443702697754, "learning_rate": 9.208030255517339e-05, "loss": 0.0309, "step": 66760 }, { "epoch": 42.55576800509879, "grad_norm": 0.37521687150001526, "learning_rate": 9.207732554736606e-05, "loss": 0.024, "step": 66770 }, { "epoch": 42.56214149139579, "grad_norm": 0.4098055958747864, "learning_rate": 9.20743480282795e-05, "loss": 0.0301, "step": 66780 }, { "epoch": 42.5685149776928, "grad_norm": 0.7972168922424316, "learning_rate": 9.207136999794987e-05, "loss": 0.0252, "step": 66790 }, { "epoch": 42.5748884639898, "grad_norm": 0.6641728281974792, "learning_rate": 9.206839145641339e-05, "loss": 0.0264, "step": 66800 }, { "epoch": 42.58126195028681, "grad_norm": 0.35140904784202576, "learning_rate": 9.20654124037062e-05, "loss": 0.0365, "step": 66810 }, { "epoch": 42.58763543658381, "grad_norm": 0.49575087428092957, "learning_rate": 9.206243283986458e-05, "loss": 0.0422, "step": 66820 }, { "epoch": 42.59400892288082, "grad_norm": 0.9227663278579712, "learning_rate": 9.205945276492465e-05, "loss": 0.0269, "step": 66830 }, { "epoch": 42.60038240917782, "grad_norm": 0.42800629138946533, "learning_rate": 9.205647217892268e-05, "loss": 0.0233, "step": 66840 }, { "epoch": 42.606755895474826, "grad_norm": 0.305634468793869, "learning_rate": 9.205349108189486e-05, "loss": 0.0224, "step": 66850 }, { "epoch": 42.61312938177183, "grad_norm": 0.5572590827941895, "learning_rate": 9.205050947387741e-05, "loss": 0.0168, "step": 66860 }, { "epoch": 42.619502868068835, "grad_norm": 0.788881242275238, "learning_rate": 9.204752735490658e-05, "loss": 0.0219, "step": 66870 }, { "epoch": 42.625876354365836, "grad_norm": 0.5477340221405029, "learning_rate": 9.204454472501859e-05, "loss": 0.022, "step": 66880 }, { "epoch": 42.632249840662844, "grad_norm": 0.4275846779346466, "learning_rate": 9.204156158424968e-05, "loss": 0.0431, "step": 66890 }, { "epoch": 42.638623326959845, "grad_norm": 0.2976972162723541, "learning_rate": 9.20385779326361e-05, "loss": 0.0174, "step": 66900 }, { "epoch": 42.64499681325685, "grad_norm": 0.31384608149528503, "learning_rate": 9.203559377021412e-05, "loss": 0.0383, "step": 66910 }, { "epoch": 42.651370299553854, "grad_norm": 0.5117910504341125, "learning_rate": 9.203260909701999e-05, "loss": 0.0197, "step": 66920 }, { "epoch": 42.65774378585086, "grad_norm": 0.42648959159851074, "learning_rate": 9.202962391308996e-05, "loss": 0.0194, "step": 66930 }, { "epoch": 42.66411727214786, "grad_norm": 0.5145031809806824, "learning_rate": 9.202663821846032e-05, "loss": 0.0227, "step": 66940 }, { "epoch": 42.67049075844487, "grad_norm": 0.5373769998550415, "learning_rate": 9.202365201316734e-05, "loss": 0.029, "step": 66950 }, { "epoch": 42.67686424474187, "grad_norm": 0.9795006513595581, "learning_rate": 9.202066529724734e-05, "loss": 0.026, "step": 66960 }, { "epoch": 42.68323773103888, "grad_norm": 0.3242267370223999, "learning_rate": 9.201767807073656e-05, "loss": 0.019, "step": 66970 }, { "epoch": 42.68961121733588, "grad_norm": 0.1857665479183197, "learning_rate": 9.201469033367132e-05, "loss": 0.0223, "step": 66980 }, { "epoch": 42.69598470363289, "grad_norm": 0.6057352423667908, "learning_rate": 9.201170208608791e-05, "loss": 0.0279, "step": 66990 }, { "epoch": 42.70235818992989, "grad_norm": 0.5301851630210876, "learning_rate": 9.200871332802267e-05, "loss": 0.0191, "step": 67000 }, { "epoch": 42.7087316762269, "grad_norm": 0.5958722233772278, "learning_rate": 9.200572405951192e-05, "loss": 0.0281, "step": 67010 }, { "epoch": 42.7151051625239, "grad_norm": 0.7547402381896973, "learning_rate": 9.200273428059193e-05, "loss": 0.0307, "step": 67020 }, { "epoch": 42.721478648820906, "grad_norm": 0.718107283115387, "learning_rate": 9.199974399129907e-05, "loss": 0.0269, "step": 67030 }, { "epoch": 42.72785213511791, "grad_norm": 0.6762949824333191, "learning_rate": 9.199675319166967e-05, "loss": 0.0312, "step": 67040 }, { "epoch": 42.734225621414915, "grad_norm": 0.4817274212837219, "learning_rate": 9.199376188174007e-05, "loss": 0.0238, "step": 67050 }, { "epoch": 42.740599107711915, "grad_norm": 0.5245202779769897, "learning_rate": 9.199077006154661e-05, "loss": 0.0272, "step": 67060 }, { "epoch": 42.74697259400892, "grad_norm": 0.9268798232078552, "learning_rate": 9.198777773112566e-05, "loss": 0.0399, "step": 67070 }, { "epoch": 42.753346080305924, "grad_norm": 0.5143446922302246, "learning_rate": 9.198478489051354e-05, "loss": 0.0299, "step": 67080 }, { "epoch": 42.75971956660293, "grad_norm": 0.525508463382721, "learning_rate": 9.198179153974667e-05, "loss": 0.0236, "step": 67090 }, { "epoch": 42.76609305289993, "grad_norm": 0.4677829444408417, "learning_rate": 9.197879767886138e-05, "loss": 0.0147, "step": 67100 }, { "epoch": 42.77246653919694, "grad_norm": 0.6096376180648804, "learning_rate": 9.197580330789407e-05, "loss": 0.0216, "step": 67110 }, { "epoch": 42.77884002549394, "grad_norm": 0.7969074249267578, "learning_rate": 9.197280842688112e-05, "loss": 0.0289, "step": 67120 }, { "epoch": 42.78521351179095, "grad_norm": 0.3099363446235657, "learning_rate": 9.196981303585892e-05, "loss": 0.0276, "step": 67130 }, { "epoch": 42.79158699808795, "grad_norm": 0.4763953685760498, "learning_rate": 9.196681713486386e-05, "loss": 0.032, "step": 67140 }, { "epoch": 42.79796048438496, "grad_norm": 0.21379682421684265, "learning_rate": 9.196382072393235e-05, "loss": 0.0199, "step": 67150 }, { "epoch": 42.80433397068197, "grad_norm": 0.3612785339355469, "learning_rate": 9.196082380310079e-05, "loss": 0.0315, "step": 67160 }, { "epoch": 42.81070745697897, "grad_norm": 0.5285971164703369, "learning_rate": 9.195782637240562e-05, "loss": 0.0249, "step": 67170 }, { "epoch": 42.817080943275975, "grad_norm": 0.5027537941932678, "learning_rate": 9.195482843188323e-05, "loss": 0.0213, "step": 67180 }, { "epoch": 42.823454429572976, "grad_norm": 0.43655627965927124, "learning_rate": 9.195182998157007e-05, "loss": 0.0236, "step": 67190 }, { "epoch": 42.829827915869984, "grad_norm": 0.28319913148880005, "learning_rate": 9.194883102150256e-05, "loss": 0.0371, "step": 67200 }, { "epoch": 42.836201402166985, "grad_norm": 0.25379395484924316, "learning_rate": 9.194583155171716e-05, "loss": 0.0218, "step": 67210 }, { "epoch": 42.84257488846399, "grad_norm": 0.36270880699157715, "learning_rate": 9.19428315722503e-05, "loss": 0.0281, "step": 67220 }, { "epoch": 42.848948374760994, "grad_norm": 0.4661734700202942, "learning_rate": 9.193983108313841e-05, "loss": 0.0264, "step": 67230 }, { "epoch": 42.855321861058, "grad_norm": 0.5283739566802979, "learning_rate": 9.193683008441801e-05, "loss": 0.0368, "step": 67240 }, { "epoch": 42.861695347355, "grad_norm": 0.3699325919151306, "learning_rate": 9.193382857612552e-05, "loss": 0.0211, "step": 67250 }, { "epoch": 42.86806883365201, "grad_norm": 0.6959346532821655, "learning_rate": 9.193082655829742e-05, "loss": 0.0295, "step": 67260 }, { "epoch": 42.87444231994901, "grad_norm": 0.6285856366157532, "learning_rate": 9.192782403097018e-05, "loss": 0.0304, "step": 67270 }, { "epoch": 42.88081580624602, "grad_norm": 0.271664023399353, "learning_rate": 9.19248209941803e-05, "loss": 0.0184, "step": 67280 }, { "epoch": 42.88718929254302, "grad_norm": 0.2527194321155548, "learning_rate": 9.192181744796425e-05, "loss": 0.0255, "step": 67290 }, { "epoch": 42.89356277884003, "grad_norm": 0.5781341195106506, "learning_rate": 9.191881339235855e-05, "loss": 0.026, "step": 67300 }, { "epoch": 42.89993626513703, "grad_norm": 0.47590798139572144, "learning_rate": 9.191580882739967e-05, "loss": 0.0165, "step": 67310 }, { "epoch": 42.90630975143404, "grad_norm": 0.7032646536827087, "learning_rate": 9.191280375312417e-05, "loss": 0.0314, "step": 67320 }, { "epoch": 42.91268323773104, "grad_norm": 0.49506446719169617, "learning_rate": 9.19097981695685e-05, "loss": 0.0203, "step": 67330 }, { "epoch": 42.919056724028046, "grad_norm": 0.5686636567115784, "learning_rate": 9.190679207676922e-05, "loss": 0.0346, "step": 67340 }, { "epoch": 42.92543021032505, "grad_norm": 0.5911718010902405, "learning_rate": 9.190378547476286e-05, "loss": 0.0241, "step": 67350 }, { "epoch": 42.931803696622055, "grad_norm": 0.42223984003067017, "learning_rate": 9.190077836358594e-05, "loss": 0.0204, "step": 67360 }, { "epoch": 42.938177182919055, "grad_norm": 0.49546900391578674, "learning_rate": 9.189777074327501e-05, "loss": 0.0173, "step": 67370 }, { "epoch": 42.94455066921606, "grad_norm": 0.32483208179473877, "learning_rate": 9.189476261386659e-05, "loss": 0.0245, "step": 67380 }, { "epoch": 42.950924155513064, "grad_norm": 0.3610183894634247, "learning_rate": 9.189175397539727e-05, "loss": 0.0249, "step": 67390 }, { "epoch": 42.95729764181007, "grad_norm": 0.6013220548629761, "learning_rate": 9.188874482790358e-05, "loss": 0.0307, "step": 67400 }, { "epoch": 42.96367112810707, "grad_norm": 0.4735114872455597, "learning_rate": 9.188573517142208e-05, "loss": 0.0198, "step": 67410 }, { "epoch": 42.97004461440408, "grad_norm": 0.5498377680778503, "learning_rate": 9.188272500598935e-05, "loss": 0.0296, "step": 67420 }, { "epoch": 42.97641810070108, "grad_norm": 0.34324944019317627, "learning_rate": 9.187971433164198e-05, "loss": 0.0257, "step": 67430 }, { "epoch": 42.98279158699809, "grad_norm": 0.40846747159957886, "learning_rate": 9.187670314841653e-05, "loss": 0.0207, "step": 67440 }, { "epoch": 42.98916507329509, "grad_norm": 0.26030123233795166, "learning_rate": 9.187369145634958e-05, "loss": 0.0296, "step": 67450 }, { "epoch": 42.9955385595921, "grad_norm": 0.6081669330596924, "learning_rate": 9.187067925547779e-05, "loss": 0.0225, "step": 67460 }, { "epoch": 43.0019120458891, "grad_norm": 0.400501549243927, "learning_rate": 9.186766654583767e-05, "loss": 0.0234, "step": 67470 }, { "epoch": 43.00828553218611, "grad_norm": 0.3854467570781708, "learning_rate": 9.18646533274659e-05, "loss": 0.0209, "step": 67480 }, { "epoch": 43.01465901848311, "grad_norm": 0.727696418762207, "learning_rate": 9.186163960039906e-05, "loss": 0.0284, "step": 67490 }, { "epoch": 43.021032504780116, "grad_norm": 0.5858085751533508, "learning_rate": 9.185862536467377e-05, "loss": 0.0314, "step": 67500 }, { "epoch": 43.02740599107712, "grad_norm": 0.3960177004337311, "learning_rate": 9.185561062032666e-05, "loss": 0.0246, "step": 67510 }, { "epoch": 43.033779477374125, "grad_norm": 0.4704405963420868, "learning_rate": 9.185259536739437e-05, "loss": 0.0279, "step": 67520 }, { "epoch": 43.040152963671126, "grad_norm": 0.3718126714229584, "learning_rate": 9.184957960591351e-05, "loss": 0.0202, "step": 67530 }, { "epoch": 43.046526449968134, "grad_norm": 0.6491569876670837, "learning_rate": 9.184656333592076e-05, "loss": 0.0261, "step": 67540 }, { "epoch": 43.052899936265135, "grad_norm": 0.38035815954208374, "learning_rate": 9.184354655745277e-05, "loss": 0.0288, "step": 67550 }, { "epoch": 43.05927342256214, "grad_norm": 0.5793561935424805, "learning_rate": 9.184052927054616e-05, "loss": 0.0369, "step": 67560 }, { "epoch": 43.065646908859144, "grad_norm": 0.45132583379745483, "learning_rate": 9.183751147523762e-05, "loss": 0.0255, "step": 67570 }, { "epoch": 43.07202039515615, "grad_norm": 0.3775477409362793, "learning_rate": 9.183449317156382e-05, "loss": 0.0342, "step": 67580 }, { "epoch": 43.07839388145315, "grad_norm": 0.4891328513622284, "learning_rate": 9.183147435956145e-05, "loss": 0.0398, "step": 67590 }, { "epoch": 43.08476736775016, "grad_norm": 0.4321242570877075, "learning_rate": 9.182845503926713e-05, "loss": 0.0364, "step": 67600 }, { "epoch": 43.09114085404716, "grad_norm": 0.6292339563369751, "learning_rate": 9.182543521071763e-05, "loss": 0.0239, "step": 67610 }, { "epoch": 43.09751434034417, "grad_norm": 0.5853735208511353, "learning_rate": 9.182241487394958e-05, "loss": 0.0439, "step": 67620 }, { "epoch": 43.10388782664117, "grad_norm": 0.5192533731460571, "learning_rate": 9.181939402899969e-05, "loss": 0.0251, "step": 67630 }, { "epoch": 43.11026131293818, "grad_norm": 0.5969426035881042, "learning_rate": 9.18163726759047e-05, "loss": 0.0361, "step": 67640 }, { "epoch": 43.11663479923518, "grad_norm": 0.7025969624519348, "learning_rate": 9.181335081470129e-05, "loss": 0.0251, "step": 67650 }, { "epoch": 43.12300828553219, "grad_norm": 0.7367541790008545, "learning_rate": 9.181032844542619e-05, "loss": 0.0193, "step": 67660 }, { "epoch": 43.12938177182919, "grad_norm": 0.45818081498146057, "learning_rate": 9.180730556811612e-05, "loss": 0.0181, "step": 67670 }, { "epoch": 43.135755258126196, "grad_norm": 0.45922696590423584, "learning_rate": 9.180428218280783e-05, "loss": 0.0232, "step": 67680 }, { "epoch": 43.1421287444232, "grad_norm": 0.4780043959617615, "learning_rate": 9.180125828953803e-05, "loss": 0.0361, "step": 67690 }, { "epoch": 43.148502230720204, "grad_norm": 0.37901586294174194, "learning_rate": 9.179823388834347e-05, "loss": 0.019, "step": 67700 }, { "epoch": 43.154875717017205, "grad_norm": 0.6662536859512329, "learning_rate": 9.179520897926092e-05, "loss": 0.0201, "step": 67710 }, { "epoch": 43.16124920331421, "grad_norm": 0.369272917509079, "learning_rate": 9.17921835623271e-05, "loss": 0.0204, "step": 67720 }, { "epoch": 43.167622689611214, "grad_norm": 0.6434751749038696, "learning_rate": 9.178915763757881e-05, "loss": 0.0224, "step": 67730 }, { "epoch": 43.17399617590822, "grad_norm": 0.6363507509231567, "learning_rate": 9.17861312050528e-05, "loss": 0.0288, "step": 67740 }, { "epoch": 43.18036966220522, "grad_norm": 0.5389083623886108, "learning_rate": 9.178310426478582e-05, "loss": 0.0196, "step": 67750 }, { "epoch": 43.18674314850223, "grad_norm": 0.915726363658905, "learning_rate": 9.17800768168147e-05, "loss": 0.0245, "step": 67760 }, { "epoch": 43.19311663479923, "grad_norm": 0.6710572242736816, "learning_rate": 9.177704886117619e-05, "loss": 0.0341, "step": 67770 }, { "epoch": 43.19949012109624, "grad_norm": 0.31386756896972656, "learning_rate": 9.177402039790708e-05, "loss": 0.023, "step": 67780 }, { "epoch": 43.20586360739325, "grad_norm": 0.4639987349510193, "learning_rate": 9.177099142704419e-05, "loss": 0.0142, "step": 67790 }, { "epoch": 43.21223709369025, "grad_norm": 0.1893891543149948, "learning_rate": 9.176796194862433e-05, "loss": 0.0177, "step": 67800 }, { "epoch": 43.218610579987256, "grad_norm": 0.5431473255157471, "learning_rate": 9.176493196268427e-05, "loss": 0.0326, "step": 67810 }, { "epoch": 43.22498406628426, "grad_norm": 0.5981893539428711, "learning_rate": 9.176190146926087e-05, "loss": 0.0259, "step": 67820 }, { "epoch": 43.231357552581265, "grad_norm": 0.32504671812057495, "learning_rate": 9.175887046839093e-05, "loss": 0.0204, "step": 67830 }, { "epoch": 43.237731038878266, "grad_norm": 0.4664250910282135, "learning_rate": 9.175583896011131e-05, "loss": 0.0392, "step": 67840 }, { "epoch": 43.244104525175274, "grad_norm": 0.2956750988960266, "learning_rate": 9.17528069444588e-05, "loss": 0.0244, "step": 67850 }, { "epoch": 43.250478011472275, "grad_norm": 0.5154459476470947, "learning_rate": 9.174977442147027e-05, "loss": 0.0211, "step": 67860 }, { "epoch": 43.25685149776928, "grad_norm": 0.3835418224334717, "learning_rate": 9.174674139118256e-05, "loss": 0.0376, "step": 67870 }, { "epoch": 43.263224984066284, "grad_norm": 0.260303258895874, "learning_rate": 9.174370785363253e-05, "loss": 0.0201, "step": 67880 }, { "epoch": 43.26959847036329, "grad_norm": 0.296442449092865, "learning_rate": 9.174067380885703e-05, "loss": 0.0206, "step": 67890 }, { "epoch": 43.27597195666029, "grad_norm": 0.8064464330673218, "learning_rate": 9.173763925689294e-05, "loss": 0.0193, "step": 67900 }, { "epoch": 43.2823454429573, "grad_norm": 0.5235943794250488, "learning_rate": 9.173460419777714e-05, "loss": 0.0209, "step": 67910 }, { "epoch": 43.2887189292543, "grad_norm": 0.5060768723487854, "learning_rate": 9.173156863154646e-05, "loss": 0.02, "step": 67920 }, { "epoch": 43.29509241555131, "grad_norm": 0.24220284819602966, "learning_rate": 9.172853255823785e-05, "loss": 0.0239, "step": 67930 }, { "epoch": 43.30146590184831, "grad_norm": 0.3179489076137543, "learning_rate": 9.172549597788815e-05, "loss": 0.0245, "step": 67940 }, { "epoch": 43.30783938814532, "grad_norm": 0.49112677574157715, "learning_rate": 9.17224588905343e-05, "loss": 0.0259, "step": 67950 }, { "epoch": 43.31421287444232, "grad_norm": 0.8284446597099304, "learning_rate": 9.171942129621317e-05, "loss": 0.0189, "step": 67960 }, { "epoch": 43.32058636073933, "grad_norm": 0.3642578721046448, "learning_rate": 9.171638319496167e-05, "loss": 0.0416, "step": 67970 }, { "epoch": 43.32695984703633, "grad_norm": 0.414824515581131, "learning_rate": 9.171334458681676e-05, "loss": 0.0264, "step": 67980 }, { "epoch": 43.333333333333336, "grad_norm": 0.45178139209747314, "learning_rate": 9.17103054718153e-05, "loss": 0.0254, "step": 67990 }, { "epoch": 43.33970681963034, "grad_norm": 0.4547591209411621, "learning_rate": 9.170726584999425e-05, "loss": 0.0248, "step": 68000 }, { "epoch": 43.346080305927345, "grad_norm": 0.6854702234268188, "learning_rate": 9.170422572139055e-05, "loss": 0.038, "step": 68010 }, { "epoch": 43.352453792224345, "grad_norm": 0.5485163927078247, "learning_rate": 9.170118508604115e-05, "loss": 0.0357, "step": 68020 }, { "epoch": 43.35882727852135, "grad_norm": 0.5728838443756104, "learning_rate": 9.169814394398296e-05, "loss": 0.0259, "step": 68030 }, { "epoch": 43.365200764818354, "grad_norm": 0.3205910623073578, "learning_rate": 9.169510229525294e-05, "loss": 0.0216, "step": 68040 }, { "epoch": 43.37157425111536, "grad_norm": 0.48292049765586853, "learning_rate": 9.169206013988807e-05, "loss": 0.0282, "step": 68050 }, { "epoch": 43.37794773741236, "grad_norm": 0.48375338315963745, "learning_rate": 9.168901747792531e-05, "loss": 0.0299, "step": 68060 }, { "epoch": 43.38432122370937, "grad_norm": 0.8962668776512146, "learning_rate": 9.168597430940162e-05, "loss": 0.0289, "step": 68070 }, { "epoch": 43.39069471000637, "grad_norm": 0.7977877855300903, "learning_rate": 9.1682930634354e-05, "loss": 0.0238, "step": 68080 }, { "epoch": 43.39706819630338, "grad_norm": 0.568190336227417, "learning_rate": 9.16798864528194e-05, "loss": 0.023, "step": 68090 }, { "epoch": 43.40344168260038, "grad_norm": 0.5441506505012512, "learning_rate": 9.167684176483483e-05, "loss": 0.0228, "step": 68100 }, { "epoch": 43.40981516889739, "grad_norm": 0.4669262766838074, "learning_rate": 9.167379657043728e-05, "loss": 0.0255, "step": 68110 }, { "epoch": 43.41618865519439, "grad_norm": 0.4530961811542511, "learning_rate": 9.167075086966376e-05, "loss": 0.0279, "step": 68120 }, { "epoch": 43.4225621414914, "grad_norm": 0.3778327405452728, "learning_rate": 9.166770466255128e-05, "loss": 0.0182, "step": 68130 }, { "epoch": 43.4289356277884, "grad_norm": 0.37270620465278625, "learning_rate": 9.166465794913685e-05, "loss": 0.0278, "step": 68140 }, { "epoch": 43.435309114085406, "grad_norm": 0.8757895231246948, "learning_rate": 9.166161072945747e-05, "loss": 0.0208, "step": 68150 }, { "epoch": 43.44168260038241, "grad_norm": 0.3630264699459076, "learning_rate": 9.16585630035502e-05, "loss": 0.0217, "step": 68160 }, { "epoch": 43.448056086679415, "grad_norm": 0.47920316457748413, "learning_rate": 9.165551477145204e-05, "loss": 0.0255, "step": 68170 }, { "epoch": 43.454429572976416, "grad_norm": 0.4455852508544922, "learning_rate": 9.165246603320006e-05, "loss": 0.0234, "step": 68180 }, { "epoch": 43.460803059273424, "grad_norm": 0.7605883479118347, "learning_rate": 9.16494167888313e-05, "loss": 0.0288, "step": 68190 }, { "epoch": 43.467176545570425, "grad_norm": 0.48727014660835266, "learning_rate": 9.164636703838278e-05, "loss": 0.0135, "step": 68200 }, { "epoch": 43.47355003186743, "grad_norm": 0.386100709438324, "learning_rate": 9.16433167818916e-05, "loss": 0.0285, "step": 68210 }, { "epoch": 43.47992351816443, "grad_norm": 0.2492709457874298, "learning_rate": 9.16402660193948e-05, "loss": 0.0171, "step": 68220 }, { "epoch": 43.48629700446144, "grad_norm": 0.5817500352859497, "learning_rate": 9.163721475092945e-05, "loss": 0.0254, "step": 68230 }, { "epoch": 43.49267049075844, "grad_norm": 0.44650399684906006, "learning_rate": 9.163416297653262e-05, "loss": 0.0227, "step": 68240 }, { "epoch": 43.49904397705545, "grad_norm": 0.6451097726821899, "learning_rate": 9.16311106962414e-05, "loss": 0.0243, "step": 68250 }, { "epoch": 43.50541746335245, "grad_norm": 0.34349969029426575, "learning_rate": 9.16280579100929e-05, "loss": 0.0177, "step": 68260 }, { "epoch": 43.51179094964946, "grad_norm": 0.14573559165000916, "learning_rate": 9.162500461812418e-05, "loss": 0.0234, "step": 68270 }, { "epoch": 43.51816443594646, "grad_norm": 0.3621068596839905, "learning_rate": 9.162195082037234e-05, "loss": 0.0243, "step": 68280 }, { "epoch": 43.52453792224347, "grad_norm": 0.4514976739883423, "learning_rate": 9.161889651687452e-05, "loss": 0.0361, "step": 68290 }, { "epoch": 43.53091140854047, "grad_norm": 0.42492860555648804, "learning_rate": 9.16158417076678e-05, "loss": 0.0426, "step": 68300 }, { "epoch": 43.53728489483748, "grad_norm": 0.49247828125953674, "learning_rate": 9.161278639278932e-05, "loss": 0.0276, "step": 68310 }, { "epoch": 43.54365838113448, "grad_norm": 0.6026017665863037, "learning_rate": 9.160973057227619e-05, "loss": 0.0298, "step": 68320 }, { "epoch": 43.550031867431485, "grad_norm": 0.49606752395629883, "learning_rate": 9.160667424616555e-05, "loss": 0.0263, "step": 68330 }, { "epoch": 43.556405353728486, "grad_norm": 0.6589988470077515, "learning_rate": 9.160361741449453e-05, "loss": 0.0287, "step": 68340 }, { "epoch": 43.562778840025494, "grad_norm": 0.7434877157211304, "learning_rate": 9.16005600773003e-05, "loss": 0.0217, "step": 68350 }, { "epoch": 43.569152326322495, "grad_norm": 0.4063299894332886, "learning_rate": 9.159750223461996e-05, "loss": 0.0159, "step": 68360 }, { "epoch": 43.5755258126195, "grad_norm": 0.2699035406112671, "learning_rate": 9.15944438864907e-05, "loss": 0.016, "step": 68370 }, { "epoch": 43.581899298916504, "grad_norm": 0.5909351110458374, "learning_rate": 9.159138503294968e-05, "loss": 0.023, "step": 68380 }, { "epoch": 43.58827278521351, "grad_norm": 0.41453972458839417, "learning_rate": 9.158832567403406e-05, "loss": 0.0324, "step": 68390 }, { "epoch": 43.59464627151051, "grad_norm": 0.28453704714775085, "learning_rate": 9.158526580978102e-05, "loss": 0.0284, "step": 68400 }, { "epoch": 43.60101975780752, "grad_norm": 0.23097805678844452, "learning_rate": 9.158220544022773e-05, "loss": 0.0171, "step": 68410 }, { "epoch": 43.60739324410453, "grad_norm": 0.4289282560348511, "learning_rate": 9.157914456541138e-05, "loss": 0.0282, "step": 68420 }, { "epoch": 43.61376673040153, "grad_norm": 0.5962466597557068, "learning_rate": 9.157608318536919e-05, "loss": 0.0221, "step": 68430 }, { "epoch": 43.62014021669854, "grad_norm": 0.22112102806568146, "learning_rate": 9.157302130013831e-05, "loss": 0.0228, "step": 68440 }, { "epoch": 43.62651370299554, "grad_norm": 0.3450978696346283, "learning_rate": 9.156995890975599e-05, "loss": 0.0235, "step": 68450 }, { "epoch": 43.632887189292546, "grad_norm": 0.6185300946235657, "learning_rate": 9.156689601425942e-05, "loss": 0.0459, "step": 68460 }, { "epoch": 43.63926067558955, "grad_norm": 0.23976293206214905, "learning_rate": 9.15638326136858e-05, "loss": 0.0229, "step": 68470 }, { "epoch": 43.645634161886555, "grad_norm": 0.7175559401512146, "learning_rate": 9.156076870807239e-05, "loss": 0.0328, "step": 68480 }, { "epoch": 43.652007648183556, "grad_norm": 0.2574958801269531, "learning_rate": 9.155770429745638e-05, "loss": 0.0264, "step": 68490 }, { "epoch": 43.658381134480564, "grad_norm": 0.6846250891685486, "learning_rate": 9.155463938187505e-05, "loss": 0.0283, "step": 68500 }, { "epoch": 43.664754620777565, "grad_norm": 0.4479522109031677, "learning_rate": 9.15515739613656e-05, "loss": 0.031, "step": 68510 }, { "epoch": 43.67112810707457, "grad_norm": 0.3911721408367157, "learning_rate": 9.154850803596529e-05, "loss": 0.0258, "step": 68520 }, { "epoch": 43.677501593371574, "grad_norm": 0.49033698439598083, "learning_rate": 9.15454416057114e-05, "loss": 0.0329, "step": 68530 }, { "epoch": 43.68387507966858, "grad_norm": 0.5527986884117126, "learning_rate": 9.154237467064115e-05, "loss": 0.027, "step": 68540 }, { "epoch": 43.69024856596558, "grad_norm": 0.8870930075645447, "learning_rate": 9.153930723079184e-05, "loss": 0.0372, "step": 68550 }, { "epoch": 43.69662205226259, "grad_norm": 0.3780331015586853, "learning_rate": 9.153623928620072e-05, "loss": 0.0375, "step": 68560 }, { "epoch": 43.70299553855959, "grad_norm": 0.3388501703739166, "learning_rate": 9.153317083690507e-05, "loss": 0.0177, "step": 68570 }, { "epoch": 43.7093690248566, "grad_norm": 0.5259823203086853, "learning_rate": 9.153010188294219e-05, "loss": 0.0166, "step": 68580 }, { "epoch": 43.7157425111536, "grad_norm": 0.6734403371810913, "learning_rate": 9.152703242434936e-05, "loss": 0.029, "step": 68590 }, { "epoch": 43.72211599745061, "grad_norm": 0.5611023306846619, "learning_rate": 9.152396246116387e-05, "loss": 0.027, "step": 68600 }, { "epoch": 43.72848948374761, "grad_norm": 0.447003573179245, "learning_rate": 9.152089199342304e-05, "loss": 0.0274, "step": 68610 }, { "epoch": 43.73486297004462, "grad_norm": 0.6926752328872681, "learning_rate": 9.151782102116417e-05, "loss": 0.036, "step": 68620 }, { "epoch": 43.74123645634162, "grad_norm": 0.5731717944145203, "learning_rate": 9.151474954442457e-05, "loss": 0.0253, "step": 68630 }, { "epoch": 43.747609942638626, "grad_norm": 0.3329644203186035, "learning_rate": 9.151167756324157e-05, "loss": 0.0196, "step": 68640 }, { "epoch": 43.753983428935626, "grad_norm": 0.5053603649139404, "learning_rate": 9.150860507765248e-05, "loss": 0.0339, "step": 68650 }, { "epoch": 43.760356915232634, "grad_norm": 0.6382958292961121, "learning_rate": 9.150553208769466e-05, "loss": 0.0284, "step": 68660 }, { "epoch": 43.766730401529635, "grad_norm": 0.4499886631965637, "learning_rate": 9.150245859340544e-05, "loss": 0.04, "step": 68670 }, { "epoch": 43.77310388782664, "grad_norm": 0.27924874424934387, "learning_rate": 9.149938459482218e-05, "loss": 0.0177, "step": 68680 }, { "epoch": 43.779477374123644, "grad_norm": 0.3499974012374878, "learning_rate": 9.149631009198218e-05, "loss": 0.0289, "step": 68690 }, { "epoch": 43.78585086042065, "grad_norm": 0.6004745960235596, "learning_rate": 9.149323508492286e-05, "loss": 0.0185, "step": 68700 }, { "epoch": 43.79222434671765, "grad_norm": 0.44475454092025757, "learning_rate": 9.149015957368155e-05, "loss": 0.0275, "step": 68710 }, { "epoch": 43.79859783301466, "grad_norm": 0.7249120473861694, "learning_rate": 9.148708355829562e-05, "loss": 0.024, "step": 68720 }, { "epoch": 43.80497131931166, "grad_norm": 0.5135484933853149, "learning_rate": 9.148400703880246e-05, "loss": 0.0299, "step": 68730 }, { "epoch": 43.81134480560867, "grad_norm": 0.6492828726768494, "learning_rate": 9.148093001523945e-05, "loss": 0.0275, "step": 68740 }, { "epoch": 43.81771829190567, "grad_norm": 0.3619154989719391, "learning_rate": 9.147785248764397e-05, "loss": 0.0245, "step": 68750 }, { "epoch": 43.82409177820268, "grad_norm": 0.4305607080459595, "learning_rate": 9.147477445605343e-05, "loss": 0.026, "step": 68760 }, { "epoch": 43.83046526449968, "grad_norm": 0.34129786491394043, "learning_rate": 9.147169592050522e-05, "loss": 0.0277, "step": 68770 }, { "epoch": 43.83683875079669, "grad_norm": 0.4091038703918457, "learning_rate": 9.146861688103674e-05, "loss": 0.0167, "step": 68780 }, { "epoch": 43.84321223709369, "grad_norm": 0.5101672410964966, "learning_rate": 9.146553733768542e-05, "loss": 0.0254, "step": 68790 }, { "epoch": 43.849585723390696, "grad_norm": 0.5365066528320312, "learning_rate": 9.146245729048865e-05, "loss": 0.0305, "step": 68800 }, { "epoch": 43.8559592096877, "grad_norm": 0.3254113495349884, "learning_rate": 9.145937673948391e-05, "loss": 0.0285, "step": 68810 }, { "epoch": 43.862332695984705, "grad_norm": 0.5276579260826111, "learning_rate": 9.145629568470857e-05, "loss": 0.0393, "step": 68820 }, { "epoch": 43.868706182281706, "grad_norm": 0.5106366872787476, "learning_rate": 9.145321412620011e-05, "loss": 0.0201, "step": 68830 }, { "epoch": 43.875079668578714, "grad_norm": 0.4448765814304352, "learning_rate": 9.145013206399597e-05, "loss": 0.033, "step": 68840 }, { "epoch": 43.881453154875715, "grad_norm": 0.4537672996520996, "learning_rate": 9.144704949813357e-05, "loss": 0.0212, "step": 68850 }, { "epoch": 43.88782664117272, "grad_norm": 0.39413580298423767, "learning_rate": 9.14439664286504e-05, "loss": 0.024, "step": 68860 }, { "epoch": 43.89420012746972, "grad_norm": 0.4654833972454071, "learning_rate": 9.144088285558389e-05, "loss": 0.0208, "step": 68870 }, { "epoch": 43.90057361376673, "grad_norm": 0.7494217157363892, "learning_rate": 9.143779877897155e-05, "loss": 0.0291, "step": 68880 }, { "epoch": 43.90694710006373, "grad_norm": 0.3948776423931122, "learning_rate": 9.143471419885082e-05, "loss": 0.0203, "step": 68890 }, { "epoch": 43.91332058636074, "grad_norm": 1.3631755113601685, "learning_rate": 9.14316291152592e-05, "loss": 0.029, "step": 68900 }, { "epoch": 43.91969407265774, "grad_norm": 0.3103019893169403, "learning_rate": 9.142854352823416e-05, "loss": 0.028, "step": 68910 }, { "epoch": 43.92606755895475, "grad_norm": 0.3118044137954712, "learning_rate": 9.14254574378132e-05, "loss": 0.0186, "step": 68920 }, { "epoch": 43.93244104525175, "grad_norm": 0.3544737696647644, "learning_rate": 9.142237084403382e-05, "loss": 0.0185, "step": 68930 }, { "epoch": 43.93881453154876, "grad_norm": 0.6615563631057739, "learning_rate": 9.141928374693353e-05, "loss": 0.028, "step": 68940 }, { "epoch": 43.94518801784576, "grad_norm": 0.3993005156517029, "learning_rate": 9.141619614654984e-05, "loss": 0.0235, "step": 68950 }, { "epoch": 43.95156150414277, "grad_norm": 0.4049597382545471, "learning_rate": 9.141310804292025e-05, "loss": 0.0257, "step": 68960 }, { "epoch": 43.95793499043977, "grad_norm": 0.43174174427986145, "learning_rate": 9.14100194360823e-05, "loss": 0.0244, "step": 68970 }, { "epoch": 43.964308476736775, "grad_norm": 0.45717301964759827, "learning_rate": 9.140693032607353e-05, "loss": 0.024, "step": 68980 }, { "epoch": 43.970681963033776, "grad_norm": 0.43140095472335815, "learning_rate": 9.140384071293145e-05, "loss": 0.0406, "step": 68990 }, { "epoch": 43.977055449330784, "grad_norm": 0.35039833188056946, "learning_rate": 9.140075059669362e-05, "loss": 0.0216, "step": 69000 }, { "epoch": 43.983428935627785, "grad_norm": 0.5256003141403198, "learning_rate": 9.139765997739758e-05, "loss": 0.0236, "step": 69010 }, { "epoch": 43.98980242192479, "grad_norm": 0.5736014246940613, "learning_rate": 9.139456885508086e-05, "loss": 0.0218, "step": 69020 }, { "epoch": 43.9961759082218, "grad_norm": 0.47632429003715515, "learning_rate": 9.139147722978108e-05, "loss": 0.0382, "step": 69030 }, { "epoch": 44.0025493945188, "grad_norm": 0.7445272207260132, "learning_rate": 9.138838510153576e-05, "loss": 0.0237, "step": 69040 }, { "epoch": 44.00892288081581, "grad_norm": 0.5885515809059143, "learning_rate": 9.138529247038247e-05, "loss": 0.0259, "step": 69050 }, { "epoch": 44.01529636711281, "grad_norm": 0.5254964828491211, "learning_rate": 9.138219933635883e-05, "loss": 0.0287, "step": 69060 }, { "epoch": 44.02166985340982, "grad_norm": 0.7452909350395203, "learning_rate": 9.13791056995024e-05, "loss": 0.0246, "step": 69070 }, { "epoch": 44.02804333970682, "grad_norm": 0.41195812821388245, "learning_rate": 9.137601155985073e-05, "loss": 0.0174, "step": 69080 }, { "epoch": 44.03441682600383, "grad_norm": 0.46608859300613403, "learning_rate": 9.137291691744147e-05, "loss": 0.0226, "step": 69090 }, { "epoch": 44.04079031230083, "grad_norm": 0.36798718571662903, "learning_rate": 9.136982177231221e-05, "loss": 0.0307, "step": 69100 }, { "epoch": 44.047163798597836, "grad_norm": 0.3875144422054291, "learning_rate": 9.136672612450057e-05, "loss": 0.0166, "step": 69110 }, { "epoch": 44.05353728489484, "grad_norm": 0.23577532172203064, "learning_rate": 9.136362997404413e-05, "loss": 0.0169, "step": 69120 }, { "epoch": 44.059910771191845, "grad_norm": 0.47972291707992554, "learning_rate": 9.136053332098054e-05, "loss": 0.0294, "step": 69130 }, { "epoch": 44.066284257488846, "grad_norm": 0.6593591570854187, "learning_rate": 9.135743616534742e-05, "loss": 0.043, "step": 69140 }, { "epoch": 44.072657743785854, "grad_norm": 0.33547520637512207, "learning_rate": 9.13543385071824e-05, "loss": 0.0308, "step": 69150 }, { "epoch": 44.079031230082855, "grad_norm": 0.6319129467010498, "learning_rate": 9.135124034652313e-05, "loss": 0.0243, "step": 69160 }, { "epoch": 44.08540471637986, "grad_norm": 0.5081955194473267, "learning_rate": 9.134814168340724e-05, "loss": 0.0204, "step": 69170 }, { "epoch": 44.09177820267686, "grad_norm": 0.5048558115959167, "learning_rate": 9.134504251787239e-05, "loss": 0.0265, "step": 69180 }, { "epoch": 44.09815168897387, "grad_norm": 0.4899298846721649, "learning_rate": 9.134194284995624e-05, "loss": 0.0319, "step": 69190 }, { "epoch": 44.10452517527087, "grad_norm": 1.1364482641220093, "learning_rate": 9.133884267969646e-05, "loss": 0.0318, "step": 69200 }, { "epoch": 44.11089866156788, "grad_norm": 0.5241957902908325, "learning_rate": 9.133574200713068e-05, "loss": 0.029, "step": 69210 }, { "epoch": 44.11727214786488, "grad_norm": 0.3964727520942688, "learning_rate": 9.133264083229665e-05, "loss": 0.0222, "step": 69220 }, { "epoch": 44.12364563416189, "grad_norm": 0.4636227786540985, "learning_rate": 9.132953915523198e-05, "loss": 0.0188, "step": 69230 }, { "epoch": 44.13001912045889, "grad_norm": 0.6302806735038757, "learning_rate": 9.132643697597439e-05, "loss": 0.0425, "step": 69240 }, { "epoch": 44.1363926067559, "grad_norm": 0.6307168006896973, "learning_rate": 9.132333429456158e-05, "loss": 0.0176, "step": 69250 }, { "epoch": 44.1427660930529, "grad_norm": 0.788242757320404, "learning_rate": 9.132023111103125e-05, "loss": 0.0414, "step": 69260 }, { "epoch": 44.14913957934991, "grad_norm": 0.3888011872768402, "learning_rate": 9.131712742542108e-05, "loss": 0.0223, "step": 69270 }, { "epoch": 44.15551306564691, "grad_norm": 0.20786574482917786, "learning_rate": 9.13140232377688e-05, "loss": 0.0197, "step": 69280 }, { "epoch": 44.161886551943915, "grad_norm": 0.34875741600990295, "learning_rate": 9.131091854811216e-05, "loss": 0.0236, "step": 69290 }, { "epoch": 44.168260038240916, "grad_norm": 0.5778881311416626, "learning_rate": 9.130781335648883e-05, "loss": 0.0215, "step": 69300 }, { "epoch": 44.174633524537924, "grad_norm": 0.5591598749160767, "learning_rate": 9.130470766293657e-05, "loss": 0.0193, "step": 69310 }, { "epoch": 44.181007010834925, "grad_norm": 0.5050633549690247, "learning_rate": 9.130160146749312e-05, "loss": 0.0297, "step": 69320 }, { "epoch": 44.18738049713193, "grad_norm": 0.6095276474952698, "learning_rate": 9.129849477019621e-05, "loss": 0.034, "step": 69330 }, { "epoch": 44.193753983428934, "grad_norm": 0.6065836548805237, "learning_rate": 9.12953875710836e-05, "loss": 0.0409, "step": 69340 }, { "epoch": 44.20012746972594, "grad_norm": 0.8167550563812256, "learning_rate": 9.129227987019304e-05, "loss": 0.0273, "step": 69350 }, { "epoch": 44.20650095602294, "grad_norm": 0.5086143016815186, "learning_rate": 9.12891716675623e-05, "loss": 0.0287, "step": 69360 }, { "epoch": 44.21287444231995, "grad_norm": 1.3218098878860474, "learning_rate": 9.128606296322915e-05, "loss": 0.0192, "step": 69370 }, { "epoch": 44.21924792861695, "grad_norm": 0.2761237323284149, "learning_rate": 9.128295375723133e-05, "loss": 0.0186, "step": 69380 }, { "epoch": 44.22562141491396, "grad_norm": 0.3746655583381653, "learning_rate": 9.127984404960666e-05, "loss": 0.0452, "step": 69390 }, { "epoch": 44.23199490121096, "grad_norm": 0.47848236560821533, "learning_rate": 9.12767338403929e-05, "loss": 0.0354, "step": 69400 }, { "epoch": 44.23836838750797, "grad_norm": 0.4695889949798584, "learning_rate": 9.127362312962786e-05, "loss": 0.0264, "step": 69410 }, { "epoch": 44.24474187380497, "grad_norm": 0.5995346307754517, "learning_rate": 9.127051191734933e-05, "loss": 0.0312, "step": 69420 }, { "epoch": 44.25111536010198, "grad_norm": 0.5833046436309814, "learning_rate": 9.126740020359512e-05, "loss": 0.031, "step": 69430 }, { "epoch": 44.25748884639898, "grad_norm": 0.3650643229484558, "learning_rate": 9.126428798840302e-05, "loss": 0.0386, "step": 69440 }, { "epoch": 44.263862332695986, "grad_norm": 0.7167417407035828, "learning_rate": 9.126117527181086e-05, "loss": 0.0178, "step": 69450 }, { "epoch": 44.27023581899299, "grad_norm": 0.3195766508579254, "learning_rate": 9.125806205385648e-05, "loss": 0.0178, "step": 69460 }, { "epoch": 44.276609305289995, "grad_norm": 0.3028687834739685, "learning_rate": 9.125494833457769e-05, "loss": 0.0293, "step": 69470 }, { "epoch": 44.282982791586996, "grad_norm": 0.3424360156059265, "learning_rate": 9.125183411401233e-05, "loss": 0.018, "step": 69480 }, { "epoch": 44.289356277884, "grad_norm": 0.40720537304878235, "learning_rate": 9.124871939219821e-05, "loss": 0.0205, "step": 69490 }, { "epoch": 44.295729764181004, "grad_norm": 0.35510939359664917, "learning_rate": 9.124560416917323e-05, "loss": 0.024, "step": 69500 }, { "epoch": 44.30210325047801, "grad_norm": 0.45656684041023254, "learning_rate": 9.12424884449752e-05, "loss": 0.0173, "step": 69510 }, { "epoch": 44.30847673677501, "grad_norm": 0.357785701751709, "learning_rate": 9.1239372219642e-05, "loss": 0.0187, "step": 69520 }, { "epoch": 44.31485022307202, "grad_norm": 0.48116904497146606, "learning_rate": 9.123625549321151e-05, "loss": 0.0258, "step": 69530 }, { "epoch": 44.32122370936902, "grad_norm": 0.8287373781204224, "learning_rate": 9.123313826572156e-05, "loss": 0.0185, "step": 69540 }, { "epoch": 44.32759719566603, "grad_norm": 0.6212223768234253, "learning_rate": 9.123002053721005e-05, "loss": 0.0303, "step": 69550 }, { "epoch": 44.33397068196303, "grad_norm": 0.3443302512168884, "learning_rate": 9.122690230771488e-05, "loss": 0.0184, "step": 69560 }, { "epoch": 44.34034416826004, "grad_norm": 0.38666781783103943, "learning_rate": 9.122378357727393e-05, "loss": 0.0151, "step": 69570 }, { "epoch": 44.34671765455704, "grad_norm": 1.0148160457611084, "learning_rate": 9.122066434592506e-05, "loss": 0.0269, "step": 69580 }, { "epoch": 44.35309114085405, "grad_norm": 0.4918820858001709, "learning_rate": 9.12175446137062e-05, "loss": 0.0198, "step": 69590 }, { "epoch": 44.35946462715105, "grad_norm": 0.4328464865684509, "learning_rate": 9.121442438065528e-05, "loss": 0.0259, "step": 69600 }, { "epoch": 44.365838113448056, "grad_norm": 0.6793479323387146, "learning_rate": 9.121130364681016e-05, "loss": 0.0208, "step": 69610 }, { "epoch": 44.37221159974506, "grad_norm": 0.5032913684844971, "learning_rate": 9.120818241220882e-05, "loss": 0.0202, "step": 69620 }, { "epoch": 44.378585086042065, "grad_norm": 0.5774743556976318, "learning_rate": 9.120506067688914e-05, "loss": 0.0237, "step": 69630 }, { "epoch": 44.384958572339066, "grad_norm": 0.3497629761695862, "learning_rate": 9.120193844088908e-05, "loss": 0.0198, "step": 69640 }, { "epoch": 44.391332058636074, "grad_norm": 0.3199697732925415, "learning_rate": 9.119881570424656e-05, "loss": 0.0232, "step": 69650 }, { "epoch": 44.397705544933075, "grad_norm": 0.30733513832092285, "learning_rate": 9.119569246699952e-05, "loss": 0.0239, "step": 69660 }, { "epoch": 44.40407903123008, "grad_norm": 0.47568878531455994, "learning_rate": 9.119256872918594e-05, "loss": 0.0285, "step": 69670 }, { "epoch": 44.41045251752709, "grad_norm": 0.2987971901893616, "learning_rate": 9.118944449084377e-05, "loss": 0.0302, "step": 69680 }, { "epoch": 44.41682600382409, "grad_norm": 0.45306047797203064, "learning_rate": 9.118631975201092e-05, "loss": 0.0287, "step": 69690 }, { "epoch": 44.4231994901211, "grad_norm": 0.3489025831222534, "learning_rate": 9.118319451272544e-05, "loss": 0.0259, "step": 69700 }, { "epoch": 44.4295729764181, "grad_norm": 0.32040104269981384, "learning_rate": 9.118006877302524e-05, "loss": 0.0285, "step": 69710 }, { "epoch": 44.43594646271511, "grad_norm": 0.3904116451740265, "learning_rate": 9.117694253294833e-05, "loss": 0.0246, "step": 69720 }, { "epoch": 44.44231994901211, "grad_norm": 0.6685142517089844, "learning_rate": 9.117381579253268e-05, "loss": 0.0375, "step": 69730 }, { "epoch": 44.44869343530912, "grad_norm": 0.2558314800262451, "learning_rate": 9.11706885518163e-05, "loss": 0.0239, "step": 69740 }, { "epoch": 44.45506692160612, "grad_norm": 0.7195074558258057, "learning_rate": 9.11675608108372e-05, "loss": 0.022, "step": 69750 }, { "epoch": 44.461440407903126, "grad_norm": 0.6453682780265808, "learning_rate": 9.116443256963333e-05, "loss": 0.0263, "step": 69760 }, { "epoch": 44.46781389420013, "grad_norm": 0.26266562938690186, "learning_rate": 9.116130382824277e-05, "loss": 0.0232, "step": 69770 }, { "epoch": 44.474187380497135, "grad_norm": 0.29991114139556885, "learning_rate": 9.11581745867035e-05, "loss": 0.0211, "step": 69780 }, { "epoch": 44.480560866794136, "grad_norm": 0.619339644908905, "learning_rate": 9.115504484505355e-05, "loss": 0.0195, "step": 69790 }, { "epoch": 44.486934353091144, "grad_norm": 0.3754052221775055, "learning_rate": 9.115191460333095e-05, "loss": 0.0326, "step": 69800 }, { "epoch": 44.493307839388144, "grad_norm": 0.6344118118286133, "learning_rate": 9.114878386157372e-05, "loss": 0.0256, "step": 69810 }, { "epoch": 44.49968132568515, "grad_norm": 0.4744247496128082, "learning_rate": 9.114565261981992e-05, "loss": 0.0198, "step": 69820 }, { "epoch": 44.50605481198215, "grad_norm": 0.30229365825653076, "learning_rate": 9.114252087810759e-05, "loss": 0.023, "step": 69830 }, { "epoch": 44.51242829827916, "grad_norm": 0.5179222822189331, "learning_rate": 9.11393886364748e-05, "loss": 0.0294, "step": 69840 }, { "epoch": 44.51880178457616, "grad_norm": 0.40976962447166443, "learning_rate": 9.113625589495958e-05, "loss": 0.0274, "step": 69850 }, { "epoch": 44.52517527087317, "grad_norm": 0.36087876558303833, "learning_rate": 9.113312265360003e-05, "loss": 0.0243, "step": 69860 }, { "epoch": 44.53154875717017, "grad_norm": 0.3029163181781769, "learning_rate": 9.112998891243419e-05, "loss": 0.0234, "step": 69870 }, { "epoch": 44.53792224346718, "grad_norm": 0.7869796752929688, "learning_rate": 9.112685467150015e-05, "loss": 0.0235, "step": 69880 }, { "epoch": 44.54429572976418, "grad_norm": 0.7458316683769226, "learning_rate": 9.1123719930836e-05, "loss": 0.0261, "step": 69890 }, { "epoch": 44.55066921606119, "grad_norm": 0.4367789328098297, "learning_rate": 9.112058469047982e-05, "loss": 0.038, "step": 69900 }, { "epoch": 44.55704270235819, "grad_norm": 0.45899897813796997, "learning_rate": 9.111744895046972e-05, "loss": 0.0182, "step": 69910 }, { "epoch": 44.5634161886552, "grad_norm": 0.29348084330558777, "learning_rate": 9.11143127108438e-05, "loss": 0.0265, "step": 69920 }, { "epoch": 44.5697896749522, "grad_norm": 0.4629710018634796, "learning_rate": 9.111117597164015e-05, "loss": 0.0295, "step": 69930 }, { "epoch": 44.576163161249205, "grad_norm": 0.4618318974971771, "learning_rate": 9.11080387328969e-05, "loss": 0.0307, "step": 69940 }, { "epoch": 44.582536647546206, "grad_norm": 0.5685538649559021, "learning_rate": 9.110490099465216e-05, "loss": 0.0284, "step": 69950 }, { "epoch": 44.588910133843214, "grad_norm": 1.0322154760360718, "learning_rate": 9.11017627569441e-05, "loss": 0.0235, "step": 69960 }, { "epoch": 44.595283620140215, "grad_norm": 0.5729140639305115, "learning_rate": 9.109862401981077e-05, "loss": 0.0241, "step": 69970 }, { "epoch": 44.60165710643722, "grad_norm": 0.32351556420326233, "learning_rate": 9.109548478329037e-05, "loss": 0.0255, "step": 69980 }, { "epoch": 44.608030592734224, "grad_norm": 0.31515541672706604, "learning_rate": 9.109234504742103e-05, "loss": 0.0194, "step": 69990 }, { "epoch": 44.61440407903123, "grad_norm": 0.6408838033676147, "learning_rate": 9.10892048122409e-05, "loss": 0.0162, "step": 70000 }, { "epoch": 44.62077756532823, "grad_norm": 0.3009513020515442, "learning_rate": 9.108606407778815e-05, "loss": 0.0225, "step": 70010 }, { "epoch": 44.62715105162524, "grad_norm": 1.4135018587112427, "learning_rate": 9.108292284410091e-05, "loss": 0.0166, "step": 70020 }, { "epoch": 44.63352453792224, "grad_norm": 0.5567973852157593, "learning_rate": 9.107978111121737e-05, "loss": 0.0244, "step": 70030 }, { "epoch": 44.63989802421925, "grad_norm": 0.44984617829322815, "learning_rate": 9.107663887917571e-05, "loss": 0.0214, "step": 70040 }, { "epoch": 44.64627151051625, "grad_norm": 0.3634715974330902, "learning_rate": 9.10734961480141e-05, "loss": 0.0208, "step": 70050 }, { "epoch": 44.65264499681326, "grad_norm": 0.4572800397872925, "learning_rate": 9.107035291777075e-05, "loss": 0.0185, "step": 70060 }, { "epoch": 44.65901848311026, "grad_norm": 0.4640272557735443, "learning_rate": 9.106720918848381e-05, "loss": 0.0321, "step": 70070 }, { "epoch": 44.66539196940727, "grad_norm": 0.3801213502883911, "learning_rate": 9.106406496019152e-05, "loss": 0.026, "step": 70080 }, { "epoch": 44.67176545570427, "grad_norm": 0.3513534665107727, "learning_rate": 9.106092023293206e-05, "loss": 0.0237, "step": 70090 }, { "epoch": 44.678138942001276, "grad_norm": 0.5299761891365051, "learning_rate": 9.105777500674366e-05, "loss": 0.0318, "step": 70100 }, { "epoch": 44.68451242829828, "grad_norm": 0.5048236846923828, "learning_rate": 9.105462928166452e-05, "loss": 0.0228, "step": 70110 }, { "epoch": 44.690885914595285, "grad_norm": 0.34549281001091003, "learning_rate": 9.10514830577329e-05, "loss": 0.021, "step": 70120 }, { "epoch": 44.697259400892285, "grad_norm": 0.5473931431770325, "learning_rate": 9.104833633498697e-05, "loss": 0.0301, "step": 70130 }, { "epoch": 44.70363288718929, "grad_norm": 0.7582613229751587, "learning_rate": 9.104518911346501e-05, "loss": 0.0327, "step": 70140 }, { "epoch": 44.710006373486294, "grad_norm": 0.5976056456565857, "learning_rate": 9.104204139320524e-05, "loss": 0.0313, "step": 70150 }, { "epoch": 44.7163798597833, "grad_norm": 0.7555539011955261, "learning_rate": 9.103889317424593e-05, "loss": 0.0452, "step": 70160 }, { "epoch": 44.7227533460803, "grad_norm": 0.40818914771080017, "learning_rate": 9.103574445662532e-05, "loss": 0.0282, "step": 70170 }, { "epoch": 44.72912683237731, "grad_norm": 0.3522981107234955, "learning_rate": 9.103259524038166e-05, "loss": 0.0274, "step": 70180 }, { "epoch": 44.73550031867431, "grad_norm": 0.4655044376850128, "learning_rate": 9.102944552555323e-05, "loss": 0.033, "step": 70190 }, { "epoch": 44.74187380497132, "grad_norm": 0.5036574602127075, "learning_rate": 9.10262953121783e-05, "loss": 0.0291, "step": 70200 }, { "epoch": 44.74824729126832, "grad_norm": 0.6618916392326355, "learning_rate": 9.102314460029516e-05, "loss": 0.0185, "step": 70210 }, { "epoch": 44.75462077756533, "grad_norm": 0.5666359663009644, "learning_rate": 9.101999338994207e-05, "loss": 0.0256, "step": 70220 }, { "epoch": 44.76099426386233, "grad_norm": 0.3985896110534668, "learning_rate": 9.101684168115732e-05, "loss": 0.0224, "step": 70230 }, { "epoch": 44.76736775015934, "grad_norm": 0.6574935913085938, "learning_rate": 9.101368947397923e-05, "loss": 0.0259, "step": 70240 }, { "epoch": 44.77374123645634, "grad_norm": 0.3646835684776306, "learning_rate": 9.101053676844608e-05, "loss": 0.0298, "step": 70250 }, { "epoch": 44.780114722753346, "grad_norm": 0.3888828754425049, "learning_rate": 9.10073835645962e-05, "loss": 0.0276, "step": 70260 }, { "epoch": 44.78648820905035, "grad_norm": 0.5211093425750732, "learning_rate": 9.100422986246788e-05, "loss": 0.0224, "step": 70270 }, { "epoch": 44.792861695347355, "grad_norm": 0.4625350832939148, "learning_rate": 9.100107566209946e-05, "loss": 0.0361, "step": 70280 }, { "epoch": 44.79923518164436, "grad_norm": 0.40586015582084656, "learning_rate": 9.099792096352926e-05, "loss": 0.0318, "step": 70290 }, { "epoch": 44.805608667941364, "grad_norm": 0.5788999795913696, "learning_rate": 9.099476576679563e-05, "loss": 0.0354, "step": 70300 }, { "epoch": 44.81198215423837, "grad_norm": 0.6112302541732788, "learning_rate": 9.099161007193687e-05, "loss": 0.0292, "step": 70310 }, { "epoch": 44.81835564053537, "grad_norm": 0.31626275181770325, "learning_rate": 9.098845387899133e-05, "loss": 0.0215, "step": 70320 }, { "epoch": 44.82472912683238, "grad_norm": 0.8335644602775574, "learning_rate": 9.09852971879974e-05, "loss": 0.0314, "step": 70330 }, { "epoch": 44.83110261312938, "grad_norm": 0.392109215259552, "learning_rate": 9.09821399989934e-05, "loss": 0.0277, "step": 70340 }, { "epoch": 44.83747609942639, "grad_norm": 0.5769377946853638, "learning_rate": 9.09789823120177e-05, "loss": 0.0218, "step": 70350 }, { "epoch": 44.84384958572339, "grad_norm": 0.4938219487667084, "learning_rate": 9.09758241271087e-05, "loss": 0.0271, "step": 70360 }, { "epoch": 44.8502230720204, "grad_norm": 0.6601452231407166, "learning_rate": 9.097266544430472e-05, "loss": 0.0204, "step": 70370 }, { "epoch": 44.8565965583174, "grad_norm": 0.4282577335834503, "learning_rate": 9.096950626364418e-05, "loss": 0.0193, "step": 70380 }, { "epoch": 44.86297004461441, "grad_norm": 0.5425766110420227, "learning_rate": 9.096634658516547e-05, "loss": 0.0192, "step": 70390 }, { "epoch": 44.86934353091141, "grad_norm": 0.2954298257827759, "learning_rate": 9.096318640890695e-05, "loss": 0.0209, "step": 70400 }, { "epoch": 44.875717017208416, "grad_norm": 0.2802054286003113, "learning_rate": 9.096002573490707e-05, "loss": 0.0238, "step": 70410 }, { "epoch": 44.88209050350542, "grad_norm": 0.3437669277191162, "learning_rate": 9.095686456320418e-05, "loss": 0.017, "step": 70420 }, { "epoch": 44.888463989802425, "grad_norm": 0.4398162066936493, "learning_rate": 9.095370289383672e-05, "loss": 0.0167, "step": 70430 }, { "epoch": 44.894837476099426, "grad_norm": 0.34611621499061584, "learning_rate": 9.09505407268431e-05, "loss": 0.0182, "step": 70440 }, { "epoch": 44.90121096239643, "grad_norm": 0.318350613117218, "learning_rate": 9.094737806226176e-05, "loss": 0.0284, "step": 70450 }, { "epoch": 44.907584448693434, "grad_norm": 0.8930920362472534, "learning_rate": 9.09442149001311e-05, "loss": 0.0191, "step": 70460 }, { "epoch": 44.91395793499044, "grad_norm": 0.8092803955078125, "learning_rate": 9.094105124048958e-05, "loss": 0.0187, "step": 70470 }, { "epoch": 44.92033142128744, "grad_norm": 0.35930967330932617, "learning_rate": 9.093788708337563e-05, "loss": 0.0277, "step": 70480 }, { "epoch": 44.92670490758445, "grad_norm": 0.6216056942939758, "learning_rate": 9.09347224288277e-05, "loss": 0.0189, "step": 70490 }, { "epoch": 44.93307839388145, "grad_norm": 0.22093836963176727, "learning_rate": 9.093155727688425e-05, "loss": 0.0222, "step": 70500 }, { "epoch": 44.93945188017846, "grad_norm": 0.5702199339866638, "learning_rate": 9.092839162758372e-05, "loss": 0.0308, "step": 70510 }, { "epoch": 44.94582536647546, "grad_norm": 0.2813703119754791, "learning_rate": 9.09252254809646e-05, "loss": 0.0216, "step": 70520 }, { "epoch": 44.95219885277247, "grad_norm": 0.3856235146522522, "learning_rate": 9.092205883706535e-05, "loss": 0.0172, "step": 70530 }, { "epoch": 44.95857233906947, "grad_norm": 0.5444739460945129, "learning_rate": 9.091889169592444e-05, "loss": 0.0234, "step": 70540 }, { "epoch": 44.96494582536648, "grad_norm": 0.5959101915359497, "learning_rate": 9.091572405758038e-05, "loss": 0.0285, "step": 70550 }, { "epoch": 44.97131931166348, "grad_norm": 0.47324520349502563, "learning_rate": 9.091255592207161e-05, "loss": 0.0283, "step": 70560 }, { "epoch": 44.977692797960486, "grad_norm": 0.5032578110694885, "learning_rate": 9.090938728943669e-05, "loss": 0.028, "step": 70570 }, { "epoch": 44.98406628425749, "grad_norm": 0.35291022062301636, "learning_rate": 9.090621815971408e-05, "loss": 0.0219, "step": 70580 }, { "epoch": 44.990439770554495, "grad_norm": 0.34902822971343994, "learning_rate": 9.09030485329423e-05, "loss": 0.0205, "step": 70590 }, { "epoch": 44.996813256851496, "grad_norm": 0.492389053106308, "learning_rate": 9.089987840915985e-05, "loss": 0.0204, "step": 70600 }, { "epoch": 45.003186743148504, "grad_norm": 0.374727725982666, "learning_rate": 9.089670778840527e-05, "loss": 0.0237, "step": 70610 }, { "epoch": 45.009560229445505, "grad_norm": 0.5217189192771912, "learning_rate": 9.089353667071707e-05, "loss": 0.0218, "step": 70620 }, { "epoch": 45.01593371574251, "grad_norm": 0.8215938210487366, "learning_rate": 9.089036505613379e-05, "loss": 0.0298, "step": 70630 }, { "epoch": 45.022307202039514, "grad_norm": 0.6175441741943359, "learning_rate": 9.088719294469397e-05, "loss": 0.0335, "step": 70640 }, { "epoch": 45.02868068833652, "grad_norm": 0.2065260410308838, "learning_rate": 9.088402033643615e-05, "loss": 0.0175, "step": 70650 }, { "epoch": 45.03505417463352, "grad_norm": 0.4584850072860718, "learning_rate": 9.088084723139889e-05, "loss": 0.0327, "step": 70660 }, { "epoch": 45.04142766093053, "grad_norm": 0.6576526761054993, "learning_rate": 9.087767362962074e-05, "loss": 0.0405, "step": 70670 }, { "epoch": 45.04780114722753, "grad_norm": 0.5055668354034424, "learning_rate": 9.087449953114025e-05, "loss": 0.0305, "step": 70680 }, { "epoch": 45.05417463352454, "grad_norm": 0.501064121723175, "learning_rate": 9.0871324935996e-05, "loss": 0.029, "step": 70690 }, { "epoch": 45.06054811982154, "grad_norm": 0.6340673565864563, "learning_rate": 9.086814984422657e-05, "loss": 0.0192, "step": 70700 }, { "epoch": 45.06692160611855, "grad_norm": 1.3853297233581543, "learning_rate": 9.086497425587053e-05, "loss": 0.0223, "step": 70710 }, { "epoch": 45.07329509241555, "grad_norm": 0.5656533241271973, "learning_rate": 9.086179817096648e-05, "loss": 0.0251, "step": 70720 }, { "epoch": 45.07966857871256, "grad_norm": 0.40715739130973816, "learning_rate": 9.085862158955299e-05, "loss": 0.0274, "step": 70730 }, { "epoch": 45.08604206500956, "grad_norm": 0.2692093849182129, "learning_rate": 9.085544451166867e-05, "loss": 0.0188, "step": 70740 }, { "epoch": 45.092415551306566, "grad_norm": 0.4276832342147827, "learning_rate": 9.085226693735211e-05, "loss": 0.0267, "step": 70750 }, { "epoch": 45.09878903760357, "grad_norm": 0.4307619631290436, "learning_rate": 9.084908886664196e-05, "loss": 0.0287, "step": 70760 }, { "epoch": 45.105162523900574, "grad_norm": 0.4087086617946625, "learning_rate": 9.084591029957681e-05, "loss": 0.0194, "step": 70770 }, { "epoch": 45.111536010197575, "grad_norm": 0.4947662949562073, "learning_rate": 9.084273123619529e-05, "loss": 0.0287, "step": 70780 }, { "epoch": 45.11790949649458, "grad_norm": 0.33628049492836, "learning_rate": 9.0839551676536e-05, "loss": 0.0298, "step": 70790 }, { "epoch": 45.124282982791584, "grad_norm": 0.5575718879699707, "learning_rate": 9.083637162063762e-05, "loss": 0.0407, "step": 70800 }, { "epoch": 45.13065646908859, "grad_norm": 0.5913023948669434, "learning_rate": 9.083319106853876e-05, "loss": 0.0372, "step": 70810 }, { "epoch": 45.13702995538559, "grad_norm": 0.6538296937942505, "learning_rate": 9.083001002027807e-05, "loss": 0.0289, "step": 70820 }, { "epoch": 45.1434034416826, "grad_norm": 0.5777148008346558, "learning_rate": 9.082682847589423e-05, "loss": 0.0236, "step": 70830 }, { "epoch": 45.1497769279796, "grad_norm": 0.363638311624527, "learning_rate": 9.082364643542585e-05, "loss": 0.0297, "step": 70840 }, { "epoch": 45.15615041427661, "grad_norm": 0.38163599371910095, "learning_rate": 9.082046389891164e-05, "loss": 0.0194, "step": 70850 }, { "epoch": 45.16252390057361, "grad_norm": 0.6348438858985901, "learning_rate": 9.081728086639025e-05, "loss": 0.0239, "step": 70860 }, { "epoch": 45.16889738687062, "grad_norm": 0.6211579442024231, "learning_rate": 9.081409733790036e-05, "loss": 0.0157, "step": 70870 }, { "epoch": 45.17527087316762, "grad_norm": 0.38792866468429565, "learning_rate": 9.081091331348064e-05, "loss": 0.037, "step": 70880 }, { "epoch": 45.18164435946463, "grad_norm": 0.41175323724746704, "learning_rate": 9.080772879316982e-05, "loss": 0.0261, "step": 70890 }, { "epoch": 45.18801784576163, "grad_norm": 0.3493616282939911, "learning_rate": 9.080454377700653e-05, "loss": 0.0297, "step": 70900 }, { "epoch": 45.194391332058636, "grad_norm": 0.4021454453468323, "learning_rate": 9.080135826502953e-05, "loss": 0.0295, "step": 70910 }, { "epoch": 45.20076481835564, "grad_norm": 0.36986207962036133, "learning_rate": 9.079817225727749e-05, "loss": 0.0248, "step": 70920 }, { "epoch": 45.207138304652645, "grad_norm": 1.6000581979751587, "learning_rate": 9.079498575378915e-05, "loss": 0.0539, "step": 70930 }, { "epoch": 45.21351179094965, "grad_norm": 0.46225082874298096, "learning_rate": 9.079179875460322e-05, "loss": 0.0236, "step": 70940 }, { "epoch": 45.219885277246654, "grad_norm": 0.35660475492477417, "learning_rate": 9.07886112597584e-05, "loss": 0.0264, "step": 70950 }, { "epoch": 45.22625876354366, "grad_norm": 0.28614985942840576, "learning_rate": 9.078542326929347e-05, "loss": 0.0273, "step": 70960 }, { "epoch": 45.23263224984066, "grad_norm": 0.39445197582244873, "learning_rate": 9.078223478324712e-05, "loss": 0.0231, "step": 70970 }, { "epoch": 45.23900573613767, "grad_norm": 0.36632266640663147, "learning_rate": 9.077904580165811e-05, "loss": 0.036, "step": 70980 }, { "epoch": 45.24537922243467, "grad_norm": 0.46131616830825806, "learning_rate": 9.07758563245652e-05, "loss": 0.0297, "step": 70990 }, { "epoch": 45.25175270873168, "grad_norm": 0.36648306250572205, "learning_rate": 9.077266635200715e-05, "loss": 0.0159, "step": 71000 }, { "epoch": 45.25812619502868, "grad_norm": 0.28990280628204346, "learning_rate": 9.076947588402268e-05, "loss": 0.0213, "step": 71010 }, { "epoch": 45.26449968132569, "grad_norm": 0.38565969467163086, "learning_rate": 9.076628492065061e-05, "loss": 0.0203, "step": 71020 }, { "epoch": 45.27087316762269, "grad_norm": 2.693587064743042, "learning_rate": 9.076309346192967e-05, "loss": 0.0243, "step": 71030 }, { "epoch": 45.2772466539197, "grad_norm": 0.37950462102890015, "learning_rate": 9.075990150789866e-05, "loss": 0.0257, "step": 71040 }, { "epoch": 45.2836201402167, "grad_norm": 0.6521598696708679, "learning_rate": 9.075670905859637e-05, "loss": 0.0313, "step": 71050 }, { "epoch": 45.289993626513706, "grad_norm": 0.6091110706329346, "learning_rate": 9.07535161140616e-05, "loss": 0.0331, "step": 71060 }, { "epoch": 45.29636711281071, "grad_norm": 0.3839631676673889, "learning_rate": 9.07503226743331e-05, "loss": 0.0467, "step": 71070 }, { "epoch": 45.302740599107715, "grad_norm": 0.34343329071998596, "learning_rate": 9.074712873944971e-05, "loss": 0.0191, "step": 71080 }, { "epoch": 45.309114085404715, "grad_norm": 0.3329848647117615, "learning_rate": 9.074393430945025e-05, "loss": 0.0203, "step": 71090 }, { "epoch": 45.31548757170172, "grad_norm": 0.5943746566772461, "learning_rate": 9.074073938437351e-05, "loss": 0.0242, "step": 71100 }, { "epoch": 45.321861057998724, "grad_norm": 0.6341114640235901, "learning_rate": 9.073754396425833e-05, "loss": 0.0199, "step": 71110 }, { "epoch": 45.32823454429573, "grad_norm": 0.390470027923584, "learning_rate": 9.073434804914352e-05, "loss": 0.0246, "step": 71120 }, { "epoch": 45.33460803059273, "grad_norm": 0.391428142786026, "learning_rate": 9.073115163906793e-05, "loss": 0.0398, "step": 71130 }, { "epoch": 45.34098151688974, "grad_norm": 0.27678635716438293, "learning_rate": 9.072795473407038e-05, "loss": 0.0314, "step": 71140 }, { "epoch": 45.34735500318674, "grad_norm": 0.3356377184391022, "learning_rate": 9.072475733418973e-05, "loss": 0.0195, "step": 71150 }, { "epoch": 45.35372848948375, "grad_norm": 0.3680795729160309, "learning_rate": 9.072155943946483e-05, "loss": 0.0209, "step": 71160 }, { "epoch": 45.36010197578075, "grad_norm": 0.4007362127304077, "learning_rate": 9.071836104993453e-05, "loss": 0.0146, "step": 71170 }, { "epoch": 45.36647546207776, "grad_norm": 0.6581327319145203, "learning_rate": 9.07151621656377e-05, "loss": 0.0233, "step": 71180 }, { "epoch": 45.37284894837476, "grad_norm": 0.33917173743247986, "learning_rate": 9.071196278661322e-05, "loss": 0.0144, "step": 71190 }, { "epoch": 45.37922243467177, "grad_norm": 0.5440648198127747, "learning_rate": 9.070876291289994e-05, "loss": 0.0374, "step": 71200 }, { "epoch": 45.38559592096877, "grad_norm": 0.636914074420929, "learning_rate": 9.070556254453676e-05, "loss": 0.0466, "step": 71210 }, { "epoch": 45.391969407265776, "grad_norm": 0.2487654834985733, "learning_rate": 9.070236168156256e-05, "loss": 0.0229, "step": 71220 }, { "epoch": 45.39834289356278, "grad_norm": 0.7482965588569641, "learning_rate": 9.069916032401624e-05, "loss": 0.0378, "step": 71230 }, { "epoch": 45.404716379859785, "grad_norm": 0.34821584820747375, "learning_rate": 9.069595847193669e-05, "loss": 0.0212, "step": 71240 }, { "epoch": 45.411089866156786, "grad_norm": 0.36341798305511475, "learning_rate": 9.069275612536283e-05, "loss": 0.0171, "step": 71250 }, { "epoch": 45.417463352453794, "grad_norm": 0.5606079697608948, "learning_rate": 9.068955328433355e-05, "loss": 0.0281, "step": 71260 }, { "epoch": 45.423836838750795, "grad_norm": 0.4267158806324005, "learning_rate": 9.06863499488878e-05, "loss": 0.0214, "step": 71270 }, { "epoch": 45.4302103250478, "grad_norm": 0.3178037703037262, "learning_rate": 9.068314611906447e-05, "loss": 0.0184, "step": 71280 }, { "epoch": 45.4365838113448, "grad_norm": 0.5893178582191467, "learning_rate": 9.067994179490251e-05, "loss": 0.0192, "step": 71290 }, { "epoch": 45.44295729764181, "grad_norm": 0.42788901925086975, "learning_rate": 9.067673697644084e-05, "loss": 0.0285, "step": 71300 }, { "epoch": 45.44933078393881, "grad_norm": 0.5455334186553955, "learning_rate": 9.067353166371841e-05, "loss": 0.0273, "step": 71310 }, { "epoch": 45.45570427023582, "grad_norm": 0.5173894166946411, "learning_rate": 9.067032585677418e-05, "loss": 0.0195, "step": 71320 }, { "epoch": 45.46207775653282, "grad_norm": 0.42463773488998413, "learning_rate": 9.066711955564708e-05, "loss": 0.0258, "step": 71330 }, { "epoch": 45.46845124282983, "grad_norm": 0.738262951374054, "learning_rate": 9.06639127603761e-05, "loss": 0.0198, "step": 71340 }, { "epoch": 45.47482472912683, "grad_norm": 0.6661819219589233, "learning_rate": 9.066070547100016e-05, "loss": 0.0362, "step": 71350 }, { "epoch": 45.48119821542384, "grad_norm": 0.641676664352417, "learning_rate": 9.065749768755826e-05, "loss": 0.0227, "step": 71360 }, { "epoch": 45.48757170172084, "grad_norm": 0.4392363727092743, "learning_rate": 9.06542894100894e-05, "loss": 0.0323, "step": 71370 }, { "epoch": 45.49394518801785, "grad_norm": 0.3640446066856384, "learning_rate": 9.065108063863253e-05, "loss": 0.0222, "step": 71380 }, { "epoch": 45.50031867431485, "grad_norm": 0.9032019972801208, "learning_rate": 9.064787137322664e-05, "loss": 0.0336, "step": 71390 }, { "epoch": 45.506692160611856, "grad_norm": 0.5808727145195007, "learning_rate": 9.064466161391075e-05, "loss": 0.0468, "step": 71400 }, { "epoch": 45.513065646908856, "grad_norm": 0.37410789728164673, "learning_rate": 9.064145136072384e-05, "loss": 0.0339, "step": 71410 }, { "epoch": 45.519439133205864, "grad_norm": 0.5197312235832214, "learning_rate": 9.063824061370491e-05, "loss": 0.03, "step": 71420 }, { "epoch": 45.525812619502865, "grad_norm": 0.5331740379333496, "learning_rate": 9.063502937289301e-05, "loss": 0.0321, "step": 71430 }, { "epoch": 45.53218610579987, "grad_norm": 0.3824875056743622, "learning_rate": 9.063181763832713e-05, "loss": 0.0257, "step": 71440 }, { "epoch": 45.538559592096874, "grad_norm": 0.4603721797466278, "learning_rate": 9.06286054100463e-05, "loss": 0.0155, "step": 71450 }, { "epoch": 45.54493307839388, "grad_norm": 0.4769715666770935, "learning_rate": 9.062539268808955e-05, "loss": 0.0267, "step": 71460 }, { "epoch": 45.55130656469088, "grad_norm": 0.3128666877746582, "learning_rate": 9.062217947249592e-05, "loss": 0.0251, "step": 71470 }, { "epoch": 45.55768005098789, "grad_norm": 0.3092348575592041, "learning_rate": 9.061896576330447e-05, "loss": 0.0217, "step": 71480 }, { "epoch": 45.56405353728489, "grad_norm": 0.7825839519500732, "learning_rate": 9.061575156055421e-05, "loss": 0.0184, "step": 71490 }, { "epoch": 45.5704270235819, "grad_norm": 0.5529515147209167, "learning_rate": 9.061253686428426e-05, "loss": 0.0279, "step": 71500 }, { "epoch": 45.5768005098789, "grad_norm": 0.4902040958404541, "learning_rate": 9.060932167453362e-05, "loss": 0.0252, "step": 71510 }, { "epoch": 45.58317399617591, "grad_norm": 0.6579200029373169, "learning_rate": 9.060610599134137e-05, "loss": 0.037, "step": 71520 }, { "epoch": 45.58954748247291, "grad_norm": 0.48239102959632874, "learning_rate": 9.060288981474661e-05, "loss": 0.0261, "step": 71530 }, { "epoch": 45.59592096876992, "grad_norm": 0.35501357913017273, "learning_rate": 9.059967314478838e-05, "loss": 0.0398, "step": 71540 }, { "epoch": 45.602294455066925, "grad_norm": 0.34689027070999146, "learning_rate": 9.059645598150582e-05, "loss": 0.0254, "step": 71550 }, { "epoch": 45.608667941363926, "grad_norm": 0.3267618417739868, "learning_rate": 9.059323832493798e-05, "loss": 0.03, "step": 71560 }, { "epoch": 45.615041427660934, "grad_norm": 0.38036423921585083, "learning_rate": 9.059002017512397e-05, "loss": 0.021, "step": 71570 }, { "epoch": 45.621414913957935, "grad_norm": 0.5270289182662964, "learning_rate": 9.058680153210289e-05, "loss": 0.0259, "step": 71580 }, { "epoch": 45.62778840025494, "grad_norm": 0.6625394225120544, "learning_rate": 9.058358239591384e-05, "loss": 0.0256, "step": 71590 }, { "epoch": 45.634161886551944, "grad_norm": 0.479209303855896, "learning_rate": 9.058036276659595e-05, "loss": 0.0317, "step": 71600 }, { "epoch": 45.64053537284895, "grad_norm": 0.25597789883613586, "learning_rate": 9.057714264418836e-05, "loss": 0.0186, "step": 71610 }, { "epoch": 45.64690885914595, "grad_norm": 0.4768673777580261, "learning_rate": 9.057392202873014e-05, "loss": 0.0237, "step": 71620 }, { "epoch": 45.65328234544296, "grad_norm": 0.27209198474884033, "learning_rate": 9.057070092026048e-05, "loss": 0.0226, "step": 71630 }, { "epoch": 45.65965583173996, "grad_norm": 0.402756929397583, "learning_rate": 9.056747931881851e-05, "loss": 0.0202, "step": 71640 }, { "epoch": 45.66602931803697, "grad_norm": 0.26985570788383484, "learning_rate": 9.056425722444335e-05, "loss": 0.0218, "step": 71650 }, { "epoch": 45.67240280433397, "grad_norm": 0.3981522023677826, "learning_rate": 9.056103463717415e-05, "loss": 0.0305, "step": 71660 }, { "epoch": 45.67877629063098, "grad_norm": 0.2474077343940735, "learning_rate": 9.055781155705011e-05, "loss": 0.0187, "step": 71670 }, { "epoch": 45.68514977692798, "grad_norm": 0.47016680240631104, "learning_rate": 9.055458798411035e-05, "loss": 0.0383, "step": 71680 }, { "epoch": 45.69152326322499, "grad_norm": 0.29472893476486206, "learning_rate": 9.055136391839407e-05, "loss": 0.0363, "step": 71690 }, { "epoch": 45.69789674952199, "grad_norm": 0.4056607782840729, "learning_rate": 9.054813935994042e-05, "loss": 0.0277, "step": 71700 }, { "epoch": 45.704270235818996, "grad_norm": 0.6031860709190369, "learning_rate": 9.05449143087886e-05, "loss": 0.0327, "step": 71710 }, { "epoch": 45.710643722116, "grad_norm": 0.467719703912735, "learning_rate": 9.054168876497778e-05, "loss": 0.0201, "step": 71720 }, { "epoch": 45.717017208413004, "grad_norm": 0.5735359787940979, "learning_rate": 9.053846272854718e-05, "loss": 0.0228, "step": 71730 }, { "epoch": 45.723390694710005, "grad_norm": 0.3167482018470764, "learning_rate": 9.053523619953596e-05, "loss": 0.0268, "step": 71740 }, { "epoch": 45.72976418100701, "grad_norm": 0.6318914294242859, "learning_rate": 9.053200917798337e-05, "loss": 0.0375, "step": 71750 }, { "epoch": 45.736137667304014, "grad_norm": 0.8060351610183716, "learning_rate": 9.05287816639286e-05, "loss": 0.0273, "step": 71760 }, { "epoch": 45.74251115360102, "grad_norm": 0.5795696973800659, "learning_rate": 9.052555365741086e-05, "loss": 0.0251, "step": 71770 }, { "epoch": 45.74888463989802, "grad_norm": 0.28991326689720154, "learning_rate": 9.052232515846936e-05, "loss": 0.0249, "step": 71780 }, { "epoch": 45.75525812619503, "grad_norm": 0.357660710811615, "learning_rate": 9.051909616714339e-05, "loss": 0.0262, "step": 71790 }, { "epoch": 45.76163161249203, "grad_norm": 0.2897481620311737, "learning_rate": 9.051586668347211e-05, "loss": 0.0202, "step": 71800 }, { "epoch": 45.76800509878904, "grad_norm": 0.25315046310424805, "learning_rate": 9.051263670749482e-05, "loss": 0.0244, "step": 71810 }, { "epoch": 45.77437858508604, "grad_norm": 0.6223523616790771, "learning_rate": 9.050940623925073e-05, "loss": 0.0227, "step": 71820 }, { "epoch": 45.78075207138305, "grad_norm": 0.3144519031047821, "learning_rate": 9.050617527877911e-05, "loss": 0.0303, "step": 71830 }, { "epoch": 45.78712555768005, "grad_norm": 0.5086790919303894, "learning_rate": 9.050294382611924e-05, "loss": 0.0298, "step": 71840 }, { "epoch": 45.79349904397706, "grad_norm": 0.3743646442890167, "learning_rate": 9.049971188131032e-05, "loss": 0.0176, "step": 71850 }, { "epoch": 45.79987253027406, "grad_norm": 0.5054063200950623, "learning_rate": 9.04964794443917e-05, "loss": 0.0376, "step": 71860 }, { "epoch": 45.806246016571066, "grad_norm": 0.5286169052124023, "learning_rate": 9.04932465154026e-05, "loss": 0.0285, "step": 71870 }, { "epoch": 45.81261950286807, "grad_norm": 0.477058082818985, "learning_rate": 9.049001309438232e-05, "loss": 0.0259, "step": 71880 }, { "epoch": 45.818992989165075, "grad_norm": 0.368804007768631, "learning_rate": 9.048677918137017e-05, "loss": 0.032, "step": 71890 }, { "epoch": 45.825366475462076, "grad_norm": 0.5150417685508728, "learning_rate": 9.04835447764054e-05, "loss": 0.0378, "step": 71900 }, { "epoch": 45.831739961759084, "grad_norm": 0.3546220362186432, "learning_rate": 9.048030987952735e-05, "loss": 0.0188, "step": 71910 }, { "epoch": 45.838113448056085, "grad_norm": 0.5040913820266724, "learning_rate": 9.047707449077533e-05, "loss": 0.0244, "step": 71920 }, { "epoch": 45.84448693435309, "grad_norm": 0.40532317757606506, "learning_rate": 9.047383861018862e-05, "loss": 0.0248, "step": 71930 }, { "epoch": 45.85086042065009, "grad_norm": 0.3688770532608032, "learning_rate": 9.047060223780657e-05, "loss": 0.0189, "step": 71940 }, { "epoch": 45.8572339069471, "grad_norm": 0.7196228504180908, "learning_rate": 9.046736537366848e-05, "loss": 0.0338, "step": 71950 }, { "epoch": 45.8636073932441, "grad_norm": 0.5405769944190979, "learning_rate": 9.04641280178137e-05, "loss": 0.0298, "step": 71960 }, { "epoch": 45.86998087954111, "grad_norm": 0.28096744418144226, "learning_rate": 9.046089017028156e-05, "loss": 0.0328, "step": 71970 }, { "epoch": 45.87635436583811, "grad_norm": 0.5554543137550354, "learning_rate": 9.04576518311114e-05, "loss": 0.027, "step": 71980 }, { "epoch": 45.88272785213512, "grad_norm": 0.6111116409301758, "learning_rate": 9.045441300034257e-05, "loss": 0.027, "step": 71990 }, { "epoch": 45.88910133843212, "grad_norm": 0.4526421129703522, "learning_rate": 9.045117367801444e-05, "loss": 0.0198, "step": 72000 }, { "epoch": 45.89547482472913, "grad_norm": 0.6276025176048279, "learning_rate": 9.044793386416635e-05, "loss": 0.0257, "step": 72010 }, { "epoch": 45.90184831102613, "grad_norm": 0.46438947319984436, "learning_rate": 9.044469355883767e-05, "loss": 0.0362, "step": 72020 }, { "epoch": 45.90822179732314, "grad_norm": 0.5982154011726379, "learning_rate": 9.044145276206778e-05, "loss": 0.0321, "step": 72030 }, { "epoch": 45.91459528362014, "grad_norm": 0.5239140391349792, "learning_rate": 9.043821147389606e-05, "loss": 0.0257, "step": 72040 }, { "epoch": 45.920968769917145, "grad_norm": 0.3900355100631714, "learning_rate": 9.043496969436188e-05, "loss": 0.0224, "step": 72050 }, { "epoch": 45.927342256214146, "grad_norm": 0.6715278029441833, "learning_rate": 9.043172742350465e-05, "loss": 0.0423, "step": 72060 }, { "epoch": 45.933715742511154, "grad_norm": 0.5185863971710205, "learning_rate": 9.042848466136376e-05, "loss": 0.0308, "step": 72070 }, { "epoch": 45.940089228808155, "grad_norm": 0.296200156211853, "learning_rate": 9.04252414079786e-05, "loss": 0.0223, "step": 72080 }, { "epoch": 45.94646271510516, "grad_norm": 0.618979811668396, "learning_rate": 9.04219976633886e-05, "loss": 0.0189, "step": 72090 }, { "epoch": 45.952836201402164, "grad_norm": 0.5431245565414429, "learning_rate": 9.041875342763315e-05, "loss": 0.0246, "step": 72100 }, { "epoch": 45.95920968769917, "grad_norm": 0.4358237087726593, "learning_rate": 9.041550870075171e-05, "loss": 0.0358, "step": 72110 }, { "epoch": 45.96558317399617, "grad_norm": 0.6650050282478333, "learning_rate": 9.041226348278365e-05, "loss": 0.0344, "step": 72120 }, { "epoch": 45.97195666029318, "grad_norm": 0.4044601321220398, "learning_rate": 9.040901777376844e-05, "loss": 0.0322, "step": 72130 }, { "epoch": 45.97833014659018, "grad_norm": 0.30342942476272583, "learning_rate": 9.040577157374553e-05, "loss": 0.0232, "step": 72140 }, { "epoch": 45.98470363288719, "grad_norm": 0.5635536313056946, "learning_rate": 9.040252488275433e-05, "loss": 0.0277, "step": 72150 }, { "epoch": 45.99107711918419, "grad_norm": 0.35416656732559204, "learning_rate": 9.039927770083432e-05, "loss": 0.0232, "step": 72160 }, { "epoch": 45.9974506054812, "grad_norm": 0.4141038656234741, "learning_rate": 9.039603002802493e-05, "loss": 0.0319, "step": 72170 }, { "epoch": 46.003824091778206, "grad_norm": 0.7099750638008118, "learning_rate": 9.039278186436563e-05, "loss": 0.023, "step": 72180 }, { "epoch": 46.01019757807521, "grad_norm": 0.33698150515556335, "learning_rate": 9.038953320989592e-05, "loss": 0.025, "step": 72190 }, { "epoch": 46.016571064372215, "grad_norm": 0.5433964133262634, "learning_rate": 9.038628406465522e-05, "loss": 0.031, "step": 72200 }, { "epoch": 46.022944550669216, "grad_norm": 0.48015496134757996, "learning_rate": 9.038303442868304e-05, "loss": 0.0342, "step": 72210 }, { "epoch": 46.029318036966224, "grad_norm": 0.5014916658401489, "learning_rate": 9.037978430201888e-05, "loss": 0.0276, "step": 72220 }, { "epoch": 46.035691523263225, "grad_norm": 0.4410669505596161, "learning_rate": 9.03765336847022e-05, "loss": 0.0177, "step": 72230 }, { "epoch": 46.04206500956023, "grad_norm": 0.6270255446434021, "learning_rate": 9.037328257677252e-05, "loss": 0.0373, "step": 72240 }, { "epoch": 46.04843849585723, "grad_norm": 0.411275178194046, "learning_rate": 9.037003097826935e-05, "loss": 0.0276, "step": 72250 }, { "epoch": 46.05481198215424, "grad_norm": 0.5081455111503601, "learning_rate": 9.036677888923217e-05, "loss": 0.0216, "step": 72260 }, { "epoch": 46.06118546845124, "grad_norm": 0.5085729360580444, "learning_rate": 9.036352630970052e-05, "loss": 0.0186, "step": 72270 }, { "epoch": 46.06755895474825, "grad_norm": 0.6005397439002991, "learning_rate": 9.036027323971392e-05, "loss": 0.023, "step": 72280 }, { "epoch": 46.07393244104525, "grad_norm": 0.3970995843410492, "learning_rate": 9.035701967931188e-05, "loss": 0.0207, "step": 72290 }, { "epoch": 46.08030592734226, "grad_norm": 0.2823580205440521, "learning_rate": 9.035376562853397e-05, "loss": 0.0207, "step": 72300 }, { "epoch": 46.08667941363926, "grad_norm": 0.5190545320510864, "learning_rate": 9.035051108741969e-05, "loss": 0.0198, "step": 72310 }, { "epoch": 46.09305289993627, "grad_norm": 0.29820096492767334, "learning_rate": 9.03472560560086e-05, "loss": 0.0252, "step": 72320 }, { "epoch": 46.09942638623327, "grad_norm": 0.25740161538124084, "learning_rate": 9.034400053434027e-05, "loss": 0.0245, "step": 72330 }, { "epoch": 46.10579987253028, "grad_norm": 0.5008454918861389, "learning_rate": 9.034074452245423e-05, "loss": 0.0297, "step": 72340 }, { "epoch": 46.11217335882728, "grad_norm": 0.3892051875591278, "learning_rate": 9.033748802039006e-05, "loss": 0.0295, "step": 72350 }, { "epoch": 46.118546845124285, "grad_norm": 0.44915884733200073, "learning_rate": 9.033423102818734e-05, "loss": 0.0151, "step": 72360 }, { "epoch": 46.124920331421286, "grad_norm": 0.46310245990753174, "learning_rate": 9.033097354588561e-05, "loss": 0.0167, "step": 72370 }, { "epoch": 46.131293817718294, "grad_norm": 0.48191672563552856, "learning_rate": 9.032771557352448e-05, "loss": 0.0484, "step": 72380 }, { "epoch": 46.137667304015295, "grad_norm": 0.32372158765792847, "learning_rate": 9.032445711114353e-05, "loss": 0.0157, "step": 72390 }, { "epoch": 46.1440407903123, "grad_norm": 0.331342488527298, "learning_rate": 9.032119815878236e-05, "loss": 0.0165, "step": 72400 }, { "epoch": 46.150414276609304, "grad_norm": 0.2668777406215668, "learning_rate": 9.031793871648056e-05, "loss": 0.0193, "step": 72410 }, { "epoch": 46.15678776290631, "grad_norm": 0.3436919152736664, "learning_rate": 9.031467878427773e-05, "loss": 0.0239, "step": 72420 }, { "epoch": 46.16316124920331, "grad_norm": 0.3849315941333771, "learning_rate": 9.03114183622135e-05, "loss": 0.0238, "step": 72430 }, { "epoch": 46.16953473550032, "grad_norm": 0.34982162714004517, "learning_rate": 9.030815745032746e-05, "loss": 0.021, "step": 72440 }, { "epoch": 46.17590822179732, "grad_norm": 0.4643505811691284, "learning_rate": 9.030489604865927e-05, "loss": 0.0213, "step": 72450 }, { "epoch": 46.18228170809433, "grad_norm": 0.46418124437332153, "learning_rate": 9.030163415724853e-05, "loss": 0.024, "step": 72460 }, { "epoch": 46.18865519439133, "grad_norm": 0.31765952706336975, "learning_rate": 9.02983717761349e-05, "loss": 0.0141, "step": 72470 }, { "epoch": 46.19502868068834, "grad_norm": 0.4196479022502899, "learning_rate": 9.029510890535797e-05, "loss": 0.0214, "step": 72480 }, { "epoch": 46.20140216698534, "grad_norm": 0.6092623472213745, "learning_rate": 9.029184554495745e-05, "loss": 0.0339, "step": 72490 }, { "epoch": 46.20777565328235, "grad_norm": 0.35390105843544006, "learning_rate": 9.028858169497295e-05, "loss": 0.0174, "step": 72500 }, { "epoch": 46.21414913957935, "grad_norm": 0.3529788851737976, "learning_rate": 9.028531735544415e-05, "loss": 0.0151, "step": 72510 }, { "epoch": 46.220522625876356, "grad_norm": 1.114708423614502, "learning_rate": 9.028205252641071e-05, "loss": 0.02, "step": 72520 }, { "epoch": 46.22689611217336, "grad_norm": 0.17391476035118103, "learning_rate": 9.027878720791229e-05, "loss": 0.0238, "step": 72530 }, { "epoch": 46.233269598470365, "grad_norm": 0.49776044487953186, "learning_rate": 9.02755213999886e-05, "loss": 0.018, "step": 72540 }, { "epoch": 46.239643084767366, "grad_norm": 0.4329456388950348, "learning_rate": 9.027225510267928e-05, "loss": 0.0263, "step": 72550 }, { "epoch": 46.246016571064374, "grad_norm": 0.4147505462169647, "learning_rate": 9.026898831602404e-05, "loss": 0.025, "step": 72560 }, { "epoch": 46.252390057361374, "grad_norm": 0.3755078911781311, "learning_rate": 9.026572104006257e-05, "loss": 0.0207, "step": 72570 }, { "epoch": 46.25876354365838, "grad_norm": 0.43202298879623413, "learning_rate": 9.026245327483456e-05, "loss": 0.0335, "step": 72580 }, { "epoch": 46.26513702995538, "grad_norm": 0.6159024238586426, "learning_rate": 9.025918502037975e-05, "loss": 0.0215, "step": 72590 }, { "epoch": 46.27151051625239, "grad_norm": 0.8211655616760254, "learning_rate": 9.025591627673782e-05, "loss": 0.0241, "step": 72600 }, { "epoch": 46.27788400254939, "grad_norm": 0.27816665172576904, "learning_rate": 9.025264704394851e-05, "loss": 0.0131, "step": 72610 }, { "epoch": 46.2842574888464, "grad_norm": 0.3577656149864197, "learning_rate": 9.024937732205152e-05, "loss": 0.0228, "step": 72620 }, { "epoch": 46.2906309751434, "grad_norm": 0.520428478717804, "learning_rate": 9.024610711108659e-05, "loss": 0.0254, "step": 72630 }, { "epoch": 46.29700446144041, "grad_norm": 0.2858421802520752, "learning_rate": 9.024283641109348e-05, "loss": 0.0296, "step": 72640 }, { "epoch": 46.30337794773741, "grad_norm": 0.5425296425819397, "learning_rate": 9.02395652221119e-05, "loss": 0.0314, "step": 72650 }, { "epoch": 46.30975143403442, "grad_norm": 0.37338337302207947, "learning_rate": 9.023629354418162e-05, "loss": 0.028, "step": 72660 }, { "epoch": 46.31612492033142, "grad_norm": 0.5902696251869202, "learning_rate": 9.023302137734238e-05, "loss": 0.0196, "step": 72670 }, { "epoch": 46.322498406628426, "grad_norm": 2.0285696983337402, "learning_rate": 9.022974872163392e-05, "loss": 0.0223, "step": 72680 }, { "epoch": 46.32887189292543, "grad_norm": 0.36439642310142517, "learning_rate": 9.022647557709604e-05, "loss": 0.0292, "step": 72690 }, { "epoch": 46.335245379222435, "grad_norm": 0.39056140184402466, "learning_rate": 9.022320194376853e-05, "loss": 0.0164, "step": 72700 }, { "epoch": 46.341618865519436, "grad_norm": 0.6784292459487915, "learning_rate": 9.021992782169112e-05, "loss": 0.0287, "step": 72710 }, { "epoch": 46.347992351816444, "grad_norm": 0.7538209557533264, "learning_rate": 9.021665321090361e-05, "loss": 0.0331, "step": 72720 }, { "epoch": 46.354365838113445, "grad_norm": 0.6099622249603271, "learning_rate": 9.02133781114458e-05, "loss": 0.0214, "step": 72730 }, { "epoch": 46.36073932441045, "grad_norm": 0.4085283577442169, "learning_rate": 9.021010252335747e-05, "loss": 0.026, "step": 72740 }, { "epoch": 46.367112810707454, "grad_norm": 0.461270272731781, "learning_rate": 9.020682644667844e-05, "loss": 0.0303, "step": 72750 }, { "epoch": 46.37348629700446, "grad_norm": 0.4061415195465088, "learning_rate": 9.02035498814485e-05, "loss": 0.0372, "step": 72760 }, { "epoch": 46.37985978330146, "grad_norm": 0.3742014467716217, "learning_rate": 9.020027282770747e-05, "loss": 0.0224, "step": 72770 }, { "epoch": 46.38623326959847, "grad_norm": 1.1309475898742676, "learning_rate": 9.019699528549518e-05, "loss": 0.0326, "step": 72780 }, { "epoch": 46.39260675589547, "grad_norm": 0.3065764009952545, "learning_rate": 9.019371725485144e-05, "loss": 0.0166, "step": 72790 }, { "epoch": 46.39898024219248, "grad_norm": 0.5502302646636963, "learning_rate": 9.019043873581607e-05, "loss": 0.0212, "step": 72800 }, { "epoch": 46.40535372848949, "grad_norm": 0.4120645523071289, "learning_rate": 9.018715972842895e-05, "loss": 0.0186, "step": 72810 }, { "epoch": 46.41172721478649, "grad_norm": 0.3674103915691376, "learning_rate": 9.018388023272987e-05, "loss": 0.0168, "step": 72820 }, { "epoch": 46.418100701083496, "grad_norm": 0.4591769874095917, "learning_rate": 9.018060024875872e-05, "loss": 0.0156, "step": 72830 }, { "epoch": 46.4244741873805, "grad_norm": 0.34724441170692444, "learning_rate": 9.017731977655535e-05, "loss": 0.0143, "step": 72840 }, { "epoch": 46.430847673677505, "grad_norm": 0.2619026005268097, "learning_rate": 9.017403881615959e-05, "loss": 0.0211, "step": 72850 }, { "epoch": 46.437221159974506, "grad_norm": 0.5359959602355957, "learning_rate": 9.017075736761133e-05, "loss": 0.0222, "step": 72860 }, { "epoch": 46.443594646271514, "grad_norm": 0.3027917146682739, "learning_rate": 9.016747543095046e-05, "loss": 0.0235, "step": 72870 }, { "epoch": 46.449968132568515, "grad_norm": 0.4403434097766876, "learning_rate": 9.016419300621683e-05, "loss": 0.0201, "step": 72880 }, { "epoch": 46.45634161886552, "grad_norm": 0.30479562282562256, "learning_rate": 9.016091009345033e-05, "loss": 0.0264, "step": 72890 }, { "epoch": 46.46271510516252, "grad_norm": 0.3653997480869293, "learning_rate": 9.015762669269085e-05, "loss": 0.0283, "step": 72900 }, { "epoch": 46.46908859145953, "grad_norm": 0.33493807911872864, "learning_rate": 9.01543428039783e-05, "loss": 0.0193, "step": 72910 }, { "epoch": 46.47546207775653, "grad_norm": 0.6176809072494507, "learning_rate": 9.015105842735257e-05, "loss": 0.031, "step": 72920 }, { "epoch": 46.48183556405354, "grad_norm": 0.4378930628299713, "learning_rate": 9.014777356285356e-05, "loss": 0.0336, "step": 72930 }, { "epoch": 46.48820905035054, "grad_norm": 0.5343567728996277, "learning_rate": 9.01444882105212e-05, "loss": 0.0213, "step": 72940 }, { "epoch": 46.49458253664755, "grad_norm": 0.25798654556274414, "learning_rate": 9.01412023703954e-05, "loss": 0.0214, "step": 72950 }, { "epoch": 46.50095602294455, "grad_norm": 0.5395998954772949, "learning_rate": 9.013791604251611e-05, "loss": 0.0312, "step": 72960 }, { "epoch": 46.50732950924156, "grad_norm": 0.2777687907218933, "learning_rate": 9.013462922692324e-05, "loss": 0.0202, "step": 72970 }, { "epoch": 46.51370299553856, "grad_norm": 0.5118763446807861, "learning_rate": 9.013134192365673e-05, "loss": 0.028, "step": 72980 }, { "epoch": 46.52007648183557, "grad_norm": 0.5927591919898987, "learning_rate": 9.01280541327565e-05, "loss": 0.0354, "step": 72990 }, { "epoch": 46.52644996813257, "grad_norm": 0.4125881493091583, "learning_rate": 9.012476585426257e-05, "loss": 0.0235, "step": 73000 }, { "epoch": 46.532823454429575, "grad_norm": 0.4646165370941162, "learning_rate": 9.012147708821483e-05, "loss": 0.0275, "step": 73010 }, { "epoch": 46.539196940726576, "grad_norm": 0.28710806369781494, "learning_rate": 9.011818783465326e-05, "loss": 0.0343, "step": 73020 }, { "epoch": 46.545570427023584, "grad_norm": 0.23979824781417847, "learning_rate": 9.011489809361782e-05, "loss": 0.0205, "step": 73030 }, { "epoch": 46.551943913320585, "grad_norm": 0.3615192770957947, "learning_rate": 9.011160786514852e-05, "loss": 0.0273, "step": 73040 }, { "epoch": 46.55831739961759, "grad_norm": 0.7982115745544434, "learning_rate": 9.01083171492853e-05, "loss": 0.0183, "step": 73050 }, { "epoch": 46.564690885914594, "grad_norm": 0.5820039510726929, "learning_rate": 9.010502594606815e-05, "loss": 0.0147, "step": 73060 }, { "epoch": 46.5710643722116, "grad_norm": 0.314201295375824, "learning_rate": 9.010173425553708e-05, "loss": 0.0279, "step": 73070 }, { "epoch": 46.5774378585086, "grad_norm": 0.4855760335922241, "learning_rate": 9.009844207773206e-05, "loss": 0.0384, "step": 73080 }, { "epoch": 46.58381134480561, "grad_norm": 0.3538670837879181, "learning_rate": 9.009514941269313e-05, "loss": 0.0195, "step": 73090 }, { "epoch": 46.59018483110261, "grad_norm": 0.5878674983978271, "learning_rate": 9.009185626046026e-05, "loss": 0.0246, "step": 73100 }, { "epoch": 46.59655831739962, "grad_norm": 0.4360893964767456, "learning_rate": 9.008856262107349e-05, "loss": 0.0267, "step": 73110 }, { "epoch": 46.60293180369662, "grad_norm": 0.538303017616272, "learning_rate": 9.008526849457284e-05, "loss": 0.0285, "step": 73120 }, { "epoch": 46.60930528999363, "grad_norm": 0.3162635862827301, "learning_rate": 9.008197388099833e-05, "loss": 0.0131, "step": 73130 }, { "epoch": 46.61567877629063, "grad_norm": 0.4767117500305176, "learning_rate": 9.007867878038998e-05, "loss": 0.0197, "step": 73140 }, { "epoch": 46.62205226258764, "grad_norm": 0.6441197991371155, "learning_rate": 9.007538319278785e-05, "loss": 0.0348, "step": 73150 }, { "epoch": 46.62842574888464, "grad_norm": 0.5303916335105896, "learning_rate": 9.007208711823198e-05, "loss": 0.0249, "step": 73160 }, { "epoch": 46.634799235181646, "grad_norm": 0.4124069809913635, "learning_rate": 9.006879055676243e-05, "loss": 0.0254, "step": 73170 }, { "epoch": 46.64117272147865, "grad_norm": 0.7156108021736145, "learning_rate": 9.006549350841923e-05, "loss": 0.029, "step": 73180 }, { "epoch": 46.647546207775655, "grad_norm": 0.27211278676986694, "learning_rate": 9.006219597324245e-05, "loss": 0.0235, "step": 73190 }, { "epoch": 46.653919694072655, "grad_norm": 0.32286497950553894, "learning_rate": 9.005889795127219e-05, "loss": 0.0278, "step": 73200 }, { "epoch": 46.66029318036966, "grad_norm": 0.5292196273803711, "learning_rate": 9.005559944254846e-05, "loss": 0.0205, "step": 73210 }, { "epoch": 46.666666666666664, "grad_norm": 0.40824341773986816, "learning_rate": 9.00523004471114e-05, "loss": 0.0306, "step": 73220 }, { "epoch": 46.67304015296367, "grad_norm": 0.5618685483932495, "learning_rate": 9.004900096500107e-05, "loss": 0.0186, "step": 73230 }, { "epoch": 46.67941363926067, "grad_norm": 0.42674723267555237, "learning_rate": 9.004570099625758e-05, "loss": 0.0274, "step": 73240 }, { "epoch": 46.68578712555768, "grad_norm": 0.9148685336112976, "learning_rate": 9.004240054092099e-05, "loss": 0.0227, "step": 73250 }, { "epoch": 46.69216061185468, "grad_norm": 0.6640389561653137, "learning_rate": 9.003909959903144e-05, "loss": 0.03, "step": 73260 }, { "epoch": 46.69853409815169, "grad_norm": 0.7129845023155212, "learning_rate": 9.003579817062901e-05, "loss": 0.0194, "step": 73270 }, { "epoch": 46.70490758444869, "grad_norm": 0.270382821559906, "learning_rate": 9.003249625575385e-05, "loss": 0.0173, "step": 73280 }, { "epoch": 46.7112810707457, "grad_norm": 0.33116793632507324, "learning_rate": 9.002919385444607e-05, "loss": 0.0237, "step": 73290 }, { "epoch": 46.7176545570427, "grad_norm": 0.5110591053962708, "learning_rate": 9.002589096674578e-05, "loss": 0.017, "step": 73300 }, { "epoch": 46.72402804333971, "grad_norm": 0.5841027498245239, "learning_rate": 9.002258759269312e-05, "loss": 0.0234, "step": 73310 }, { "epoch": 46.73040152963671, "grad_norm": 0.29364582896232605, "learning_rate": 9.001928373232825e-05, "loss": 0.0204, "step": 73320 }, { "epoch": 46.736775015933716, "grad_norm": 0.49264290928840637, "learning_rate": 9.001597938569128e-05, "loss": 0.0207, "step": 73330 }, { "epoch": 46.74314850223072, "grad_norm": 0.24101972579956055, "learning_rate": 9.001267455282239e-05, "loss": 0.0271, "step": 73340 }, { "epoch": 46.749521988527725, "grad_norm": 0.4318997263908386, "learning_rate": 9.000936923376171e-05, "loss": 0.0201, "step": 73350 }, { "epoch": 46.755895474824726, "grad_norm": 0.47395551204681396, "learning_rate": 9.000606342854942e-05, "loss": 0.0214, "step": 73360 }, { "epoch": 46.762268961121734, "grad_norm": 0.3183608949184418, "learning_rate": 9.00027571372257e-05, "loss": 0.0261, "step": 73370 }, { "epoch": 46.768642447418735, "grad_norm": 0.4427333474159241, "learning_rate": 8.999945035983072e-05, "loss": 0.0163, "step": 73380 }, { "epoch": 46.77501593371574, "grad_norm": 0.43970638513565063, "learning_rate": 8.999614309640464e-05, "loss": 0.0278, "step": 73390 }, { "epoch": 46.781389420012744, "grad_norm": 0.38535189628601074, "learning_rate": 8.999283534698766e-05, "loss": 0.025, "step": 73400 }, { "epoch": 46.78776290630975, "grad_norm": 0.24869051575660706, "learning_rate": 8.998952711161996e-05, "loss": 0.0194, "step": 73410 }, { "epoch": 46.79413639260676, "grad_norm": 0.5560080409049988, "learning_rate": 8.998621839034177e-05, "loss": 0.0184, "step": 73420 }, { "epoch": 46.80050987890376, "grad_norm": 0.35762229561805725, "learning_rate": 8.998290918319327e-05, "loss": 0.0246, "step": 73430 }, { "epoch": 46.80688336520077, "grad_norm": 0.3314529061317444, "learning_rate": 8.997959949021466e-05, "loss": 0.0218, "step": 73440 }, { "epoch": 46.81325685149777, "grad_norm": 0.6552419066429138, "learning_rate": 8.997628931144619e-05, "loss": 0.0376, "step": 73450 }, { "epoch": 46.81963033779478, "grad_norm": 0.5220251679420471, "learning_rate": 8.997297864692803e-05, "loss": 0.0313, "step": 73460 }, { "epoch": 46.82600382409178, "grad_norm": 0.3886709213256836, "learning_rate": 8.996966749670046e-05, "loss": 0.0232, "step": 73470 }, { "epoch": 46.832377310388786, "grad_norm": 0.462576299905777, "learning_rate": 8.996635586080369e-05, "loss": 0.021, "step": 73480 }, { "epoch": 46.83875079668579, "grad_norm": 0.43018224835395813, "learning_rate": 8.996304373927795e-05, "loss": 0.025, "step": 73490 }, { "epoch": 46.845124282982795, "grad_norm": 0.3941669762134552, "learning_rate": 8.995973113216352e-05, "loss": 0.0368, "step": 73500 }, { "epoch": 46.851497769279796, "grad_norm": 0.5081268548965454, "learning_rate": 8.995641803950062e-05, "loss": 0.0173, "step": 73510 }, { "epoch": 46.8578712555768, "grad_norm": 0.3971559405326843, "learning_rate": 8.99531044613295e-05, "loss": 0.026, "step": 73520 }, { "epoch": 46.864244741873804, "grad_norm": 0.3755420446395874, "learning_rate": 8.994979039769046e-05, "loss": 0.025, "step": 73530 }, { "epoch": 46.87061822817081, "grad_norm": 0.3128087520599365, "learning_rate": 8.994647584862374e-05, "loss": 0.0288, "step": 73540 }, { "epoch": 46.87699171446781, "grad_norm": 0.4109993577003479, "learning_rate": 8.994316081416961e-05, "loss": 0.0206, "step": 73550 }, { "epoch": 46.88336520076482, "grad_norm": 0.6940052509307861, "learning_rate": 8.993984529436839e-05, "loss": 0.0225, "step": 73560 }, { "epoch": 46.88973868706182, "grad_norm": 0.5909662246704102, "learning_rate": 8.99365292892603e-05, "loss": 0.0202, "step": 73570 }, { "epoch": 46.89611217335883, "grad_norm": 0.358241468667984, "learning_rate": 8.993321279888571e-05, "loss": 0.0206, "step": 73580 }, { "epoch": 46.90248565965583, "grad_norm": 0.42378413677215576, "learning_rate": 8.992989582328485e-05, "loss": 0.0473, "step": 73590 }, { "epoch": 46.90885914595284, "grad_norm": 0.4851742386817932, "learning_rate": 8.992657836249809e-05, "loss": 0.0234, "step": 73600 }, { "epoch": 46.91523263224984, "grad_norm": 0.42154306173324585, "learning_rate": 8.992326041656568e-05, "loss": 0.019, "step": 73610 }, { "epoch": 46.92160611854685, "grad_norm": 0.426662802696228, "learning_rate": 8.991994198552798e-05, "loss": 0.0354, "step": 73620 }, { "epoch": 46.92797960484385, "grad_norm": 0.3295361399650574, "learning_rate": 8.991662306942529e-05, "loss": 0.0229, "step": 73630 }, { "epoch": 46.934353091140856, "grad_norm": 0.3389422297477722, "learning_rate": 8.991330366829794e-05, "loss": 0.0229, "step": 73640 }, { "epoch": 46.94072657743786, "grad_norm": 0.6542728543281555, "learning_rate": 8.990998378218627e-05, "loss": 0.0348, "step": 73650 }, { "epoch": 46.947100063734865, "grad_norm": 0.6729058027267456, "learning_rate": 8.990666341113061e-05, "loss": 0.0239, "step": 73660 }, { "epoch": 46.953473550031866, "grad_norm": 0.9905174970626831, "learning_rate": 8.99033425551713e-05, "loss": 0.0304, "step": 73670 }, { "epoch": 46.959847036328874, "grad_norm": 0.33427032828330994, "learning_rate": 8.990002121434874e-05, "loss": 0.0286, "step": 73680 }, { "epoch": 46.966220522625875, "grad_norm": 0.5040664076805115, "learning_rate": 8.989669938870321e-05, "loss": 0.0321, "step": 73690 }, { "epoch": 46.97259400892288, "grad_norm": 0.49050062894821167, "learning_rate": 8.989337707827513e-05, "loss": 0.0169, "step": 73700 }, { "epoch": 46.978967495219884, "grad_norm": 0.28033921122550964, "learning_rate": 8.989005428310483e-05, "loss": 0.028, "step": 73710 }, { "epoch": 46.98534098151689, "grad_norm": 0.56724613904953, "learning_rate": 8.988673100323274e-05, "loss": 0.03, "step": 73720 }, { "epoch": 46.99171446781389, "grad_norm": 0.3086795508861542, "learning_rate": 8.988340723869922e-05, "loss": 0.0205, "step": 73730 }, { "epoch": 46.9980879541109, "grad_norm": 0.30637165904045105, "learning_rate": 8.988008298954461e-05, "loss": 0.0235, "step": 73740 }, { "epoch": 47.0044614404079, "grad_norm": 0.5054007768630981, "learning_rate": 8.987675825580936e-05, "loss": 0.0222, "step": 73750 }, { "epoch": 47.01083492670491, "grad_norm": 0.7684495449066162, "learning_rate": 8.987343303753384e-05, "loss": 0.0199, "step": 73760 }, { "epoch": 47.01720841300191, "grad_norm": 0.3937392830848694, "learning_rate": 8.987010733475846e-05, "loss": 0.0239, "step": 73770 }, { "epoch": 47.02358189929892, "grad_norm": 0.3889522850513458, "learning_rate": 8.986678114752363e-05, "loss": 0.0357, "step": 73780 }, { "epoch": 47.02995538559592, "grad_norm": 0.7441005706787109, "learning_rate": 8.986345447586978e-05, "loss": 0.0234, "step": 73790 }, { "epoch": 47.03632887189293, "grad_norm": 0.3977629542350769, "learning_rate": 8.98601273198373e-05, "loss": 0.0267, "step": 73800 }, { "epoch": 47.04270235818993, "grad_norm": 0.453766793012619, "learning_rate": 8.985679967946666e-05, "loss": 0.019, "step": 73810 }, { "epoch": 47.049075844486936, "grad_norm": 0.5155267715454102, "learning_rate": 8.985347155479827e-05, "loss": 0.0257, "step": 73820 }, { "epoch": 47.05544933078394, "grad_norm": 0.4805154800415039, "learning_rate": 8.985014294587258e-05, "loss": 0.0259, "step": 73830 }, { "epoch": 47.061822817080945, "grad_norm": 0.4487517476081848, "learning_rate": 8.984681385273001e-05, "loss": 0.0195, "step": 73840 }, { "epoch": 47.068196303377945, "grad_norm": 0.4629252254962921, "learning_rate": 8.984348427541103e-05, "loss": 0.0268, "step": 73850 }, { "epoch": 47.07456978967495, "grad_norm": 0.44507139921188354, "learning_rate": 8.98401542139561e-05, "loss": 0.0216, "step": 73860 }, { "epoch": 47.080943275971954, "grad_norm": 0.6372135281562805, "learning_rate": 8.98368236684057e-05, "loss": 0.0259, "step": 73870 }, { "epoch": 47.08731676226896, "grad_norm": 0.854559063911438, "learning_rate": 8.983349263880026e-05, "loss": 0.0266, "step": 73880 }, { "epoch": 47.09369024856596, "grad_norm": 0.6928684711456299, "learning_rate": 8.983016112518028e-05, "loss": 0.0198, "step": 73890 }, { "epoch": 47.10006373486297, "grad_norm": 0.24849557876586914, "learning_rate": 8.982682912758624e-05, "loss": 0.02, "step": 73900 }, { "epoch": 47.10643722115997, "grad_norm": 0.32001641392707825, "learning_rate": 8.982349664605863e-05, "loss": 0.0146, "step": 73910 }, { "epoch": 47.11281070745698, "grad_norm": 0.39378491044044495, "learning_rate": 8.982016368063792e-05, "loss": 0.0176, "step": 73920 }, { "epoch": 47.11918419375398, "grad_norm": 0.3528071343898773, "learning_rate": 8.981683023136462e-05, "loss": 0.0193, "step": 73930 }, { "epoch": 47.12555768005099, "grad_norm": 0.3132566809654236, "learning_rate": 8.981349629827926e-05, "loss": 0.0275, "step": 73940 }, { "epoch": 47.13193116634799, "grad_norm": 0.49145981669425964, "learning_rate": 8.981016188142233e-05, "loss": 0.0259, "step": 73950 }, { "epoch": 47.138304652645, "grad_norm": 0.4064086377620697, "learning_rate": 8.980682698083433e-05, "loss": 0.0282, "step": 73960 }, { "epoch": 47.144678138942, "grad_norm": 0.7351500391960144, "learning_rate": 8.980349159655581e-05, "loss": 0.0153, "step": 73970 }, { "epoch": 47.151051625239006, "grad_norm": 0.5894414782524109, "learning_rate": 8.980015572862727e-05, "loss": 0.0253, "step": 73980 }, { "epoch": 47.15742511153601, "grad_norm": 0.280489057302475, "learning_rate": 8.979681937708928e-05, "loss": 0.0207, "step": 73990 }, { "epoch": 47.163798597833015, "grad_norm": 0.42340391874313354, "learning_rate": 8.979348254198235e-05, "loss": 0.0257, "step": 74000 }, { "epoch": 47.170172084130016, "grad_norm": 0.3818224370479584, "learning_rate": 8.979014522334704e-05, "loss": 0.0225, "step": 74010 }, { "epoch": 47.176545570427024, "grad_norm": 0.43937116861343384, "learning_rate": 8.97868074212239e-05, "loss": 0.0207, "step": 74020 }, { "epoch": 47.182919056724025, "grad_norm": 0.3366369307041168, "learning_rate": 8.978346913565348e-05, "loss": 0.023, "step": 74030 }, { "epoch": 47.18929254302103, "grad_norm": 0.6811960339546204, "learning_rate": 8.978013036667634e-05, "loss": 0.0307, "step": 74040 }, { "epoch": 47.19566602931803, "grad_norm": 0.2967533767223358, "learning_rate": 8.977679111433308e-05, "loss": 0.0168, "step": 74050 }, { "epoch": 47.20203951561504, "grad_norm": 0.4854520261287689, "learning_rate": 8.977345137866422e-05, "loss": 0.0433, "step": 74060 }, { "epoch": 47.20841300191205, "grad_norm": 0.31170758605003357, "learning_rate": 8.97701111597104e-05, "loss": 0.0214, "step": 74070 }, { "epoch": 47.21478648820905, "grad_norm": 0.8534892797470093, "learning_rate": 8.976677045751217e-05, "loss": 0.0259, "step": 74080 }, { "epoch": 47.22115997450606, "grad_norm": 0.5970094203948975, "learning_rate": 8.976342927211014e-05, "loss": 0.0275, "step": 74090 }, { "epoch": 47.22753346080306, "grad_norm": 0.7391400337219238, "learning_rate": 8.976008760354488e-05, "loss": 0.0322, "step": 74100 }, { "epoch": 47.23390694710007, "grad_norm": 0.38938653469085693, "learning_rate": 8.975674545185703e-05, "loss": 0.0242, "step": 74110 }, { "epoch": 47.24028043339707, "grad_norm": 0.5006521344184875, "learning_rate": 8.97534028170872e-05, "loss": 0.0191, "step": 74120 }, { "epoch": 47.246653919694076, "grad_norm": 0.40289443731307983, "learning_rate": 8.975005969927599e-05, "loss": 0.028, "step": 74130 }, { "epoch": 47.25302740599108, "grad_norm": 0.513492226600647, "learning_rate": 8.974671609846402e-05, "loss": 0.031, "step": 74140 }, { "epoch": 47.259400892288085, "grad_norm": 0.6376273036003113, "learning_rate": 8.97433720146919e-05, "loss": 0.0431, "step": 74150 }, { "epoch": 47.265774378585085, "grad_norm": 0.5154752731323242, "learning_rate": 8.974002744800033e-05, "loss": 0.0347, "step": 74160 }, { "epoch": 47.27214786488209, "grad_norm": 0.5322593450546265, "learning_rate": 8.973668239842986e-05, "loss": 0.0194, "step": 74170 }, { "epoch": 47.278521351179094, "grad_norm": 0.7278199195861816, "learning_rate": 8.97333368660212e-05, "loss": 0.0302, "step": 74180 }, { "epoch": 47.2848948374761, "grad_norm": 0.4009280502796173, "learning_rate": 8.972999085081498e-05, "loss": 0.0223, "step": 74190 }, { "epoch": 47.2912683237731, "grad_norm": 0.8206236362457275, "learning_rate": 8.972664435285188e-05, "loss": 0.0391, "step": 74200 }, { "epoch": 47.29764181007011, "grad_norm": 0.5483067631721497, "learning_rate": 8.972329737217251e-05, "loss": 0.0162, "step": 74210 }, { "epoch": 47.30401529636711, "grad_norm": 0.3799150884151459, "learning_rate": 8.971994990881759e-05, "loss": 0.0167, "step": 74220 }, { "epoch": 47.31038878266412, "grad_norm": 0.5094002485275269, "learning_rate": 8.971660196282779e-05, "loss": 0.0418, "step": 74230 }, { "epoch": 47.31676226896112, "grad_norm": 0.4540728032588959, "learning_rate": 8.971325353424374e-05, "loss": 0.0175, "step": 74240 }, { "epoch": 47.32313575525813, "grad_norm": 0.7567378878593445, "learning_rate": 8.97099046231062e-05, "loss": 0.0348, "step": 74250 }, { "epoch": 47.32950924155513, "grad_norm": 0.5790677070617676, "learning_rate": 8.970655522945581e-05, "loss": 0.0153, "step": 74260 }, { "epoch": 47.33588272785214, "grad_norm": 0.6918450593948364, "learning_rate": 8.970320535333327e-05, "loss": 0.0224, "step": 74270 }, { "epoch": 47.34225621414914, "grad_norm": 0.5672913193702698, "learning_rate": 8.969985499477933e-05, "loss": 0.0208, "step": 74280 }, { "epoch": 47.348629700446146, "grad_norm": 0.36000218987464905, "learning_rate": 8.969650415383466e-05, "loss": 0.0304, "step": 74290 }, { "epoch": 47.35500318674315, "grad_norm": 0.4054889976978302, "learning_rate": 8.969315283053997e-05, "loss": 0.0175, "step": 74300 }, { "epoch": 47.361376673040155, "grad_norm": 0.22036977112293243, "learning_rate": 8.9689801024936e-05, "loss": 0.015, "step": 74310 }, { "epoch": 47.367750159337156, "grad_norm": 0.48379793763160706, "learning_rate": 8.96864487370635e-05, "loss": 0.02, "step": 74320 }, { "epoch": 47.374123645634164, "grad_norm": 0.27999424934387207, "learning_rate": 8.968309596696315e-05, "loss": 0.0282, "step": 74330 }, { "epoch": 47.380497131931165, "grad_norm": 0.43151018023490906, "learning_rate": 8.967974271467572e-05, "loss": 0.0283, "step": 74340 }, { "epoch": 47.38687061822817, "grad_norm": 0.5295888781547546, "learning_rate": 8.967638898024195e-05, "loss": 0.0218, "step": 74350 }, { "epoch": 47.393244104525174, "grad_norm": 0.426802396774292, "learning_rate": 8.967303476370261e-05, "loss": 0.0218, "step": 74360 }, { "epoch": 47.39961759082218, "grad_norm": 0.47229161858558655, "learning_rate": 8.966968006509844e-05, "loss": 0.0299, "step": 74370 }, { "epoch": 47.40599107711918, "grad_norm": 0.4737984538078308, "learning_rate": 8.966632488447018e-05, "loss": 0.028, "step": 74380 }, { "epoch": 47.41236456341619, "grad_norm": 1.129098653793335, "learning_rate": 8.966296922185864e-05, "loss": 0.0389, "step": 74390 }, { "epoch": 47.41873804971319, "grad_norm": 0.3622661530971527, "learning_rate": 8.965961307730457e-05, "loss": 0.0242, "step": 74400 }, { "epoch": 47.4251115360102, "grad_norm": 0.5072131156921387, "learning_rate": 8.965625645084877e-05, "loss": 0.0213, "step": 74410 }, { "epoch": 47.4314850223072, "grad_norm": 0.6081326603889465, "learning_rate": 8.965289934253198e-05, "loss": 0.0362, "step": 74420 }, { "epoch": 47.43785850860421, "grad_norm": 0.5486676096916199, "learning_rate": 8.964954175239506e-05, "loss": 0.023, "step": 74430 }, { "epoch": 47.44423199490121, "grad_norm": 0.5025041699409485, "learning_rate": 8.964618368047875e-05, "loss": 0.0224, "step": 74440 }, { "epoch": 47.45060548119822, "grad_norm": 0.43800249695777893, "learning_rate": 8.964282512682387e-05, "loss": 0.0258, "step": 74450 }, { "epoch": 47.45697896749522, "grad_norm": 0.5994106531143188, "learning_rate": 8.963946609147126e-05, "loss": 0.0234, "step": 74460 }, { "epoch": 47.463352453792226, "grad_norm": 0.45363786816596985, "learning_rate": 8.963610657446169e-05, "loss": 0.0246, "step": 74470 }, { "epoch": 47.469725940089226, "grad_norm": 0.35016995668411255, "learning_rate": 8.963274657583602e-05, "loss": 0.0166, "step": 74480 }, { "epoch": 47.476099426386234, "grad_norm": 0.5277679562568665, "learning_rate": 8.962938609563504e-05, "loss": 0.0216, "step": 74490 }, { "epoch": 47.482472912683235, "grad_norm": 0.44823870062828064, "learning_rate": 8.962602513389961e-05, "loss": 0.0216, "step": 74500 }, { "epoch": 47.48884639898024, "grad_norm": 0.650191605091095, "learning_rate": 8.962266369067058e-05, "loss": 0.0155, "step": 74510 }, { "epoch": 47.495219885277244, "grad_norm": 0.6429213285446167, "learning_rate": 8.961930176598877e-05, "loss": 0.033, "step": 74520 }, { "epoch": 47.50159337157425, "grad_norm": 0.46865591406822205, "learning_rate": 8.961593935989502e-05, "loss": 0.0233, "step": 74530 }, { "epoch": 47.50796685787125, "grad_norm": 0.37498950958251953, "learning_rate": 8.961257647243021e-05, "loss": 0.0162, "step": 74540 }, { "epoch": 47.51434034416826, "grad_norm": 0.3605881929397583, "learning_rate": 8.960921310363519e-05, "loss": 0.0223, "step": 74550 }, { "epoch": 47.52071383046526, "grad_norm": 0.322897344827652, "learning_rate": 8.960584925355085e-05, "loss": 0.018, "step": 74560 }, { "epoch": 47.52708731676227, "grad_norm": 0.3989594280719757, "learning_rate": 8.960248492221804e-05, "loss": 0.0317, "step": 74570 }, { "epoch": 47.53346080305927, "grad_norm": 0.41885626316070557, "learning_rate": 8.959912010967764e-05, "loss": 0.0225, "step": 74580 }, { "epoch": 47.53983428935628, "grad_norm": 0.3031647205352783, "learning_rate": 8.959575481597055e-05, "loss": 0.0216, "step": 74590 }, { "epoch": 47.54620777565328, "grad_norm": 0.313166081905365, "learning_rate": 8.959238904113765e-05, "loss": 0.0245, "step": 74600 }, { "epoch": 47.55258126195029, "grad_norm": 0.4249531030654907, "learning_rate": 8.958902278521984e-05, "loss": 0.0212, "step": 74610 }, { "epoch": 47.55895474824729, "grad_norm": 0.27459752559661865, "learning_rate": 8.958565604825803e-05, "loss": 0.016, "step": 74620 }, { "epoch": 47.565328234544296, "grad_norm": 0.23955778777599335, "learning_rate": 8.958228883029312e-05, "loss": 0.0243, "step": 74630 }, { "epoch": 47.5717017208413, "grad_norm": 0.5925716757774353, "learning_rate": 8.957892113136604e-05, "loss": 0.0301, "step": 74640 }, { "epoch": 47.578075207138305, "grad_norm": 0.4620061218738556, "learning_rate": 8.95755529515177e-05, "loss": 0.025, "step": 74650 }, { "epoch": 47.584448693435306, "grad_norm": 0.40409213304519653, "learning_rate": 8.9572184290789e-05, "loss": 0.0253, "step": 74660 }, { "epoch": 47.590822179732314, "grad_norm": 0.5958356261253357, "learning_rate": 8.956881514922092e-05, "loss": 0.0392, "step": 74670 }, { "epoch": 47.59719566602932, "grad_norm": 0.23627549409866333, "learning_rate": 8.956544552685437e-05, "loss": 0.0284, "step": 74680 }, { "epoch": 47.60356915232632, "grad_norm": 0.5844635963439941, "learning_rate": 8.95620754237303e-05, "loss": 0.0445, "step": 74690 }, { "epoch": 47.60994263862333, "grad_norm": 0.6666755080223083, "learning_rate": 8.955870483988966e-05, "loss": 0.0253, "step": 74700 }, { "epoch": 47.61631612492033, "grad_norm": 0.7627792954444885, "learning_rate": 8.955533377537342e-05, "loss": 0.0289, "step": 74710 }, { "epoch": 47.62268961121734, "grad_norm": 0.5118746757507324, "learning_rate": 8.955196223022251e-05, "loss": 0.03, "step": 74720 }, { "epoch": 47.62906309751434, "grad_norm": 0.46787917613983154, "learning_rate": 8.954859020447793e-05, "loss": 0.0253, "step": 74730 }, { "epoch": 47.63543658381135, "grad_norm": 0.4045558273792267, "learning_rate": 8.954521769818063e-05, "loss": 0.0213, "step": 74740 }, { "epoch": 47.64181007010835, "grad_norm": 0.6612014174461365, "learning_rate": 8.954184471137161e-05, "loss": 0.03, "step": 74750 }, { "epoch": 47.64818355640536, "grad_norm": 0.5688436627388, "learning_rate": 8.953847124409183e-05, "loss": 0.0247, "step": 74760 }, { "epoch": 47.65455704270236, "grad_norm": 0.5196619033813477, "learning_rate": 8.95350972963823e-05, "loss": 0.0298, "step": 74770 }, { "epoch": 47.660930528999366, "grad_norm": 0.5087519884109497, "learning_rate": 8.953172286828401e-05, "loss": 0.0303, "step": 74780 }, { "epoch": 47.66730401529637, "grad_norm": 0.5044441223144531, "learning_rate": 8.952834795983796e-05, "loss": 0.0235, "step": 74790 }, { "epoch": 47.673677501593374, "grad_norm": 0.28789567947387695, "learning_rate": 8.952497257108517e-05, "loss": 0.0295, "step": 74800 }, { "epoch": 47.680050987890375, "grad_norm": 0.623721182346344, "learning_rate": 8.952159670206664e-05, "loss": 0.0443, "step": 74810 }, { "epoch": 47.68642447418738, "grad_norm": 0.5846380591392517, "learning_rate": 8.95182203528234e-05, "loss": 0.0386, "step": 74820 }, { "epoch": 47.692797960484384, "grad_norm": 0.44565868377685547, "learning_rate": 8.951484352339645e-05, "loss": 0.0308, "step": 74830 }, { "epoch": 47.69917144678139, "grad_norm": 1.4342551231384277, "learning_rate": 8.951146621382687e-05, "loss": 0.0256, "step": 74840 }, { "epoch": 47.70554493307839, "grad_norm": 0.25130876898765564, "learning_rate": 8.950808842415565e-05, "loss": 0.0218, "step": 74850 }, { "epoch": 47.7119184193754, "grad_norm": 0.6003896594047546, "learning_rate": 8.950471015442386e-05, "loss": 0.0193, "step": 74860 }, { "epoch": 47.7182919056724, "grad_norm": 0.38983699679374695, "learning_rate": 8.950133140467255e-05, "loss": 0.0276, "step": 74870 }, { "epoch": 47.72466539196941, "grad_norm": 0.29974237084388733, "learning_rate": 8.949795217494276e-05, "loss": 0.0302, "step": 74880 }, { "epoch": 47.73103887826641, "grad_norm": 0.4310964345932007, "learning_rate": 8.949457246527558e-05, "loss": 0.0168, "step": 74890 }, { "epoch": 47.73741236456342, "grad_norm": 0.7860645055770874, "learning_rate": 8.949119227571204e-05, "loss": 0.0264, "step": 74900 }, { "epoch": 47.74378585086042, "grad_norm": 0.3847382068634033, "learning_rate": 8.948781160629322e-05, "loss": 0.0157, "step": 74910 }, { "epoch": 47.75015933715743, "grad_norm": 0.41175761818885803, "learning_rate": 8.948443045706021e-05, "loss": 0.0205, "step": 74920 }, { "epoch": 47.75653282345443, "grad_norm": 0.30427005887031555, "learning_rate": 8.948104882805409e-05, "loss": 0.0213, "step": 74930 }, { "epoch": 47.762906309751436, "grad_norm": 0.2957451641559601, "learning_rate": 8.947766671931596e-05, "loss": 0.0247, "step": 74940 }, { "epoch": 47.76927979604844, "grad_norm": 0.290268212556839, "learning_rate": 8.94742841308869e-05, "loss": 0.0321, "step": 74950 }, { "epoch": 47.775653282345445, "grad_norm": 0.547484278678894, "learning_rate": 8.947090106280802e-05, "loss": 0.0276, "step": 74960 }, { "epoch": 47.782026768642446, "grad_norm": 0.476398766040802, "learning_rate": 8.946751751512041e-05, "loss": 0.0195, "step": 74970 }, { "epoch": 47.788400254939454, "grad_norm": 0.5073164105415344, "learning_rate": 8.946413348786522e-05, "loss": 0.024, "step": 74980 }, { "epoch": 47.794773741236455, "grad_norm": 0.2715703845024109, "learning_rate": 8.946074898108355e-05, "loss": 0.0183, "step": 74990 }, { "epoch": 47.80114722753346, "grad_norm": 0.49503853917121887, "learning_rate": 8.94573639948165e-05, "loss": 0.0215, "step": 75000 }, { "epoch": 47.80752071383046, "grad_norm": 0.3227577805519104, "learning_rate": 8.945397852910524e-05, "loss": 0.0238, "step": 75010 }, { "epoch": 47.81389420012747, "grad_norm": 0.7431999444961548, "learning_rate": 8.945059258399089e-05, "loss": 0.0162, "step": 75020 }, { "epoch": 47.82026768642447, "grad_norm": 0.48027366399765015, "learning_rate": 8.944720615951459e-05, "loss": 0.0227, "step": 75030 }, { "epoch": 47.82664117272148, "grad_norm": 0.4501822292804718, "learning_rate": 8.944381925571748e-05, "loss": 0.0221, "step": 75040 }, { "epoch": 47.83301465901848, "grad_norm": 0.5772968530654907, "learning_rate": 8.944043187264073e-05, "loss": 0.022, "step": 75050 }, { "epoch": 47.83938814531549, "grad_norm": 0.40132200717926025, "learning_rate": 8.943704401032551e-05, "loss": 0.0226, "step": 75060 }, { "epoch": 47.84576163161249, "grad_norm": 0.5347130298614502, "learning_rate": 8.943365566881295e-05, "loss": 0.0187, "step": 75070 }, { "epoch": 47.8521351179095, "grad_norm": 0.41869693994522095, "learning_rate": 8.943026684814426e-05, "loss": 0.0262, "step": 75080 }, { "epoch": 47.8585086042065, "grad_norm": 0.2702358663082123, "learning_rate": 8.942687754836059e-05, "loss": 0.0279, "step": 75090 }, { "epoch": 47.86488209050351, "grad_norm": 0.4160465598106384, "learning_rate": 8.942348776950313e-05, "loss": 0.0271, "step": 75100 }, { "epoch": 47.87125557680051, "grad_norm": 0.2977065443992615, "learning_rate": 8.942009751161308e-05, "loss": 0.0164, "step": 75110 }, { "epoch": 47.877629063097515, "grad_norm": 0.2532102167606354, "learning_rate": 8.941670677473162e-05, "loss": 0.0356, "step": 75120 }, { "epoch": 47.884002549394516, "grad_norm": 0.4975897967815399, "learning_rate": 8.941331555889995e-05, "loss": 0.0228, "step": 75130 }, { "epoch": 47.890376035691524, "grad_norm": 0.3004886209964752, "learning_rate": 8.94099238641593e-05, "loss": 0.0306, "step": 75140 }, { "epoch": 47.896749521988525, "grad_norm": 0.3163139820098877, "learning_rate": 8.940653169055087e-05, "loss": 0.0225, "step": 75150 }, { "epoch": 47.90312300828553, "grad_norm": 0.5864865183830261, "learning_rate": 8.940313903811586e-05, "loss": 0.0235, "step": 75160 }, { "epoch": 47.909496494582534, "grad_norm": 0.6935971975326538, "learning_rate": 8.939974590689551e-05, "loss": 0.0267, "step": 75170 }, { "epoch": 47.91586998087954, "grad_norm": 0.5338389873504639, "learning_rate": 8.939635229693104e-05, "loss": 0.026, "step": 75180 }, { "epoch": 47.92224346717654, "grad_norm": 0.35723260045051575, "learning_rate": 8.93929582082637e-05, "loss": 0.0347, "step": 75190 }, { "epoch": 47.92861695347355, "grad_norm": 0.3422093689441681, "learning_rate": 8.938956364093474e-05, "loss": 0.0188, "step": 75200 }, { "epoch": 47.93499043977055, "grad_norm": 0.25266242027282715, "learning_rate": 8.938616859498536e-05, "loss": 0.0157, "step": 75210 }, { "epoch": 47.94136392606756, "grad_norm": 0.9404997229576111, "learning_rate": 8.938277307045687e-05, "loss": 0.032, "step": 75220 }, { "epoch": 47.94773741236456, "grad_norm": 0.5825783610343933, "learning_rate": 8.937937706739051e-05, "loss": 0.0328, "step": 75230 }, { "epoch": 47.95411089866157, "grad_norm": 0.4462839961051941, "learning_rate": 8.937598058582751e-05, "loss": 0.0283, "step": 75240 }, { "epoch": 47.96048438495857, "grad_norm": 0.4645649790763855, "learning_rate": 8.937258362580918e-05, "loss": 0.0287, "step": 75250 }, { "epoch": 47.96685787125558, "grad_norm": 0.42328014969825745, "learning_rate": 8.93691861873768e-05, "loss": 0.0234, "step": 75260 }, { "epoch": 47.97323135755258, "grad_norm": 0.48635467886924744, "learning_rate": 8.936578827057163e-05, "loss": 0.0235, "step": 75270 }, { "epoch": 47.979604843849586, "grad_norm": 0.3894951641559601, "learning_rate": 8.936238987543496e-05, "loss": 0.022, "step": 75280 }, { "epoch": 47.98597833014659, "grad_norm": 0.5891890525817871, "learning_rate": 8.935899100200808e-05, "loss": 0.0345, "step": 75290 }, { "epoch": 47.992351816443595, "grad_norm": 0.4908401370048523, "learning_rate": 8.935559165033231e-05, "loss": 0.0233, "step": 75300 }, { "epoch": 47.998725302740596, "grad_norm": 0.5012348294258118, "learning_rate": 8.935219182044894e-05, "loss": 0.0252, "step": 75310 }, { "epoch": 48.0050987890376, "grad_norm": 1.4121203422546387, "learning_rate": 8.93487915123993e-05, "loss": 0.0244, "step": 75320 }, { "epoch": 48.01147227533461, "grad_norm": 0.45563772320747375, "learning_rate": 8.934539072622467e-05, "loss": 0.0204, "step": 75330 }, { "epoch": 48.01784576163161, "grad_norm": 0.5013715028762817, "learning_rate": 8.93419894619664e-05, "loss": 0.0217, "step": 75340 }, { "epoch": 48.02421924792862, "grad_norm": 0.5847480893135071, "learning_rate": 8.933858771966581e-05, "loss": 0.0174, "step": 75350 }, { "epoch": 48.03059273422562, "grad_norm": 0.547662615776062, "learning_rate": 8.933518549936427e-05, "loss": 0.025, "step": 75360 }, { "epoch": 48.03696622052263, "grad_norm": 0.2706519067287445, "learning_rate": 8.933178280110305e-05, "loss": 0.0193, "step": 75370 }, { "epoch": 48.04333970681963, "grad_norm": 0.5265756249427795, "learning_rate": 8.932837962492355e-05, "loss": 0.0207, "step": 75380 }, { "epoch": 48.04971319311664, "grad_norm": 0.7416671514511108, "learning_rate": 8.932497597086708e-05, "loss": 0.0231, "step": 75390 }, { "epoch": 48.05608667941364, "grad_norm": 0.52782142162323, "learning_rate": 8.932157183897505e-05, "loss": 0.0334, "step": 75400 }, { "epoch": 48.06246016571065, "grad_norm": 0.7380556464195251, "learning_rate": 8.931816722928878e-05, "loss": 0.0263, "step": 75410 }, { "epoch": 48.06883365200765, "grad_norm": 0.5533484220504761, "learning_rate": 8.931476214184968e-05, "loss": 0.0232, "step": 75420 }, { "epoch": 48.075207138304656, "grad_norm": 0.5061846375465393, "learning_rate": 8.931135657669909e-05, "loss": 0.0198, "step": 75430 }, { "epoch": 48.081580624601656, "grad_norm": 0.3923509120941162, "learning_rate": 8.930795053387839e-05, "loss": 0.0262, "step": 75440 }, { "epoch": 48.087954110898664, "grad_norm": 0.4133387804031372, "learning_rate": 8.930454401342899e-05, "loss": 0.0188, "step": 75450 }, { "epoch": 48.094327597195665, "grad_norm": 0.32283154129981995, "learning_rate": 8.930113701539227e-05, "loss": 0.0243, "step": 75460 }, { "epoch": 48.10070108349267, "grad_norm": 0.5459981560707092, "learning_rate": 8.929772953980962e-05, "loss": 0.0238, "step": 75470 }, { "epoch": 48.107074569789674, "grad_norm": 1.0120607614517212, "learning_rate": 8.929432158672245e-05, "loss": 0.0185, "step": 75480 }, { "epoch": 48.11344805608668, "grad_norm": 0.4208924472332001, "learning_rate": 8.929091315617219e-05, "loss": 0.0245, "step": 75490 }, { "epoch": 48.11982154238368, "grad_norm": 0.6432639956474304, "learning_rate": 8.928750424820023e-05, "loss": 0.0272, "step": 75500 }, { "epoch": 48.12619502868069, "grad_norm": 0.49686241149902344, "learning_rate": 8.928409486284799e-05, "loss": 0.0169, "step": 75510 }, { "epoch": 48.13256851497769, "grad_norm": 0.5811657309532166, "learning_rate": 8.928068500015691e-05, "loss": 0.0201, "step": 75520 }, { "epoch": 48.1389420012747, "grad_norm": 0.43043413758277893, "learning_rate": 8.927727466016843e-05, "loss": 0.0294, "step": 75530 }, { "epoch": 48.1453154875717, "grad_norm": 0.596208930015564, "learning_rate": 8.927386384292397e-05, "loss": 0.0338, "step": 75540 }, { "epoch": 48.15168897386871, "grad_norm": 0.5121863484382629, "learning_rate": 8.9270452548465e-05, "loss": 0.0195, "step": 75550 }, { "epoch": 48.15806246016571, "grad_norm": 0.597659170627594, "learning_rate": 8.926704077683294e-05, "loss": 0.0216, "step": 75560 }, { "epoch": 48.16443594646272, "grad_norm": 0.23879288136959076, "learning_rate": 8.926362852806926e-05, "loss": 0.0198, "step": 75570 }, { "epoch": 48.17080943275972, "grad_norm": 0.35727983713150024, "learning_rate": 8.926021580221542e-05, "loss": 0.0156, "step": 75580 }, { "epoch": 48.177182919056726, "grad_norm": 0.46775326132774353, "learning_rate": 8.925680259931291e-05, "loss": 0.0165, "step": 75590 }, { "epoch": 48.18355640535373, "grad_norm": 0.511959969997406, "learning_rate": 8.925338891940316e-05, "loss": 0.0167, "step": 75600 }, { "epoch": 48.189929891650735, "grad_norm": 0.3565732538700104, "learning_rate": 8.924997476252769e-05, "loss": 0.0296, "step": 75610 }, { "epoch": 48.196303377947736, "grad_norm": 0.4447449743747711, "learning_rate": 8.924656012872797e-05, "loss": 0.0213, "step": 75620 }, { "epoch": 48.202676864244744, "grad_norm": 0.6001524329185486, "learning_rate": 8.924314501804548e-05, "loss": 0.0232, "step": 75630 }, { "epoch": 48.209050350541744, "grad_norm": 0.41357290744781494, "learning_rate": 8.923972943052171e-05, "loss": 0.0209, "step": 75640 }, { "epoch": 48.21542383683875, "grad_norm": 0.2810899019241333, "learning_rate": 8.92363133661982e-05, "loss": 0.0138, "step": 75650 }, { "epoch": 48.22179732313575, "grad_norm": 0.39860740303993225, "learning_rate": 8.923289682511644e-05, "loss": 0.0234, "step": 75660 }, { "epoch": 48.22817080943276, "grad_norm": 0.3649909198284149, "learning_rate": 8.922947980731792e-05, "loss": 0.0309, "step": 75670 }, { "epoch": 48.23454429572976, "grad_norm": 0.8226469159126282, "learning_rate": 8.92260623128442e-05, "loss": 0.0224, "step": 75680 }, { "epoch": 48.24091778202677, "grad_norm": 0.6315596103668213, "learning_rate": 8.92226443417368e-05, "loss": 0.0274, "step": 75690 }, { "epoch": 48.24729126832377, "grad_norm": 0.47064143419265747, "learning_rate": 8.921922589403721e-05, "loss": 0.0327, "step": 75700 }, { "epoch": 48.25366475462078, "grad_norm": 0.5965138077735901, "learning_rate": 8.921580696978702e-05, "loss": 0.026, "step": 75710 }, { "epoch": 48.26003824091778, "grad_norm": 0.3230724632740021, "learning_rate": 8.921238756902773e-05, "loss": 0.0276, "step": 75720 }, { "epoch": 48.26641172721479, "grad_norm": 0.44348838925361633, "learning_rate": 8.920896769180092e-05, "loss": 0.0153, "step": 75730 }, { "epoch": 48.27278521351179, "grad_norm": 0.48899027705192566, "learning_rate": 8.920554733814812e-05, "loss": 0.0278, "step": 75740 }, { "epoch": 48.2791586998088, "grad_norm": 0.35626277327537537, "learning_rate": 8.920212650811093e-05, "loss": 0.0225, "step": 75750 }, { "epoch": 48.2855321861058, "grad_norm": 0.5385358929634094, "learning_rate": 8.919870520173085e-05, "loss": 0.0264, "step": 75760 }, { "epoch": 48.291905672402805, "grad_norm": 0.27742716670036316, "learning_rate": 8.919528341904953e-05, "loss": 0.0263, "step": 75770 }, { "epoch": 48.298279158699806, "grad_norm": 0.3483431935310364, "learning_rate": 8.91918611601085e-05, "loss": 0.0153, "step": 75780 }, { "epoch": 48.304652644996814, "grad_norm": 0.21175138652324677, "learning_rate": 8.918843842494935e-05, "loss": 0.0242, "step": 75790 }, { "epoch": 48.311026131293815, "grad_norm": 0.3293132483959198, "learning_rate": 8.918501521361366e-05, "loss": 0.0199, "step": 75800 }, { "epoch": 48.31739961759082, "grad_norm": 0.36684665083885193, "learning_rate": 8.918159152614306e-05, "loss": 0.0176, "step": 75810 }, { "epoch": 48.323773103887824, "grad_norm": 0.5040516257286072, "learning_rate": 8.917816736257911e-05, "loss": 0.0188, "step": 75820 }, { "epoch": 48.33014659018483, "grad_norm": 0.4116906225681305, "learning_rate": 8.917474272296345e-05, "loss": 0.0252, "step": 75830 }, { "epoch": 48.33652007648183, "grad_norm": 1.108932614326477, "learning_rate": 8.917131760733768e-05, "loss": 0.0428, "step": 75840 }, { "epoch": 48.34289356277884, "grad_norm": 0.6733537316322327, "learning_rate": 8.916789201574341e-05, "loss": 0.021, "step": 75850 }, { "epoch": 48.34926704907584, "grad_norm": 0.5461486577987671, "learning_rate": 8.916446594822228e-05, "loss": 0.0248, "step": 75860 }, { "epoch": 48.35564053537285, "grad_norm": 0.40634480118751526, "learning_rate": 8.91610394048159e-05, "loss": 0.0305, "step": 75870 }, { "epoch": 48.36201402166985, "grad_norm": 0.38543078303337097, "learning_rate": 8.915761238556592e-05, "loss": 0.0306, "step": 75880 }, { "epoch": 48.36838750796686, "grad_norm": 0.47182533144950867, "learning_rate": 8.9154184890514e-05, "loss": 0.0236, "step": 75890 }, { "epoch": 48.37476099426386, "grad_norm": 0.300396591424942, "learning_rate": 8.915075691970173e-05, "loss": 0.0145, "step": 75900 }, { "epoch": 48.38113448056087, "grad_norm": 0.5201929211616516, "learning_rate": 8.914732847317081e-05, "loss": 0.0322, "step": 75910 }, { "epoch": 48.38750796685787, "grad_norm": 0.3313845098018646, "learning_rate": 8.91438995509629e-05, "loss": 0.0178, "step": 75920 }, { "epoch": 48.393881453154876, "grad_norm": 0.24322479963302612, "learning_rate": 8.914047015311966e-05, "loss": 0.0229, "step": 75930 }, { "epoch": 48.400254939451884, "grad_norm": 0.5620152354240417, "learning_rate": 8.913704027968273e-05, "loss": 0.0208, "step": 75940 }, { "epoch": 48.406628425748885, "grad_norm": 0.4299810826778412, "learning_rate": 8.913360993069382e-05, "loss": 0.0162, "step": 75950 }, { "epoch": 48.41300191204589, "grad_norm": 0.4556967318058014, "learning_rate": 8.91301791061946e-05, "loss": 0.0217, "step": 75960 }, { "epoch": 48.41937539834289, "grad_norm": 0.29739418625831604, "learning_rate": 8.912674780622676e-05, "loss": 0.0156, "step": 75970 }, { "epoch": 48.4257488846399, "grad_norm": 0.27627599239349365, "learning_rate": 8.912331603083199e-05, "loss": 0.0157, "step": 75980 }, { "epoch": 48.4321223709369, "grad_norm": 0.6144986748695374, "learning_rate": 8.911988378005199e-05, "loss": 0.0234, "step": 75990 }, { "epoch": 48.43849585723391, "grad_norm": 0.3971882462501526, "learning_rate": 8.911645105392848e-05, "loss": 0.0291, "step": 76000 }, { "epoch": 48.44486934353091, "grad_norm": 0.5654749870300293, "learning_rate": 8.911301785250314e-05, "loss": 0.0163, "step": 76010 }, { "epoch": 48.45124282982792, "grad_norm": 0.6646308302879333, "learning_rate": 8.910958417581772e-05, "loss": 0.0163, "step": 76020 }, { "epoch": 48.45761631612492, "grad_norm": 0.28484082221984863, "learning_rate": 8.910615002391392e-05, "loss": 0.0225, "step": 76030 }, { "epoch": 48.46398980242193, "grad_norm": 0.4190088212490082, "learning_rate": 8.910271539683346e-05, "loss": 0.0197, "step": 76040 }, { "epoch": 48.47036328871893, "grad_norm": 0.7240654230117798, "learning_rate": 8.90992802946181e-05, "loss": 0.0243, "step": 76050 }, { "epoch": 48.47673677501594, "grad_norm": 0.4968497157096863, "learning_rate": 8.909584471730957e-05, "loss": 0.032, "step": 76060 }, { "epoch": 48.48311026131294, "grad_norm": 0.9180871248245239, "learning_rate": 8.909240866494961e-05, "loss": 0.0309, "step": 76070 }, { "epoch": 48.489483747609945, "grad_norm": 0.6101062297821045, "learning_rate": 8.908897213757999e-05, "loss": 0.0252, "step": 76080 }, { "epoch": 48.495857233906946, "grad_norm": 0.47214585542678833, "learning_rate": 8.908553513524245e-05, "loss": 0.0177, "step": 76090 }, { "epoch": 48.502230720203954, "grad_norm": 0.4123691916465759, "learning_rate": 8.908209765797873e-05, "loss": 0.0257, "step": 76100 }, { "epoch": 48.508604206500955, "grad_norm": 0.2346596121788025, "learning_rate": 8.907865970583064e-05, "loss": 0.0357, "step": 76110 }, { "epoch": 48.51497769279796, "grad_norm": 0.26319506764411926, "learning_rate": 8.907522127883995e-05, "loss": 0.0219, "step": 76120 }, { "epoch": 48.521351179094964, "grad_norm": 0.7729846239089966, "learning_rate": 8.907178237704842e-05, "loss": 0.0189, "step": 76130 }, { "epoch": 48.52772466539197, "grad_norm": 0.5811246037483215, "learning_rate": 8.906834300049785e-05, "loss": 0.0268, "step": 76140 }, { "epoch": 48.53409815168897, "grad_norm": 0.47579431533813477, "learning_rate": 8.906490314923e-05, "loss": 0.0212, "step": 76150 }, { "epoch": 48.54047163798598, "grad_norm": 0.49991121888160706, "learning_rate": 8.906146282328671e-05, "loss": 0.0256, "step": 76160 }, { "epoch": 48.54684512428298, "grad_norm": 0.41751834750175476, "learning_rate": 8.905802202270977e-05, "loss": 0.0231, "step": 76170 }, { "epoch": 48.55321861057999, "grad_norm": 0.7454197406768799, "learning_rate": 8.905458074754098e-05, "loss": 0.0273, "step": 76180 }, { "epoch": 48.55959209687699, "grad_norm": 0.4572162330150604, "learning_rate": 8.905113899782218e-05, "loss": 0.0192, "step": 76190 }, { "epoch": 48.565965583174, "grad_norm": 0.5171675682067871, "learning_rate": 8.904769677359515e-05, "loss": 0.0262, "step": 76200 }, { "epoch": 48.572339069471, "grad_norm": 0.6909810304641724, "learning_rate": 8.904425407490174e-05, "loss": 0.0222, "step": 76210 }, { "epoch": 48.57871255576801, "grad_norm": 0.5221519470214844, "learning_rate": 8.904081090178378e-05, "loss": 0.0188, "step": 76220 }, { "epoch": 48.58508604206501, "grad_norm": 0.4986597001552582, "learning_rate": 8.90373672542831e-05, "loss": 0.0167, "step": 76230 }, { "epoch": 48.591459528362016, "grad_norm": 0.9982235431671143, "learning_rate": 8.903392313244157e-05, "loss": 0.0207, "step": 76240 }, { "epoch": 48.59783301465902, "grad_norm": 0.3603552579879761, "learning_rate": 8.9030478536301e-05, "loss": 0.0196, "step": 76250 }, { "epoch": 48.604206500956025, "grad_norm": 0.45641520619392395, "learning_rate": 8.902703346590327e-05, "loss": 0.018, "step": 76260 }, { "epoch": 48.610579987253026, "grad_norm": 0.3008120059967041, "learning_rate": 8.902358792129024e-05, "loss": 0.0175, "step": 76270 }, { "epoch": 48.61695347355003, "grad_norm": 0.3910423219203949, "learning_rate": 8.902014190250377e-05, "loss": 0.0236, "step": 76280 }, { "epoch": 48.623326959847034, "grad_norm": 0.2935914099216461, "learning_rate": 8.901669540958574e-05, "loss": 0.0254, "step": 76290 }, { "epoch": 48.62970044614404, "grad_norm": 0.28027310967445374, "learning_rate": 8.901324844257802e-05, "loss": 0.0223, "step": 76300 }, { "epoch": 48.63607393244104, "grad_norm": 0.3035241961479187, "learning_rate": 8.90098010015225e-05, "loss": 0.025, "step": 76310 }, { "epoch": 48.64244741873805, "grad_norm": 0.24401028454303741, "learning_rate": 8.900635308646106e-05, "loss": 0.0221, "step": 76320 }, { "epoch": 48.64882090503505, "grad_norm": 0.30767902731895447, "learning_rate": 8.900290469743559e-05, "loss": 0.0234, "step": 76330 }, { "epoch": 48.65519439133206, "grad_norm": 0.3489046096801758, "learning_rate": 8.899945583448803e-05, "loss": 0.0267, "step": 76340 }, { "epoch": 48.66156787762906, "grad_norm": 0.6133447289466858, "learning_rate": 8.899600649766024e-05, "loss": 0.0245, "step": 76350 }, { "epoch": 48.66794136392607, "grad_norm": 0.779419481754303, "learning_rate": 8.899255668699415e-05, "loss": 0.0241, "step": 76360 }, { "epoch": 48.67431485022307, "grad_norm": 0.9319865107536316, "learning_rate": 8.898910640253171e-05, "loss": 0.0244, "step": 76370 }, { "epoch": 48.68068833652008, "grad_norm": 0.4049406349658966, "learning_rate": 8.898565564431478e-05, "loss": 0.0246, "step": 76380 }, { "epoch": 48.68706182281708, "grad_norm": 0.959450364112854, "learning_rate": 8.898220441238534e-05, "loss": 0.0302, "step": 76390 }, { "epoch": 48.693435309114086, "grad_norm": 0.4261667728424072, "learning_rate": 8.897875270678532e-05, "loss": 0.0186, "step": 76400 }, { "epoch": 48.69980879541109, "grad_norm": 0.6010489463806152, "learning_rate": 8.897530052755665e-05, "loss": 0.0241, "step": 76410 }, { "epoch": 48.706182281708095, "grad_norm": 0.3465341329574585, "learning_rate": 8.897184787474127e-05, "loss": 0.0348, "step": 76420 }, { "epoch": 48.712555768005096, "grad_norm": 0.35874900221824646, "learning_rate": 8.896839474838115e-05, "loss": 0.0249, "step": 76430 }, { "epoch": 48.718929254302104, "grad_norm": 0.4277724623680115, "learning_rate": 8.896494114851823e-05, "loss": 0.0239, "step": 76440 }, { "epoch": 48.725302740599105, "grad_norm": 0.41374707221984863, "learning_rate": 8.89614870751945e-05, "loss": 0.0194, "step": 76450 }, { "epoch": 48.73167622689611, "grad_norm": 0.3206664025783539, "learning_rate": 8.895803252845192e-05, "loss": 0.0249, "step": 76460 }, { "epoch": 48.738049713193114, "grad_norm": 0.3847193419933319, "learning_rate": 8.895457750833244e-05, "loss": 0.0191, "step": 76470 }, { "epoch": 48.74442319949012, "grad_norm": 0.236180379986763, "learning_rate": 8.895112201487808e-05, "loss": 0.0274, "step": 76480 }, { "epoch": 48.75079668578712, "grad_norm": 0.3643247187137604, "learning_rate": 8.89476660481308e-05, "loss": 0.0211, "step": 76490 }, { "epoch": 48.75717017208413, "grad_norm": 0.2318602204322815, "learning_rate": 8.894420960813263e-05, "loss": 0.0158, "step": 76500 }, { "epoch": 48.76354365838113, "grad_norm": 0.4744431972503662, "learning_rate": 8.894075269492552e-05, "loss": 0.0273, "step": 76510 }, { "epoch": 48.76991714467814, "grad_norm": 0.3607683777809143, "learning_rate": 8.89372953085515e-05, "loss": 0.0197, "step": 76520 }, { "epoch": 48.77629063097514, "grad_norm": 0.39541012048721313, "learning_rate": 8.893383744905259e-05, "loss": 0.0233, "step": 76530 }, { "epoch": 48.78266411727215, "grad_norm": 0.8160480856895447, "learning_rate": 8.893037911647079e-05, "loss": 0.0268, "step": 76540 }, { "epoch": 48.78903760356915, "grad_norm": 0.30967459082603455, "learning_rate": 8.892692031084814e-05, "loss": 0.028, "step": 76550 }, { "epoch": 48.79541108986616, "grad_norm": 0.700473964214325, "learning_rate": 8.892346103222664e-05, "loss": 0.029, "step": 76560 }, { "epoch": 48.80178457616316, "grad_norm": 0.3504991829395294, "learning_rate": 8.892000128064834e-05, "loss": 0.0228, "step": 76570 }, { "epoch": 48.808158062460166, "grad_norm": 0.31831562519073486, "learning_rate": 8.891654105615528e-05, "loss": 0.0154, "step": 76580 }, { "epoch": 48.814531548757174, "grad_norm": 0.44795316457748413, "learning_rate": 8.891308035878951e-05, "loss": 0.0265, "step": 76590 }, { "epoch": 48.820905035054174, "grad_norm": 0.35997816920280457, "learning_rate": 8.890961918859306e-05, "loss": 0.0243, "step": 76600 }, { "epoch": 48.82727852135118, "grad_norm": 0.5554684400558472, "learning_rate": 8.890615754560801e-05, "loss": 0.0303, "step": 76610 }, { "epoch": 48.83365200764818, "grad_norm": 0.5572734475135803, "learning_rate": 8.890269542987642e-05, "loss": 0.0223, "step": 76620 }, { "epoch": 48.84002549394519, "grad_norm": 0.4752501845359802, "learning_rate": 8.889923284144033e-05, "loss": 0.034, "step": 76630 }, { "epoch": 48.84639898024219, "grad_norm": 0.26858168840408325, "learning_rate": 8.889576978034184e-05, "loss": 0.0188, "step": 76640 }, { "epoch": 48.8527724665392, "grad_norm": 0.3426305055618286, "learning_rate": 8.889230624662303e-05, "loss": 0.0169, "step": 76650 }, { "epoch": 48.8591459528362, "grad_norm": 0.4153650403022766, "learning_rate": 8.888884224032598e-05, "loss": 0.0329, "step": 76660 }, { "epoch": 48.86551943913321, "grad_norm": 0.30574050545692444, "learning_rate": 8.888537776149277e-05, "loss": 0.0271, "step": 76670 }, { "epoch": 48.87189292543021, "grad_norm": 0.4339156746864319, "learning_rate": 8.888191281016551e-05, "loss": 0.0158, "step": 76680 }, { "epoch": 48.87826641172722, "grad_norm": 0.3017025589942932, "learning_rate": 8.88784473863863e-05, "loss": 0.0293, "step": 76690 }, { "epoch": 48.88463989802422, "grad_norm": 0.40968942642211914, "learning_rate": 8.887498149019725e-05, "loss": 0.0253, "step": 76700 }, { "epoch": 48.89101338432123, "grad_norm": 0.883637011051178, "learning_rate": 8.887151512164048e-05, "loss": 0.0212, "step": 76710 }, { "epoch": 48.89738687061823, "grad_norm": 0.4192010462284088, "learning_rate": 8.886804828075807e-05, "loss": 0.0275, "step": 76720 }, { "epoch": 48.903760356915235, "grad_norm": 0.3322286605834961, "learning_rate": 8.88645809675922e-05, "loss": 0.0227, "step": 76730 }, { "epoch": 48.910133843212236, "grad_norm": 0.3261466920375824, "learning_rate": 8.886111318218498e-05, "loss": 0.0191, "step": 76740 }, { "epoch": 48.916507329509244, "grad_norm": 0.3612256944179535, "learning_rate": 8.885764492457854e-05, "loss": 0.0218, "step": 76750 }, { "epoch": 48.922880815806245, "grad_norm": 0.6655124425888062, "learning_rate": 8.8854176194815e-05, "loss": 0.0264, "step": 76760 }, { "epoch": 48.92925430210325, "grad_norm": 0.29641762375831604, "learning_rate": 8.885070699293656e-05, "loss": 0.0199, "step": 76770 }, { "epoch": 48.935627788400254, "grad_norm": 0.5378060340881348, "learning_rate": 8.884723731898536e-05, "loss": 0.0371, "step": 76780 }, { "epoch": 48.94200127469726, "grad_norm": 0.46489793062210083, "learning_rate": 8.884376717300354e-05, "loss": 0.0196, "step": 76790 }, { "epoch": 48.94837476099426, "grad_norm": 0.37712621688842773, "learning_rate": 8.884029655503325e-05, "loss": 0.0262, "step": 76800 }, { "epoch": 48.95474824729127, "grad_norm": 0.43126267194747925, "learning_rate": 8.883682546511671e-05, "loss": 0.0143, "step": 76810 }, { "epoch": 48.96112173358827, "grad_norm": 0.22369031608104706, "learning_rate": 8.883335390329607e-05, "loss": 0.0166, "step": 76820 }, { "epoch": 48.96749521988528, "grad_norm": 0.6358847618103027, "learning_rate": 8.88298818696135e-05, "loss": 0.0153, "step": 76830 }, { "epoch": 48.97386870618228, "grad_norm": 0.35620570182800293, "learning_rate": 8.882640936411122e-05, "loss": 0.0203, "step": 76840 }, { "epoch": 48.98024219247929, "grad_norm": 0.27171969413757324, "learning_rate": 8.882293638683141e-05, "loss": 0.0312, "step": 76850 }, { "epoch": 48.98661567877629, "grad_norm": 0.47724440693855286, "learning_rate": 8.881946293781626e-05, "loss": 0.018, "step": 76860 }, { "epoch": 48.9929891650733, "grad_norm": 0.6344947814941406, "learning_rate": 8.881598901710798e-05, "loss": 0.0366, "step": 76870 }, { "epoch": 48.9993626513703, "grad_norm": 0.3690052628517151, "learning_rate": 8.88125146247488e-05, "loss": 0.0218, "step": 76880 }, { "epoch": 49.005736137667306, "grad_norm": 0.3196150064468384, "learning_rate": 8.88090397607809e-05, "loss": 0.02, "step": 76890 }, { "epoch": 49.01210962396431, "grad_norm": 0.2323322743177414, "learning_rate": 8.880556442524655e-05, "loss": 0.018, "step": 76900 }, { "epoch": 49.018483110261315, "grad_norm": 0.5877742171287537, "learning_rate": 8.880208861818792e-05, "loss": 0.0312, "step": 76910 }, { "epoch": 49.024856596558315, "grad_norm": 0.46226146817207336, "learning_rate": 8.879861233964732e-05, "loss": 0.0227, "step": 76920 }, { "epoch": 49.03123008285532, "grad_norm": 0.770878791809082, "learning_rate": 8.879513558966693e-05, "loss": 0.0215, "step": 76930 }, { "epoch": 49.037603569152324, "grad_norm": 0.6134500503540039, "learning_rate": 8.8791658368289e-05, "loss": 0.029, "step": 76940 }, { "epoch": 49.04397705544933, "grad_norm": 0.2950141727924347, "learning_rate": 8.87881806755558e-05, "loss": 0.0182, "step": 76950 }, { "epoch": 49.05035054174633, "grad_norm": 0.47258442640304565, "learning_rate": 8.878470251150959e-05, "loss": 0.0142, "step": 76960 }, { "epoch": 49.05672402804334, "grad_norm": 0.536088764667511, "learning_rate": 8.878122387619264e-05, "loss": 0.0293, "step": 76970 }, { "epoch": 49.06309751434034, "grad_norm": 0.4390946924686432, "learning_rate": 8.877774476964718e-05, "loss": 0.0311, "step": 76980 }, { "epoch": 49.06947100063735, "grad_norm": 0.5162464380264282, "learning_rate": 8.877426519191552e-05, "loss": 0.0187, "step": 76990 }, { "epoch": 49.07584448693435, "grad_norm": 0.6299289464950562, "learning_rate": 8.877078514303993e-05, "loss": 0.0364, "step": 77000 }, { "epoch": 49.08221797323136, "grad_norm": 0.5679315328598022, "learning_rate": 8.876730462306268e-05, "loss": 0.0266, "step": 77010 }, { "epoch": 49.08859145952836, "grad_norm": 0.29127684235572815, "learning_rate": 8.87638236320261e-05, "loss": 0.024, "step": 77020 }, { "epoch": 49.09496494582537, "grad_norm": 0.3191298842430115, "learning_rate": 8.876034216997245e-05, "loss": 0.0259, "step": 77030 }, { "epoch": 49.10133843212237, "grad_norm": 0.20821388065814972, "learning_rate": 8.875686023694404e-05, "loss": 0.0231, "step": 77040 }, { "epoch": 49.107711918419376, "grad_norm": 0.4768683612346649, "learning_rate": 8.875337783298321e-05, "loss": 0.0268, "step": 77050 }, { "epoch": 49.11408540471638, "grad_norm": 0.4767087399959564, "learning_rate": 8.874989495813223e-05, "loss": 0.0199, "step": 77060 }, { "epoch": 49.120458891013385, "grad_norm": 0.21784226596355438, "learning_rate": 8.874641161243345e-05, "loss": 0.0181, "step": 77070 }, { "epoch": 49.126832377310386, "grad_norm": 0.647822916507721, "learning_rate": 8.874292779592918e-05, "loss": 0.0171, "step": 77080 }, { "epoch": 49.133205863607394, "grad_norm": 0.6063865423202515, "learning_rate": 8.873944350866175e-05, "loss": 0.0197, "step": 77090 }, { "epoch": 49.139579349904395, "grad_norm": 0.4788207709789276, "learning_rate": 8.873595875067352e-05, "loss": 0.0304, "step": 77100 }, { "epoch": 49.1459528362014, "grad_norm": 0.1634960174560547, "learning_rate": 8.87324735220068e-05, "loss": 0.0202, "step": 77110 }, { "epoch": 49.1523263224984, "grad_norm": 0.6162344217300415, "learning_rate": 8.872898782270397e-05, "loss": 0.0251, "step": 77120 }, { "epoch": 49.15869980879541, "grad_norm": 0.5550015568733215, "learning_rate": 8.872550165280738e-05, "loss": 0.0218, "step": 77130 }, { "epoch": 49.16507329509241, "grad_norm": 0.6461644172668457, "learning_rate": 8.872201501235937e-05, "loss": 0.0279, "step": 77140 }, { "epoch": 49.17144678138942, "grad_norm": 0.38927990198135376, "learning_rate": 8.871852790140233e-05, "loss": 0.0329, "step": 77150 }, { "epoch": 49.17782026768642, "grad_norm": 0.2511272430419922, "learning_rate": 8.871504031997862e-05, "loss": 0.0194, "step": 77160 }, { "epoch": 49.18419375398343, "grad_norm": 0.5065391659736633, "learning_rate": 8.87115522681306e-05, "loss": 0.0173, "step": 77170 }, { "epoch": 49.19056724028043, "grad_norm": 0.44868627190589905, "learning_rate": 8.870806374590068e-05, "loss": 0.0262, "step": 77180 }, { "epoch": 49.19694072657744, "grad_norm": 0.7002864480018616, "learning_rate": 8.870457475333123e-05, "loss": 0.0203, "step": 77190 }, { "epoch": 49.203314212874446, "grad_norm": 0.36071133613586426, "learning_rate": 8.870108529046466e-05, "loss": 0.0209, "step": 77200 }, { "epoch": 49.20968769917145, "grad_norm": 0.2789326310157776, "learning_rate": 8.869759535734336e-05, "loss": 0.0167, "step": 77210 }, { "epoch": 49.216061185468455, "grad_norm": 0.43922707438468933, "learning_rate": 8.869410495400976e-05, "loss": 0.0179, "step": 77220 }, { "epoch": 49.222434671765456, "grad_norm": 0.23701466619968414, "learning_rate": 8.869061408050623e-05, "loss": 0.0187, "step": 77230 }, { "epoch": 49.22880815806246, "grad_norm": 0.5033986568450928, "learning_rate": 8.868712273687521e-05, "loss": 0.0248, "step": 77240 }, { "epoch": 49.235181644359464, "grad_norm": 0.4085116386413574, "learning_rate": 8.868363092315914e-05, "loss": 0.0168, "step": 77250 }, { "epoch": 49.24155513065647, "grad_norm": 0.9206916689872742, "learning_rate": 8.868013863940041e-05, "loss": 0.0211, "step": 77260 }, { "epoch": 49.24792861695347, "grad_norm": 0.262408584356308, "learning_rate": 8.867664588564148e-05, "loss": 0.0209, "step": 77270 }, { "epoch": 49.25430210325048, "grad_norm": 0.2950029671192169, "learning_rate": 8.867315266192479e-05, "loss": 0.0167, "step": 77280 }, { "epoch": 49.26067558954748, "grad_norm": 0.5051262974739075, "learning_rate": 8.866965896829278e-05, "loss": 0.0231, "step": 77290 }, { "epoch": 49.26704907584449, "grad_norm": 0.47850650548934937, "learning_rate": 8.866616480478791e-05, "loss": 0.0192, "step": 77300 }, { "epoch": 49.27342256214149, "grad_norm": 0.23890072107315063, "learning_rate": 8.866267017145262e-05, "loss": 0.0134, "step": 77310 }, { "epoch": 49.2797960484385, "grad_norm": 0.24545453488826752, "learning_rate": 8.865917506832939e-05, "loss": 0.028, "step": 77320 }, { "epoch": 49.2861695347355, "grad_norm": 0.2781742513179779, "learning_rate": 8.865567949546068e-05, "loss": 0.0259, "step": 77330 }, { "epoch": 49.29254302103251, "grad_norm": 0.6485303640365601, "learning_rate": 8.865218345288897e-05, "loss": 0.0174, "step": 77340 }, { "epoch": 49.29891650732951, "grad_norm": 0.7008166909217834, "learning_rate": 8.864868694065674e-05, "loss": 0.0267, "step": 77350 }, { "epoch": 49.305289993626516, "grad_norm": 0.48867708444595337, "learning_rate": 8.864518995880646e-05, "loss": 0.0324, "step": 77360 }, { "epoch": 49.31166347992352, "grad_norm": 0.3192881643772125, "learning_rate": 8.864169250738064e-05, "loss": 0.0214, "step": 77370 }, { "epoch": 49.318036966220525, "grad_norm": 0.3369359076023102, "learning_rate": 8.863819458642179e-05, "loss": 0.0142, "step": 77380 }, { "epoch": 49.324410452517526, "grad_norm": 0.41567790508270264, "learning_rate": 8.863469619597238e-05, "loss": 0.022, "step": 77390 }, { "epoch": 49.330783938814534, "grad_norm": 0.578463077545166, "learning_rate": 8.863119733607493e-05, "loss": 0.0185, "step": 77400 }, { "epoch": 49.337157425111535, "grad_norm": 0.4225253164768219, "learning_rate": 8.862769800677197e-05, "loss": 0.0283, "step": 77410 }, { "epoch": 49.34353091140854, "grad_norm": 0.2911081612110138, "learning_rate": 8.8624198208106e-05, "loss": 0.0307, "step": 77420 }, { "epoch": 49.349904397705544, "grad_norm": 0.37398651242256165, "learning_rate": 8.862069794011956e-05, "loss": 0.0164, "step": 77430 }, { "epoch": 49.35627788400255, "grad_norm": 0.541105329990387, "learning_rate": 8.861719720285517e-05, "loss": 0.0233, "step": 77440 }, { "epoch": 49.36265137029955, "grad_norm": 0.4153154790401459, "learning_rate": 8.861369599635538e-05, "loss": 0.0168, "step": 77450 }, { "epoch": 49.36902485659656, "grad_norm": 0.7136523127555847, "learning_rate": 8.861019432066272e-05, "loss": 0.0177, "step": 77460 }, { "epoch": 49.37539834289356, "grad_norm": 0.4418129622936249, "learning_rate": 8.860669217581975e-05, "loss": 0.0231, "step": 77470 }, { "epoch": 49.38177182919057, "grad_norm": 0.30205056071281433, "learning_rate": 8.860318956186901e-05, "loss": 0.0163, "step": 77480 }, { "epoch": 49.38814531548757, "grad_norm": 0.40903952717781067, "learning_rate": 8.859968647885309e-05, "loss": 0.0229, "step": 77490 }, { "epoch": 49.39451880178458, "grad_norm": 0.5170226097106934, "learning_rate": 8.859618292681451e-05, "loss": 0.0264, "step": 77500 }, { "epoch": 49.40089228808158, "grad_norm": 0.42853203415870667, "learning_rate": 8.859267890579587e-05, "loss": 0.0208, "step": 77510 }, { "epoch": 49.40726577437859, "grad_norm": 0.22943678498268127, "learning_rate": 8.858917441583976e-05, "loss": 0.0272, "step": 77520 }, { "epoch": 49.41363926067559, "grad_norm": 0.27444449067115784, "learning_rate": 8.858566945698874e-05, "loss": 0.0261, "step": 77530 }, { "epoch": 49.420012746972596, "grad_norm": 0.4407459795475006, "learning_rate": 8.858216402928541e-05, "loss": 0.0189, "step": 77540 }, { "epoch": 49.4263862332696, "grad_norm": 0.5925467014312744, "learning_rate": 8.857865813277235e-05, "loss": 0.019, "step": 77550 }, { "epoch": 49.432759719566604, "grad_norm": 0.4955221116542816, "learning_rate": 8.857515176749217e-05, "loss": 0.0262, "step": 77560 }, { "epoch": 49.439133205863605, "grad_norm": 0.4943479299545288, "learning_rate": 8.857164493348749e-05, "loss": 0.0372, "step": 77570 }, { "epoch": 49.44550669216061, "grad_norm": 0.4927414655685425, "learning_rate": 8.856813763080089e-05, "loss": 0.0266, "step": 77580 }, { "epoch": 49.451880178457614, "grad_norm": 0.669206976890564, "learning_rate": 8.8564629859475e-05, "loss": 0.0195, "step": 77590 }, { "epoch": 49.45825366475462, "grad_norm": 0.30219265818595886, "learning_rate": 8.856112161955245e-05, "loss": 0.0254, "step": 77600 }, { "epoch": 49.46462715105162, "grad_norm": 0.3279782831668854, "learning_rate": 8.855761291107587e-05, "loss": 0.0187, "step": 77610 }, { "epoch": 49.47100063734863, "grad_norm": 0.5357823371887207, "learning_rate": 8.85541037340879e-05, "loss": 0.023, "step": 77620 }, { "epoch": 49.47737412364563, "grad_norm": 0.9298370480537415, "learning_rate": 8.855059408863114e-05, "loss": 0.0257, "step": 77630 }, { "epoch": 49.48374760994264, "grad_norm": 0.4157021939754486, "learning_rate": 8.854708397474827e-05, "loss": 0.0224, "step": 77640 }, { "epoch": 49.49012109623964, "grad_norm": 0.453798770904541, "learning_rate": 8.854357339248196e-05, "loss": 0.034, "step": 77650 }, { "epoch": 49.49649458253665, "grad_norm": 0.5312669277191162, "learning_rate": 8.854006234187482e-05, "loss": 0.0231, "step": 77660 }, { "epoch": 49.50286806883365, "grad_norm": 0.7626932263374329, "learning_rate": 8.853655082296955e-05, "loss": 0.0251, "step": 77670 }, { "epoch": 49.50924155513066, "grad_norm": 0.4287428855895996, "learning_rate": 8.85330388358088e-05, "loss": 0.0228, "step": 77680 }, { "epoch": 49.51561504142766, "grad_norm": 0.5255755186080933, "learning_rate": 8.852952638043523e-05, "loss": 0.018, "step": 77690 }, { "epoch": 49.521988527724666, "grad_norm": 0.372689425945282, "learning_rate": 8.852601345689155e-05, "loss": 0.026, "step": 77700 }, { "epoch": 49.52836201402167, "grad_norm": 0.3128395974636078, "learning_rate": 8.852250006522042e-05, "loss": 0.0217, "step": 77710 }, { "epoch": 49.534735500318675, "grad_norm": 0.30768460035324097, "learning_rate": 8.851898620546456e-05, "loss": 0.0227, "step": 77720 }, { "epoch": 49.541108986615676, "grad_norm": 0.3918452262878418, "learning_rate": 8.851547187766665e-05, "loss": 0.0241, "step": 77730 }, { "epoch": 49.547482472912684, "grad_norm": 0.300589919090271, "learning_rate": 8.851195708186938e-05, "loss": 0.0167, "step": 77740 }, { "epoch": 49.553855959209685, "grad_norm": 0.35560616850852966, "learning_rate": 8.850844181811547e-05, "loss": 0.0198, "step": 77750 }, { "epoch": 49.56022944550669, "grad_norm": 0.403573215007782, "learning_rate": 8.850492608644763e-05, "loss": 0.0224, "step": 77760 }, { "epoch": 49.56660293180369, "grad_norm": 0.22804976999759674, "learning_rate": 8.85014098869086e-05, "loss": 0.0283, "step": 77770 }, { "epoch": 49.5729764181007, "grad_norm": 0.9392104148864746, "learning_rate": 8.849789321954109e-05, "loss": 0.0332, "step": 77780 }, { "epoch": 49.5793499043977, "grad_norm": 0.4760918915271759, "learning_rate": 8.849437608438782e-05, "loss": 0.0297, "step": 77790 }, { "epoch": 49.58572339069471, "grad_norm": 0.36500418186187744, "learning_rate": 8.849085848149152e-05, "loss": 0.0265, "step": 77800 }, { "epoch": 49.59209687699171, "grad_norm": 0.2690048813819885, "learning_rate": 8.848734041089497e-05, "loss": 0.0199, "step": 77810 }, { "epoch": 49.59847036328872, "grad_norm": 0.6772710680961609, "learning_rate": 8.84838218726409e-05, "loss": 0.027, "step": 77820 }, { "epoch": 49.60484384958572, "grad_norm": 0.5453587770462036, "learning_rate": 8.848030286677205e-05, "loss": 0.0208, "step": 77830 }, { "epoch": 49.61121733588273, "grad_norm": 1.0597739219665527, "learning_rate": 8.847678339333119e-05, "loss": 0.0169, "step": 77840 }, { "epoch": 49.617590822179736, "grad_norm": 0.5157111883163452, "learning_rate": 8.847326345236106e-05, "loss": 0.0203, "step": 77850 }, { "epoch": 49.62396430847674, "grad_norm": 0.5466475486755371, "learning_rate": 8.846974304390449e-05, "loss": 0.0214, "step": 77860 }, { "epoch": 49.630337794773745, "grad_norm": 0.6174289584159851, "learning_rate": 8.84662221680042e-05, "loss": 0.0249, "step": 77870 }, { "epoch": 49.636711281070745, "grad_norm": 0.5043948292732239, "learning_rate": 8.8462700824703e-05, "loss": 0.0302, "step": 77880 }, { "epoch": 49.64308476736775, "grad_norm": 0.551700234413147, "learning_rate": 8.845917901404368e-05, "loss": 0.027, "step": 77890 }, { "epoch": 49.649458253664754, "grad_norm": 0.5174895524978638, "learning_rate": 8.8455656736069e-05, "loss": 0.0217, "step": 77900 }, { "epoch": 49.65583173996176, "grad_norm": 0.5218281149864197, "learning_rate": 8.84521339908218e-05, "loss": 0.0314, "step": 77910 }, { "epoch": 49.66220522625876, "grad_norm": 0.72174072265625, "learning_rate": 8.844861077834486e-05, "loss": 0.0438, "step": 77920 }, { "epoch": 49.66857871255577, "grad_norm": 0.2980845272541046, "learning_rate": 8.8445087098681e-05, "loss": 0.0289, "step": 77930 }, { "epoch": 49.67495219885277, "grad_norm": 0.6927303671836853, "learning_rate": 8.844156295187303e-05, "loss": 0.0244, "step": 77940 }, { "epoch": 49.68132568514978, "grad_norm": 0.6235808730125427, "learning_rate": 8.843803833796378e-05, "loss": 0.0377, "step": 77950 }, { "epoch": 49.68769917144678, "grad_norm": 0.5893200635910034, "learning_rate": 8.843451325699608e-05, "loss": 0.0316, "step": 77960 }, { "epoch": 49.69407265774379, "grad_norm": 0.5723037719726562, "learning_rate": 8.843098770901274e-05, "loss": 0.0166, "step": 77970 }, { "epoch": 49.70044614404079, "grad_norm": 0.6646707653999329, "learning_rate": 8.842746169405663e-05, "loss": 0.0264, "step": 77980 }, { "epoch": 49.7068196303378, "grad_norm": 0.3388667404651642, "learning_rate": 8.842393521217057e-05, "loss": 0.0189, "step": 77990 }, { "epoch": 49.7131931166348, "grad_norm": 0.5286176204681396, "learning_rate": 8.842040826339741e-05, "loss": 0.0172, "step": 78000 }, { "epoch": 49.719566602931806, "grad_norm": 0.46457430720329285, "learning_rate": 8.841688084778001e-05, "loss": 0.0177, "step": 78010 }, { "epoch": 49.72594008922881, "grad_norm": 0.3973242938518524, "learning_rate": 8.841335296536127e-05, "loss": 0.0246, "step": 78020 }, { "epoch": 49.732313575525815, "grad_norm": 0.4042084217071533, "learning_rate": 8.8409824616184e-05, "loss": 0.0232, "step": 78030 }, { "epoch": 49.738687061822816, "grad_norm": 0.4817942976951599, "learning_rate": 8.840629580029109e-05, "loss": 0.0211, "step": 78040 }, { "epoch": 49.745060548119824, "grad_norm": 0.22444985806941986, "learning_rate": 8.840276651772544e-05, "loss": 0.0174, "step": 78050 }, { "epoch": 49.751434034416825, "grad_norm": 0.41225841641426086, "learning_rate": 8.839923676852992e-05, "loss": 0.0326, "step": 78060 }, { "epoch": 49.75780752071383, "grad_norm": 0.5309574604034424, "learning_rate": 8.83957065527474e-05, "loss": 0.0247, "step": 78070 }, { "epoch": 49.76418100701083, "grad_norm": 0.24984565377235413, "learning_rate": 8.839217587042081e-05, "loss": 0.0199, "step": 78080 }, { "epoch": 49.77055449330784, "grad_norm": 0.5951162576675415, "learning_rate": 8.838864472159303e-05, "loss": 0.0245, "step": 78090 }, { "epoch": 49.77692797960484, "grad_norm": 0.2755894362926483, "learning_rate": 8.838511310630697e-05, "loss": 0.0132, "step": 78100 }, { "epoch": 49.78330146590185, "grad_norm": 0.795556902885437, "learning_rate": 8.838158102460554e-05, "loss": 0.0362, "step": 78110 }, { "epoch": 49.78967495219885, "grad_norm": 0.3600795865058899, "learning_rate": 8.837804847653168e-05, "loss": 0.0262, "step": 78120 }, { "epoch": 49.79604843849586, "grad_norm": 0.3459025025367737, "learning_rate": 8.837451546212827e-05, "loss": 0.0308, "step": 78130 }, { "epoch": 49.80242192479286, "grad_norm": 0.28365886211395264, "learning_rate": 8.83709819814383e-05, "loss": 0.0241, "step": 78140 }, { "epoch": 49.80879541108987, "grad_norm": 0.317590594291687, "learning_rate": 8.836744803450464e-05, "loss": 0.0197, "step": 78150 }, { "epoch": 49.81516889738687, "grad_norm": 0.5156533718109131, "learning_rate": 8.836391362137027e-05, "loss": 0.0274, "step": 78160 }, { "epoch": 49.82154238368388, "grad_norm": 0.5411946773529053, "learning_rate": 8.836037874207813e-05, "loss": 0.0197, "step": 78170 }, { "epoch": 49.82791586998088, "grad_norm": 0.5274532437324524, "learning_rate": 8.835684339667117e-05, "loss": 0.0288, "step": 78180 }, { "epoch": 49.834289356277885, "grad_norm": 0.31607282161712646, "learning_rate": 8.835330758519236e-05, "loss": 0.0243, "step": 78190 }, { "epoch": 49.840662842574886, "grad_norm": 0.5689270496368408, "learning_rate": 8.834977130768462e-05, "loss": 0.0273, "step": 78200 }, { "epoch": 49.847036328871894, "grad_norm": 0.6697741150856018, "learning_rate": 8.834623456419097e-05, "loss": 0.0395, "step": 78210 }, { "epoch": 49.853409815168895, "grad_norm": 0.47497016191482544, "learning_rate": 8.834269735475438e-05, "loss": 0.0242, "step": 78220 }, { "epoch": 49.8597833014659, "grad_norm": 0.35743266344070435, "learning_rate": 8.83391596794178e-05, "loss": 0.0174, "step": 78230 }, { "epoch": 49.866156787762904, "grad_norm": 0.4587726294994354, "learning_rate": 8.833562153822422e-05, "loss": 0.0288, "step": 78240 }, { "epoch": 49.87253027405991, "grad_norm": 0.43858957290649414, "learning_rate": 8.833208293121667e-05, "loss": 0.0252, "step": 78250 }, { "epoch": 49.87890376035691, "grad_norm": 0.36850830912590027, "learning_rate": 8.83285438584381e-05, "loss": 0.0217, "step": 78260 }, { "epoch": 49.88527724665392, "grad_norm": 0.3949533700942993, "learning_rate": 8.832500431993153e-05, "loss": 0.0323, "step": 78270 }, { "epoch": 49.89165073295092, "grad_norm": 0.5647585391998291, "learning_rate": 8.832146431573999e-05, "loss": 0.0309, "step": 78280 }, { "epoch": 49.89802421924793, "grad_norm": 0.5529283881187439, "learning_rate": 8.831792384590647e-05, "loss": 0.031, "step": 78290 }, { "epoch": 49.90439770554493, "grad_norm": 0.3740663230419159, "learning_rate": 8.831438291047401e-05, "loss": 0.0189, "step": 78300 }, { "epoch": 49.91077119184194, "grad_norm": 0.5241754055023193, "learning_rate": 8.83108415094856e-05, "loss": 0.0196, "step": 78310 }, { "epoch": 49.91714467813894, "grad_norm": 0.34823673963546753, "learning_rate": 8.83072996429843e-05, "loss": 0.0248, "step": 78320 }, { "epoch": 49.92351816443595, "grad_norm": 0.4125213623046875, "learning_rate": 8.830375731101315e-05, "loss": 0.0245, "step": 78330 }, { "epoch": 49.92989165073295, "grad_norm": 0.4358454942703247, "learning_rate": 8.830021451361518e-05, "loss": 0.0242, "step": 78340 }, { "epoch": 49.936265137029956, "grad_norm": 0.4816979467868805, "learning_rate": 8.829667125083344e-05, "loss": 0.0148, "step": 78350 }, { "epoch": 49.94263862332696, "grad_norm": 0.5312575697898865, "learning_rate": 8.8293127522711e-05, "loss": 0.0134, "step": 78360 }, { "epoch": 49.949012109623965, "grad_norm": 1.1517137289047241, "learning_rate": 8.828958332929088e-05, "loss": 0.0288, "step": 78370 }, { "epoch": 49.955385595920966, "grad_norm": 0.6083505749702454, "learning_rate": 8.828603867061619e-05, "loss": 0.0261, "step": 78380 }, { "epoch": 49.961759082217974, "grad_norm": 0.3646330237388611, "learning_rate": 8.828249354672997e-05, "loss": 0.0167, "step": 78390 }, { "epoch": 49.968132568514974, "grad_norm": 0.5330575704574585, "learning_rate": 8.827894795767533e-05, "loss": 0.0206, "step": 78400 }, { "epoch": 49.97450605481198, "grad_norm": 0.36092036962509155, "learning_rate": 8.82754019034953e-05, "loss": 0.0201, "step": 78410 }, { "epoch": 49.98087954110898, "grad_norm": 0.3796204924583435, "learning_rate": 8.827185538423302e-05, "loss": 0.0223, "step": 78420 }, { "epoch": 49.98725302740599, "grad_norm": 0.5716899633407593, "learning_rate": 8.826830839993157e-05, "loss": 0.0233, "step": 78430 }, { "epoch": 49.99362651370299, "grad_norm": 0.45647770166397095, "learning_rate": 8.826476095063403e-05, "loss": 0.0172, "step": 78440 }, { "epoch": 50.0, "grad_norm": 0.6249266266822815, "learning_rate": 8.826121303638351e-05, "loss": 0.0096, "step": 78450 }, { "epoch": 50.00637348629701, "grad_norm": 0.3381504416465759, "learning_rate": 8.825766465722313e-05, "loss": 0.0174, "step": 78460 }, { "epoch": 50.01274697259401, "grad_norm": 0.3201063573360443, "learning_rate": 8.825411581319601e-05, "loss": 0.0233, "step": 78470 }, { "epoch": 50.01912045889102, "grad_norm": 0.48725444078445435, "learning_rate": 8.825056650434527e-05, "loss": 0.0166, "step": 78480 }, { "epoch": 50.02549394518802, "grad_norm": 0.5825637578964233, "learning_rate": 8.824701673071402e-05, "loss": 0.0216, "step": 78490 }, { "epoch": 50.031867431485026, "grad_norm": 0.433925598859787, "learning_rate": 8.82434664923454e-05, "loss": 0.0291, "step": 78500 }, { "epoch": 50.038240917782026, "grad_norm": 0.4949580132961273, "learning_rate": 8.823991578928256e-05, "loss": 0.0252, "step": 78510 }, { "epoch": 50.044614404079034, "grad_norm": 0.680374026298523, "learning_rate": 8.823636462156866e-05, "loss": 0.0177, "step": 78520 }, { "epoch": 50.050987890376035, "grad_norm": 0.41085904836654663, "learning_rate": 8.823281298924681e-05, "loss": 0.021, "step": 78530 }, { "epoch": 50.05736137667304, "grad_norm": 0.31770628690719604, "learning_rate": 8.822926089236019e-05, "loss": 0.0198, "step": 78540 }, { "epoch": 50.063734862970044, "grad_norm": 0.42102426290512085, "learning_rate": 8.822570833095193e-05, "loss": 0.0236, "step": 78550 }, { "epoch": 50.07010834926705, "grad_norm": 0.270768404006958, "learning_rate": 8.822215530506526e-05, "loss": 0.013, "step": 78560 }, { "epoch": 50.07648183556405, "grad_norm": 0.7138969302177429, "learning_rate": 8.82186018147433e-05, "loss": 0.0148, "step": 78570 }, { "epoch": 50.08285532186106, "grad_norm": 0.2616616487503052, "learning_rate": 8.821504786002925e-05, "loss": 0.0142, "step": 78580 }, { "epoch": 50.08922880815806, "grad_norm": 0.3224712908267975, "learning_rate": 8.821149344096628e-05, "loss": 0.0183, "step": 78590 }, { "epoch": 50.09560229445507, "grad_norm": 0.49666711688041687, "learning_rate": 8.820793855759758e-05, "loss": 0.0175, "step": 78600 }, { "epoch": 50.10197578075207, "grad_norm": 0.6033188104629517, "learning_rate": 8.820438320996637e-05, "loss": 0.0182, "step": 78610 }, { "epoch": 50.10834926704908, "grad_norm": 0.6989789009094238, "learning_rate": 8.820082739811582e-05, "loss": 0.025, "step": 78620 }, { "epoch": 50.11472275334608, "grad_norm": 0.3877696990966797, "learning_rate": 8.819727112208916e-05, "loss": 0.0353, "step": 78630 }, { "epoch": 50.12109623964309, "grad_norm": 0.3612739145755768, "learning_rate": 8.819371438192958e-05, "loss": 0.0185, "step": 78640 }, { "epoch": 50.12746972594009, "grad_norm": 0.4553041458129883, "learning_rate": 8.819015717768032e-05, "loss": 0.0231, "step": 78650 }, { "epoch": 50.133843212237096, "grad_norm": 0.25178971886634827, "learning_rate": 8.818659950938457e-05, "loss": 0.0253, "step": 78660 }, { "epoch": 50.1402166985341, "grad_norm": 0.5894650220870972, "learning_rate": 8.81830413770856e-05, "loss": 0.0293, "step": 78670 }, { "epoch": 50.146590184831105, "grad_norm": 0.3403888940811157, "learning_rate": 8.817948278082663e-05, "loss": 0.0256, "step": 78680 }, { "epoch": 50.152963671128106, "grad_norm": 0.652225136756897, "learning_rate": 8.817592372065089e-05, "loss": 0.0215, "step": 78690 }, { "epoch": 50.159337157425114, "grad_norm": 0.531622588634491, "learning_rate": 8.817236419660164e-05, "loss": 0.0188, "step": 78700 }, { "epoch": 50.165710643722115, "grad_norm": 0.2072288691997528, "learning_rate": 8.816880420872212e-05, "loss": 0.0161, "step": 78710 }, { "epoch": 50.17208413001912, "grad_norm": 0.4642229378223419, "learning_rate": 8.816524375705559e-05, "loss": 0.0217, "step": 78720 }, { "epoch": 50.17845761631612, "grad_norm": 0.2581069767475128, "learning_rate": 8.816168284164531e-05, "loss": 0.0209, "step": 78730 }, { "epoch": 50.18483110261313, "grad_norm": 0.3714772164821625, "learning_rate": 8.815812146253456e-05, "loss": 0.0153, "step": 78740 }, { "epoch": 50.19120458891013, "grad_norm": 0.5910926461219788, "learning_rate": 8.815455961976661e-05, "loss": 0.0199, "step": 78750 }, { "epoch": 50.19757807520714, "grad_norm": 0.4600505828857422, "learning_rate": 8.815099731338472e-05, "loss": 0.0221, "step": 78760 }, { "epoch": 50.20395156150414, "grad_norm": 0.4295461177825928, "learning_rate": 8.814743454343221e-05, "loss": 0.0211, "step": 78770 }, { "epoch": 50.21032504780115, "grad_norm": 0.698404848575592, "learning_rate": 8.814387130995234e-05, "loss": 0.0231, "step": 78780 }, { "epoch": 50.21669853409815, "grad_norm": 0.2891003489494324, "learning_rate": 8.814030761298843e-05, "loss": 0.0326, "step": 78790 }, { "epoch": 50.22307202039516, "grad_norm": 0.43265002965927124, "learning_rate": 8.813674345258378e-05, "loss": 0.0217, "step": 78800 }, { "epoch": 50.22944550669216, "grad_norm": 0.40825155377388, "learning_rate": 8.813317882878168e-05, "loss": 0.0164, "step": 78810 }, { "epoch": 50.23581899298917, "grad_norm": 0.3375880718231201, "learning_rate": 8.812961374162546e-05, "loss": 0.0299, "step": 78820 }, { "epoch": 50.24219247928617, "grad_norm": 0.5209476947784424, "learning_rate": 8.812604819115842e-05, "loss": 0.0215, "step": 78830 }, { "epoch": 50.248565965583175, "grad_norm": 0.6618402600288391, "learning_rate": 8.812248217742389e-05, "loss": 0.019, "step": 78840 }, { "epoch": 50.254939451880176, "grad_norm": 0.36991608142852783, "learning_rate": 8.811891570046523e-05, "loss": 0.0277, "step": 78850 }, { "epoch": 50.261312938177184, "grad_norm": 0.4550458490848541, "learning_rate": 8.811534876032575e-05, "loss": 0.0194, "step": 78860 }, { "epoch": 50.267686424474185, "grad_norm": 0.49720388650894165, "learning_rate": 8.811178135704879e-05, "loss": 0.0341, "step": 78870 }, { "epoch": 50.27405991077119, "grad_norm": 0.43963623046875, "learning_rate": 8.81082134906777e-05, "loss": 0.0199, "step": 78880 }, { "epoch": 50.280433397068194, "grad_norm": 0.7907137274742126, "learning_rate": 8.810464516125582e-05, "loss": 0.0189, "step": 78890 }, { "epoch": 50.2868068833652, "grad_norm": 0.3150918483734131, "learning_rate": 8.810107636882656e-05, "loss": 0.0221, "step": 78900 }, { "epoch": 50.2931803696622, "grad_norm": 0.4583601653575897, "learning_rate": 8.809750711343323e-05, "loss": 0.0161, "step": 78910 }, { "epoch": 50.29955385595921, "grad_norm": 0.2740141451358795, "learning_rate": 8.809393739511922e-05, "loss": 0.0432, "step": 78920 }, { "epoch": 50.30592734225621, "grad_norm": 0.4254174828529358, "learning_rate": 8.80903672139279e-05, "loss": 0.0207, "step": 78930 }, { "epoch": 50.31230082855322, "grad_norm": 0.3557502031326294, "learning_rate": 8.808679656990266e-05, "loss": 0.0169, "step": 78940 }, { "epoch": 50.31867431485022, "grad_norm": 0.4067841172218323, "learning_rate": 8.80832254630869e-05, "loss": 0.031, "step": 78950 }, { "epoch": 50.32504780114723, "grad_norm": 0.6919642686843872, "learning_rate": 8.807965389352397e-05, "loss": 0.0178, "step": 78960 }, { "epoch": 50.33142128744423, "grad_norm": 0.4111323654651642, "learning_rate": 8.807608186125729e-05, "loss": 0.021, "step": 78970 }, { "epoch": 50.33779477374124, "grad_norm": 0.6904878616333008, "learning_rate": 8.807250936633026e-05, "loss": 0.0275, "step": 78980 }, { "epoch": 50.34416826003824, "grad_norm": 0.9003934264183044, "learning_rate": 8.806893640878632e-05, "loss": 0.0256, "step": 78990 }, { "epoch": 50.350541746335246, "grad_norm": 0.4523234963417053, "learning_rate": 8.806536298866886e-05, "loss": 0.0344, "step": 79000 }, { "epoch": 50.35691523263225, "grad_norm": 0.5343559980392456, "learning_rate": 8.806178910602127e-05, "loss": 0.0231, "step": 79010 }, { "epoch": 50.363288718929255, "grad_norm": 0.3758242130279541, "learning_rate": 8.805821476088704e-05, "loss": 0.0242, "step": 79020 }, { "epoch": 50.369662205226255, "grad_norm": 0.3731200098991394, "learning_rate": 8.805463995330955e-05, "loss": 0.0236, "step": 79030 }, { "epoch": 50.37603569152326, "grad_norm": 0.3420630693435669, "learning_rate": 8.805106468333225e-05, "loss": 0.0253, "step": 79040 }, { "epoch": 50.382409177820264, "grad_norm": 0.5027868151664734, "learning_rate": 8.804748895099859e-05, "loss": 0.0168, "step": 79050 }, { "epoch": 50.38878266411727, "grad_norm": 0.4659039378166199, "learning_rate": 8.804391275635202e-05, "loss": 0.0312, "step": 79060 }, { "epoch": 50.39515615041427, "grad_norm": 0.35768458247184753, "learning_rate": 8.8040336099436e-05, "loss": 0.0207, "step": 79070 }, { "epoch": 50.40152963671128, "grad_norm": 0.7950899004936218, "learning_rate": 8.803675898029397e-05, "loss": 0.0292, "step": 79080 }, { "epoch": 50.40790312300829, "grad_norm": 0.23811520636081696, "learning_rate": 8.803318139896942e-05, "loss": 0.0245, "step": 79090 }, { "epoch": 50.41427660930529, "grad_norm": 0.43506231904029846, "learning_rate": 8.802960335550578e-05, "loss": 0.0252, "step": 79100 }, { "epoch": 50.4206500956023, "grad_norm": 0.22509349882602692, "learning_rate": 8.802602484994658e-05, "loss": 0.0289, "step": 79110 }, { "epoch": 50.4270235818993, "grad_norm": 0.34668299555778503, "learning_rate": 8.802244588233526e-05, "loss": 0.0235, "step": 79120 }, { "epoch": 50.43339706819631, "grad_norm": 0.43714094161987305, "learning_rate": 8.801886645271533e-05, "loss": 0.0278, "step": 79130 }, { "epoch": 50.43977055449331, "grad_norm": 0.40661999583244324, "learning_rate": 8.801528656113027e-05, "loss": 0.0161, "step": 79140 }, { "epoch": 50.446144040790315, "grad_norm": 0.3066137433052063, "learning_rate": 8.801170620762358e-05, "loss": 0.0178, "step": 79150 }, { "epoch": 50.452517527087316, "grad_norm": 0.4663435220718384, "learning_rate": 8.800812539223879e-05, "loss": 0.03, "step": 79160 }, { "epoch": 50.458891013384324, "grad_norm": 0.3341331481933594, "learning_rate": 8.800454411501938e-05, "loss": 0.0289, "step": 79170 }, { "epoch": 50.465264499681325, "grad_norm": 0.5391457676887512, "learning_rate": 8.800096237600888e-05, "loss": 0.0277, "step": 79180 }, { "epoch": 50.47163798597833, "grad_norm": 0.7094258069992065, "learning_rate": 8.799738017525081e-05, "loss": 0.0269, "step": 79190 }, { "epoch": 50.478011472275334, "grad_norm": 0.5689120888710022, "learning_rate": 8.799379751278869e-05, "loss": 0.0191, "step": 79200 }, { "epoch": 50.48438495857234, "grad_norm": 0.5518429279327393, "learning_rate": 8.799021438866606e-05, "loss": 0.0265, "step": 79210 }, { "epoch": 50.49075844486934, "grad_norm": 0.558884859085083, "learning_rate": 8.798663080292646e-05, "loss": 0.0217, "step": 79220 }, { "epoch": 50.49713193116635, "grad_norm": 0.760184645652771, "learning_rate": 8.798304675561342e-05, "loss": 0.0322, "step": 79230 }, { "epoch": 50.50350541746335, "grad_norm": 0.49008604884147644, "learning_rate": 8.797946224677052e-05, "loss": 0.033, "step": 79240 }, { "epoch": 50.50987890376036, "grad_norm": 0.5365548133850098, "learning_rate": 8.797587727644128e-05, "loss": 0.023, "step": 79250 }, { "epoch": 50.51625239005736, "grad_norm": 0.4222807288169861, "learning_rate": 8.797229184466927e-05, "loss": 0.0213, "step": 79260 }, { "epoch": 50.52262587635437, "grad_norm": 0.7180867791175842, "learning_rate": 8.796870595149806e-05, "loss": 0.0224, "step": 79270 }, { "epoch": 50.52899936265137, "grad_norm": 0.3896630108356476, "learning_rate": 8.796511959697124e-05, "loss": 0.016, "step": 79280 }, { "epoch": 50.53537284894838, "grad_norm": 0.4448973834514618, "learning_rate": 8.796153278113237e-05, "loss": 0.0189, "step": 79290 }, { "epoch": 50.54174633524538, "grad_norm": 0.27137893438339233, "learning_rate": 8.795794550402503e-05, "loss": 0.0232, "step": 79300 }, { "epoch": 50.548119821542386, "grad_norm": 0.6930257678031921, "learning_rate": 8.795435776569282e-05, "loss": 0.04, "step": 79310 }, { "epoch": 50.55449330783939, "grad_norm": 0.6881492137908936, "learning_rate": 8.795076956617932e-05, "loss": 0.0313, "step": 79320 }, { "epoch": 50.560866794136395, "grad_norm": 0.611892580986023, "learning_rate": 8.794718090552815e-05, "loss": 0.0224, "step": 79330 }, { "epoch": 50.567240280433396, "grad_norm": 0.39108553528785706, "learning_rate": 8.794359178378289e-05, "loss": 0.0163, "step": 79340 }, { "epoch": 50.5736137667304, "grad_norm": 0.6892337203025818, "learning_rate": 8.794000220098717e-05, "loss": 0.032, "step": 79350 }, { "epoch": 50.579987253027404, "grad_norm": 0.24506813287734985, "learning_rate": 8.79364121571846e-05, "loss": 0.0221, "step": 79360 }, { "epoch": 50.58636073932441, "grad_norm": 0.3120301365852356, "learning_rate": 8.79328216524188e-05, "loss": 0.0398, "step": 79370 }, { "epoch": 50.59273422562141, "grad_norm": 0.446262001991272, "learning_rate": 8.792923068673342e-05, "loss": 0.0249, "step": 79380 }, { "epoch": 50.59910771191842, "grad_norm": 0.5298762321472168, "learning_rate": 8.792563926017207e-05, "loss": 0.0246, "step": 79390 }, { "epoch": 50.60548119821542, "grad_norm": 0.5015619993209839, "learning_rate": 8.792204737277838e-05, "loss": 0.0333, "step": 79400 }, { "epoch": 50.61185468451243, "grad_norm": 0.6664139628410339, "learning_rate": 8.791845502459604e-05, "loss": 0.0422, "step": 79410 }, { "epoch": 50.61822817080943, "grad_norm": 0.5999925136566162, "learning_rate": 8.791486221566865e-05, "loss": 0.036, "step": 79420 }, { "epoch": 50.62460165710644, "grad_norm": 0.6568979620933533, "learning_rate": 8.791126894603987e-05, "loss": 0.0344, "step": 79430 }, { "epoch": 50.63097514340344, "grad_norm": 0.7174100279808044, "learning_rate": 8.790767521575341e-05, "loss": 0.0247, "step": 79440 }, { "epoch": 50.63734862970045, "grad_norm": 0.8572925925254822, "learning_rate": 8.790408102485287e-05, "loss": 0.026, "step": 79450 }, { "epoch": 50.64372211599745, "grad_norm": 0.31181514263153076, "learning_rate": 8.7900486373382e-05, "loss": 0.0276, "step": 79460 }, { "epoch": 50.650095602294456, "grad_norm": 0.39411067962646484, "learning_rate": 8.789689126138441e-05, "loss": 0.0231, "step": 79470 }, { "epoch": 50.65646908859146, "grad_norm": 0.5730124711990356, "learning_rate": 8.789329568890382e-05, "loss": 0.0238, "step": 79480 }, { "epoch": 50.662842574888465, "grad_norm": 0.4170571565628052, "learning_rate": 8.78896996559839e-05, "loss": 0.0315, "step": 79490 }, { "epoch": 50.669216061185466, "grad_norm": 0.3468725085258484, "learning_rate": 8.788610316266837e-05, "loss": 0.0311, "step": 79500 }, { "epoch": 50.675589547482474, "grad_norm": 0.4995209574699402, "learning_rate": 8.788250620900091e-05, "loss": 0.0174, "step": 79510 }, { "epoch": 50.681963033779475, "grad_norm": 0.366370290517807, "learning_rate": 8.787890879502522e-05, "loss": 0.0231, "step": 79520 }, { "epoch": 50.68833652007648, "grad_norm": 0.473082959651947, "learning_rate": 8.787531092078504e-05, "loss": 0.0169, "step": 79530 }, { "epoch": 50.694710006373484, "grad_norm": 0.3006952702999115, "learning_rate": 8.787171258632407e-05, "loss": 0.0174, "step": 79540 }, { "epoch": 50.70108349267049, "grad_norm": 0.4600659906864166, "learning_rate": 8.786811379168604e-05, "loss": 0.0443, "step": 79550 }, { "epoch": 50.70745697896749, "grad_norm": 0.32899153232574463, "learning_rate": 8.786451453691466e-05, "loss": 0.0151, "step": 79560 }, { "epoch": 50.7138304652645, "grad_norm": 0.5269507169723511, "learning_rate": 8.78609148220537e-05, "loss": 0.0242, "step": 79570 }, { "epoch": 50.7202039515615, "grad_norm": 0.5984697937965393, "learning_rate": 8.785731464714686e-05, "loss": 0.0324, "step": 79580 }, { "epoch": 50.72657743785851, "grad_norm": 0.558231770992279, "learning_rate": 8.78537140122379e-05, "loss": 0.0164, "step": 79590 }, { "epoch": 50.73295092415551, "grad_norm": 0.6354507207870483, "learning_rate": 8.785011291737058e-05, "loss": 0.0208, "step": 79600 }, { "epoch": 50.73932441045252, "grad_norm": 0.5819699168205261, "learning_rate": 8.784651136258865e-05, "loss": 0.0282, "step": 79610 }, { "epoch": 50.74569789674952, "grad_norm": 0.3994032144546509, "learning_rate": 8.784290934793589e-05, "loss": 0.0218, "step": 79620 }, { "epoch": 50.75207138304653, "grad_norm": 0.4292277693748474, "learning_rate": 8.783930687345603e-05, "loss": 0.024, "step": 79630 }, { "epoch": 50.75844486934353, "grad_norm": 1.104720950126648, "learning_rate": 8.783570393919288e-05, "loss": 0.0282, "step": 79640 }, { "epoch": 50.764818355640536, "grad_norm": 0.41390469670295715, "learning_rate": 8.78321005451902e-05, "loss": 0.0261, "step": 79650 }, { "epoch": 50.77119184193754, "grad_norm": 0.5971600413322449, "learning_rate": 8.782849669149178e-05, "loss": 0.0166, "step": 79660 }, { "epoch": 50.777565328234544, "grad_norm": 0.38011690974235535, "learning_rate": 8.78248923781414e-05, "loss": 0.0192, "step": 79670 }, { "epoch": 50.783938814531545, "grad_norm": 1.0127087831497192, "learning_rate": 8.782128760518287e-05, "loss": 0.0255, "step": 79680 }, { "epoch": 50.79031230082855, "grad_norm": 0.6223517060279846, "learning_rate": 8.781768237266e-05, "loss": 0.0237, "step": 79690 }, { "epoch": 50.796685787125554, "grad_norm": 0.43792593479156494, "learning_rate": 8.781407668061656e-05, "loss": 0.0244, "step": 79700 }, { "epoch": 50.80305927342256, "grad_norm": 0.4628196060657501, "learning_rate": 8.781047052909641e-05, "loss": 0.021, "step": 79710 }, { "epoch": 50.80943275971957, "grad_norm": 0.4288725256919861, "learning_rate": 8.780686391814333e-05, "loss": 0.0235, "step": 79720 }, { "epoch": 50.81580624601657, "grad_norm": 0.5100308656692505, "learning_rate": 8.780325684780115e-05, "loss": 0.0198, "step": 79730 }, { "epoch": 50.82217973231358, "grad_norm": 0.34135833382606506, "learning_rate": 8.779964931811373e-05, "loss": 0.0226, "step": 79740 }, { "epoch": 50.82855321861058, "grad_norm": 0.45776090025901794, "learning_rate": 8.779604132912487e-05, "loss": 0.0179, "step": 79750 }, { "epoch": 50.83492670490759, "grad_norm": 0.5911512970924377, "learning_rate": 8.779243288087842e-05, "loss": 0.0194, "step": 79760 }, { "epoch": 50.84130019120459, "grad_norm": 0.7046938538551331, "learning_rate": 8.778882397341823e-05, "loss": 0.0268, "step": 79770 }, { "epoch": 50.8476736775016, "grad_norm": 0.659308910369873, "learning_rate": 8.778521460678815e-05, "loss": 0.0232, "step": 79780 }, { "epoch": 50.8540471637986, "grad_norm": 0.30571916699409485, "learning_rate": 8.778160478103204e-05, "loss": 0.0194, "step": 79790 }, { "epoch": 50.860420650095605, "grad_norm": 0.2518063485622406, "learning_rate": 8.777799449619375e-05, "loss": 0.0185, "step": 79800 }, { "epoch": 50.866794136392606, "grad_norm": 0.6893629431724548, "learning_rate": 8.777438375231717e-05, "loss": 0.0279, "step": 79810 }, { "epoch": 50.873167622689614, "grad_norm": 0.45029497146606445, "learning_rate": 8.777077254944616e-05, "loss": 0.0276, "step": 79820 }, { "epoch": 50.879541108986615, "grad_norm": 0.4098087549209595, "learning_rate": 8.77671608876246e-05, "loss": 0.0209, "step": 79830 }, { "epoch": 50.88591459528362, "grad_norm": 0.4715461730957031, "learning_rate": 8.776354876689638e-05, "loss": 0.0196, "step": 79840 }, { "epoch": 50.892288081580624, "grad_norm": 0.7570445537567139, "learning_rate": 8.775993618730538e-05, "loss": 0.0244, "step": 79850 }, { "epoch": 50.89866156787763, "grad_norm": 0.44060954451560974, "learning_rate": 8.775632314889552e-05, "loss": 0.0252, "step": 79860 }, { "epoch": 50.90503505417463, "grad_norm": 0.4500226676464081, "learning_rate": 8.775270965171067e-05, "loss": 0.0249, "step": 79870 }, { "epoch": 50.91140854047164, "grad_norm": 0.5883831977844238, "learning_rate": 8.774909569579476e-05, "loss": 0.0238, "step": 79880 }, { "epoch": 50.91778202676864, "grad_norm": 0.9299945831298828, "learning_rate": 8.774548128119169e-05, "loss": 0.033, "step": 79890 }, { "epoch": 50.92415551306565, "grad_norm": 0.5143661499023438, "learning_rate": 8.77418664079454e-05, "loss": 0.0236, "step": 79900 }, { "epoch": 50.93052899936265, "grad_norm": 0.4556948244571686, "learning_rate": 8.773825107609979e-05, "loss": 0.0176, "step": 79910 }, { "epoch": 50.93690248565966, "grad_norm": 0.2951599061489105, "learning_rate": 8.773463528569879e-05, "loss": 0.016, "step": 79920 }, { "epoch": 50.94327597195666, "grad_norm": 0.4766428470611572, "learning_rate": 8.773101903678635e-05, "loss": 0.0218, "step": 79930 }, { "epoch": 50.94964945825367, "grad_norm": 0.23115359246730804, "learning_rate": 8.77274023294064e-05, "loss": 0.0235, "step": 79940 }, { "epoch": 50.95602294455067, "grad_norm": 0.4037020802497864, "learning_rate": 8.77237851636029e-05, "loss": 0.0204, "step": 79950 }, { "epoch": 50.962396430847676, "grad_norm": 0.3885551989078522, "learning_rate": 8.772016753941979e-05, "loss": 0.0255, "step": 79960 }, { "epoch": 50.96876991714468, "grad_norm": 0.6576778888702393, "learning_rate": 8.771654945690104e-05, "loss": 0.026, "step": 79970 }, { "epoch": 50.975143403441685, "grad_norm": 0.3049602210521698, "learning_rate": 8.771293091609058e-05, "loss": 0.0285, "step": 79980 }, { "epoch": 50.981516889738685, "grad_norm": 0.3631782531738281, "learning_rate": 8.770931191703243e-05, "loss": 0.014, "step": 79990 }, { "epoch": 50.98789037603569, "grad_norm": 0.7584400773048401, "learning_rate": 8.770569245977051e-05, "loss": 0.0211, "step": 80000 }, { "epoch": 50.994263862332694, "grad_norm": 0.8856493234634399, "learning_rate": 8.770207254434886e-05, "loss": 0.0248, "step": 80010 }, { "epoch": 51.0006373486297, "grad_norm": 0.3686481714248657, "learning_rate": 8.769845217081141e-05, "loss": 0.0178, "step": 80020 }, { "epoch": 51.0070108349267, "grad_norm": 0.5448732376098633, "learning_rate": 8.769483133920218e-05, "loss": 0.022, "step": 80030 }, { "epoch": 51.01338432122371, "grad_norm": 0.21480068564414978, "learning_rate": 8.769121004956516e-05, "loss": 0.0183, "step": 80040 }, { "epoch": 51.01975780752071, "grad_norm": 0.34940415620803833, "learning_rate": 8.768758830194434e-05, "loss": 0.0244, "step": 80050 }, { "epoch": 51.02613129381772, "grad_norm": 0.47757336497306824, "learning_rate": 8.768396609638374e-05, "loss": 0.0208, "step": 80060 }, { "epoch": 51.03250478011472, "grad_norm": 0.4913415014743805, "learning_rate": 8.768034343292737e-05, "loss": 0.0185, "step": 80070 }, { "epoch": 51.03887826641173, "grad_norm": 0.4642592966556549, "learning_rate": 8.767672031161926e-05, "loss": 0.0197, "step": 80080 }, { "epoch": 51.04525175270873, "grad_norm": 0.6955509781837463, "learning_rate": 8.767309673250344e-05, "loss": 0.0183, "step": 80090 }, { "epoch": 51.05162523900574, "grad_norm": 0.5535367727279663, "learning_rate": 8.76694726956239e-05, "loss": 0.0205, "step": 80100 }, { "epoch": 51.05799872530274, "grad_norm": 0.24094238877296448, "learning_rate": 8.76658482010247e-05, "loss": 0.0235, "step": 80110 }, { "epoch": 51.064372211599746, "grad_norm": 0.3038710057735443, "learning_rate": 8.766222324874989e-05, "loss": 0.0214, "step": 80120 }, { "epoch": 51.07074569789675, "grad_norm": 0.49538812041282654, "learning_rate": 8.765859783884351e-05, "loss": 0.0249, "step": 80130 }, { "epoch": 51.077119184193755, "grad_norm": 0.5099695324897766, "learning_rate": 8.765497197134962e-05, "loss": 0.0253, "step": 80140 }, { "epoch": 51.083492670490756, "grad_norm": 0.43623775243759155, "learning_rate": 8.765134564631225e-05, "loss": 0.0201, "step": 80150 }, { "epoch": 51.089866156787764, "grad_norm": 0.7550836801528931, "learning_rate": 8.76477188637755e-05, "loss": 0.0157, "step": 80160 }, { "epoch": 51.096239643084765, "grad_norm": 0.6480455994606018, "learning_rate": 8.764409162378342e-05, "loss": 0.0238, "step": 80170 }, { "epoch": 51.10261312938177, "grad_norm": 0.7898861169815063, "learning_rate": 8.764046392638008e-05, "loss": 0.028, "step": 80180 }, { "epoch": 51.10898661567877, "grad_norm": 0.6691673398017883, "learning_rate": 8.763683577160955e-05, "loss": 0.0294, "step": 80190 }, { "epoch": 51.11536010197578, "grad_norm": 0.5142627954483032, "learning_rate": 8.763320715951595e-05, "loss": 0.0143, "step": 80200 }, { "epoch": 51.12173358827278, "grad_norm": 0.34042656421661377, "learning_rate": 8.762957809014334e-05, "loss": 0.0256, "step": 80210 }, { "epoch": 51.12810707456979, "grad_norm": 0.6345179080963135, "learning_rate": 8.762594856353583e-05, "loss": 0.0278, "step": 80220 }, { "epoch": 51.13448056086679, "grad_norm": 0.29842260479927063, "learning_rate": 8.762231857973752e-05, "loss": 0.0209, "step": 80230 }, { "epoch": 51.1408540471638, "grad_norm": 0.5378405451774597, "learning_rate": 8.761868813879253e-05, "loss": 0.0272, "step": 80240 }, { "epoch": 51.1472275334608, "grad_norm": 0.534474790096283, "learning_rate": 8.761505724074495e-05, "loss": 0.0225, "step": 80250 }, { "epoch": 51.15360101975781, "grad_norm": 0.4871113896369934, "learning_rate": 8.761142588563891e-05, "loss": 0.0178, "step": 80260 }, { "epoch": 51.15997450605481, "grad_norm": 0.3796258270740509, "learning_rate": 8.760779407351853e-05, "loss": 0.0283, "step": 80270 }, { "epoch": 51.16634799235182, "grad_norm": 0.2343824803829193, "learning_rate": 8.760416180442795e-05, "loss": 0.0209, "step": 80280 }, { "epoch": 51.17272147864882, "grad_norm": 0.4615975320339203, "learning_rate": 8.760052907841129e-05, "loss": 0.0232, "step": 80290 }, { "epoch": 51.179094964945826, "grad_norm": 0.4010646343231201, "learning_rate": 8.759689589551272e-05, "loss": 0.0221, "step": 80300 }, { "epoch": 51.185468451242826, "grad_norm": 0.5545438528060913, "learning_rate": 8.759326225577635e-05, "loss": 0.0246, "step": 80310 }, { "epoch": 51.191841937539834, "grad_norm": 0.6877827048301697, "learning_rate": 8.758962815924636e-05, "loss": 0.0189, "step": 80320 }, { "epoch": 51.198215423836835, "grad_norm": 0.4778510332107544, "learning_rate": 8.75859936059669e-05, "loss": 0.015, "step": 80330 }, { "epoch": 51.20458891013384, "grad_norm": 0.751682460308075, "learning_rate": 8.758235859598212e-05, "loss": 0.0196, "step": 80340 }, { "epoch": 51.21096239643085, "grad_norm": 0.3829878568649292, "learning_rate": 8.757872312933622e-05, "loss": 0.0172, "step": 80350 }, { "epoch": 51.21733588272785, "grad_norm": 0.5323362946510315, "learning_rate": 8.757508720607332e-05, "loss": 0.0233, "step": 80360 }, { "epoch": 51.22370936902486, "grad_norm": 0.7504252195358276, "learning_rate": 8.757145082623766e-05, "loss": 0.0323, "step": 80370 }, { "epoch": 51.23008285532186, "grad_norm": 0.42253565788269043, "learning_rate": 8.75678139898734e-05, "loss": 0.0171, "step": 80380 }, { "epoch": 51.23645634161887, "grad_norm": 0.5518625378608704, "learning_rate": 8.756417669702472e-05, "loss": 0.0264, "step": 80390 }, { "epoch": 51.24282982791587, "grad_norm": 0.3475361764431, "learning_rate": 8.756053894773583e-05, "loss": 0.0264, "step": 80400 }, { "epoch": 51.24920331421288, "grad_norm": 0.35568803548812866, "learning_rate": 8.755690074205092e-05, "loss": 0.0152, "step": 80410 }, { "epoch": 51.25557680050988, "grad_norm": 0.40979093313217163, "learning_rate": 8.755326208001422e-05, "loss": 0.0188, "step": 80420 }, { "epoch": 51.261950286806886, "grad_norm": 0.5376774668693542, "learning_rate": 8.754962296166992e-05, "loss": 0.03, "step": 80430 }, { "epoch": 51.26832377310389, "grad_norm": 0.7888513207435608, "learning_rate": 8.754598338706224e-05, "loss": 0.0172, "step": 80440 }, { "epoch": 51.274697259400895, "grad_norm": 0.38959643244743347, "learning_rate": 8.754234335623544e-05, "loss": 0.0214, "step": 80450 }, { "epoch": 51.281070745697896, "grad_norm": 0.29616448283195496, "learning_rate": 8.75387028692337e-05, "loss": 0.0173, "step": 80460 }, { "epoch": 51.287444231994904, "grad_norm": 0.4908255338668823, "learning_rate": 8.753506192610129e-05, "loss": 0.0207, "step": 80470 }, { "epoch": 51.293817718291905, "grad_norm": 0.57209312915802, "learning_rate": 8.753142052688243e-05, "loss": 0.0158, "step": 80480 }, { "epoch": 51.30019120458891, "grad_norm": 0.3021024465560913, "learning_rate": 8.752777867162137e-05, "loss": 0.0226, "step": 80490 }, { "epoch": 51.306564690885914, "grad_norm": 0.21389898657798767, "learning_rate": 8.752413636036237e-05, "loss": 0.0136, "step": 80500 }, { "epoch": 51.31293817718292, "grad_norm": 0.3558087944984436, "learning_rate": 8.752049359314969e-05, "loss": 0.011, "step": 80510 }, { "epoch": 51.31931166347992, "grad_norm": 0.38619810342788696, "learning_rate": 8.751685037002759e-05, "loss": 0.0183, "step": 80520 }, { "epoch": 51.32568514977693, "grad_norm": 0.31868284940719604, "learning_rate": 8.751320669104034e-05, "loss": 0.0157, "step": 80530 }, { "epoch": 51.33205863607393, "grad_norm": 0.5700975656509399, "learning_rate": 8.75095625562322e-05, "loss": 0.0231, "step": 80540 }, { "epoch": 51.33843212237094, "grad_norm": 0.3430424630641937, "learning_rate": 8.750591796564746e-05, "loss": 0.0224, "step": 80550 }, { "epoch": 51.34480560866794, "grad_norm": 0.6331754922866821, "learning_rate": 8.750227291933042e-05, "loss": 0.0266, "step": 80560 }, { "epoch": 51.35117909496495, "grad_norm": 0.4514833092689514, "learning_rate": 8.749862741732532e-05, "loss": 0.0191, "step": 80570 }, { "epoch": 51.35755258126195, "grad_norm": 0.47303831577301025, "learning_rate": 8.749498145967654e-05, "loss": 0.0301, "step": 80580 }, { "epoch": 51.36392606755896, "grad_norm": 0.46819037199020386, "learning_rate": 8.74913350464283e-05, "loss": 0.023, "step": 80590 }, { "epoch": 51.37029955385596, "grad_norm": 0.45215895771980286, "learning_rate": 8.748768817762496e-05, "loss": 0.0311, "step": 80600 }, { "epoch": 51.376673040152966, "grad_norm": 0.4320497214794159, "learning_rate": 8.748404085331081e-05, "loss": 0.0353, "step": 80610 }, { "epoch": 51.38304652644997, "grad_norm": 0.7134445905685425, "learning_rate": 8.748039307353017e-05, "loss": 0.0198, "step": 80620 }, { "epoch": 51.389420012746974, "grad_norm": 0.5475758910179138, "learning_rate": 8.747674483832737e-05, "loss": 0.0215, "step": 80630 }, { "epoch": 51.395793499043975, "grad_norm": 0.42564475536346436, "learning_rate": 8.747309614774673e-05, "loss": 0.0355, "step": 80640 }, { "epoch": 51.40216698534098, "grad_norm": 0.32626304030418396, "learning_rate": 8.746944700183259e-05, "loss": 0.0216, "step": 80650 }, { "epoch": 51.408540471637984, "grad_norm": 0.22941851615905762, "learning_rate": 8.74657974006293e-05, "loss": 0.0239, "step": 80660 }, { "epoch": 51.41491395793499, "grad_norm": 0.4409743845462799, "learning_rate": 8.746214734418119e-05, "loss": 0.0219, "step": 80670 }, { "epoch": 51.42128744423199, "grad_norm": 0.3632958233356476, "learning_rate": 8.745849683253262e-05, "loss": 0.0191, "step": 80680 }, { "epoch": 51.427660930529, "grad_norm": 0.18139545619487762, "learning_rate": 8.745484586572796e-05, "loss": 0.0271, "step": 80690 }, { "epoch": 51.434034416826, "grad_norm": 0.528667151927948, "learning_rate": 8.745119444381154e-05, "loss": 0.0305, "step": 80700 }, { "epoch": 51.44040790312301, "grad_norm": 0.31671062111854553, "learning_rate": 8.744754256682776e-05, "loss": 0.0149, "step": 80710 }, { "epoch": 51.44678138942001, "grad_norm": 0.490806519985199, "learning_rate": 8.744389023482098e-05, "loss": 0.0382, "step": 80720 }, { "epoch": 51.45315487571702, "grad_norm": 0.3629381060600281, "learning_rate": 8.744023744783558e-05, "loss": 0.0303, "step": 80730 }, { "epoch": 51.45952836201402, "grad_norm": 0.4655114710330963, "learning_rate": 8.743658420591595e-05, "loss": 0.0279, "step": 80740 }, { "epoch": 51.46590184831103, "grad_norm": 0.6693554520606995, "learning_rate": 8.743293050910647e-05, "loss": 0.0248, "step": 80750 }, { "epoch": 51.47227533460803, "grad_norm": 0.35395073890686035, "learning_rate": 8.742927635745155e-05, "loss": 0.0205, "step": 80760 }, { "epoch": 51.478648820905036, "grad_norm": 0.38700437545776367, "learning_rate": 8.742562175099558e-05, "loss": 0.0311, "step": 80770 }, { "epoch": 51.48502230720204, "grad_norm": 0.39843520522117615, "learning_rate": 8.742196668978297e-05, "loss": 0.0233, "step": 80780 }, { "epoch": 51.491395793499045, "grad_norm": 0.5317781567573547, "learning_rate": 8.741831117385812e-05, "loss": 0.0124, "step": 80790 }, { "epoch": 51.497769279796046, "grad_norm": 0.3286365270614624, "learning_rate": 8.741465520326548e-05, "loss": 0.0132, "step": 80800 }, { "epoch": 51.504142766093054, "grad_norm": 0.4258129596710205, "learning_rate": 8.741099877804943e-05, "loss": 0.0513, "step": 80810 }, { "epoch": 51.510516252390055, "grad_norm": 0.3511577844619751, "learning_rate": 8.740734189825443e-05, "loss": 0.0251, "step": 80820 }, { "epoch": 51.51688973868706, "grad_norm": 0.32276976108551025, "learning_rate": 8.740368456392493e-05, "loss": 0.0261, "step": 80830 }, { "epoch": 51.52326322498406, "grad_norm": 0.3797406256198883, "learning_rate": 8.740002677510532e-05, "loss": 0.022, "step": 80840 }, { "epoch": 51.52963671128107, "grad_norm": 0.3691493272781372, "learning_rate": 8.739636853184008e-05, "loss": 0.021, "step": 80850 }, { "epoch": 51.53601019757807, "grad_norm": 0.4678703248500824, "learning_rate": 8.739270983417366e-05, "loss": 0.0187, "step": 80860 }, { "epoch": 51.54238368387508, "grad_norm": 0.7122126221656799, "learning_rate": 8.738905068215052e-05, "loss": 0.0239, "step": 80870 }, { "epoch": 51.54875717017208, "grad_norm": 0.34224560856819153, "learning_rate": 8.73853910758151e-05, "loss": 0.0213, "step": 80880 }, { "epoch": 51.55513065646909, "grad_norm": 0.6818312406539917, "learning_rate": 8.738173101521189e-05, "loss": 0.0314, "step": 80890 }, { "epoch": 51.56150414276609, "grad_norm": 0.48439544439315796, "learning_rate": 8.737807050038534e-05, "loss": 0.0357, "step": 80900 }, { "epoch": 51.5678776290631, "grad_norm": 0.49195408821105957, "learning_rate": 8.737440953137997e-05, "loss": 0.0351, "step": 80910 }, { "epoch": 51.5742511153601, "grad_norm": 0.5004852414131165, "learning_rate": 8.737074810824019e-05, "loss": 0.0146, "step": 80920 }, { "epoch": 51.58062460165711, "grad_norm": 0.7326125502586365, "learning_rate": 8.736708623101058e-05, "loss": 0.0212, "step": 80930 }, { "epoch": 51.58699808795411, "grad_norm": 0.31113579869270325, "learning_rate": 8.736342389973557e-05, "loss": 0.0139, "step": 80940 }, { "epoch": 51.593371574251115, "grad_norm": 0.5144744515419006, "learning_rate": 8.73597611144597e-05, "loss": 0.0182, "step": 80950 }, { "epoch": 51.599745060548116, "grad_norm": 0.4252970218658447, "learning_rate": 8.735609787522745e-05, "loss": 0.0228, "step": 80960 }, { "epoch": 51.606118546845124, "grad_norm": 0.26974132657051086, "learning_rate": 8.735243418208333e-05, "loss": 0.0234, "step": 80970 }, { "epoch": 51.61249203314213, "grad_norm": 0.3215683698654175, "learning_rate": 8.734877003507188e-05, "loss": 0.0296, "step": 80980 }, { "epoch": 51.61886551943913, "grad_norm": 0.3875557780265808, "learning_rate": 8.73451054342376e-05, "loss": 0.0248, "step": 80990 }, { "epoch": 51.62523900573614, "grad_norm": 0.40797123312950134, "learning_rate": 8.734144037962504e-05, "loss": 0.0192, "step": 81000 }, { "epoch": 51.63161249203314, "grad_norm": 0.49092209339141846, "learning_rate": 8.733777487127872e-05, "loss": 0.0162, "step": 81010 }, { "epoch": 51.63798597833015, "grad_norm": 0.19122110307216644, "learning_rate": 8.733410890924318e-05, "loss": 0.0148, "step": 81020 }, { "epoch": 51.64435946462715, "grad_norm": 0.4005959630012512, "learning_rate": 8.733044249356298e-05, "loss": 0.0224, "step": 81030 }, { "epoch": 51.65073295092416, "grad_norm": 0.4438916742801666, "learning_rate": 8.732677562428264e-05, "loss": 0.0267, "step": 81040 }, { "epoch": 51.65710643722116, "grad_norm": 0.515895426273346, "learning_rate": 8.732310830144675e-05, "loss": 0.0218, "step": 81050 }, { "epoch": 51.66347992351817, "grad_norm": 0.3400057256221771, "learning_rate": 8.731944052509985e-05, "loss": 0.0144, "step": 81060 }, { "epoch": 51.66985340981517, "grad_norm": 0.3960772454738617, "learning_rate": 8.731577229528652e-05, "loss": 0.0242, "step": 81070 }, { "epoch": 51.676226896112176, "grad_norm": 0.37429627776145935, "learning_rate": 8.73121036120513e-05, "loss": 0.0265, "step": 81080 }, { "epoch": 51.68260038240918, "grad_norm": 0.3340900242328644, "learning_rate": 8.730843447543882e-05, "loss": 0.0244, "step": 81090 }, { "epoch": 51.688973868706185, "grad_norm": 0.386396586894989, "learning_rate": 8.730476488549362e-05, "loss": 0.0175, "step": 81100 }, { "epoch": 51.695347355003186, "grad_norm": 0.511213481426239, "learning_rate": 8.73010948422603e-05, "loss": 0.0235, "step": 81110 }, { "epoch": 51.701720841300194, "grad_norm": 0.48626789450645447, "learning_rate": 8.729742434578348e-05, "loss": 0.0139, "step": 81120 }, { "epoch": 51.708094327597195, "grad_norm": 0.45288586616516113, "learning_rate": 8.729375339610773e-05, "loss": 0.0177, "step": 81130 }, { "epoch": 51.7144678138942, "grad_norm": 0.19390925765037537, "learning_rate": 8.729008199327766e-05, "loss": 0.0226, "step": 81140 }, { "epoch": 51.7208413001912, "grad_norm": 0.5896400213241577, "learning_rate": 8.728641013733789e-05, "loss": 0.0154, "step": 81150 }, { "epoch": 51.72721478648821, "grad_norm": 0.504985511302948, "learning_rate": 8.728273782833303e-05, "loss": 0.0147, "step": 81160 }, { "epoch": 51.73358827278521, "grad_norm": 0.7524563670158386, "learning_rate": 8.727906506630769e-05, "loss": 0.0209, "step": 81170 }, { "epoch": 51.73996175908222, "grad_norm": 0.22519922256469727, "learning_rate": 8.727539185130654e-05, "loss": 0.0248, "step": 81180 }, { "epoch": 51.74633524537922, "grad_norm": 0.4732171893119812, "learning_rate": 8.727171818337417e-05, "loss": 0.0205, "step": 81190 }, { "epoch": 51.75270873167623, "grad_norm": 0.5426949858665466, "learning_rate": 8.726804406255525e-05, "loss": 0.0294, "step": 81200 }, { "epoch": 51.75908221797323, "grad_norm": 0.46127068996429443, "learning_rate": 8.726436948889438e-05, "loss": 0.0241, "step": 81210 }, { "epoch": 51.76545570427024, "grad_norm": 0.40696483850479126, "learning_rate": 8.726069446243625e-05, "loss": 0.0128, "step": 81220 }, { "epoch": 51.77182919056724, "grad_norm": 0.35702329874038696, "learning_rate": 8.725701898322551e-05, "loss": 0.0174, "step": 81230 }, { "epoch": 51.77820267686425, "grad_norm": 0.28448814153671265, "learning_rate": 8.725334305130682e-05, "loss": 0.0173, "step": 81240 }, { "epoch": 51.78457616316125, "grad_norm": 0.3182666599750519, "learning_rate": 8.724966666672481e-05, "loss": 0.0157, "step": 81250 }, { "epoch": 51.790949649458256, "grad_norm": 0.3836197555065155, "learning_rate": 8.724598982952421e-05, "loss": 0.0196, "step": 81260 }, { "epoch": 51.797323135755256, "grad_norm": 0.5858541131019592, "learning_rate": 8.724231253974964e-05, "loss": 0.0233, "step": 81270 }, { "epoch": 51.803696622052264, "grad_norm": 0.41204991936683655, "learning_rate": 8.723863479744583e-05, "loss": 0.0215, "step": 81280 }, { "epoch": 51.810070108349265, "grad_norm": 0.4540404975414276, "learning_rate": 8.723495660265745e-05, "loss": 0.0141, "step": 81290 }, { "epoch": 51.81644359464627, "grad_norm": 0.5977151989936829, "learning_rate": 8.723127795542919e-05, "loss": 0.0183, "step": 81300 }, { "epoch": 51.822817080943274, "grad_norm": 0.3443578779697418, "learning_rate": 8.722759885580573e-05, "loss": 0.0212, "step": 81310 }, { "epoch": 51.82919056724028, "grad_norm": 0.4912019371986389, "learning_rate": 8.722391930383182e-05, "loss": 0.0321, "step": 81320 }, { "epoch": 51.83556405353728, "grad_norm": 0.3972175419330597, "learning_rate": 8.722023929955213e-05, "loss": 0.0219, "step": 81330 }, { "epoch": 51.84193753983429, "grad_norm": 0.4738822281360626, "learning_rate": 8.72165588430114e-05, "loss": 0.0233, "step": 81340 }, { "epoch": 51.84831102613129, "grad_norm": 0.6504071354866028, "learning_rate": 8.721287793425434e-05, "loss": 0.0305, "step": 81350 }, { "epoch": 51.8546845124283, "grad_norm": 0.6047699451446533, "learning_rate": 8.720919657332566e-05, "loss": 0.0319, "step": 81360 }, { "epoch": 51.8610579987253, "grad_norm": 0.47295376658439636, "learning_rate": 8.720551476027015e-05, "loss": 0.0347, "step": 81370 }, { "epoch": 51.86743148502231, "grad_norm": 0.5706363916397095, "learning_rate": 8.720183249513248e-05, "loss": 0.0418, "step": 81380 }, { "epoch": 51.87380497131931, "grad_norm": 0.40852758288383484, "learning_rate": 8.719814977795741e-05, "loss": 0.0174, "step": 81390 }, { "epoch": 51.88017845761632, "grad_norm": 0.48534706234931946, "learning_rate": 8.719446660878973e-05, "loss": 0.0223, "step": 81400 }, { "epoch": 51.88655194391332, "grad_norm": 0.6410535573959351, "learning_rate": 8.719078298767415e-05, "loss": 0.0184, "step": 81410 }, { "epoch": 51.892925430210326, "grad_norm": 0.4877389967441559, "learning_rate": 8.718709891465544e-05, "loss": 0.024, "step": 81420 }, { "epoch": 51.89929891650733, "grad_norm": 0.5058755874633789, "learning_rate": 8.718341438977837e-05, "loss": 0.0194, "step": 81430 }, { "epoch": 51.905672402804335, "grad_norm": 0.31806501746177673, "learning_rate": 8.71797294130877e-05, "loss": 0.035, "step": 81440 }, { "epoch": 51.912045889101336, "grad_norm": 0.3037961721420288, "learning_rate": 8.717604398462822e-05, "loss": 0.0179, "step": 81450 }, { "epoch": 51.918419375398344, "grad_norm": 0.5935440063476562, "learning_rate": 8.717235810444471e-05, "loss": 0.0216, "step": 81460 }, { "epoch": 51.924792861695344, "grad_norm": 0.5404843091964722, "learning_rate": 8.716867177258194e-05, "loss": 0.0142, "step": 81470 }, { "epoch": 51.93116634799235, "grad_norm": 0.5413027405738831, "learning_rate": 8.716498498908473e-05, "loss": 0.0224, "step": 81480 }, { "epoch": 51.93753983428935, "grad_norm": 0.3893929421901703, "learning_rate": 8.716129775399786e-05, "loss": 0.0206, "step": 81490 }, { "epoch": 51.94391332058636, "grad_norm": 0.3883205056190491, "learning_rate": 8.715761006736614e-05, "loss": 0.0208, "step": 81500 }, { "epoch": 51.95028680688336, "grad_norm": 0.38589707016944885, "learning_rate": 8.715392192923437e-05, "loss": 0.0243, "step": 81510 }, { "epoch": 51.95666029318037, "grad_norm": 0.1656983643770218, "learning_rate": 8.715023333964736e-05, "loss": 0.0225, "step": 81520 }, { "epoch": 51.96303377947737, "grad_norm": 0.5262556076049805, "learning_rate": 8.714654429864994e-05, "loss": 0.0148, "step": 81530 }, { "epoch": 51.96940726577438, "grad_norm": 0.308219850063324, "learning_rate": 8.714285480628694e-05, "loss": 0.0243, "step": 81540 }, { "epoch": 51.97578075207138, "grad_norm": 0.6634693145751953, "learning_rate": 8.713916486260319e-05, "loss": 0.0235, "step": 81550 }, { "epoch": 51.98215423836839, "grad_norm": 1.0083328485488892, "learning_rate": 8.713547446764352e-05, "loss": 0.0326, "step": 81560 }, { "epoch": 51.98852772466539, "grad_norm": 0.46199601888656616, "learning_rate": 8.713178362145278e-05, "loss": 0.0208, "step": 81570 }, { "epoch": 51.9949012109624, "grad_norm": 0.5431836843490601, "learning_rate": 8.712809232407579e-05, "loss": 0.0211, "step": 81580 }, { "epoch": 52.001274697259404, "grad_norm": 0.9592965841293335, "learning_rate": 8.712440057555744e-05, "loss": 0.0247, "step": 81590 }, { "epoch": 52.007648183556405, "grad_norm": 0.3221082389354706, "learning_rate": 8.712070837594257e-05, "loss": 0.016, "step": 81600 }, { "epoch": 52.01402166985341, "grad_norm": 0.5910373330116272, "learning_rate": 8.711701572527605e-05, "loss": 0.0214, "step": 81610 }, { "epoch": 52.020395156150414, "grad_norm": 0.2582823634147644, "learning_rate": 8.711332262360274e-05, "loss": 0.0159, "step": 81620 }, { "epoch": 52.02676864244742, "grad_norm": 0.3554055988788605, "learning_rate": 8.710962907096751e-05, "loss": 0.0187, "step": 81630 }, { "epoch": 52.03314212874442, "grad_norm": 0.3477265536785126, "learning_rate": 8.710593506741526e-05, "loss": 0.0209, "step": 81640 }, { "epoch": 52.03951561504143, "grad_norm": 0.3982335925102234, "learning_rate": 8.710224061299086e-05, "loss": 0.0254, "step": 81650 }, { "epoch": 52.04588910133843, "grad_norm": 0.3781326413154602, "learning_rate": 8.709854570773921e-05, "loss": 0.0348, "step": 81660 }, { "epoch": 52.05226258763544, "grad_norm": 0.5378394722938538, "learning_rate": 8.70948503517052e-05, "loss": 0.0174, "step": 81670 }, { "epoch": 52.05863607393244, "grad_norm": 0.8060970902442932, "learning_rate": 8.709115454493375e-05, "loss": 0.0227, "step": 81680 }, { "epoch": 52.06500956022945, "grad_norm": 0.4416046738624573, "learning_rate": 8.708745828746973e-05, "loss": 0.0186, "step": 81690 }, { "epoch": 52.07138304652645, "grad_norm": 0.6488023996353149, "learning_rate": 8.708376157935808e-05, "loss": 0.019, "step": 81700 }, { "epoch": 52.07775653282346, "grad_norm": 0.3122628927230835, "learning_rate": 8.708006442064372e-05, "loss": 0.032, "step": 81710 }, { "epoch": 52.08413001912046, "grad_norm": 0.5989399552345276, "learning_rate": 8.707636681137156e-05, "loss": 0.0189, "step": 81720 }, { "epoch": 52.090503505417466, "grad_norm": 0.5266742706298828, "learning_rate": 8.707266875158655e-05, "loss": 0.0168, "step": 81730 }, { "epoch": 52.09687699171447, "grad_norm": 0.3032207489013672, "learning_rate": 8.70689702413336e-05, "loss": 0.0275, "step": 81740 }, { "epoch": 52.103250478011475, "grad_norm": 0.5930258631706238, "learning_rate": 8.706527128065767e-05, "loss": 0.0333, "step": 81750 }, { "epoch": 52.109623964308476, "grad_norm": 0.5194653868675232, "learning_rate": 8.706157186960371e-05, "loss": 0.0199, "step": 81760 }, { "epoch": 52.115997450605484, "grad_norm": 0.705303966999054, "learning_rate": 8.705787200821664e-05, "loss": 0.0206, "step": 81770 }, { "epoch": 52.122370936902485, "grad_norm": 0.29638293385505676, "learning_rate": 8.705417169654145e-05, "loss": 0.0323, "step": 81780 }, { "epoch": 52.12874442319949, "grad_norm": 0.21216413378715515, "learning_rate": 8.705047093462307e-05, "loss": 0.0181, "step": 81790 }, { "epoch": 52.13511790949649, "grad_norm": 0.5359810590744019, "learning_rate": 8.704676972250651e-05, "loss": 0.0151, "step": 81800 }, { "epoch": 52.1414913957935, "grad_norm": 0.2751602232456207, "learning_rate": 8.70430680602367e-05, "loss": 0.0323, "step": 81810 }, { "epoch": 52.1478648820905, "grad_norm": 0.4097042679786682, "learning_rate": 8.703936594785865e-05, "loss": 0.0192, "step": 81820 }, { "epoch": 52.15423836838751, "grad_norm": 0.37493911385536194, "learning_rate": 8.703566338541733e-05, "loss": 0.0166, "step": 81830 }, { "epoch": 52.16061185468451, "grad_norm": 0.6698934435844421, "learning_rate": 8.703196037295773e-05, "loss": 0.026, "step": 81840 }, { "epoch": 52.16698534098152, "grad_norm": 0.36260277032852173, "learning_rate": 8.702825691052486e-05, "loss": 0.0309, "step": 81850 }, { "epoch": 52.17335882727852, "grad_norm": 0.1413009911775589, "learning_rate": 8.70245529981637e-05, "loss": 0.0101, "step": 81860 }, { "epoch": 52.17973231357553, "grad_norm": 0.4117658734321594, "learning_rate": 8.702084863591926e-05, "loss": 0.0273, "step": 81870 }, { "epoch": 52.18610579987253, "grad_norm": 0.47552376985549927, "learning_rate": 8.701714382383655e-05, "loss": 0.0185, "step": 81880 }, { "epoch": 52.19247928616954, "grad_norm": 0.6598080992698669, "learning_rate": 8.701343856196061e-05, "loss": 0.021, "step": 81890 }, { "epoch": 52.19885277246654, "grad_norm": 0.3817600607872009, "learning_rate": 8.700973285033643e-05, "loss": 0.0289, "step": 81900 }, { "epoch": 52.205226258763545, "grad_norm": 0.5875312089920044, "learning_rate": 8.700602668900906e-05, "loss": 0.0217, "step": 81910 }, { "epoch": 52.211599745060546, "grad_norm": 0.37648648023605347, "learning_rate": 8.700232007802351e-05, "loss": 0.0226, "step": 81920 }, { "epoch": 52.217973231357554, "grad_norm": 0.5557661056518555, "learning_rate": 8.699861301742486e-05, "loss": 0.0194, "step": 81930 }, { "epoch": 52.224346717654555, "grad_norm": 0.7103478312492371, "learning_rate": 8.699490550725811e-05, "loss": 0.0197, "step": 81940 }, { "epoch": 52.23072020395156, "grad_norm": 0.3458067774772644, "learning_rate": 8.699119754756835e-05, "loss": 0.0196, "step": 81950 }, { "epoch": 52.237093690248564, "grad_norm": 0.5022388100624084, "learning_rate": 8.69874891384006e-05, "loss": 0.0181, "step": 81960 }, { "epoch": 52.24346717654557, "grad_norm": 0.4194103479385376, "learning_rate": 8.698378027979994e-05, "loss": 0.0202, "step": 81970 }, { "epoch": 52.24984066284257, "grad_norm": 0.3981282114982605, "learning_rate": 8.698007097181142e-05, "loss": 0.0306, "step": 81980 }, { "epoch": 52.25621414913958, "grad_norm": 0.5014811158180237, "learning_rate": 8.697636121448014e-05, "loss": 0.0189, "step": 81990 }, { "epoch": 52.26258763543658, "grad_norm": 0.3814089596271515, "learning_rate": 8.697265100785115e-05, "loss": 0.0254, "step": 82000 }, { "epoch": 52.26896112173359, "grad_norm": 0.4989509880542755, "learning_rate": 8.696894035196953e-05, "loss": 0.0155, "step": 82010 }, { "epoch": 52.27533460803059, "grad_norm": 0.18871840834617615, "learning_rate": 8.696522924688041e-05, "loss": 0.0206, "step": 82020 }, { "epoch": 52.2817080943276, "grad_norm": 0.2247210443019867, "learning_rate": 8.696151769262883e-05, "loss": 0.027, "step": 82030 }, { "epoch": 52.2880815806246, "grad_norm": 0.3103176951408386, "learning_rate": 8.695780568925992e-05, "loss": 0.0327, "step": 82040 }, { "epoch": 52.29445506692161, "grad_norm": 0.6794428825378418, "learning_rate": 8.695409323681877e-05, "loss": 0.0204, "step": 82050 }, { "epoch": 52.30082855321861, "grad_norm": 0.3984158933162689, "learning_rate": 8.695038033535052e-05, "loss": 0.0168, "step": 82060 }, { "epoch": 52.307202039515616, "grad_norm": 0.33990049362182617, "learning_rate": 8.694666698490024e-05, "loss": 0.0217, "step": 82070 }, { "epoch": 52.31357552581262, "grad_norm": 0.3129803538322449, "learning_rate": 8.694295318551309e-05, "loss": 0.0202, "step": 82080 }, { "epoch": 52.319949012109625, "grad_norm": 0.21619196236133575, "learning_rate": 8.693923893723416e-05, "loss": 0.0222, "step": 82090 }, { "epoch": 52.326322498406626, "grad_norm": 0.24451348185539246, "learning_rate": 8.69355242401086e-05, "loss": 0.0229, "step": 82100 }, { "epoch": 52.33269598470363, "grad_norm": 0.28829383850097656, "learning_rate": 8.693180909418156e-05, "loss": 0.0209, "step": 82110 }, { "epoch": 52.339069471000634, "grad_norm": 0.6541208028793335, "learning_rate": 8.692809349949816e-05, "loss": 0.0183, "step": 82120 }, { "epoch": 52.34544295729764, "grad_norm": 0.5769757032394409, "learning_rate": 8.692437745610357e-05, "loss": 0.0213, "step": 82130 }, { "epoch": 52.35181644359464, "grad_norm": 0.40215086936950684, "learning_rate": 8.692066096404293e-05, "loss": 0.0151, "step": 82140 }, { "epoch": 52.35818992989165, "grad_norm": 0.4064716696739197, "learning_rate": 8.691694402336139e-05, "loss": 0.0358, "step": 82150 }, { "epoch": 52.36456341618865, "grad_norm": 0.5725083351135254, "learning_rate": 8.691322663410413e-05, "loss": 0.0183, "step": 82160 }, { "epoch": 52.37093690248566, "grad_norm": 0.5562969446182251, "learning_rate": 8.690950879631632e-05, "loss": 0.0228, "step": 82170 }, { "epoch": 52.37731038878266, "grad_norm": 0.4773479998111725, "learning_rate": 8.690579051004312e-05, "loss": 0.0212, "step": 82180 }, { "epoch": 52.38368387507967, "grad_norm": 0.36255648732185364, "learning_rate": 8.690207177532971e-05, "loss": 0.0239, "step": 82190 }, { "epoch": 52.39005736137667, "grad_norm": 0.25858187675476074, "learning_rate": 8.689835259222132e-05, "loss": 0.0125, "step": 82200 }, { "epoch": 52.39643084767368, "grad_norm": 0.41720646619796753, "learning_rate": 8.689463296076308e-05, "loss": 0.0382, "step": 82210 }, { "epoch": 52.40280433397068, "grad_norm": 0.35158684849739075, "learning_rate": 8.689091288100025e-05, "loss": 0.0183, "step": 82220 }, { "epoch": 52.409177820267686, "grad_norm": 0.47380325198173523, "learning_rate": 8.688719235297796e-05, "loss": 0.0183, "step": 82230 }, { "epoch": 52.415551306564694, "grad_norm": 0.42066654562950134, "learning_rate": 8.688347137674146e-05, "loss": 0.0255, "step": 82240 }, { "epoch": 52.421924792861695, "grad_norm": 0.6499806642532349, "learning_rate": 8.687974995233597e-05, "loss": 0.0325, "step": 82250 }, { "epoch": 52.4282982791587, "grad_norm": 0.33543363213539124, "learning_rate": 8.68760280798067e-05, "loss": 0.0188, "step": 82260 }, { "epoch": 52.434671765455704, "grad_norm": 0.25506868958473206, "learning_rate": 8.687230575919886e-05, "loss": 0.0277, "step": 82270 }, { "epoch": 52.44104525175271, "grad_norm": 0.347199410200119, "learning_rate": 8.68685829905577e-05, "loss": 0.0201, "step": 82280 }, { "epoch": 52.44741873804971, "grad_norm": 0.5167614221572876, "learning_rate": 8.686485977392844e-05, "loss": 0.0169, "step": 82290 }, { "epoch": 52.45379222434672, "grad_norm": 0.5606223940849304, "learning_rate": 8.686113610935634e-05, "loss": 0.0391, "step": 82300 }, { "epoch": 52.46016571064372, "grad_norm": 0.38621944189071655, "learning_rate": 8.685741199688662e-05, "loss": 0.0271, "step": 82310 }, { "epoch": 52.46653919694073, "grad_norm": 0.4576461911201477, "learning_rate": 8.685368743656456e-05, "loss": 0.0151, "step": 82320 }, { "epoch": 52.47291268323773, "grad_norm": 0.493071973323822, "learning_rate": 8.684996242843539e-05, "loss": 0.0232, "step": 82330 }, { "epoch": 52.47928616953474, "grad_norm": 0.346667617559433, "learning_rate": 8.684623697254439e-05, "loss": 0.0276, "step": 82340 }, { "epoch": 52.48565965583174, "grad_norm": 0.4078003764152527, "learning_rate": 8.684251106893681e-05, "loss": 0.0229, "step": 82350 }, { "epoch": 52.49203314212875, "grad_norm": 0.6541142463684082, "learning_rate": 8.683878471765796e-05, "loss": 0.0139, "step": 82360 }, { "epoch": 52.49840662842575, "grad_norm": 0.6681216359138489, "learning_rate": 8.683505791875308e-05, "loss": 0.0222, "step": 82370 }, { "epoch": 52.504780114722756, "grad_norm": 0.3047804832458496, "learning_rate": 8.683133067226748e-05, "loss": 0.0181, "step": 82380 }, { "epoch": 52.51115360101976, "grad_norm": 0.24645927548408508, "learning_rate": 8.682760297824641e-05, "loss": 0.0192, "step": 82390 }, { "epoch": 52.517527087316765, "grad_norm": 0.4493977427482605, "learning_rate": 8.682387483673522e-05, "loss": 0.0237, "step": 82400 }, { "epoch": 52.523900573613766, "grad_norm": 0.4152500331401825, "learning_rate": 8.682014624777917e-05, "loss": 0.0236, "step": 82410 }, { "epoch": 52.530274059910774, "grad_norm": 0.4852845072746277, "learning_rate": 8.681641721142358e-05, "loss": 0.0321, "step": 82420 }, { "epoch": 52.536647546207774, "grad_norm": 0.7115806341171265, "learning_rate": 8.681268772771376e-05, "loss": 0.0306, "step": 82430 }, { "epoch": 52.54302103250478, "grad_norm": 0.4402046501636505, "learning_rate": 8.680895779669502e-05, "loss": 0.0205, "step": 82440 }, { "epoch": 52.54939451880178, "grad_norm": 0.24096879363059998, "learning_rate": 8.680522741841271e-05, "loss": 0.0212, "step": 82450 }, { "epoch": 52.55576800509879, "grad_norm": 0.3535328209400177, "learning_rate": 8.680149659291212e-05, "loss": 0.0247, "step": 82460 }, { "epoch": 52.56214149139579, "grad_norm": 0.6806064248085022, "learning_rate": 8.679776532023861e-05, "loss": 0.0214, "step": 82470 }, { "epoch": 52.5685149776928, "grad_norm": 0.27704957127571106, "learning_rate": 8.679403360043749e-05, "loss": 0.0159, "step": 82480 }, { "epoch": 52.5748884639898, "grad_norm": 0.4703468680381775, "learning_rate": 8.679030143355416e-05, "loss": 0.0215, "step": 82490 }, { "epoch": 52.58126195028681, "grad_norm": 0.3734597861766815, "learning_rate": 8.67865688196339e-05, "loss": 0.0177, "step": 82500 }, { "epoch": 52.58763543658381, "grad_norm": 0.37866517901420593, "learning_rate": 8.67828357587221e-05, "loss": 0.0195, "step": 82510 }, { "epoch": 52.59400892288082, "grad_norm": 0.18547266721725464, "learning_rate": 8.677910225086412e-05, "loss": 0.0174, "step": 82520 }, { "epoch": 52.60038240917782, "grad_norm": 0.6179351210594177, "learning_rate": 8.677536829610533e-05, "loss": 0.0231, "step": 82530 }, { "epoch": 52.606755895474826, "grad_norm": 0.28108468651771545, "learning_rate": 8.677163389449108e-05, "loss": 0.0274, "step": 82540 }, { "epoch": 52.61312938177183, "grad_norm": 0.48772165179252625, "learning_rate": 8.676789904606676e-05, "loss": 0.0262, "step": 82550 }, { "epoch": 52.619502868068835, "grad_norm": 0.271126925945282, "learning_rate": 8.676416375087776e-05, "loss": 0.0207, "step": 82560 }, { "epoch": 52.625876354365836, "grad_norm": 0.2954702079296112, "learning_rate": 8.676042800896947e-05, "loss": 0.0163, "step": 82570 }, { "epoch": 52.632249840662844, "grad_norm": 0.3048221170902252, "learning_rate": 8.675669182038725e-05, "loss": 0.0449, "step": 82580 }, { "epoch": 52.638623326959845, "grad_norm": 0.4199792742729187, "learning_rate": 8.675295518517655e-05, "loss": 0.0228, "step": 82590 }, { "epoch": 52.64499681325685, "grad_norm": 0.2763124108314514, "learning_rate": 8.674921810338272e-05, "loss": 0.015, "step": 82600 }, { "epoch": 52.651370299553854, "grad_norm": 0.6224765181541443, "learning_rate": 8.674548057505119e-05, "loss": 0.0261, "step": 82610 }, { "epoch": 52.65774378585086, "grad_norm": 0.3449157178401947, "learning_rate": 8.674174260022738e-05, "loss": 0.0249, "step": 82620 }, { "epoch": 52.66411727214786, "grad_norm": 0.2401883900165558, "learning_rate": 8.673800417895672e-05, "loss": 0.0291, "step": 82630 }, { "epoch": 52.67049075844487, "grad_norm": 0.35523858666419983, "learning_rate": 8.67342653112846e-05, "loss": 0.0272, "step": 82640 }, { "epoch": 52.67686424474187, "grad_norm": 0.5049643516540527, "learning_rate": 8.673052599725651e-05, "loss": 0.0239, "step": 82650 }, { "epoch": 52.68323773103888, "grad_norm": 0.46251335740089417, "learning_rate": 8.672678623691783e-05, "loss": 0.0299, "step": 82660 }, { "epoch": 52.68961121733588, "grad_norm": 0.5315220952033997, "learning_rate": 8.672304603031402e-05, "loss": 0.0272, "step": 82670 }, { "epoch": 52.69598470363289, "grad_norm": 0.6401378512382507, "learning_rate": 8.671930537749053e-05, "loss": 0.0187, "step": 82680 }, { "epoch": 52.70235818992989, "grad_norm": 0.6892049312591553, "learning_rate": 8.67155642784928e-05, "loss": 0.0248, "step": 82690 }, { "epoch": 52.7087316762269, "grad_norm": 0.5373736023902893, "learning_rate": 8.671182273336632e-05, "loss": 0.0204, "step": 82700 }, { "epoch": 52.7151051625239, "grad_norm": 0.45758479833602905, "learning_rate": 8.67080807421565e-05, "loss": 0.0222, "step": 82710 }, { "epoch": 52.721478648820906, "grad_norm": 0.26840701699256897, "learning_rate": 8.670433830490887e-05, "loss": 0.022, "step": 82720 }, { "epoch": 52.72785213511791, "grad_norm": 0.6195713877677917, "learning_rate": 8.670059542166887e-05, "loss": 0.0182, "step": 82730 }, { "epoch": 52.734225621414915, "grad_norm": 0.4386936128139496, "learning_rate": 8.669685209248198e-05, "loss": 0.0192, "step": 82740 }, { "epoch": 52.740599107711915, "grad_norm": 0.23305873572826385, "learning_rate": 8.669310831739369e-05, "loss": 0.0172, "step": 82750 }, { "epoch": 52.74697259400892, "grad_norm": 0.43298399448394775, "learning_rate": 8.66893640964495e-05, "loss": 0.0237, "step": 82760 }, { "epoch": 52.753346080305924, "grad_norm": 0.6129809617996216, "learning_rate": 8.668561942969488e-05, "loss": 0.0235, "step": 82770 }, { "epoch": 52.75971956660293, "grad_norm": 0.4076831042766571, "learning_rate": 8.668187431717535e-05, "loss": 0.016, "step": 82780 }, { "epoch": 52.76609305289993, "grad_norm": 0.558911919593811, "learning_rate": 8.667812875893641e-05, "loss": 0.0197, "step": 82790 }, { "epoch": 52.77246653919694, "grad_norm": 0.540917694568634, "learning_rate": 8.66743827550236e-05, "loss": 0.0239, "step": 82800 }, { "epoch": 52.77884002549394, "grad_norm": 0.2936572730541229, "learning_rate": 8.66706363054824e-05, "loss": 0.0186, "step": 82810 }, { "epoch": 52.78521351179095, "grad_norm": 0.46870535612106323, "learning_rate": 8.666688941035832e-05, "loss": 0.0274, "step": 82820 }, { "epoch": 52.79158699808795, "grad_norm": 0.3715057969093323, "learning_rate": 8.666314206969694e-05, "loss": 0.0182, "step": 82830 }, { "epoch": 52.79796048438496, "grad_norm": 0.5497382879257202, "learning_rate": 8.665939428354377e-05, "loss": 0.0181, "step": 82840 }, { "epoch": 52.80433397068197, "grad_norm": 0.19983555376529694, "learning_rate": 8.665564605194434e-05, "loss": 0.0189, "step": 82850 }, { "epoch": 52.81070745697897, "grad_norm": 0.6503185033798218, "learning_rate": 8.66518973749442e-05, "loss": 0.0302, "step": 82860 }, { "epoch": 52.817080943275975, "grad_norm": 0.4079761505126953, "learning_rate": 8.664814825258892e-05, "loss": 0.0186, "step": 82870 }, { "epoch": 52.823454429572976, "grad_norm": 0.6760707497596741, "learning_rate": 8.664439868492402e-05, "loss": 0.03, "step": 82880 }, { "epoch": 52.829827915869984, "grad_norm": 0.5822170972824097, "learning_rate": 8.664064867199508e-05, "loss": 0.03, "step": 82890 }, { "epoch": 52.836201402166985, "grad_norm": 0.21028010547161102, "learning_rate": 8.663689821384766e-05, "loss": 0.0175, "step": 82900 }, { "epoch": 52.84257488846399, "grad_norm": 1.0702921152114868, "learning_rate": 8.663314731052734e-05, "loss": 0.0365, "step": 82910 }, { "epoch": 52.848948374760994, "grad_norm": 0.6527359485626221, "learning_rate": 8.662939596207968e-05, "loss": 0.0246, "step": 82920 }, { "epoch": 52.855321861058, "grad_norm": 0.8144825100898743, "learning_rate": 8.66256441685503e-05, "loss": 0.0306, "step": 82930 }, { "epoch": 52.861695347355, "grad_norm": 0.3406814932823181, "learning_rate": 8.662189192998474e-05, "loss": 0.0278, "step": 82940 }, { "epoch": 52.86806883365201, "grad_norm": 0.3905585706233978, "learning_rate": 8.661813924642863e-05, "loss": 0.0207, "step": 82950 }, { "epoch": 52.87444231994901, "grad_norm": 0.24959905445575714, "learning_rate": 8.661438611792756e-05, "loss": 0.0174, "step": 82960 }, { "epoch": 52.88081580624602, "grad_norm": 0.9980722665786743, "learning_rate": 8.661063254452712e-05, "loss": 0.032, "step": 82970 }, { "epoch": 52.88718929254302, "grad_norm": 0.4151591360569, "learning_rate": 8.660687852627292e-05, "loss": 0.019, "step": 82980 }, { "epoch": 52.89356277884003, "grad_norm": 1.07515549659729, "learning_rate": 8.660312406321059e-05, "loss": 0.0188, "step": 82990 }, { "epoch": 52.89993626513703, "grad_norm": 0.6609443426132202, "learning_rate": 8.659936915538576e-05, "loss": 0.0414, "step": 83000 }, { "epoch": 52.90630975143404, "grad_norm": 0.4041118919849396, "learning_rate": 8.659561380284402e-05, "loss": 0.0175, "step": 83010 }, { "epoch": 52.91268323773104, "grad_norm": 0.5503021478652954, "learning_rate": 8.659185800563102e-05, "loss": 0.022, "step": 83020 }, { "epoch": 52.919056724028046, "grad_norm": 0.5025366544723511, "learning_rate": 8.658810176379241e-05, "loss": 0.027, "step": 83030 }, { "epoch": 52.92543021032505, "grad_norm": 0.4865161180496216, "learning_rate": 8.65843450773738e-05, "loss": 0.0164, "step": 83040 }, { "epoch": 52.931803696622055, "grad_norm": 0.30988237261772156, "learning_rate": 8.658058794642088e-05, "loss": 0.0262, "step": 83050 }, { "epoch": 52.938177182919055, "grad_norm": 0.44865894317626953, "learning_rate": 8.657683037097926e-05, "loss": 0.0185, "step": 83060 }, { "epoch": 52.94455066921606, "grad_norm": 0.5063388347625732, "learning_rate": 8.657307235109462e-05, "loss": 0.0284, "step": 83070 }, { "epoch": 52.950924155513064, "grad_norm": 0.3172318935394287, "learning_rate": 8.656931388681262e-05, "loss": 0.0172, "step": 83080 }, { "epoch": 52.95729764181007, "grad_norm": 0.5125734806060791, "learning_rate": 8.656555497817893e-05, "loss": 0.0212, "step": 83090 }, { "epoch": 52.96367112810707, "grad_norm": 0.694328784942627, "learning_rate": 8.656179562523922e-05, "loss": 0.0194, "step": 83100 }, { "epoch": 52.97004461440408, "grad_norm": 0.551508903503418, "learning_rate": 8.655803582803919e-05, "loss": 0.025, "step": 83110 }, { "epoch": 52.97641810070108, "grad_norm": 0.4099421799182892, "learning_rate": 8.65542755866245e-05, "loss": 0.026, "step": 83120 }, { "epoch": 52.98279158699809, "grad_norm": 0.22747386991977692, "learning_rate": 8.655051490104085e-05, "loss": 0.0267, "step": 83130 }, { "epoch": 52.98916507329509, "grad_norm": 0.2044578194618225, "learning_rate": 8.654675377133392e-05, "loss": 0.0217, "step": 83140 }, { "epoch": 52.9955385595921, "grad_norm": 0.6027621030807495, "learning_rate": 8.654299219754945e-05, "loss": 0.0279, "step": 83150 }, { "epoch": 53.0019120458891, "grad_norm": 0.4147590100765228, "learning_rate": 8.65392301797331e-05, "loss": 0.0244, "step": 83160 }, { "epoch": 53.00828553218611, "grad_norm": 0.2936416566371918, "learning_rate": 8.653546771793061e-05, "loss": 0.0378, "step": 83170 }, { "epoch": 53.01465901848311, "grad_norm": 0.48172932863235474, "learning_rate": 8.653170481218769e-05, "loss": 0.0227, "step": 83180 }, { "epoch": 53.021032504780116, "grad_norm": 0.19088655710220337, "learning_rate": 8.652794146255008e-05, "loss": 0.0353, "step": 83190 }, { "epoch": 53.02740599107712, "grad_norm": 0.35283172130584717, "learning_rate": 8.652417766906348e-05, "loss": 0.0234, "step": 83200 }, { "epoch": 53.033779477374125, "grad_norm": 0.353640615940094, "learning_rate": 8.652041343177364e-05, "loss": 0.0249, "step": 83210 }, { "epoch": 53.040152963671126, "grad_norm": 0.38593441247940063, "learning_rate": 8.651664875072629e-05, "loss": 0.0282, "step": 83220 }, { "epoch": 53.046526449968134, "grad_norm": 0.43414387106895447, "learning_rate": 8.651288362596718e-05, "loss": 0.0166, "step": 83230 }, { "epoch": 53.052899936265135, "grad_norm": 0.3523552119731903, "learning_rate": 8.650911805754208e-05, "loss": 0.0172, "step": 83240 }, { "epoch": 53.05927342256214, "grad_norm": 0.4244237542152405, "learning_rate": 8.65053520454967e-05, "loss": 0.0278, "step": 83250 }, { "epoch": 53.065646908859144, "grad_norm": 0.22153964638710022, "learning_rate": 8.650158558987685e-05, "loss": 0.0272, "step": 83260 }, { "epoch": 53.07202039515615, "grad_norm": 0.3531704545021057, "learning_rate": 8.649781869072826e-05, "loss": 0.0235, "step": 83270 }, { "epoch": 53.07839388145315, "grad_norm": 0.2647213339805603, "learning_rate": 8.64940513480967e-05, "loss": 0.0175, "step": 83280 }, { "epoch": 53.08476736775016, "grad_norm": 0.32380354404449463, "learning_rate": 8.649028356202798e-05, "loss": 0.0241, "step": 83290 }, { "epoch": 53.09114085404716, "grad_norm": 0.4932224452495575, "learning_rate": 8.648651533256785e-05, "loss": 0.0308, "step": 83300 }, { "epoch": 53.09751434034417, "grad_norm": 0.2854040861129761, "learning_rate": 8.648274665976211e-05, "loss": 0.0131, "step": 83310 }, { "epoch": 53.10388782664117, "grad_norm": 0.34878554940223694, "learning_rate": 8.647897754365656e-05, "loss": 0.0214, "step": 83320 }, { "epoch": 53.11026131293818, "grad_norm": 0.40313050150871277, "learning_rate": 8.647520798429698e-05, "loss": 0.0159, "step": 83330 }, { "epoch": 53.11663479923518, "grad_norm": 0.6901426911354065, "learning_rate": 8.647143798172919e-05, "loss": 0.0302, "step": 83340 }, { "epoch": 53.12300828553219, "grad_norm": 0.32066458463668823, "learning_rate": 8.6467667535999e-05, "loss": 0.0234, "step": 83350 }, { "epoch": 53.12938177182919, "grad_norm": 0.6121717095375061, "learning_rate": 8.646389664715223e-05, "loss": 0.0231, "step": 83360 }, { "epoch": 53.135755258126196, "grad_norm": 0.3708300292491913, "learning_rate": 8.646012531523467e-05, "loss": 0.028, "step": 83370 }, { "epoch": 53.1421287444232, "grad_norm": 0.46055150032043457, "learning_rate": 8.645635354029216e-05, "loss": 0.0269, "step": 83380 }, { "epoch": 53.148502230720204, "grad_norm": 0.5853471755981445, "learning_rate": 8.645258132237054e-05, "loss": 0.0262, "step": 83390 }, { "epoch": 53.154875717017205, "grad_norm": 0.6062769889831543, "learning_rate": 8.644880866151564e-05, "loss": 0.0201, "step": 83400 }, { "epoch": 53.16124920331421, "grad_norm": 0.510599672794342, "learning_rate": 8.64450355577733e-05, "loss": 0.024, "step": 83410 }, { "epoch": 53.167622689611214, "grad_norm": 0.322971373796463, "learning_rate": 8.644126201118936e-05, "loss": 0.018, "step": 83420 }, { "epoch": 53.17399617590822, "grad_norm": 0.46966373920440674, "learning_rate": 8.64374880218097e-05, "loss": 0.0226, "step": 83430 }, { "epoch": 53.18036966220522, "grad_norm": 0.6031379103660583, "learning_rate": 8.643371358968015e-05, "loss": 0.0241, "step": 83440 }, { "epoch": 53.18674314850223, "grad_norm": 0.3685668110847473, "learning_rate": 8.642993871484657e-05, "loss": 0.0146, "step": 83450 }, { "epoch": 53.19311663479923, "grad_norm": 0.5559024810791016, "learning_rate": 8.642616339735485e-05, "loss": 0.015, "step": 83460 }, { "epoch": 53.19949012109624, "grad_norm": 0.3526475727558136, "learning_rate": 8.642238763725085e-05, "loss": 0.0178, "step": 83470 }, { "epoch": 53.20586360739325, "grad_norm": 0.2626326382160187, "learning_rate": 8.641861143458045e-05, "loss": 0.0234, "step": 83480 }, { "epoch": 53.21223709369025, "grad_norm": 0.7630456686019897, "learning_rate": 8.641483478938954e-05, "loss": 0.0207, "step": 83490 }, { "epoch": 53.218610579987256, "grad_norm": 0.5011307001113892, "learning_rate": 8.6411057701724e-05, "loss": 0.017, "step": 83500 }, { "epoch": 53.22498406628426, "grad_norm": 0.3697773814201355, "learning_rate": 8.640728017162974e-05, "loss": 0.0144, "step": 83510 }, { "epoch": 53.231357552581265, "grad_norm": 0.30675068497657776, "learning_rate": 8.640350219915266e-05, "loss": 0.0203, "step": 83520 }, { "epoch": 53.237731038878266, "grad_norm": 0.5024585723876953, "learning_rate": 8.639972378433863e-05, "loss": 0.0229, "step": 83530 }, { "epoch": 53.244104525175274, "grad_norm": 0.5399240255355835, "learning_rate": 8.63959449272336e-05, "loss": 0.0234, "step": 83540 }, { "epoch": 53.250478011472275, "grad_norm": 0.5785936713218689, "learning_rate": 8.639216562788349e-05, "loss": 0.0158, "step": 83550 }, { "epoch": 53.25685149776928, "grad_norm": 0.6639803051948547, "learning_rate": 8.638838588633419e-05, "loss": 0.0133, "step": 83560 }, { "epoch": 53.263224984066284, "grad_norm": 0.6767406463623047, "learning_rate": 8.638460570263165e-05, "loss": 0.0172, "step": 83570 }, { "epoch": 53.26959847036329, "grad_norm": 0.33116331696510315, "learning_rate": 8.63808250768218e-05, "loss": 0.0256, "step": 83580 }, { "epoch": 53.27597195666029, "grad_norm": 0.24956843256950378, "learning_rate": 8.637704400895058e-05, "loss": 0.0253, "step": 83590 }, { "epoch": 53.2823454429573, "grad_norm": 0.3254590928554535, "learning_rate": 8.637326249906392e-05, "loss": 0.0218, "step": 83600 }, { "epoch": 53.2887189292543, "grad_norm": 0.30613407492637634, "learning_rate": 8.636948054720779e-05, "loss": 0.0188, "step": 83610 }, { "epoch": 53.29509241555131, "grad_norm": 0.37428730726242065, "learning_rate": 8.636569815342813e-05, "loss": 0.0213, "step": 83620 }, { "epoch": 53.30146590184831, "grad_norm": 0.24845999479293823, "learning_rate": 8.636191531777089e-05, "loss": 0.021, "step": 83630 }, { "epoch": 53.30783938814532, "grad_norm": 0.7117323279380798, "learning_rate": 8.635813204028207e-05, "loss": 0.0294, "step": 83640 }, { "epoch": 53.31421287444232, "grad_norm": 0.3941926062107086, "learning_rate": 8.63543483210076e-05, "loss": 0.0178, "step": 83650 }, { "epoch": 53.32058636073933, "grad_norm": 0.27202844619750977, "learning_rate": 8.635056415999347e-05, "loss": 0.0212, "step": 83660 }, { "epoch": 53.32695984703633, "grad_norm": 0.33432695269584656, "learning_rate": 8.634677955728567e-05, "loss": 0.0221, "step": 83670 }, { "epoch": 53.333333333333336, "grad_norm": 0.40294480323791504, "learning_rate": 8.63429945129302e-05, "loss": 0.0233, "step": 83680 }, { "epoch": 53.33970681963034, "grad_norm": 0.5967270135879517, "learning_rate": 8.6339209026973e-05, "loss": 0.0293, "step": 83690 }, { "epoch": 53.346080305927345, "grad_norm": 0.4534532129764557, "learning_rate": 8.633542309946014e-05, "loss": 0.0272, "step": 83700 }, { "epoch": 53.352453792224345, "grad_norm": 0.6571350693702698, "learning_rate": 8.633163673043756e-05, "loss": 0.0269, "step": 83710 }, { "epoch": 53.35882727852135, "grad_norm": 0.5004278421401978, "learning_rate": 8.632784991995129e-05, "loss": 0.0206, "step": 83720 }, { "epoch": 53.365200764818354, "grad_norm": 0.2790112793445587, "learning_rate": 8.632406266804733e-05, "loss": 0.0131, "step": 83730 }, { "epoch": 53.37157425111536, "grad_norm": 0.4587191343307495, "learning_rate": 8.632027497477174e-05, "loss": 0.031, "step": 83740 }, { "epoch": 53.37794773741236, "grad_norm": 0.27395695447921753, "learning_rate": 8.63164868401705e-05, "loss": 0.0195, "step": 83750 }, { "epoch": 53.38432122370937, "grad_norm": 0.2730388641357422, "learning_rate": 8.631269826428967e-05, "loss": 0.0188, "step": 83760 }, { "epoch": 53.39069471000637, "grad_norm": 0.5676767230033875, "learning_rate": 8.630890924717527e-05, "loss": 0.0278, "step": 83770 }, { "epoch": 53.39706819630338, "grad_norm": 0.4027407765388489, "learning_rate": 8.630511978887334e-05, "loss": 0.0196, "step": 83780 }, { "epoch": 53.40344168260038, "grad_norm": 0.42916440963745117, "learning_rate": 8.630132988942991e-05, "loss": 0.0216, "step": 83790 }, { "epoch": 53.40981516889739, "grad_norm": 0.45874109864234924, "learning_rate": 8.629753954889107e-05, "loss": 0.0181, "step": 83800 }, { "epoch": 53.41618865519439, "grad_norm": 0.6286768317222595, "learning_rate": 8.629374876730284e-05, "loss": 0.0235, "step": 83810 }, { "epoch": 53.4225621414914, "grad_norm": 0.35537102818489075, "learning_rate": 8.62899575447113e-05, "loss": 0.0232, "step": 83820 }, { "epoch": 53.4289356277884, "grad_norm": 0.3062276840209961, "learning_rate": 8.628616588116252e-05, "loss": 0.0226, "step": 83830 }, { "epoch": 53.435309114085406, "grad_norm": 0.5648360252380371, "learning_rate": 8.628237377670257e-05, "loss": 0.032, "step": 83840 }, { "epoch": 53.44168260038241, "grad_norm": 0.45258596539497375, "learning_rate": 8.627858123137748e-05, "loss": 0.0212, "step": 83850 }, { "epoch": 53.448056086679415, "grad_norm": 0.63763427734375, "learning_rate": 8.627478824523342e-05, "loss": 0.0227, "step": 83860 }, { "epoch": 53.454429572976416, "grad_norm": 0.5500563383102417, "learning_rate": 8.627099481831641e-05, "loss": 0.0252, "step": 83870 }, { "epoch": 53.460803059273424, "grad_norm": 0.35673612356185913, "learning_rate": 8.626720095067257e-05, "loss": 0.0328, "step": 83880 }, { "epoch": 53.467176545570425, "grad_norm": 0.5232762694358826, "learning_rate": 8.626340664234802e-05, "loss": 0.027, "step": 83890 }, { "epoch": 53.47355003186743, "grad_norm": 0.33171963691711426, "learning_rate": 8.625961189338882e-05, "loss": 0.0215, "step": 83900 }, { "epoch": 53.47992351816443, "grad_norm": 0.5208948254585266, "learning_rate": 8.62558167038411e-05, "loss": 0.0346, "step": 83910 }, { "epoch": 53.48629700446144, "grad_norm": 0.844214141368866, "learning_rate": 8.6252021073751e-05, "loss": 0.0213, "step": 83920 }, { "epoch": 53.49267049075844, "grad_norm": 0.44236111640930176, "learning_rate": 8.624822500316458e-05, "loss": 0.029, "step": 83930 }, { "epoch": 53.49904397705545, "grad_norm": 0.4460776746273041, "learning_rate": 8.624442849212802e-05, "loss": 0.0203, "step": 83940 }, { "epoch": 53.50541746335245, "grad_norm": 0.329765647649765, "learning_rate": 8.624063154068744e-05, "loss": 0.0254, "step": 83950 }, { "epoch": 53.51179094964946, "grad_norm": 0.39859193563461304, "learning_rate": 8.623683414888895e-05, "loss": 0.0312, "step": 83960 }, { "epoch": 53.51816443594646, "grad_norm": 0.47122275829315186, "learning_rate": 8.623303631677873e-05, "loss": 0.0222, "step": 83970 }, { "epoch": 53.52453792224347, "grad_norm": 0.4777257442474365, "learning_rate": 8.62292380444029e-05, "loss": 0.0205, "step": 83980 }, { "epoch": 53.53091140854047, "grad_norm": 0.6008886098861694, "learning_rate": 8.622543933180762e-05, "loss": 0.0156, "step": 83990 }, { "epoch": 53.53728489483748, "grad_norm": 1.3685667514801025, "learning_rate": 8.622164017903907e-05, "loss": 0.043, "step": 84000 }, { "epoch": 53.54365838113448, "grad_norm": 0.4811289310455322, "learning_rate": 8.621784058614337e-05, "loss": 0.0302, "step": 84010 }, { "epoch": 53.550031867431485, "grad_norm": 0.4748082160949707, "learning_rate": 8.621404055316673e-05, "loss": 0.0208, "step": 84020 }, { "epoch": 53.556405353728486, "grad_norm": 0.29795607924461365, "learning_rate": 8.621024008015528e-05, "loss": 0.0165, "step": 84030 }, { "epoch": 53.562778840025494, "grad_norm": 0.43697237968444824, "learning_rate": 8.620643916715525e-05, "loss": 0.0212, "step": 84040 }, { "epoch": 53.569152326322495, "grad_norm": 0.40295693278312683, "learning_rate": 8.620263781421278e-05, "loss": 0.0147, "step": 84050 }, { "epoch": 53.5755258126195, "grad_norm": 0.6573575139045715, "learning_rate": 8.619883602137408e-05, "loss": 0.0276, "step": 84060 }, { "epoch": 53.581899298916504, "grad_norm": 0.2965167164802551, "learning_rate": 8.619503378868536e-05, "loss": 0.018, "step": 84070 }, { "epoch": 53.58827278521351, "grad_norm": 0.3197731375694275, "learning_rate": 8.619123111619281e-05, "loss": 0.0192, "step": 84080 }, { "epoch": 53.59464627151051, "grad_norm": 0.41930368542671204, "learning_rate": 8.618742800394261e-05, "loss": 0.0186, "step": 84090 }, { "epoch": 53.60101975780752, "grad_norm": 0.338462233543396, "learning_rate": 8.618362445198101e-05, "loss": 0.0188, "step": 84100 }, { "epoch": 53.60739324410453, "grad_norm": 0.3289066553115845, "learning_rate": 8.61798204603542e-05, "loss": 0.031, "step": 84110 }, { "epoch": 53.61376673040153, "grad_norm": 0.4314407706260681, "learning_rate": 8.617601602910842e-05, "loss": 0.0214, "step": 84120 }, { "epoch": 53.62014021669854, "grad_norm": 0.3921735882759094, "learning_rate": 8.617221115828988e-05, "loss": 0.0149, "step": 84130 }, { "epoch": 53.62651370299554, "grad_norm": 0.3524188995361328, "learning_rate": 8.616840584794483e-05, "loss": 0.0119, "step": 84140 }, { "epoch": 53.632887189292546, "grad_norm": 0.8854230642318726, "learning_rate": 8.61646000981195e-05, "loss": 0.0212, "step": 84150 }, { "epoch": 53.63926067558955, "grad_norm": 0.2818962037563324, "learning_rate": 8.616079390886014e-05, "loss": 0.0161, "step": 84160 }, { "epoch": 53.645634161886555, "grad_norm": 0.34340497851371765, "learning_rate": 8.615698728021298e-05, "loss": 0.0183, "step": 84170 }, { "epoch": 53.652007648183556, "grad_norm": 0.5003302097320557, "learning_rate": 8.61531802122243e-05, "loss": 0.0199, "step": 84180 }, { "epoch": 53.658381134480564, "grad_norm": 0.5449504852294922, "learning_rate": 8.614937270494032e-05, "loss": 0.0235, "step": 84190 }, { "epoch": 53.664754620777565, "grad_norm": 0.48928606510162354, "learning_rate": 8.614556475840735e-05, "loss": 0.0232, "step": 84200 }, { "epoch": 53.67112810707457, "grad_norm": 0.38711628317832947, "learning_rate": 8.614175637267165e-05, "loss": 0.02, "step": 84210 }, { "epoch": 53.677501593371574, "grad_norm": 0.35434842109680176, "learning_rate": 8.613794754777946e-05, "loss": 0.0248, "step": 84220 }, { "epoch": 53.68387507966858, "grad_norm": 0.4495094418525696, "learning_rate": 8.613413828377711e-05, "loss": 0.026, "step": 84230 }, { "epoch": 53.69024856596558, "grad_norm": 0.3084827661514282, "learning_rate": 8.613032858071086e-05, "loss": 0.0251, "step": 84240 }, { "epoch": 53.69662205226259, "grad_norm": 0.35163170099258423, "learning_rate": 8.6126518438627e-05, "loss": 0.0317, "step": 84250 }, { "epoch": 53.70299553855959, "grad_norm": 0.41589632630348206, "learning_rate": 8.612270785757183e-05, "loss": 0.0276, "step": 84260 }, { "epoch": 53.7093690248566, "grad_norm": 0.3462205231189728, "learning_rate": 8.611889683759166e-05, "loss": 0.0184, "step": 84270 }, { "epoch": 53.7157425111536, "grad_norm": 0.34634271264076233, "learning_rate": 8.61150853787328e-05, "loss": 0.0199, "step": 84280 }, { "epoch": 53.72211599745061, "grad_norm": 0.2034803032875061, "learning_rate": 8.611127348104153e-05, "loss": 0.0162, "step": 84290 }, { "epoch": 53.72848948374761, "grad_norm": 0.40219178795814514, "learning_rate": 8.61074611445642e-05, "loss": 0.0199, "step": 84300 }, { "epoch": 53.73486297004462, "grad_norm": 0.28426632285118103, "learning_rate": 8.610364836934714e-05, "loss": 0.0283, "step": 84310 }, { "epoch": 53.74123645634162, "grad_norm": 0.3836933374404907, "learning_rate": 8.609983515543667e-05, "loss": 0.0345, "step": 84320 }, { "epoch": 53.747609942638626, "grad_norm": 0.4653191864490509, "learning_rate": 8.60960215028791e-05, "loss": 0.0331, "step": 84330 }, { "epoch": 53.753983428935626, "grad_norm": 0.4005078375339508, "learning_rate": 8.609220741172079e-05, "loss": 0.0195, "step": 84340 }, { "epoch": 53.760356915232634, "grad_norm": 0.5394477248191833, "learning_rate": 8.60883928820081e-05, "loss": 0.0255, "step": 84350 }, { "epoch": 53.766730401529635, "grad_norm": 0.29108813405036926, "learning_rate": 8.608457791378734e-05, "loss": 0.0222, "step": 84360 }, { "epoch": 53.77310388782664, "grad_norm": 0.7882189154624939, "learning_rate": 8.60807625071049e-05, "loss": 0.0187, "step": 84370 }, { "epoch": 53.779477374123644, "grad_norm": 0.6536107659339905, "learning_rate": 8.607694666200715e-05, "loss": 0.0196, "step": 84380 }, { "epoch": 53.78585086042065, "grad_norm": 0.39699968695640564, "learning_rate": 8.60731303785404e-05, "loss": 0.0323, "step": 84390 }, { "epoch": 53.79222434671765, "grad_norm": 0.376953125, "learning_rate": 8.606931365675108e-05, "loss": 0.0197, "step": 84400 }, { "epoch": 53.79859783301466, "grad_norm": 0.4711611270904541, "learning_rate": 8.606549649668553e-05, "loss": 0.0273, "step": 84410 }, { "epoch": 53.80497131931166, "grad_norm": 0.5453488230705261, "learning_rate": 8.606167889839016e-05, "loss": 0.0309, "step": 84420 }, { "epoch": 53.81134480560867, "grad_norm": 0.7160468101501465, "learning_rate": 8.605786086191132e-05, "loss": 0.0287, "step": 84430 }, { "epoch": 53.81771829190567, "grad_norm": 0.9054581522941589, "learning_rate": 8.605404238729544e-05, "loss": 0.0195, "step": 84440 }, { "epoch": 53.82409177820268, "grad_norm": 0.5454513430595398, "learning_rate": 8.605022347458891e-05, "loss": 0.0229, "step": 84450 }, { "epoch": 53.83046526449968, "grad_norm": 0.4035387337207794, "learning_rate": 8.604640412383814e-05, "loss": 0.0178, "step": 84460 }, { "epoch": 53.83683875079669, "grad_norm": 0.354356586933136, "learning_rate": 8.60425843350895e-05, "loss": 0.0209, "step": 84470 }, { "epoch": 53.84321223709369, "grad_norm": 0.616362988948822, "learning_rate": 8.603876410838944e-05, "loss": 0.0241, "step": 84480 }, { "epoch": 53.849585723390696, "grad_norm": 0.4769623875617981, "learning_rate": 8.603494344378438e-05, "loss": 0.0187, "step": 84490 }, { "epoch": 53.8559592096877, "grad_norm": 0.3293713629245758, "learning_rate": 8.603112234132073e-05, "loss": 0.0165, "step": 84500 }, { "epoch": 53.862332695984705, "grad_norm": 0.5047313570976257, "learning_rate": 8.60273008010449e-05, "loss": 0.0168, "step": 84510 }, { "epoch": 53.868706182281706, "grad_norm": 0.6762397885322571, "learning_rate": 8.60234788230034e-05, "loss": 0.0189, "step": 84520 }, { "epoch": 53.875079668578714, "grad_norm": 0.7818517088890076, "learning_rate": 8.601965640724257e-05, "loss": 0.0218, "step": 84530 }, { "epoch": 53.881453154875715, "grad_norm": 0.4257866144180298, "learning_rate": 8.601583355380894e-05, "loss": 0.0208, "step": 84540 }, { "epoch": 53.88782664117272, "grad_norm": 0.3315708339214325, "learning_rate": 8.60120102627489e-05, "loss": 0.0291, "step": 84550 }, { "epoch": 53.89420012746972, "grad_norm": 0.3500559329986572, "learning_rate": 8.600818653410895e-05, "loss": 0.0219, "step": 84560 }, { "epoch": 53.90057361376673, "grad_norm": 0.48855504393577576, "learning_rate": 8.600436236793554e-05, "loss": 0.0251, "step": 84570 }, { "epoch": 53.90694710006373, "grad_norm": 0.4539107382297516, "learning_rate": 8.600053776427513e-05, "loss": 0.0167, "step": 84580 }, { "epoch": 53.91332058636074, "grad_norm": 0.2650492787361145, "learning_rate": 8.59967127231742e-05, "loss": 0.0211, "step": 84590 }, { "epoch": 53.91969407265774, "grad_norm": 0.2661218047142029, "learning_rate": 8.599288724467921e-05, "loss": 0.0175, "step": 84600 }, { "epoch": 53.92606755895475, "grad_norm": 0.37709036469459534, "learning_rate": 8.598906132883667e-05, "loss": 0.0246, "step": 84610 }, { "epoch": 53.93244104525175, "grad_norm": 0.4451025724411011, "learning_rate": 8.598523497569305e-05, "loss": 0.0295, "step": 84620 }, { "epoch": 53.93881453154876, "grad_norm": 0.5607625842094421, "learning_rate": 8.598140818529485e-05, "loss": 0.0244, "step": 84630 }, { "epoch": 53.94518801784576, "grad_norm": 0.6989381909370422, "learning_rate": 8.597758095768858e-05, "loss": 0.0224, "step": 84640 }, { "epoch": 53.95156150414277, "grad_norm": 0.3150734603404999, "learning_rate": 8.597375329292073e-05, "loss": 0.016, "step": 84650 }, { "epoch": 53.95793499043977, "grad_norm": 0.36495548486709595, "learning_rate": 8.59699251910378e-05, "loss": 0.0155, "step": 84660 }, { "epoch": 53.964308476736775, "grad_norm": 0.5768551826477051, "learning_rate": 8.596609665208631e-05, "loss": 0.0254, "step": 84670 }, { "epoch": 53.970681963033776, "grad_norm": 0.5511453747749329, "learning_rate": 8.59622676761128e-05, "loss": 0.0207, "step": 84680 }, { "epoch": 53.977055449330784, "grad_norm": 0.4071067273616791, "learning_rate": 8.595843826316379e-05, "loss": 0.0262, "step": 84690 }, { "epoch": 53.983428935627785, "grad_norm": 0.6202547550201416, "learning_rate": 8.59546084132858e-05, "loss": 0.0216, "step": 84700 }, { "epoch": 53.98980242192479, "grad_norm": 0.32500600814819336, "learning_rate": 8.595077812652536e-05, "loss": 0.0195, "step": 84710 }, { "epoch": 53.9961759082218, "grad_norm": 0.2570093870162964, "learning_rate": 8.594694740292905e-05, "loss": 0.0178, "step": 84720 }, { "epoch": 54.0025493945188, "grad_norm": 0.5094832181930542, "learning_rate": 8.594311624254337e-05, "loss": 0.0242, "step": 84730 }, { "epoch": 54.00892288081581, "grad_norm": 0.18848976492881775, "learning_rate": 8.59392846454149e-05, "loss": 0.0191, "step": 84740 }, { "epoch": 54.01529636711281, "grad_norm": 0.3133816719055176, "learning_rate": 8.593545261159017e-05, "loss": 0.0123, "step": 84750 }, { "epoch": 54.02166985340982, "grad_norm": 0.23563435673713684, "learning_rate": 8.593162014111577e-05, "loss": 0.0147, "step": 84760 }, { "epoch": 54.02804333970682, "grad_norm": 0.43539753556251526, "learning_rate": 8.592778723403827e-05, "loss": 0.0167, "step": 84770 }, { "epoch": 54.03441682600383, "grad_norm": 0.4799574911594391, "learning_rate": 8.592395389040424e-05, "loss": 0.0241, "step": 84780 }, { "epoch": 54.04079031230083, "grad_norm": 0.2808975875377655, "learning_rate": 8.592012011026023e-05, "loss": 0.0196, "step": 84790 }, { "epoch": 54.047163798597836, "grad_norm": 0.5799266695976257, "learning_rate": 8.591628589365286e-05, "loss": 0.0274, "step": 84800 }, { "epoch": 54.05353728489484, "grad_norm": 0.39807814359664917, "learning_rate": 8.59124512406287e-05, "loss": 0.011, "step": 84810 }, { "epoch": 54.059910771191845, "grad_norm": 0.32962679862976074, "learning_rate": 8.590861615123435e-05, "loss": 0.0307, "step": 84820 }, { "epoch": 54.066284257488846, "grad_norm": 0.31680044531822205, "learning_rate": 8.590478062551643e-05, "loss": 0.0189, "step": 84830 }, { "epoch": 54.072657743785854, "grad_norm": 0.3967878818511963, "learning_rate": 8.59009446635215e-05, "loss": 0.0212, "step": 84840 }, { "epoch": 54.079031230082855, "grad_norm": 0.37752580642700195, "learning_rate": 8.589710826529622e-05, "loss": 0.0207, "step": 84850 }, { "epoch": 54.08540471637986, "grad_norm": 0.5211849808692932, "learning_rate": 8.589327143088715e-05, "loss": 0.0188, "step": 84860 }, { "epoch": 54.09177820267686, "grad_norm": 0.44455671310424805, "learning_rate": 8.588943416034097e-05, "loss": 0.017, "step": 84870 }, { "epoch": 54.09815168897387, "grad_norm": 0.26508474349975586, "learning_rate": 8.588559645370428e-05, "loss": 0.0291, "step": 84880 }, { "epoch": 54.10452517527087, "grad_norm": 0.521621584892273, "learning_rate": 8.58817583110237e-05, "loss": 0.0241, "step": 84890 }, { "epoch": 54.11089866156788, "grad_norm": 0.6650809645652771, "learning_rate": 8.587791973234588e-05, "loss": 0.0171, "step": 84900 }, { "epoch": 54.11727214786488, "grad_norm": 0.44355618953704834, "learning_rate": 8.587408071771745e-05, "loss": 0.0194, "step": 84910 }, { "epoch": 54.12364563416189, "grad_norm": 0.4106016755104065, "learning_rate": 8.58702412671851e-05, "loss": 0.0251, "step": 84920 }, { "epoch": 54.13001912045889, "grad_norm": 0.26764363050460815, "learning_rate": 8.586640138079543e-05, "loss": 0.0306, "step": 84930 }, { "epoch": 54.1363926067559, "grad_norm": 0.3683658242225647, "learning_rate": 8.586256105859512e-05, "loss": 0.0263, "step": 84940 }, { "epoch": 54.1427660930529, "grad_norm": 0.35346463322639465, "learning_rate": 8.585872030063082e-05, "loss": 0.033, "step": 84950 }, { "epoch": 54.14913957934991, "grad_norm": 0.42187610268592834, "learning_rate": 8.585487910694925e-05, "loss": 0.0198, "step": 84960 }, { "epoch": 54.15551306564691, "grad_norm": 0.4995281994342804, "learning_rate": 8.585103747759702e-05, "loss": 0.0194, "step": 84970 }, { "epoch": 54.161886551943915, "grad_norm": 0.4050264358520508, "learning_rate": 8.584719541262082e-05, "loss": 0.0201, "step": 84980 }, { "epoch": 54.168260038240916, "grad_norm": 0.44719401001930237, "learning_rate": 8.584335291206739e-05, "loss": 0.0285, "step": 84990 }, { "epoch": 54.174633524537924, "grad_norm": 0.4625746011734009, "learning_rate": 8.583950997598336e-05, "loss": 0.0127, "step": 85000 }, { "epoch": 54.181007010834925, "grad_norm": 0.4877275228500366, "learning_rate": 8.583566660441545e-05, "loss": 0.0198, "step": 85010 }, { "epoch": 54.18738049713193, "grad_norm": 0.5660766959190369, "learning_rate": 8.583182279741035e-05, "loss": 0.0222, "step": 85020 }, { "epoch": 54.193753983428934, "grad_norm": 0.25721949338912964, "learning_rate": 8.582797855501476e-05, "loss": 0.0154, "step": 85030 }, { "epoch": 54.20012746972594, "grad_norm": 0.6928658485412598, "learning_rate": 8.582413387727542e-05, "loss": 0.0262, "step": 85040 }, { "epoch": 54.20650095602294, "grad_norm": 0.4689923822879791, "learning_rate": 8.582028876423904e-05, "loss": 0.0279, "step": 85050 }, { "epoch": 54.21287444231995, "grad_norm": 0.32293230295181274, "learning_rate": 8.581644321595231e-05, "loss": 0.0186, "step": 85060 }, { "epoch": 54.21924792861695, "grad_norm": 0.45877373218536377, "learning_rate": 8.5812597232462e-05, "loss": 0.0227, "step": 85070 }, { "epoch": 54.22562141491396, "grad_norm": 0.4153297245502472, "learning_rate": 8.580875081381481e-05, "loss": 0.032, "step": 85080 }, { "epoch": 54.23199490121096, "grad_norm": 0.3995732069015503, "learning_rate": 8.580490396005746e-05, "loss": 0.0232, "step": 85090 }, { "epoch": 54.23836838750797, "grad_norm": 0.351686954498291, "learning_rate": 8.580105667123677e-05, "loss": 0.0287, "step": 85100 }, { "epoch": 54.24474187380497, "grad_norm": 0.37945473194122314, "learning_rate": 8.579720894739943e-05, "loss": 0.0171, "step": 85110 }, { "epoch": 54.25111536010198, "grad_norm": 0.4682920277118683, "learning_rate": 8.579336078859217e-05, "loss": 0.0278, "step": 85120 }, { "epoch": 54.25748884639898, "grad_norm": 0.4148190915584564, "learning_rate": 8.578951219486182e-05, "loss": 0.0238, "step": 85130 }, { "epoch": 54.263862332695986, "grad_norm": 0.31452199816703796, "learning_rate": 8.578566316625507e-05, "loss": 0.0135, "step": 85140 }, { "epoch": 54.27023581899299, "grad_norm": 0.5568601489067078, "learning_rate": 8.578181370281875e-05, "loss": 0.0225, "step": 85150 }, { "epoch": 54.276609305289995, "grad_norm": 0.5864519476890564, "learning_rate": 8.577796380459961e-05, "loss": 0.0177, "step": 85160 }, { "epoch": 54.282982791586996, "grad_norm": 0.5937750339508057, "learning_rate": 8.577411347164442e-05, "loss": 0.0225, "step": 85170 }, { "epoch": 54.289356277884, "grad_norm": 0.3964175581932068, "learning_rate": 8.577026270399997e-05, "loss": 0.0291, "step": 85180 }, { "epoch": 54.295729764181004, "grad_norm": 0.5911518931388855, "learning_rate": 8.576641150171307e-05, "loss": 0.0238, "step": 85190 }, { "epoch": 54.30210325047801, "grad_norm": 0.8719543218612671, "learning_rate": 8.57625598648305e-05, "loss": 0.0305, "step": 85200 }, { "epoch": 54.30847673677501, "grad_norm": 0.2648394703865051, "learning_rate": 8.575870779339905e-05, "loss": 0.0303, "step": 85210 }, { "epoch": 54.31485022307202, "grad_norm": 0.5186700820922852, "learning_rate": 8.575485528746556e-05, "loss": 0.0251, "step": 85220 }, { "epoch": 54.32122370936902, "grad_norm": 0.47704845666885376, "learning_rate": 8.57510023470768e-05, "loss": 0.0173, "step": 85230 }, { "epoch": 54.32759719566603, "grad_norm": 0.3839057683944702, "learning_rate": 8.574714897227963e-05, "loss": 0.0118, "step": 85240 }, { "epoch": 54.33397068196303, "grad_norm": 0.70790696144104, "learning_rate": 8.574329516312081e-05, "loss": 0.0207, "step": 85250 }, { "epoch": 54.34034416826004, "grad_norm": 0.33637383580207825, "learning_rate": 8.573944091964723e-05, "loss": 0.0225, "step": 85260 }, { "epoch": 54.34671765455704, "grad_norm": 0.404680460691452, "learning_rate": 8.573558624190569e-05, "loss": 0.0187, "step": 85270 }, { "epoch": 54.35309114085405, "grad_norm": 0.473209410905838, "learning_rate": 8.573173112994306e-05, "loss": 0.0253, "step": 85280 }, { "epoch": 54.35946462715105, "grad_norm": 0.44637513160705566, "learning_rate": 8.572787558380614e-05, "loss": 0.024, "step": 85290 }, { "epoch": 54.365838113448056, "grad_norm": 0.890125572681427, "learning_rate": 8.572401960354182e-05, "loss": 0.0283, "step": 85300 }, { "epoch": 54.37221159974506, "grad_norm": 0.4701524078845978, "learning_rate": 8.572016318919691e-05, "loss": 0.0204, "step": 85310 }, { "epoch": 54.378585086042065, "grad_norm": 0.2882644236087799, "learning_rate": 8.57163063408183e-05, "loss": 0.0162, "step": 85320 }, { "epoch": 54.384958572339066, "grad_norm": 0.38176703453063965, "learning_rate": 8.571244905845284e-05, "loss": 0.0127, "step": 85330 }, { "epoch": 54.391332058636074, "grad_norm": 0.4173128604888916, "learning_rate": 8.570859134214741e-05, "loss": 0.0225, "step": 85340 }, { "epoch": 54.397705544933075, "grad_norm": 0.7550672888755798, "learning_rate": 8.570473319194889e-05, "loss": 0.0233, "step": 85350 }, { "epoch": 54.40407903123008, "grad_norm": 0.30151137709617615, "learning_rate": 8.570087460790415e-05, "loss": 0.0189, "step": 85360 }, { "epoch": 54.41045251752709, "grad_norm": 0.77811199426651, "learning_rate": 8.569701559006008e-05, "loss": 0.0261, "step": 85370 }, { "epoch": 54.41682600382409, "grad_norm": 0.6901721954345703, "learning_rate": 8.569315613846354e-05, "loss": 0.0169, "step": 85380 }, { "epoch": 54.4231994901211, "grad_norm": 0.3597675859928131, "learning_rate": 8.568929625316146e-05, "loss": 0.0179, "step": 85390 }, { "epoch": 54.4295729764181, "grad_norm": 0.4583025872707367, "learning_rate": 8.568543593420076e-05, "loss": 0.0216, "step": 85400 }, { "epoch": 54.43594646271511, "grad_norm": 0.41436609625816345, "learning_rate": 8.568157518162829e-05, "loss": 0.024, "step": 85410 }, { "epoch": 54.44231994901211, "grad_norm": 0.2920903265476227, "learning_rate": 8.5677713995491e-05, "loss": 0.011, "step": 85420 }, { "epoch": 54.44869343530912, "grad_norm": 0.36646533012390137, "learning_rate": 8.567385237583581e-05, "loss": 0.0246, "step": 85430 }, { "epoch": 54.45506692160612, "grad_norm": 0.5572330355644226, "learning_rate": 8.566999032270964e-05, "loss": 0.0327, "step": 85440 }, { "epoch": 54.461440407903126, "grad_norm": 0.6074442267417908, "learning_rate": 8.56661278361594e-05, "loss": 0.0312, "step": 85450 }, { "epoch": 54.46781389420013, "grad_norm": 0.5684938430786133, "learning_rate": 8.566226491623203e-05, "loss": 0.0181, "step": 85460 }, { "epoch": 54.474187380497135, "grad_norm": 0.3168732523918152, "learning_rate": 8.565840156297447e-05, "loss": 0.0245, "step": 85470 }, { "epoch": 54.480560866794136, "grad_norm": 0.5349665284156799, "learning_rate": 8.565453777643366e-05, "loss": 0.0238, "step": 85480 }, { "epoch": 54.486934353091144, "grad_norm": 0.3802516758441925, "learning_rate": 8.565067355665656e-05, "loss": 0.0197, "step": 85490 }, { "epoch": 54.493307839388144, "grad_norm": 0.3255811333656311, "learning_rate": 8.564680890369011e-05, "loss": 0.025, "step": 85500 }, { "epoch": 54.49968132568515, "grad_norm": 0.5623638033866882, "learning_rate": 8.564294381758128e-05, "loss": 0.0205, "step": 85510 }, { "epoch": 54.50605481198215, "grad_norm": 0.6858962774276733, "learning_rate": 8.563907829837703e-05, "loss": 0.0241, "step": 85520 }, { "epoch": 54.51242829827916, "grad_norm": 0.46164995431900024, "learning_rate": 8.563521234612433e-05, "loss": 0.019, "step": 85530 }, { "epoch": 54.51880178457616, "grad_norm": 0.36765310168266296, "learning_rate": 8.563134596087015e-05, "loss": 0.027, "step": 85540 }, { "epoch": 54.52517527087317, "grad_norm": 0.35005107522010803, "learning_rate": 8.562747914266149e-05, "loss": 0.0253, "step": 85550 }, { "epoch": 54.53154875717017, "grad_norm": 0.7256959080696106, "learning_rate": 8.56236118915453e-05, "loss": 0.0192, "step": 85560 }, { "epoch": 54.53792224346718, "grad_norm": 0.8625956773757935, "learning_rate": 8.56197442075686e-05, "loss": 0.0157, "step": 85570 }, { "epoch": 54.54429572976418, "grad_norm": 0.33602285385131836, "learning_rate": 8.561587609077837e-05, "loss": 0.0228, "step": 85580 }, { "epoch": 54.55066921606119, "grad_norm": 0.498575896024704, "learning_rate": 8.561200754122163e-05, "loss": 0.0278, "step": 85590 }, { "epoch": 54.55704270235819, "grad_norm": 0.5766869783401489, "learning_rate": 8.560813855894539e-05, "loss": 0.0297, "step": 85600 }, { "epoch": 54.5634161886552, "grad_norm": 0.7085694670677185, "learning_rate": 8.560426914399662e-05, "loss": 0.023, "step": 85610 }, { "epoch": 54.5697896749522, "grad_norm": 0.6201806664466858, "learning_rate": 8.560039929642236e-05, "loss": 0.0206, "step": 85620 }, { "epoch": 54.576163161249205, "grad_norm": 0.42374396324157715, "learning_rate": 8.559652901626966e-05, "loss": 0.0174, "step": 85630 }, { "epoch": 54.582536647546206, "grad_norm": 0.4321466386318207, "learning_rate": 8.559265830358552e-05, "loss": 0.0288, "step": 85640 }, { "epoch": 54.588910133843214, "grad_norm": 0.5083127617835999, "learning_rate": 8.558878715841696e-05, "loss": 0.0166, "step": 85650 }, { "epoch": 54.595283620140215, "grad_norm": 0.5091572999954224, "learning_rate": 8.558491558081104e-05, "loss": 0.0259, "step": 85660 }, { "epoch": 54.60165710643722, "grad_norm": 0.5440375804901123, "learning_rate": 8.55810435708148e-05, "loss": 0.0222, "step": 85670 }, { "epoch": 54.608030592734224, "grad_norm": 0.4063795208930969, "learning_rate": 8.557717112847529e-05, "loss": 0.0306, "step": 85680 }, { "epoch": 54.61440407903123, "grad_norm": 0.4484158456325531, "learning_rate": 8.557329825383956e-05, "loss": 0.0174, "step": 85690 }, { "epoch": 54.62077756532823, "grad_norm": 0.5302240252494812, "learning_rate": 8.556942494695467e-05, "loss": 0.0249, "step": 85700 }, { "epoch": 54.62715105162524, "grad_norm": 0.3472258448600769, "learning_rate": 8.556555120786769e-05, "loss": 0.0191, "step": 85710 }, { "epoch": 54.63352453792224, "grad_norm": 0.35999906063079834, "learning_rate": 8.556167703662566e-05, "loss": 0.0309, "step": 85720 }, { "epoch": 54.63989802421925, "grad_norm": 0.7518576979637146, "learning_rate": 8.555780243327569e-05, "loss": 0.0247, "step": 85730 }, { "epoch": 54.64627151051625, "grad_norm": 0.5930815935134888, "learning_rate": 8.555392739786485e-05, "loss": 0.0272, "step": 85740 }, { "epoch": 54.65264499681326, "grad_norm": 0.3392670452594757, "learning_rate": 8.555005193044023e-05, "loss": 0.0186, "step": 85750 }, { "epoch": 54.65901848311026, "grad_norm": 0.6263693571090698, "learning_rate": 8.55461760310489e-05, "loss": 0.0272, "step": 85760 }, { "epoch": 54.66539196940727, "grad_norm": 0.43870407342910767, "learning_rate": 8.5542299699738e-05, "loss": 0.0306, "step": 85770 }, { "epoch": 54.67176545570427, "grad_norm": 0.4818042516708374, "learning_rate": 8.553842293655459e-05, "loss": 0.0227, "step": 85780 }, { "epoch": 54.678138942001276, "grad_norm": 0.28415727615356445, "learning_rate": 8.553454574154577e-05, "loss": 0.0228, "step": 85790 }, { "epoch": 54.68451242829828, "grad_norm": 1.2254512310028076, "learning_rate": 8.553066811475867e-05, "loss": 0.0328, "step": 85800 }, { "epoch": 54.690885914595285, "grad_norm": 0.5818784832954407, "learning_rate": 8.552679005624041e-05, "loss": 0.0247, "step": 85810 }, { "epoch": 54.697259400892285, "grad_norm": 0.4756927192211151, "learning_rate": 8.55229115660381e-05, "loss": 0.032, "step": 85820 }, { "epoch": 54.70363288718929, "grad_norm": 0.25622543692588806, "learning_rate": 8.55190326441989e-05, "loss": 0.0172, "step": 85830 }, { "epoch": 54.710006373486294, "grad_norm": 0.3521977663040161, "learning_rate": 8.551515329076991e-05, "loss": 0.0225, "step": 85840 }, { "epoch": 54.7163798597833, "grad_norm": 0.5041458010673523, "learning_rate": 8.551127350579827e-05, "loss": 0.0287, "step": 85850 }, { "epoch": 54.7227533460803, "grad_norm": 0.3370957374572754, "learning_rate": 8.550739328933112e-05, "loss": 0.0169, "step": 85860 }, { "epoch": 54.72912683237731, "grad_norm": 0.333385169506073, "learning_rate": 8.550351264141563e-05, "loss": 0.0251, "step": 85870 }, { "epoch": 54.73550031867431, "grad_norm": 0.3394824266433716, "learning_rate": 8.549963156209894e-05, "loss": 0.0168, "step": 85880 }, { "epoch": 54.74187380497132, "grad_norm": 2.013967752456665, "learning_rate": 8.54957500514282e-05, "loss": 0.0321, "step": 85890 }, { "epoch": 54.74824729126832, "grad_norm": 0.4192601442337036, "learning_rate": 8.549186810945059e-05, "loss": 0.0277, "step": 85900 }, { "epoch": 54.75462077756533, "grad_norm": 0.4651796817779541, "learning_rate": 8.548798573621327e-05, "loss": 0.0173, "step": 85910 }, { "epoch": 54.76099426386233, "grad_norm": 0.49102360010147095, "learning_rate": 8.548410293176341e-05, "loss": 0.021, "step": 85920 }, { "epoch": 54.76736775015934, "grad_norm": 0.7323653697967529, "learning_rate": 8.548021969614821e-05, "loss": 0.0323, "step": 85930 }, { "epoch": 54.77374123645634, "grad_norm": 0.4036193788051605, "learning_rate": 8.547633602941484e-05, "loss": 0.0208, "step": 85940 }, { "epoch": 54.780114722753346, "grad_norm": 0.35377243161201477, "learning_rate": 8.547245193161049e-05, "loss": 0.0213, "step": 85950 }, { "epoch": 54.78648820905035, "grad_norm": 0.45204293727874756, "learning_rate": 8.546856740278235e-05, "loss": 0.0193, "step": 85960 }, { "epoch": 54.792861695347355, "grad_norm": 0.5348866581916809, "learning_rate": 8.546468244297765e-05, "loss": 0.0236, "step": 85970 }, { "epoch": 54.79923518164436, "grad_norm": 0.6440789699554443, "learning_rate": 8.546079705224354e-05, "loss": 0.025, "step": 85980 }, { "epoch": 54.805608667941364, "grad_norm": 0.48172369599342346, "learning_rate": 8.54569112306273e-05, "loss": 0.0235, "step": 85990 }, { "epoch": 54.81198215423837, "grad_norm": 0.7067767381668091, "learning_rate": 8.545302497817609e-05, "loss": 0.0351, "step": 86000 }, { "epoch": 54.81835564053537, "grad_norm": 0.4462307095527649, "learning_rate": 8.544913829493718e-05, "loss": 0.0203, "step": 86010 }, { "epoch": 54.82472912683238, "grad_norm": 0.35432422161102295, "learning_rate": 8.544525118095774e-05, "loss": 0.0195, "step": 86020 }, { "epoch": 54.83110261312938, "grad_norm": 0.7455963492393494, "learning_rate": 8.544136363628503e-05, "loss": 0.0232, "step": 86030 }, { "epoch": 54.83747609942639, "grad_norm": 0.28518256545066833, "learning_rate": 8.54374756609663e-05, "loss": 0.0273, "step": 86040 }, { "epoch": 54.84384958572339, "grad_norm": 0.5440710783004761, "learning_rate": 8.54335872550488e-05, "loss": 0.0216, "step": 86050 }, { "epoch": 54.8502230720204, "grad_norm": 0.583558976650238, "learning_rate": 8.542969841857972e-05, "loss": 0.0168, "step": 86060 }, { "epoch": 54.8565965583174, "grad_norm": 0.4299478530883789, "learning_rate": 8.542580915160638e-05, "loss": 0.0275, "step": 86070 }, { "epoch": 54.86297004461441, "grad_norm": 0.26359421014785767, "learning_rate": 8.542191945417601e-05, "loss": 0.0172, "step": 86080 }, { "epoch": 54.86934353091141, "grad_norm": 0.4829738736152649, "learning_rate": 8.541802932633588e-05, "loss": 0.0281, "step": 86090 }, { "epoch": 54.875717017208416, "grad_norm": 0.4335806667804718, "learning_rate": 8.541413876813325e-05, "loss": 0.0191, "step": 86100 }, { "epoch": 54.88209050350542, "grad_norm": 0.43728089332580566, "learning_rate": 8.54102477796154e-05, "loss": 0.0186, "step": 86110 }, { "epoch": 54.888463989802425, "grad_norm": 0.4246975779533386, "learning_rate": 8.540635636082959e-05, "loss": 0.019, "step": 86120 }, { "epoch": 54.894837476099426, "grad_norm": 0.5081120729446411, "learning_rate": 8.540246451182314e-05, "loss": 0.0238, "step": 86130 }, { "epoch": 54.90121096239643, "grad_norm": 0.6616365313529968, "learning_rate": 8.539857223264331e-05, "loss": 0.0173, "step": 86140 }, { "epoch": 54.907584448693434, "grad_norm": 0.31998535990715027, "learning_rate": 8.53946795233374e-05, "loss": 0.0205, "step": 86150 }, { "epoch": 54.91395793499044, "grad_norm": 0.48805829882621765, "learning_rate": 8.539078638395273e-05, "loss": 0.0291, "step": 86160 }, { "epoch": 54.92033142128744, "grad_norm": 0.5109148621559143, "learning_rate": 8.538689281453657e-05, "loss": 0.0163, "step": 86170 }, { "epoch": 54.92670490758445, "grad_norm": 0.3015347421169281, "learning_rate": 8.538299881513628e-05, "loss": 0.0188, "step": 86180 }, { "epoch": 54.93307839388145, "grad_norm": 0.5622946619987488, "learning_rate": 8.537910438579913e-05, "loss": 0.0181, "step": 86190 }, { "epoch": 54.93945188017846, "grad_norm": 0.3560231029987335, "learning_rate": 8.537520952657246e-05, "loss": 0.0139, "step": 86200 }, { "epoch": 54.94582536647546, "grad_norm": 0.39006224274635315, "learning_rate": 8.537131423750359e-05, "loss": 0.0159, "step": 86210 }, { "epoch": 54.95219885277247, "grad_norm": 0.6075239181518555, "learning_rate": 8.536741851863988e-05, "loss": 0.0219, "step": 86220 }, { "epoch": 54.95857233906947, "grad_norm": 0.2855777144432068, "learning_rate": 8.536352237002861e-05, "loss": 0.0185, "step": 86230 }, { "epoch": 54.96494582536648, "grad_norm": 0.5669342875480652, "learning_rate": 8.535962579171717e-05, "loss": 0.0176, "step": 86240 }, { "epoch": 54.97131931166348, "grad_norm": 0.45056989789009094, "learning_rate": 8.535572878375288e-05, "loss": 0.0221, "step": 86250 }, { "epoch": 54.977692797960486, "grad_norm": 0.446866899728775, "learning_rate": 8.535183134618311e-05, "loss": 0.0178, "step": 86260 }, { "epoch": 54.98406628425749, "grad_norm": 0.44455230236053467, "learning_rate": 8.534793347905522e-05, "loss": 0.0212, "step": 86270 }, { "epoch": 54.990439770554495, "grad_norm": 0.45189395546913147, "learning_rate": 8.534403518241657e-05, "loss": 0.0265, "step": 86280 }, { "epoch": 54.996813256851496, "grad_norm": 0.40467384457588196, "learning_rate": 8.534013645631452e-05, "loss": 0.0168, "step": 86290 }, { "epoch": 55.003186743148504, "grad_norm": 0.5539917945861816, "learning_rate": 8.533623730079643e-05, "loss": 0.0155, "step": 86300 }, { "epoch": 55.009560229445505, "grad_norm": 0.5893488526344299, "learning_rate": 8.533233771590971e-05, "loss": 0.0254, "step": 86310 }, { "epoch": 55.01593371574251, "grad_norm": 0.5297513008117676, "learning_rate": 8.532843770170171e-05, "loss": 0.0159, "step": 86320 }, { "epoch": 55.022307202039514, "grad_norm": 0.3981828987598419, "learning_rate": 8.532453725821984e-05, "loss": 0.0155, "step": 86330 }, { "epoch": 55.02868068833652, "grad_norm": 0.4786800742149353, "learning_rate": 8.532063638551151e-05, "loss": 0.0221, "step": 86340 }, { "epoch": 55.03505417463352, "grad_norm": 0.7285570502281189, "learning_rate": 8.531673508362408e-05, "loss": 0.0244, "step": 86350 }, { "epoch": 55.04142766093053, "grad_norm": 0.402475506067276, "learning_rate": 8.531283335260498e-05, "loss": 0.0359, "step": 86360 }, { "epoch": 55.04780114722753, "grad_norm": 0.4064553678035736, "learning_rate": 8.530893119250161e-05, "loss": 0.0245, "step": 86370 }, { "epoch": 55.05417463352454, "grad_norm": 0.6936973929405212, "learning_rate": 8.53050286033614e-05, "loss": 0.0236, "step": 86380 }, { "epoch": 55.06054811982154, "grad_norm": 0.4192628562450409, "learning_rate": 8.530112558523175e-05, "loss": 0.0198, "step": 86390 }, { "epoch": 55.06692160611855, "grad_norm": 0.7908431887626648, "learning_rate": 8.529722213816009e-05, "loss": 0.0224, "step": 86400 }, { "epoch": 55.07329509241555, "grad_norm": 0.6014593839645386, "learning_rate": 8.529331826219387e-05, "loss": 0.0163, "step": 86410 }, { "epoch": 55.07966857871256, "grad_norm": 0.6179099678993225, "learning_rate": 8.52894139573805e-05, "loss": 0.0201, "step": 86420 }, { "epoch": 55.08604206500956, "grad_norm": 0.7082214951515198, "learning_rate": 8.528550922376744e-05, "loss": 0.0367, "step": 86430 }, { "epoch": 55.092415551306566, "grad_norm": 0.5604935884475708, "learning_rate": 8.528160406140211e-05, "loss": 0.0196, "step": 86440 }, { "epoch": 55.09878903760357, "grad_norm": 0.4169777035713196, "learning_rate": 8.5277698470332e-05, "loss": 0.0238, "step": 86450 }, { "epoch": 55.105162523900574, "grad_norm": 0.6768692135810852, "learning_rate": 8.527379245060453e-05, "loss": 0.0348, "step": 86460 }, { "epoch": 55.111536010197575, "grad_norm": 0.5148084759712219, "learning_rate": 8.52698860022672e-05, "loss": 0.0353, "step": 86470 }, { "epoch": 55.11790949649458, "grad_norm": 0.5138073563575745, "learning_rate": 8.526597912536743e-05, "loss": 0.0201, "step": 86480 }, { "epoch": 55.124282982791584, "grad_norm": 0.4618106782436371, "learning_rate": 8.526207181995272e-05, "loss": 0.0147, "step": 86490 }, { "epoch": 55.13065646908859, "grad_norm": 0.3576715588569641, "learning_rate": 8.525816408607055e-05, "loss": 0.0304, "step": 86500 }, { "epoch": 55.13702995538559, "grad_norm": 0.5648069977760315, "learning_rate": 8.52542559237684e-05, "loss": 0.0266, "step": 86510 }, { "epoch": 55.1434034416826, "grad_norm": 0.4206564724445343, "learning_rate": 8.525034733309376e-05, "loss": 0.029, "step": 86520 }, { "epoch": 55.1497769279796, "grad_norm": 0.45659393072128296, "learning_rate": 8.524643831409412e-05, "loss": 0.0161, "step": 86530 }, { "epoch": 55.15615041427661, "grad_norm": 0.39672553539276123, "learning_rate": 8.524252886681697e-05, "loss": 0.0254, "step": 86540 }, { "epoch": 55.16252390057361, "grad_norm": 0.23461787402629852, "learning_rate": 8.52386189913098e-05, "loss": 0.0237, "step": 86550 }, { "epoch": 55.16889738687062, "grad_norm": 0.3028956651687622, "learning_rate": 8.523470868762016e-05, "loss": 0.0228, "step": 86560 }, { "epoch": 55.17527087316762, "grad_norm": 0.6393052935600281, "learning_rate": 8.523079795579554e-05, "loss": 0.0199, "step": 86570 }, { "epoch": 55.18164435946463, "grad_norm": 0.5493573546409607, "learning_rate": 8.522688679588346e-05, "loss": 0.0257, "step": 86580 }, { "epoch": 55.18801784576163, "grad_norm": 0.46611249446868896, "learning_rate": 8.522297520793144e-05, "loss": 0.0147, "step": 86590 }, { "epoch": 55.194391332058636, "grad_norm": 0.4382973611354828, "learning_rate": 8.521906319198703e-05, "loss": 0.0183, "step": 86600 }, { "epoch": 55.20076481835564, "grad_norm": 0.500516951084137, "learning_rate": 8.521515074809772e-05, "loss": 0.0196, "step": 86610 }, { "epoch": 55.207138304652645, "grad_norm": 0.6276620030403137, "learning_rate": 8.521123787631112e-05, "loss": 0.0241, "step": 86620 }, { "epoch": 55.21351179094965, "grad_norm": 0.3739742338657379, "learning_rate": 8.520732457667471e-05, "loss": 0.0204, "step": 86630 }, { "epoch": 55.219885277246654, "grad_norm": 0.6306214928627014, "learning_rate": 8.520341084923607e-05, "loss": 0.0195, "step": 86640 }, { "epoch": 55.22625876354366, "grad_norm": 0.43090885877609253, "learning_rate": 8.519949669404274e-05, "loss": 0.0237, "step": 86650 }, { "epoch": 55.23263224984066, "grad_norm": 0.36384761333465576, "learning_rate": 8.519558211114232e-05, "loss": 0.0208, "step": 86660 }, { "epoch": 55.23900573613767, "grad_norm": 0.34764453768730164, "learning_rate": 8.519166710058231e-05, "loss": 0.013, "step": 86670 }, { "epoch": 55.24537922243467, "grad_norm": 0.4838601052761078, "learning_rate": 8.518775166241034e-05, "loss": 0.0216, "step": 86680 }, { "epoch": 55.25175270873168, "grad_norm": 0.4001941680908203, "learning_rate": 8.518383579667396e-05, "loss": 0.0243, "step": 86690 }, { "epoch": 55.25812619502868, "grad_norm": 0.7490586638450623, "learning_rate": 8.517991950342077e-05, "loss": 0.0282, "step": 86700 }, { "epoch": 55.26449968132569, "grad_norm": 0.5147513747215271, "learning_rate": 8.517600278269833e-05, "loss": 0.0249, "step": 86710 }, { "epoch": 55.27087316762269, "grad_norm": 0.2288786917924881, "learning_rate": 8.517208563455423e-05, "loss": 0.0234, "step": 86720 }, { "epoch": 55.2772466539197, "grad_norm": 0.18899662792682648, "learning_rate": 8.516816805903611e-05, "loss": 0.0198, "step": 86730 }, { "epoch": 55.2836201402167, "grad_norm": 0.3168834447860718, "learning_rate": 8.516425005619153e-05, "loss": 0.0169, "step": 86740 }, { "epoch": 55.289993626513706, "grad_norm": 0.5765627026557922, "learning_rate": 8.516033162606811e-05, "loss": 0.0322, "step": 86750 }, { "epoch": 55.29636711281071, "grad_norm": 0.6220970153808594, "learning_rate": 8.515641276871346e-05, "loss": 0.032, "step": 86760 }, { "epoch": 55.302740599107715, "grad_norm": 0.4723188877105713, "learning_rate": 8.515249348417522e-05, "loss": 0.0218, "step": 86770 }, { "epoch": 55.309114085404715, "grad_norm": 0.3541519045829773, "learning_rate": 8.514857377250098e-05, "loss": 0.015, "step": 86780 }, { "epoch": 55.31548757170172, "grad_norm": 0.4373086988925934, "learning_rate": 8.514465363373839e-05, "loss": 0.023, "step": 86790 }, { "epoch": 55.321861057998724, "grad_norm": 0.4673744738101959, "learning_rate": 8.514073306793506e-05, "loss": 0.0311, "step": 86800 }, { "epoch": 55.32823454429573, "grad_norm": 0.5443738698959351, "learning_rate": 8.513681207513867e-05, "loss": 0.0174, "step": 86810 }, { "epoch": 55.33460803059273, "grad_norm": 0.48015719652175903, "learning_rate": 8.513289065539683e-05, "loss": 0.0302, "step": 86820 }, { "epoch": 55.34098151688974, "grad_norm": 0.2767954170703888, "learning_rate": 8.512896880875718e-05, "loss": 0.0198, "step": 86830 }, { "epoch": 55.34735500318674, "grad_norm": 0.45080387592315674, "learning_rate": 8.51250465352674e-05, "loss": 0.0233, "step": 86840 }, { "epoch": 55.35372848948375, "grad_norm": 0.35147109627723694, "learning_rate": 8.512112383497515e-05, "loss": 0.0152, "step": 86850 }, { "epoch": 55.36010197578075, "grad_norm": 0.4502158463001251, "learning_rate": 8.511720070792808e-05, "loss": 0.0214, "step": 86860 }, { "epoch": 55.36647546207776, "grad_norm": 0.5294784307479858, "learning_rate": 8.511327715417388e-05, "loss": 0.0294, "step": 86870 }, { "epoch": 55.37284894837476, "grad_norm": 0.33427414298057556, "learning_rate": 8.510935317376019e-05, "loss": 0.0267, "step": 86880 }, { "epoch": 55.37922243467177, "grad_norm": 0.21311397850513458, "learning_rate": 8.51054287667347e-05, "loss": 0.0132, "step": 86890 }, { "epoch": 55.38559592096877, "grad_norm": 1.059682846069336, "learning_rate": 8.510150393314513e-05, "loss": 0.0212, "step": 86900 }, { "epoch": 55.391969407265776, "grad_norm": 0.2483416497707367, "learning_rate": 8.509757867303914e-05, "loss": 0.0187, "step": 86910 }, { "epoch": 55.39834289356278, "grad_norm": 0.474150151014328, "learning_rate": 8.509365298646444e-05, "loss": 0.0265, "step": 86920 }, { "epoch": 55.404716379859785, "grad_norm": 0.24896462261676788, "learning_rate": 8.50897268734687e-05, "loss": 0.015, "step": 86930 }, { "epoch": 55.411089866156786, "grad_norm": 0.508091390132904, "learning_rate": 8.508580033409967e-05, "loss": 0.019, "step": 86940 }, { "epoch": 55.417463352453794, "grad_norm": 0.2760290503501892, "learning_rate": 8.508187336840502e-05, "loss": 0.0295, "step": 86950 }, { "epoch": 55.423836838750795, "grad_norm": 0.26464876532554626, "learning_rate": 8.50779459764325e-05, "loss": 0.0167, "step": 86960 }, { "epoch": 55.4302103250478, "grad_norm": 0.4598115384578705, "learning_rate": 8.50740181582298e-05, "loss": 0.019, "step": 86970 }, { "epoch": 55.4365838113448, "grad_norm": 0.28399914503097534, "learning_rate": 8.507008991384469e-05, "loss": 0.0184, "step": 86980 }, { "epoch": 55.44295729764181, "grad_norm": 0.5353851914405823, "learning_rate": 8.506616124332485e-05, "loss": 0.0184, "step": 86990 }, { "epoch": 55.44933078393881, "grad_norm": 0.2473720759153366, "learning_rate": 8.506223214671807e-05, "loss": 0.0285, "step": 87000 }, { "epoch": 55.45570427023582, "grad_norm": 0.2210370898246765, "learning_rate": 8.505830262407204e-05, "loss": 0.0312, "step": 87010 }, { "epoch": 55.46207775653282, "grad_norm": 1.7098239660263062, "learning_rate": 8.505437267543454e-05, "loss": 0.0242, "step": 87020 }, { "epoch": 55.46845124282983, "grad_norm": 0.3326781094074249, "learning_rate": 8.505044230085333e-05, "loss": 0.0321, "step": 87030 }, { "epoch": 55.47482472912683, "grad_norm": 0.5609923005104065, "learning_rate": 8.504651150037614e-05, "loss": 0.0225, "step": 87040 }, { "epoch": 55.48119821542384, "grad_norm": 0.5638129711151123, "learning_rate": 8.504258027405074e-05, "loss": 0.021, "step": 87050 }, { "epoch": 55.48757170172084, "grad_norm": 0.507244884967804, "learning_rate": 8.503864862192491e-05, "loss": 0.0215, "step": 87060 }, { "epoch": 55.49394518801785, "grad_norm": 0.37580406665802, "learning_rate": 8.503471654404643e-05, "loss": 0.0204, "step": 87070 }, { "epoch": 55.50031867431485, "grad_norm": 0.39622536301612854, "learning_rate": 8.503078404046305e-05, "loss": 0.0231, "step": 87080 }, { "epoch": 55.506692160611856, "grad_norm": 0.2698856592178345, "learning_rate": 8.502685111122258e-05, "loss": 0.0192, "step": 87090 }, { "epoch": 55.513065646908856, "grad_norm": 0.43262979388237, "learning_rate": 8.502291775637278e-05, "loss": 0.0126, "step": 87100 }, { "epoch": 55.519439133205864, "grad_norm": 0.2894722819328308, "learning_rate": 8.501898397596148e-05, "loss": 0.0161, "step": 87110 }, { "epoch": 55.525812619502865, "grad_norm": 0.6817557215690613, "learning_rate": 8.501504977003645e-05, "loss": 0.0217, "step": 87120 }, { "epoch": 55.53218610579987, "grad_norm": 0.421288400888443, "learning_rate": 8.501111513864551e-05, "loss": 0.0192, "step": 87130 }, { "epoch": 55.538559592096874, "grad_norm": 0.23434068262577057, "learning_rate": 8.500718008183648e-05, "loss": 0.0251, "step": 87140 }, { "epoch": 55.54493307839388, "grad_norm": 0.49615976214408875, "learning_rate": 8.500324459965714e-05, "loss": 0.0333, "step": 87150 }, { "epoch": 55.55130656469088, "grad_norm": 0.3419979214668274, "learning_rate": 8.499930869215534e-05, "loss": 0.0165, "step": 87160 }, { "epoch": 55.55768005098789, "grad_norm": 0.5330997109413147, "learning_rate": 8.499537235937888e-05, "loss": 0.022, "step": 87170 }, { "epoch": 55.56405353728489, "grad_norm": 0.6238991618156433, "learning_rate": 8.499143560137561e-05, "loss": 0.022, "step": 87180 }, { "epoch": 55.5704270235819, "grad_norm": 0.3130239248275757, "learning_rate": 8.498749841819337e-05, "loss": 0.0173, "step": 87190 }, { "epoch": 55.5768005098789, "grad_norm": 0.5819651484489441, "learning_rate": 8.498356080987999e-05, "loss": 0.0167, "step": 87200 }, { "epoch": 55.58317399617591, "grad_norm": 0.4238913655281067, "learning_rate": 8.49796227764833e-05, "loss": 0.0215, "step": 87210 }, { "epoch": 55.58954748247291, "grad_norm": 0.46811988949775696, "learning_rate": 8.497568431805119e-05, "loss": 0.0206, "step": 87220 }, { "epoch": 55.59592096876992, "grad_norm": 0.5295667052268982, "learning_rate": 8.497174543463147e-05, "loss": 0.014, "step": 87230 }, { "epoch": 55.602294455066925, "grad_norm": 0.22549764811992645, "learning_rate": 8.496780612627205e-05, "loss": 0.0268, "step": 87240 }, { "epoch": 55.608667941363926, "grad_norm": 0.4042595624923706, "learning_rate": 8.496386639302076e-05, "loss": 0.019, "step": 87250 }, { "epoch": 55.615041427660934, "grad_norm": 0.6040352582931519, "learning_rate": 8.495992623492547e-05, "loss": 0.0444, "step": 87260 }, { "epoch": 55.621414913957935, "grad_norm": 0.7099024057388306, "learning_rate": 8.495598565203407e-05, "loss": 0.0234, "step": 87270 }, { "epoch": 55.62778840025494, "grad_norm": 0.35963934659957886, "learning_rate": 8.495204464439446e-05, "loss": 0.0186, "step": 87280 }, { "epoch": 55.634161886551944, "grad_norm": 0.3433648943901062, "learning_rate": 8.494810321205449e-05, "loss": 0.0161, "step": 87290 }, { "epoch": 55.64053537284895, "grad_norm": 0.34662947058677673, "learning_rate": 8.494416135506208e-05, "loss": 0.0255, "step": 87300 }, { "epoch": 55.64690885914595, "grad_norm": 0.6556997299194336, "learning_rate": 8.49402190734651e-05, "loss": 0.0265, "step": 87310 }, { "epoch": 55.65328234544296, "grad_norm": 0.34683218598365784, "learning_rate": 8.493627636731147e-05, "loss": 0.0237, "step": 87320 }, { "epoch": 55.65965583173996, "grad_norm": 0.38685324788093567, "learning_rate": 8.493233323664911e-05, "loss": 0.0257, "step": 87330 }, { "epoch": 55.66602931803697, "grad_norm": 0.3549465835094452, "learning_rate": 8.492838968152592e-05, "loss": 0.028, "step": 87340 }, { "epoch": 55.67240280433397, "grad_norm": 0.29948002099990845, "learning_rate": 8.492444570198982e-05, "loss": 0.0159, "step": 87350 }, { "epoch": 55.67877629063098, "grad_norm": 0.40896064043045044, "learning_rate": 8.492050129808871e-05, "loss": 0.0179, "step": 87360 }, { "epoch": 55.68514977692798, "grad_norm": 0.3966725766658783, "learning_rate": 8.491655646987056e-05, "loss": 0.0142, "step": 87370 }, { "epoch": 55.69152326322499, "grad_norm": 0.34421515464782715, "learning_rate": 8.491261121738328e-05, "loss": 0.0215, "step": 87380 }, { "epoch": 55.69789674952199, "grad_norm": 0.5204769372940063, "learning_rate": 8.49086655406748e-05, "loss": 0.0268, "step": 87390 }, { "epoch": 55.704270235818996, "grad_norm": 0.2608213722705841, "learning_rate": 8.490471943979308e-05, "loss": 0.0224, "step": 87400 }, { "epoch": 55.710643722116, "grad_norm": 0.3870161175727844, "learning_rate": 8.490077291478606e-05, "loss": 0.0213, "step": 87410 }, { "epoch": 55.717017208413004, "grad_norm": 0.488418847322464, "learning_rate": 8.489682596570171e-05, "loss": 0.0324, "step": 87420 }, { "epoch": 55.723390694710005, "grad_norm": 0.5423234105110168, "learning_rate": 8.489287859258796e-05, "loss": 0.0172, "step": 87430 }, { "epoch": 55.72976418100701, "grad_norm": 0.4633110463619232, "learning_rate": 8.488893079549281e-05, "loss": 0.0175, "step": 87440 }, { "epoch": 55.736137667304014, "grad_norm": 0.2761323153972626, "learning_rate": 8.48849825744642e-05, "loss": 0.024, "step": 87450 }, { "epoch": 55.74251115360102, "grad_norm": 0.5709750056266785, "learning_rate": 8.488103392955012e-05, "loss": 0.0244, "step": 87460 }, { "epoch": 55.74888463989802, "grad_norm": 0.4144880473613739, "learning_rate": 8.487708486079853e-05, "loss": 0.0198, "step": 87470 }, { "epoch": 55.75525812619503, "grad_norm": 0.41861483454704285, "learning_rate": 8.487313536825744e-05, "loss": 0.0152, "step": 87480 }, { "epoch": 55.76163161249203, "grad_norm": 0.45658978819847107, "learning_rate": 8.486918545197483e-05, "loss": 0.0134, "step": 87490 }, { "epoch": 55.76800509878904, "grad_norm": 0.42662158608436584, "learning_rate": 8.486523511199869e-05, "loss": 0.0223, "step": 87500 }, { "epoch": 55.77437858508604, "grad_norm": 0.38616234064102173, "learning_rate": 8.486128434837704e-05, "loss": 0.0224, "step": 87510 }, { "epoch": 55.78075207138305, "grad_norm": 0.46418219804763794, "learning_rate": 8.485733316115785e-05, "loss": 0.0177, "step": 87520 }, { "epoch": 55.78712555768005, "grad_norm": 0.4100997745990753, "learning_rate": 8.485338155038915e-05, "loss": 0.021, "step": 87530 }, { "epoch": 55.79349904397706, "grad_norm": 0.32803383469581604, "learning_rate": 8.484942951611896e-05, "loss": 0.0197, "step": 87540 }, { "epoch": 55.79987253027406, "grad_norm": 0.40545889735221863, "learning_rate": 8.484547705839532e-05, "loss": 0.0173, "step": 87550 }, { "epoch": 55.806246016571066, "grad_norm": 0.3022302985191345, "learning_rate": 8.48415241772662e-05, "loss": 0.0235, "step": 87560 }, { "epoch": 55.81261950286807, "grad_norm": 0.5484441518783569, "learning_rate": 8.483757087277969e-05, "loss": 0.0161, "step": 87570 }, { "epoch": 55.818992989165075, "grad_norm": 0.31769126653671265, "learning_rate": 8.483361714498379e-05, "loss": 0.0205, "step": 87580 }, { "epoch": 55.825366475462076, "grad_norm": 0.6617003679275513, "learning_rate": 8.482966299392656e-05, "loss": 0.0243, "step": 87590 }, { "epoch": 55.831739961759084, "grad_norm": 0.6437682509422302, "learning_rate": 8.482570841965604e-05, "loss": 0.0159, "step": 87600 }, { "epoch": 55.838113448056085, "grad_norm": 0.5488786101341248, "learning_rate": 8.482175342222029e-05, "loss": 0.0319, "step": 87610 }, { "epoch": 55.84448693435309, "grad_norm": 0.40729546546936035, "learning_rate": 8.481779800166735e-05, "loss": 0.0215, "step": 87620 }, { "epoch": 55.85086042065009, "grad_norm": 0.38643932342529297, "learning_rate": 8.481384215804529e-05, "loss": 0.0219, "step": 87630 }, { "epoch": 55.8572339069471, "grad_norm": 0.5844265818595886, "learning_rate": 8.480988589140217e-05, "loss": 0.0357, "step": 87640 }, { "epoch": 55.8636073932441, "grad_norm": 0.3872585594654083, "learning_rate": 8.480592920178608e-05, "loss": 0.0176, "step": 87650 }, { "epoch": 55.86998087954111, "grad_norm": 0.2649463415145874, "learning_rate": 8.48019720892451e-05, "loss": 0.0247, "step": 87660 }, { "epoch": 55.87635436583811, "grad_norm": 0.47644591331481934, "learning_rate": 8.479801455382727e-05, "loss": 0.02, "step": 87670 }, { "epoch": 55.88272785213512, "grad_norm": 0.32054460048675537, "learning_rate": 8.479405659558072e-05, "loss": 0.01, "step": 87680 }, { "epoch": 55.88910133843212, "grad_norm": 0.2663976550102234, "learning_rate": 8.479009821455355e-05, "loss": 0.0193, "step": 87690 }, { "epoch": 55.89547482472913, "grad_norm": 0.30819013714790344, "learning_rate": 8.478613941079383e-05, "loss": 0.0139, "step": 87700 }, { "epoch": 55.90184831102613, "grad_norm": 0.49763163924217224, "learning_rate": 8.478218018434968e-05, "loss": 0.0298, "step": 87710 }, { "epoch": 55.90822179732314, "grad_norm": 0.4986898601055145, "learning_rate": 8.477822053526917e-05, "loss": 0.0219, "step": 87720 }, { "epoch": 55.91459528362014, "grad_norm": 0.4842838943004608, "learning_rate": 8.477426046360046e-05, "loss": 0.026, "step": 87730 }, { "epoch": 55.920968769917145, "grad_norm": 0.719170093536377, "learning_rate": 8.477029996939167e-05, "loss": 0.0214, "step": 87740 }, { "epoch": 55.927342256214146, "grad_norm": 0.7870925664901733, "learning_rate": 8.476633905269089e-05, "loss": 0.0293, "step": 87750 }, { "epoch": 55.933715742511154, "grad_norm": 0.4626970887184143, "learning_rate": 8.476237771354627e-05, "loss": 0.0204, "step": 87760 }, { "epoch": 55.940089228808155, "grad_norm": 0.617261528968811, "learning_rate": 8.475841595200593e-05, "loss": 0.0253, "step": 87770 }, { "epoch": 55.94646271510516, "grad_norm": 0.6017148494720459, "learning_rate": 8.475445376811803e-05, "loss": 0.0206, "step": 87780 }, { "epoch": 55.952836201402164, "grad_norm": 0.4775473177433014, "learning_rate": 8.475049116193071e-05, "loss": 0.0189, "step": 87790 }, { "epoch": 55.95920968769917, "grad_norm": 0.23234955966472626, "learning_rate": 8.47465281334921e-05, "loss": 0.0269, "step": 87800 }, { "epoch": 55.96558317399617, "grad_norm": 0.5189412236213684, "learning_rate": 8.474256468285037e-05, "loss": 0.0266, "step": 87810 }, { "epoch": 55.97195666029318, "grad_norm": 0.44525763392448425, "learning_rate": 8.473860081005369e-05, "loss": 0.019, "step": 87820 }, { "epoch": 55.97833014659018, "grad_norm": 0.30188095569610596, "learning_rate": 8.473463651515019e-05, "loss": 0.0186, "step": 87830 }, { "epoch": 55.98470363288719, "grad_norm": 0.8558187484741211, "learning_rate": 8.473067179818808e-05, "loss": 0.0313, "step": 87840 }, { "epoch": 55.99107711918419, "grad_norm": 0.2960856258869171, "learning_rate": 8.47267066592155e-05, "loss": 0.0161, "step": 87850 }, { "epoch": 55.9974506054812, "grad_norm": 0.7070679068565369, "learning_rate": 8.472274109828065e-05, "loss": 0.0221, "step": 87860 }, { "epoch": 56.003824091778206, "grad_norm": 0.44926151633262634, "learning_rate": 8.471877511543172e-05, "loss": 0.0148, "step": 87870 }, { "epoch": 56.01019757807521, "grad_norm": 0.42732515931129456, "learning_rate": 8.471480871071689e-05, "loss": 0.0157, "step": 87880 }, { "epoch": 56.016571064372215, "grad_norm": 0.5049242973327637, "learning_rate": 8.471084188418436e-05, "loss": 0.0217, "step": 87890 }, { "epoch": 56.022944550669216, "grad_norm": 0.4717085659503937, "learning_rate": 8.470687463588231e-05, "loss": 0.0161, "step": 87900 }, { "epoch": 56.029318036966224, "grad_norm": 0.5494597554206848, "learning_rate": 8.470290696585899e-05, "loss": 0.0271, "step": 87910 }, { "epoch": 56.035691523263225, "grad_norm": 0.41327694058418274, "learning_rate": 8.469893887416259e-05, "loss": 0.0302, "step": 87920 }, { "epoch": 56.04206500956023, "grad_norm": 0.6733253598213196, "learning_rate": 8.469497036084128e-05, "loss": 0.0463, "step": 87930 }, { "epoch": 56.04843849585723, "grad_norm": 0.6372947692871094, "learning_rate": 8.469100142594337e-05, "loss": 0.0233, "step": 87940 }, { "epoch": 56.05481198215424, "grad_norm": 0.46489372849464417, "learning_rate": 8.4687032069517e-05, "loss": 0.0177, "step": 87950 }, { "epoch": 56.06118546845124, "grad_norm": 0.5751542448997498, "learning_rate": 8.468306229161047e-05, "loss": 0.0263, "step": 87960 }, { "epoch": 56.06755895474825, "grad_norm": 0.8481424450874329, "learning_rate": 8.467909209227197e-05, "loss": 0.0188, "step": 87970 }, { "epoch": 56.07393244104525, "grad_norm": 0.3661065697669983, "learning_rate": 8.467512147154976e-05, "loss": 0.019, "step": 87980 }, { "epoch": 56.08030592734226, "grad_norm": 0.4513249099254608, "learning_rate": 8.46711504294921e-05, "loss": 0.0152, "step": 87990 }, { "epoch": 56.08667941363926, "grad_norm": 0.4545053541660309, "learning_rate": 8.46671789661472e-05, "loss": 0.0282, "step": 88000 }, { "epoch": 56.09305289993627, "grad_norm": 0.6917222142219543, "learning_rate": 8.466320708156336e-05, "loss": 0.0234, "step": 88010 }, { "epoch": 56.09942638623327, "grad_norm": 0.9094108939170837, "learning_rate": 8.465923477578883e-05, "loss": 0.0219, "step": 88020 }, { "epoch": 56.10579987253028, "grad_norm": 0.5615320205688477, "learning_rate": 8.465526204887187e-05, "loss": 0.0237, "step": 88030 }, { "epoch": 56.11217335882728, "grad_norm": 0.5545759201049805, "learning_rate": 8.465128890086075e-05, "loss": 0.0239, "step": 88040 }, { "epoch": 56.118546845124285, "grad_norm": 0.3962620198726654, "learning_rate": 8.464731533180376e-05, "loss": 0.0175, "step": 88050 }, { "epoch": 56.124920331421286, "grad_norm": 0.4922754764556885, "learning_rate": 8.464334134174919e-05, "loss": 0.0239, "step": 88060 }, { "epoch": 56.131293817718294, "grad_norm": 0.4500688910484314, "learning_rate": 8.46393669307453e-05, "loss": 0.0199, "step": 88070 }, { "epoch": 56.137667304015295, "grad_norm": 0.44794538617134094, "learning_rate": 8.46353920988404e-05, "loss": 0.0332, "step": 88080 }, { "epoch": 56.1440407903123, "grad_norm": 0.3919735848903656, "learning_rate": 8.463141684608278e-05, "loss": 0.0183, "step": 88090 }, { "epoch": 56.150414276609304, "grad_norm": 0.491375207901001, "learning_rate": 8.462744117252074e-05, "loss": 0.0307, "step": 88100 }, { "epoch": 56.15678776290631, "grad_norm": 0.535897970199585, "learning_rate": 8.46234650782026e-05, "loss": 0.0304, "step": 88110 }, { "epoch": 56.16316124920331, "grad_norm": 0.5247924327850342, "learning_rate": 8.461948856317669e-05, "loss": 0.0195, "step": 88120 }, { "epoch": 56.16953473550032, "grad_norm": 0.2518024742603302, "learning_rate": 8.461551162749128e-05, "loss": 0.0183, "step": 88130 }, { "epoch": 56.17590822179732, "grad_norm": 0.22354301810264587, "learning_rate": 8.461153427119474e-05, "loss": 0.0233, "step": 88140 }, { "epoch": 56.18228170809433, "grad_norm": 0.2721497714519501, "learning_rate": 8.460755649433538e-05, "loss": 0.02, "step": 88150 }, { "epoch": 56.18865519439133, "grad_norm": 0.4538951814174652, "learning_rate": 8.460357829696152e-05, "loss": 0.0171, "step": 88160 }, { "epoch": 56.19502868068834, "grad_norm": 0.7810394167900085, "learning_rate": 8.459959967912153e-05, "loss": 0.0252, "step": 88170 }, { "epoch": 56.20140216698534, "grad_norm": 0.38765275478363037, "learning_rate": 8.459562064086372e-05, "loss": 0.0352, "step": 88180 }, { "epoch": 56.20777565328235, "grad_norm": 0.5936357975006104, "learning_rate": 8.459164118223646e-05, "loss": 0.0247, "step": 88190 }, { "epoch": 56.21414913957935, "grad_norm": 0.5780360102653503, "learning_rate": 8.45876613032881e-05, "loss": 0.0168, "step": 88200 }, { "epoch": 56.220522625876356, "grad_norm": 0.3825396001338959, "learning_rate": 8.458368100406701e-05, "loss": 0.0166, "step": 88210 }, { "epoch": 56.22689611217336, "grad_norm": 0.36209428310394287, "learning_rate": 8.457970028462153e-05, "loss": 0.0303, "step": 88220 }, { "epoch": 56.233269598470365, "grad_norm": 0.5339956879615784, "learning_rate": 8.457571914500004e-05, "loss": 0.0257, "step": 88230 }, { "epoch": 56.239643084767366, "grad_norm": 0.5305065512657166, "learning_rate": 8.457173758525095e-05, "loss": 0.0251, "step": 88240 }, { "epoch": 56.246016571064374, "grad_norm": 0.5893843770027161, "learning_rate": 8.456775560542256e-05, "loss": 0.0205, "step": 88250 }, { "epoch": 56.252390057361374, "grad_norm": 0.25814586877822876, "learning_rate": 8.456377320556332e-05, "loss": 0.0122, "step": 88260 }, { "epoch": 56.25876354365838, "grad_norm": 0.3158847987651825, "learning_rate": 8.455979038572162e-05, "loss": 0.0139, "step": 88270 }, { "epoch": 56.26513702995538, "grad_norm": 0.4592151939868927, "learning_rate": 8.455580714594582e-05, "loss": 0.0147, "step": 88280 }, { "epoch": 56.27151051625239, "grad_norm": 0.4977256953716278, "learning_rate": 8.455182348628434e-05, "loss": 0.0301, "step": 88290 }, { "epoch": 56.27788400254939, "grad_norm": 0.402782142162323, "learning_rate": 8.454783940678559e-05, "loss": 0.0195, "step": 88300 }, { "epoch": 56.2842574888464, "grad_norm": 0.6122016310691833, "learning_rate": 8.454385490749795e-05, "loss": 0.0161, "step": 88310 }, { "epoch": 56.2906309751434, "grad_norm": 0.6580690741539001, "learning_rate": 8.453986998846988e-05, "loss": 0.0353, "step": 88320 }, { "epoch": 56.29700446144041, "grad_norm": 0.2752344608306885, "learning_rate": 8.453588464974978e-05, "loss": 0.0136, "step": 88330 }, { "epoch": 56.30337794773741, "grad_norm": 0.37171950936317444, "learning_rate": 8.453189889138607e-05, "loss": 0.0288, "step": 88340 }, { "epoch": 56.30975143403442, "grad_norm": 4.347799301147461, "learning_rate": 8.452791271342717e-05, "loss": 0.03, "step": 88350 }, { "epoch": 56.31612492033142, "grad_norm": 0.6892208456993103, "learning_rate": 8.452392611592153e-05, "loss": 0.0239, "step": 88360 }, { "epoch": 56.322498406628426, "grad_norm": 0.5005861520767212, "learning_rate": 8.451993909891762e-05, "loss": 0.0202, "step": 88370 }, { "epoch": 56.32887189292543, "grad_norm": 0.2772069275379181, "learning_rate": 8.451595166246384e-05, "loss": 0.0213, "step": 88380 }, { "epoch": 56.335245379222435, "grad_norm": 0.6833258271217346, "learning_rate": 8.451196380660865e-05, "loss": 0.0144, "step": 88390 }, { "epoch": 56.341618865519436, "grad_norm": 0.789817750453949, "learning_rate": 8.450797553140053e-05, "loss": 0.0289, "step": 88400 }, { "epoch": 56.347992351816444, "grad_norm": 0.33028924465179443, "learning_rate": 8.450398683688792e-05, "loss": 0.0336, "step": 88410 }, { "epoch": 56.354365838113445, "grad_norm": 0.6567124128341675, "learning_rate": 8.44999977231193e-05, "loss": 0.0184, "step": 88420 }, { "epoch": 56.36073932441045, "grad_norm": 0.27108368277549744, "learning_rate": 8.449600819014312e-05, "loss": 0.0227, "step": 88430 }, { "epoch": 56.367112810707454, "grad_norm": 0.44227170944213867, "learning_rate": 8.449201823800787e-05, "loss": 0.0197, "step": 88440 }, { "epoch": 56.37348629700446, "grad_norm": 0.29102951288223267, "learning_rate": 8.448802786676205e-05, "loss": 0.0215, "step": 88450 }, { "epoch": 56.37985978330146, "grad_norm": 0.45583412051200867, "learning_rate": 8.448403707645412e-05, "loss": 0.0385, "step": 88460 }, { "epoch": 56.38623326959847, "grad_norm": 0.4818269610404968, "learning_rate": 8.448004586713258e-05, "loss": 0.0218, "step": 88470 }, { "epoch": 56.39260675589547, "grad_norm": 0.38694676756858826, "learning_rate": 8.447605423884593e-05, "loss": 0.0148, "step": 88480 }, { "epoch": 56.39898024219248, "grad_norm": 0.46213018894195557, "learning_rate": 8.447206219164268e-05, "loss": 0.022, "step": 88490 }, { "epoch": 56.40535372848949, "grad_norm": 0.37967929244041443, "learning_rate": 8.446806972557131e-05, "loss": 0.0272, "step": 88500 }, { "epoch": 56.41172721478649, "grad_norm": 0.7696411609649658, "learning_rate": 8.446407684068037e-05, "loss": 0.0271, "step": 88510 }, { "epoch": 56.418100701083496, "grad_norm": 0.6426995396614075, "learning_rate": 8.446008353701833e-05, "loss": 0.0217, "step": 88520 }, { "epoch": 56.4244741873805, "grad_norm": 0.5220997333526611, "learning_rate": 8.445608981463375e-05, "loss": 0.0161, "step": 88530 }, { "epoch": 56.430847673677505, "grad_norm": 0.37450170516967773, "learning_rate": 8.445209567357515e-05, "loss": 0.0117, "step": 88540 }, { "epoch": 56.437221159974506, "grad_norm": 0.38822221755981445, "learning_rate": 8.444810111389108e-05, "loss": 0.0193, "step": 88550 }, { "epoch": 56.443594646271514, "grad_norm": 0.27645304799079895, "learning_rate": 8.444410613563004e-05, "loss": 0.0212, "step": 88560 }, { "epoch": 56.449968132568515, "grad_norm": 0.46493765711784363, "learning_rate": 8.444011073884057e-05, "loss": 0.0245, "step": 88570 }, { "epoch": 56.45634161886552, "grad_norm": 0.6677665114402771, "learning_rate": 8.443611492357126e-05, "loss": 0.0245, "step": 88580 }, { "epoch": 56.46271510516252, "grad_norm": 0.3798067569732666, "learning_rate": 8.443211868987063e-05, "loss": 0.0224, "step": 88590 }, { "epoch": 56.46908859145953, "grad_norm": 0.3099798262119293, "learning_rate": 8.442812203778726e-05, "loss": 0.0134, "step": 88600 }, { "epoch": 56.47546207775653, "grad_norm": 0.4550483524799347, "learning_rate": 8.44241249673697e-05, "loss": 0.0201, "step": 88610 }, { "epoch": 56.48183556405354, "grad_norm": 0.37614384293556213, "learning_rate": 8.44201274786665e-05, "loss": 0.0236, "step": 88620 }, { "epoch": 56.48820905035054, "grad_norm": 0.4918666183948517, "learning_rate": 8.441612957172626e-05, "loss": 0.0233, "step": 88630 }, { "epoch": 56.49458253664755, "grad_norm": 0.365041047334671, "learning_rate": 8.441213124659757e-05, "loss": 0.0217, "step": 88640 }, { "epoch": 56.50095602294455, "grad_norm": 0.31699469685554504, "learning_rate": 8.4408132503329e-05, "loss": 0.0212, "step": 88650 }, { "epoch": 56.50732950924156, "grad_norm": 0.371976375579834, "learning_rate": 8.44041333419691e-05, "loss": 0.0207, "step": 88660 }, { "epoch": 56.51370299553856, "grad_norm": 0.3743080496788025, "learning_rate": 8.440013376256653e-05, "loss": 0.0145, "step": 88670 }, { "epoch": 56.52007648183557, "grad_norm": 0.6124394536018372, "learning_rate": 8.439613376516984e-05, "loss": 0.0143, "step": 88680 }, { "epoch": 56.52644996813257, "grad_norm": 0.42020294070243835, "learning_rate": 8.439213334982765e-05, "loss": 0.0261, "step": 88690 }, { "epoch": 56.532823454429575, "grad_norm": 0.31843769550323486, "learning_rate": 8.438813251658857e-05, "loss": 0.0146, "step": 88700 }, { "epoch": 56.539196940726576, "grad_norm": 0.6491100788116455, "learning_rate": 8.438413126550119e-05, "loss": 0.0255, "step": 88710 }, { "epoch": 56.545570427023584, "grad_norm": 0.2174498736858368, "learning_rate": 8.438012959661419e-05, "loss": 0.0137, "step": 88720 }, { "epoch": 56.551943913320585, "grad_norm": 0.40119218826293945, "learning_rate": 8.437612750997614e-05, "loss": 0.0222, "step": 88730 }, { "epoch": 56.55831739961759, "grad_norm": 0.40927138924598694, "learning_rate": 8.437212500563567e-05, "loss": 0.0175, "step": 88740 }, { "epoch": 56.564690885914594, "grad_norm": 0.605706512928009, "learning_rate": 8.436812208364144e-05, "loss": 0.03, "step": 88750 }, { "epoch": 56.5710643722116, "grad_norm": 0.4160982072353363, "learning_rate": 8.436411874404209e-05, "loss": 0.0247, "step": 88760 }, { "epoch": 56.5774378585086, "grad_norm": 0.3002631962299347, "learning_rate": 8.436011498688624e-05, "loss": 0.0209, "step": 88770 }, { "epoch": 56.58381134480561, "grad_norm": 0.22086432576179504, "learning_rate": 8.435611081222255e-05, "loss": 0.0142, "step": 88780 }, { "epoch": 56.59018483110261, "grad_norm": 0.30750617384910583, "learning_rate": 8.435210622009967e-05, "loss": 0.0208, "step": 88790 }, { "epoch": 56.59655831739962, "grad_norm": 0.6891700029373169, "learning_rate": 8.434810121056627e-05, "loss": 0.0339, "step": 88800 }, { "epoch": 56.60293180369662, "grad_norm": 0.4078996479511261, "learning_rate": 8.434409578367102e-05, "loss": 0.0197, "step": 88810 }, { "epoch": 56.60930528999363, "grad_norm": 0.6693564057350159, "learning_rate": 8.434008993946256e-05, "loss": 0.0276, "step": 88820 }, { "epoch": 56.61567877629063, "grad_norm": 0.3512659966945648, "learning_rate": 8.433608367798961e-05, "loss": 0.0122, "step": 88830 }, { "epoch": 56.62205226258764, "grad_norm": 0.21837028861045837, "learning_rate": 8.433207699930081e-05, "loss": 0.0124, "step": 88840 }, { "epoch": 56.62842574888464, "grad_norm": 0.4864201545715332, "learning_rate": 8.432806990344486e-05, "loss": 0.0203, "step": 88850 }, { "epoch": 56.634799235181646, "grad_norm": 0.25383928418159485, "learning_rate": 8.432406239047046e-05, "loss": 0.02, "step": 88860 }, { "epoch": 56.64117272147865, "grad_norm": 0.502954363822937, "learning_rate": 8.432005446042627e-05, "loss": 0.0319, "step": 88870 }, { "epoch": 56.647546207775655, "grad_norm": 0.7558793425559998, "learning_rate": 8.431604611336105e-05, "loss": 0.0213, "step": 88880 }, { "epoch": 56.653919694072655, "grad_norm": 0.30104249715805054, "learning_rate": 8.431203734932345e-05, "loss": 0.0193, "step": 88890 }, { "epoch": 56.66029318036966, "grad_norm": 0.4844406843185425, "learning_rate": 8.430802816836221e-05, "loss": 0.0221, "step": 88900 }, { "epoch": 56.666666666666664, "grad_norm": 0.3854568898677826, "learning_rate": 8.430401857052601e-05, "loss": 0.0156, "step": 88910 }, { "epoch": 56.67304015296367, "grad_norm": 0.5751428008079529, "learning_rate": 8.430000855586362e-05, "loss": 0.035, "step": 88920 }, { "epoch": 56.67941363926067, "grad_norm": 0.5251092314720154, "learning_rate": 8.429599812442372e-05, "loss": 0.0184, "step": 88930 }, { "epoch": 56.68578712555768, "grad_norm": 0.5491990447044373, "learning_rate": 8.429198727625509e-05, "loss": 0.0232, "step": 88940 }, { "epoch": 56.69216061185468, "grad_norm": 0.4346757233142853, "learning_rate": 8.428797601140642e-05, "loss": 0.0214, "step": 88950 }, { "epoch": 56.69853409815169, "grad_norm": 0.24910549819469452, "learning_rate": 8.428396432992648e-05, "loss": 0.0127, "step": 88960 }, { "epoch": 56.70490758444869, "grad_norm": 0.4894624948501587, "learning_rate": 8.4279952231864e-05, "loss": 0.0258, "step": 88970 }, { "epoch": 56.7112810707457, "grad_norm": 0.47326216101646423, "learning_rate": 8.427593971726772e-05, "loss": 0.0207, "step": 88980 }, { "epoch": 56.7176545570427, "grad_norm": 0.2441207468509674, "learning_rate": 8.427192678618643e-05, "loss": 0.0266, "step": 88990 }, { "epoch": 56.72402804333971, "grad_norm": 0.4305455982685089, "learning_rate": 8.426791343866887e-05, "loss": 0.0163, "step": 89000 }, { "epoch": 56.73040152963671, "grad_norm": 0.4559779167175293, "learning_rate": 8.426389967476381e-05, "loss": 0.0176, "step": 89010 }, { "epoch": 56.736775015933716, "grad_norm": 0.4351821541786194, "learning_rate": 8.425988549452e-05, "loss": 0.027, "step": 89020 }, { "epoch": 56.74314850223072, "grad_norm": 0.3989706039428711, "learning_rate": 8.425587089798624e-05, "loss": 0.017, "step": 89030 }, { "epoch": 56.749521988527725, "grad_norm": 0.45699042081832886, "learning_rate": 8.425185588521131e-05, "loss": 0.0401, "step": 89040 }, { "epoch": 56.755895474824726, "grad_norm": 0.3551419973373413, "learning_rate": 8.424784045624398e-05, "loss": 0.018, "step": 89050 }, { "epoch": 56.762268961121734, "grad_norm": 0.26561206579208374, "learning_rate": 8.424382461113307e-05, "loss": 0.0193, "step": 89060 }, { "epoch": 56.768642447418735, "grad_norm": 0.3921823799610138, "learning_rate": 8.423980834992736e-05, "loss": 0.0138, "step": 89070 }, { "epoch": 56.77501593371574, "grad_norm": 0.6603374481201172, "learning_rate": 8.423579167267564e-05, "loss": 0.0235, "step": 89080 }, { "epoch": 56.781389420012744, "grad_norm": 0.425673246383667, "learning_rate": 8.423177457942674e-05, "loss": 0.0281, "step": 89090 }, { "epoch": 56.78776290630975, "grad_norm": 0.5114652514457703, "learning_rate": 8.422775707022946e-05, "loss": 0.0212, "step": 89100 }, { "epoch": 56.79413639260676, "grad_norm": 0.4565389156341553, "learning_rate": 8.42237391451326e-05, "loss": 0.029, "step": 89110 }, { "epoch": 56.80050987890376, "grad_norm": 0.31695836782455444, "learning_rate": 8.4219720804185e-05, "loss": 0.0163, "step": 89120 }, { "epoch": 56.80688336520077, "grad_norm": 0.3238644301891327, "learning_rate": 8.421570204743549e-05, "loss": 0.0195, "step": 89130 }, { "epoch": 56.81325685149777, "grad_norm": 0.5549646019935608, "learning_rate": 8.421168287493288e-05, "loss": 0.0177, "step": 89140 }, { "epoch": 56.81963033779478, "grad_norm": 0.785845160484314, "learning_rate": 8.420766328672604e-05, "loss": 0.0144, "step": 89150 }, { "epoch": 56.82600382409178, "grad_norm": 0.6765007376670837, "learning_rate": 8.420364328286379e-05, "loss": 0.0161, "step": 89160 }, { "epoch": 56.832377310388786, "grad_norm": 0.29179131984710693, "learning_rate": 8.419962286339497e-05, "loss": 0.0129, "step": 89170 }, { "epoch": 56.83875079668579, "grad_norm": 0.24377037584781647, "learning_rate": 8.419560202836844e-05, "loss": 0.0205, "step": 89180 }, { "epoch": 56.845124282982795, "grad_norm": 0.5405519604682922, "learning_rate": 8.419158077783309e-05, "loss": 0.0214, "step": 89190 }, { "epoch": 56.851497769279796, "grad_norm": 0.16153636574745178, "learning_rate": 8.418755911183773e-05, "loss": 0.0143, "step": 89200 }, { "epoch": 56.8578712555768, "grad_norm": 0.3196083903312683, "learning_rate": 8.418353703043125e-05, "loss": 0.0139, "step": 89210 }, { "epoch": 56.864244741873804, "grad_norm": 0.8624732494354248, "learning_rate": 8.417951453366251e-05, "loss": 0.0183, "step": 89220 }, { "epoch": 56.87061822817081, "grad_norm": 0.6127678751945496, "learning_rate": 8.41754916215804e-05, "loss": 0.0181, "step": 89230 }, { "epoch": 56.87699171446781, "grad_norm": 0.38532915711402893, "learning_rate": 8.417146829423383e-05, "loss": 0.0243, "step": 89240 }, { "epoch": 56.88336520076482, "grad_norm": 0.5425240397453308, "learning_rate": 8.416744455167162e-05, "loss": 0.0188, "step": 89250 }, { "epoch": 56.88973868706182, "grad_norm": 0.4425295889377594, "learning_rate": 8.416342039394272e-05, "loss": 0.0227, "step": 89260 }, { "epoch": 56.89611217335883, "grad_norm": 0.3195371925830841, "learning_rate": 8.4159395821096e-05, "loss": 0.02, "step": 89270 }, { "epoch": 56.90248565965583, "grad_norm": 0.7239754796028137, "learning_rate": 8.415537083318038e-05, "loss": 0.0216, "step": 89280 }, { "epoch": 56.90885914595284, "grad_norm": 0.2908647060394287, "learning_rate": 8.415134543024474e-05, "loss": 0.0227, "step": 89290 }, { "epoch": 56.91523263224984, "grad_norm": 0.5402952432632446, "learning_rate": 8.4147319612338e-05, "loss": 0.0345, "step": 89300 }, { "epoch": 56.92160611854685, "grad_norm": 0.4413334131240845, "learning_rate": 8.41432933795091e-05, "loss": 0.0193, "step": 89310 }, { "epoch": 56.92797960484385, "grad_norm": 0.4403780400753021, "learning_rate": 8.413926673180696e-05, "loss": 0.0189, "step": 89320 }, { "epoch": 56.934353091140856, "grad_norm": 0.39796262979507446, "learning_rate": 8.413523966928048e-05, "loss": 0.026, "step": 89330 }, { "epoch": 56.94072657743786, "grad_norm": 0.41868263483047485, "learning_rate": 8.413121219197861e-05, "loss": 0.0178, "step": 89340 }, { "epoch": 56.947100063734865, "grad_norm": 0.671564519405365, "learning_rate": 8.41271842999503e-05, "loss": 0.0219, "step": 89350 }, { "epoch": 56.953473550031866, "grad_norm": 0.4594668745994568, "learning_rate": 8.412315599324448e-05, "loss": 0.0211, "step": 89360 }, { "epoch": 56.959847036328874, "grad_norm": 0.316904753446579, "learning_rate": 8.411912727191006e-05, "loss": 0.0212, "step": 89370 }, { "epoch": 56.966220522625875, "grad_norm": 0.3670274615287781, "learning_rate": 8.411509813599608e-05, "loss": 0.0192, "step": 89380 }, { "epoch": 56.97259400892288, "grad_norm": 0.3125549256801605, "learning_rate": 8.411106858555142e-05, "loss": 0.0189, "step": 89390 }, { "epoch": 56.978967495219884, "grad_norm": 0.30466461181640625, "learning_rate": 8.410703862062508e-05, "loss": 0.0246, "step": 89400 }, { "epoch": 56.98534098151689, "grad_norm": 0.30217137932777405, "learning_rate": 8.4103008241266e-05, "loss": 0.0207, "step": 89410 }, { "epoch": 56.99171446781389, "grad_norm": 0.45970234274864197, "learning_rate": 8.40989774475232e-05, "loss": 0.0157, "step": 89420 }, { "epoch": 56.9980879541109, "grad_norm": 0.272697389125824, "learning_rate": 8.40949462394456e-05, "loss": 0.0173, "step": 89430 }, { "epoch": 57.0044614404079, "grad_norm": 0.2503385841846466, "learning_rate": 8.409091461708224e-05, "loss": 0.0173, "step": 89440 }, { "epoch": 57.01083492670491, "grad_norm": 0.31914013624191284, "learning_rate": 8.408688258048207e-05, "loss": 0.0177, "step": 89450 }, { "epoch": 57.01720841300191, "grad_norm": 0.4664101302623749, "learning_rate": 8.408285012969408e-05, "loss": 0.0233, "step": 89460 }, { "epoch": 57.02358189929892, "grad_norm": 0.4527615010738373, "learning_rate": 8.40788172647673e-05, "loss": 0.0169, "step": 89470 }, { "epoch": 57.02995538559592, "grad_norm": 0.1816396415233612, "learning_rate": 8.40747839857507e-05, "loss": 0.0245, "step": 89480 }, { "epoch": 57.03632887189293, "grad_norm": 0.5130508542060852, "learning_rate": 8.407075029269332e-05, "loss": 0.0286, "step": 89490 }, { "epoch": 57.04270235818993, "grad_norm": 0.5528929829597473, "learning_rate": 8.406671618564413e-05, "loss": 0.02, "step": 89500 }, { "epoch": 57.049075844486936, "grad_norm": 0.6410553455352783, "learning_rate": 8.406268166465218e-05, "loss": 0.0285, "step": 89510 }, { "epoch": 57.05544933078394, "grad_norm": 0.5955047011375427, "learning_rate": 8.405864672976652e-05, "loss": 0.031, "step": 89520 }, { "epoch": 57.061822817080945, "grad_norm": 0.35587820410728455, "learning_rate": 8.405461138103612e-05, "loss": 0.0222, "step": 89530 }, { "epoch": 57.068196303377945, "grad_norm": 0.49654316902160645, "learning_rate": 8.405057561851005e-05, "loss": 0.0191, "step": 89540 }, { "epoch": 57.07456978967495, "grad_norm": 0.2847869098186493, "learning_rate": 8.40465394422373e-05, "loss": 0.0239, "step": 89550 }, { "epoch": 57.080943275971954, "grad_norm": 0.29756835103034973, "learning_rate": 8.4042502852267e-05, "loss": 0.0165, "step": 89560 }, { "epoch": 57.08731676226896, "grad_norm": 0.31888434290885925, "learning_rate": 8.403846584864812e-05, "loss": 0.0171, "step": 89570 }, { "epoch": 57.09369024856596, "grad_norm": 0.6209361553192139, "learning_rate": 8.403442843142974e-05, "loss": 0.0198, "step": 89580 }, { "epoch": 57.10006373486297, "grad_norm": 0.24462798237800598, "learning_rate": 8.403039060066094e-05, "loss": 0.0146, "step": 89590 }, { "epoch": 57.10643722115997, "grad_norm": 0.30650609731674194, "learning_rate": 8.402635235639075e-05, "loss": 0.0311, "step": 89600 }, { "epoch": 57.11281070745698, "grad_norm": 0.411437451839447, "learning_rate": 8.402231369866827e-05, "loss": 0.0133, "step": 89610 }, { "epoch": 57.11918419375398, "grad_norm": 0.2467217892408371, "learning_rate": 8.401827462754253e-05, "loss": 0.0247, "step": 89620 }, { "epoch": 57.12555768005099, "grad_norm": 0.3094038665294647, "learning_rate": 8.401423514306265e-05, "loss": 0.0173, "step": 89630 }, { "epoch": 57.13193116634799, "grad_norm": 0.7158586978912354, "learning_rate": 8.401019524527769e-05, "loss": 0.0268, "step": 89640 }, { "epoch": 57.138304652645, "grad_norm": 0.42708465456962585, "learning_rate": 8.400615493423675e-05, "loss": 0.0298, "step": 89650 }, { "epoch": 57.144678138942, "grad_norm": 0.40915343165397644, "learning_rate": 8.400211420998893e-05, "loss": 0.0187, "step": 89660 }, { "epoch": 57.151051625239006, "grad_norm": 0.2856992781162262, "learning_rate": 8.399807307258332e-05, "loss": 0.017, "step": 89670 }, { "epoch": 57.15742511153601, "grad_norm": 0.5486305356025696, "learning_rate": 8.399403152206901e-05, "loss": 0.0231, "step": 89680 }, { "epoch": 57.163798597833015, "grad_norm": 0.4538962244987488, "learning_rate": 8.398998955849513e-05, "loss": 0.0261, "step": 89690 }, { "epoch": 57.170172084130016, "grad_norm": 0.40554606914520264, "learning_rate": 8.398594718191079e-05, "loss": 0.0143, "step": 89700 }, { "epoch": 57.176545570427024, "grad_norm": 0.31516873836517334, "learning_rate": 8.398190439236508e-05, "loss": 0.0182, "step": 89710 }, { "epoch": 57.182919056724025, "grad_norm": 0.4766589403152466, "learning_rate": 8.397786118990717e-05, "loss": 0.0264, "step": 89720 }, { "epoch": 57.18929254302103, "grad_norm": 0.5214773416519165, "learning_rate": 8.397381757458616e-05, "loss": 0.0253, "step": 89730 }, { "epoch": 57.19566602931803, "grad_norm": 0.45385462045669556, "learning_rate": 8.396977354645119e-05, "loss": 0.0241, "step": 89740 }, { "epoch": 57.20203951561504, "grad_norm": 0.30090734362602234, "learning_rate": 8.396572910555138e-05, "loss": 0.0247, "step": 89750 }, { "epoch": 57.20841300191205, "grad_norm": 0.35574355721473694, "learning_rate": 8.396168425193592e-05, "loss": 0.0183, "step": 89760 }, { "epoch": 57.21478648820905, "grad_norm": 0.7694368958473206, "learning_rate": 8.395763898565392e-05, "loss": 0.0185, "step": 89770 }, { "epoch": 57.22115997450606, "grad_norm": 0.5543959140777588, "learning_rate": 8.395359330675454e-05, "loss": 0.0153, "step": 89780 }, { "epoch": 57.22753346080306, "grad_norm": 0.3384433388710022, "learning_rate": 8.394954721528696e-05, "loss": 0.0209, "step": 89790 }, { "epoch": 57.23390694710007, "grad_norm": 0.699780285358429, "learning_rate": 8.394550071130031e-05, "loss": 0.0398, "step": 89800 }, { "epoch": 57.24028043339707, "grad_norm": 0.32391348481178284, "learning_rate": 8.39414537948438e-05, "loss": 0.0289, "step": 89810 }, { "epoch": 57.246653919694076, "grad_norm": 0.4683291018009186, "learning_rate": 8.393740646596655e-05, "loss": 0.035, "step": 89820 }, { "epoch": 57.25302740599108, "grad_norm": 0.6120004653930664, "learning_rate": 8.393335872471777e-05, "loss": 0.0275, "step": 89830 }, { "epoch": 57.259400892288085, "grad_norm": 0.419186532497406, "learning_rate": 8.392931057114667e-05, "loss": 0.0257, "step": 89840 }, { "epoch": 57.265774378585085, "grad_norm": 0.5148308277130127, "learning_rate": 8.39252620053024e-05, "loss": 0.0195, "step": 89850 }, { "epoch": 57.27214786488209, "grad_norm": 0.5679914355278015, "learning_rate": 8.392121302723417e-05, "loss": 0.0289, "step": 89860 }, { "epoch": 57.278521351179094, "grad_norm": 0.4626162648200989, "learning_rate": 8.391716363699115e-05, "loss": 0.0232, "step": 89870 }, { "epoch": 57.2848948374761, "grad_norm": 0.31322181224823, "learning_rate": 8.391311383462259e-05, "loss": 0.015, "step": 89880 }, { "epoch": 57.2912683237731, "grad_norm": 0.3781730830669403, "learning_rate": 8.390906362017767e-05, "loss": 0.0182, "step": 89890 }, { "epoch": 57.29764181007011, "grad_norm": 0.4961112439632416, "learning_rate": 8.390501299370563e-05, "loss": 0.0212, "step": 89900 }, { "epoch": 57.30401529636711, "grad_norm": 0.6286005973815918, "learning_rate": 8.390096195525565e-05, "loss": 0.0166, "step": 89910 }, { "epoch": 57.31038878266412, "grad_norm": 0.5444197058677673, "learning_rate": 8.389691050487699e-05, "loss": 0.0223, "step": 89920 }, { "epoch": 57.31676226896112, "grad_norm": 0.2742018401622772, "learning_rate": 8.389285864261887e-05, "loss": 0.0176, "step": 89930 }, { "epoch": 57.32313575525813, "grad_norm": 0.49162429571151733, "learning_rate": 8.38888063685305e-05, "loss": 0.0239, "step": 89940 }, { "epoch": 57.32950924155513, "grad_norm": 0.5490507483482361, "learning_rate": 8.388475368266113e-05, "loss": 0.014, "step": 89950 }, { "epoch": 57.33588272785214, "grad_norm": 0.3573302626609802, "learning_rate": 8.388070058506001e-05, "loss": 0.0151, "step": 89960 }, { "epoch": 57.34225621414914, "grad_norm": 0.4907649755477905, "learning_rate": 8.38766470757764e-05, "loss": 0.0243, "step": 89970 }, { "epoch": 57.348629700446146, "grad_norm": 0.7088059782981873, "learning_rate": 8.387259315485954e-05, "loss": 0.0348, "step": 89980 }, { "epoch": 57.35500318674315, "grad_norm": 0.36850523948669434, "learning_rate": 8.38685388223587e-05, "loss": 0.0184, "step": 89990 }, { "epoch": 57.361376673040155, "grad_norm": 0.3041794002056122, "learning_rate": 8.386448407832313e-05, "loss": 0.0147, "step": 90000 }, { "epoch": 57.367750159337156, "grad_norm": 0.3795422315597534, "learning_rate": 8.38604289228021e-05, "loss": 0.02, "step": 90010 }, { "epoch": 57.374123645634164, "grad_norm": 0.33706802129745483, "learning_rate": 8.38563733558449e-05, "loss": 0.0219, "step": 90020 }, { "epoch": 57.380497131931165, "grad_norm": 0.47220587730407715, "learning_rate": 8.385231737750078e-05, "loss": 0.0203, "step": 90030 }, { "epoch": 57.38687061822817, "grad_norm": 0.2156868427991867, "learning_rate": 8.384826098781906e-05, "loss": 0.0134, "step": 90040 }, { "epoch": 57.393244104525174, "grad_norm": 0.3545942008495331, "learning_rate": 8.384420418684901e-05, "loss": 0.0164, "step": 90050 }, { "epoch": 57.39961759082218, "grad_norm": 0.3200230002403259, "learning_rate": 8.384014697463992e-05, "loss": 0.0251, "step": 90060 }, { "epoch": 57.40599107711918, "grad_norm": 0.34192800521850586, "learning_rate": 8.383608935124109e-05, "loss": 0.0163, "step": 90070 }, { "epoch": 57.41236456341619, "grad_norm": 0.7476663589477539, "learning_rate": 8.383203131670183e-05, "loss": 0.0285, "step": 90080 }, { "epoch": 57.41873804971319, "grad_norm": 0.5134480595588684, "learning_rate": 8.382797287107145e-05, "loss": 0.0211, "step": 90090 }, { "epoch": 57.4251115360102, "grad_norm": 0.40954017639160156, "learning_rate": 8.382391401439927e-05, "loss": 0.0227, "step": 90100 }, { "epoch": 57.4314850223072, "grad_norm": 0.6072366237640381, "learning_rate": 8.381985474673458e-05, "loss": 0.0193, "step": 90110 }, { "epoch": 57.43785850860421, "grad_norm": 0.4288114011287689, "learning_rate": 8.381579506812673e-05, "loss": 0.0195, "step": 90120 }, { "epoch": 57.44423199490121, "grad_norm": 0.27396345138549805, "learning_rate": 8.381173497862504e-05, "loss": 0.0176, "step": 90130 }, { "epoch": 57.45060548119822, "grad_norm": 0.6597847938537598, "learning_rate": 8.380767447827886e-05, "loss": 0.023, "step": 90140 }, { "epoch": 57.45697896749522, "grad_norm": 0.2794700562953949, "learning_rate": 8.380361356713751e-05, "loss": 0.0164, "step": 90150 }, { "epoch": 57.463352453792226, "grad_norm": 0.277604877948761, "learning_rate": 8.379955224525032e-05, "loss": 0.0136, "step": 90160 }, { "epoch": 57.469725940089226, "grad_norm": 0.3635536730289459, "learning_rate": 8.379549051266669e-05, "loss": 0.0171, "step": 90170 }, { "epoch": 57.476099426386234, "grad_norm": 0.2913624048233032, "learning_rate": 8.379142836943592e-05, "loss": 0.0128, "step": 90180 }, { "epoch": 57.482472912683235, "grad_norm": 0.6907758116722107, "learning_rate": 8.37873658156074e-05, "loss": 0.0262, "step": 90190 }, { "epoch": 57.48884639898024, "grad_norm": 0.2994072139263153, "learning_rate": 8.378330285123048e-05, "loss": 0.019, "step": 90200 }, { "epoch": 57.495219885277244, "grad_norm": 0.5289950966835022, "learning_rate": 8.377923947635453e-05, "loss": 0.015, "step": 90210 }, { "epoch": 57.50159337157425, "grad_norm": 0.26163166761398315, "learning_rate": 8.377517569102894e-05, "loss": 0.0136, "step": 90220 }, { "epoch": 57.50796685787125, "grad_norm": 0.3765043616294861, "learning_rate": 8.377111149530306e-05, "loss": 0.0186, "step": 90230 }, { "epoch": 57.51434034416826, "grad_norm": 0.6472104787826538, "learning_rate": 8.376704688922629e-05, "loss": 0.0147, "step": 90240 }, { "epoch": 57.52071383046526, "grad_norm": 0.34595000743865967, "learning_rate": 8.376298187284803e-05, "loss": 0.0228, "step": 90250 }, { "epoch": 57.52708731676227, "grad_norm": 0.6250975728034973, "learning_rate": 8.375891644621767e-05, "loss": 0.023, "step": 90260 }, { "epoch": 57.53346080305927, "grad_norm": 0.4491671919822693, "learning_rate": 8.375485060938459e-05, "loss": 0.0145, "step": 90270 }, { "epoch": 57.53983428935628, "grad_norm": 0.5883461833000183, "learning_rate": 8.375078436239821e-05, "loss": 0.0223, "step": 90280 }, { "epoch": 57.54620777565328, "grad_norm": 0.3602447211742401, "learning_rate": 8.374671770530793e-05, "loss": 0.0222, "step": 90290 }, { "epoch": 57.55258126195029, "grad_norm": 0.43664276599884033, "learning_rate": 8.374265063816316e-05, "loss": 0.0262, "step": 90300 }, { "epoch": 57.55895474824729, "grad_norm": 0.3652399778366089, "learning_rate": 8.373858316101335e-05, "loss": 0.0183, "step": 90310 }, { "epoch": 57.565328234544296, "grad_norm": 0.342951238155365, "learning_rate": 8.373451527390787e-05, "loss": 0.0158, "step": 90320 }, { "epoch": 57.5717017208413, "grad_norm": 0.2996302545070648, "learning_rate": 8.373044697689621e-05, "loss": 0.0207, "step": 90330 }, { "epoch": 57.578075207138305, "grad_norm": 0.470356822013855, "learning_rate": 8.372637827002776e-05, "loss": 0.0229, "step": 90340 }, { "epoch": 57.584448693435306, "grad_norm": 0.48482009768486023, "learning_rate": 8.372230915335195e-05, "loss": 0.0187, "step": 90350 }, { "epoch": 57.590822179732314, "grad_norm": 0.6500172019004822, "learning_rate": 8.371823962691827e-05, "loss": 0.0178, "step": 90360 }, { "epoch": 57.59719566602932, "grad_norm": 0.3064844608306885, "learning_rate": 8.371416969077614e-05, "loss": 0.0165, "step": 90370 }, { "epoch": 57.60356915232632, "grad_norm": 0.49580442905426025, "learning_rate": 8.371009934497502e-05, "loss": 0.0201, "step": 90380 }, { "epoch": 57.60994263862333, "grad_norm": 0.3165915608406067, "learning_rate": 8.370602858956434e-05, "loss": 0.0202, "step": 90390 }, { "epoch": 57.61631612492033, "grad_norm": 0.3350183665752411, "learning_rate": 8.370195742459362e-05, "loss": 0.0224, "step": 90400 }, { "epoch": 57.62268961121734, "grad_norm": 0.9619952440261841, "learning_rate": 8.36978858501123e-05, "loss": 0.0296, "step": 90410 }, { "epoch": 57.62906309751434, "grad_norm": 0.39266541600227356, "learning_rate": 8.369381386616981e-05, "loss": 0.0174, "step": 90420 }, { "epoch": 57.63543658381135, "grad_norm": 0.43098047375679016, "learning_rate": 8.36897414728157e-05, "loss": 0.0167, "step": 90430 }, { "epoch": 57.64181007010835, "grad_norm": 0.6389949321746826, "learning_rate": 8.368566867009942e-05, "loss": 0.0166, "step": 90440 }, { "epoch": 57.64818355640536, "grad_norm": 0.2967068552970886, "learning_rate": 8.368159545807047e-05, "loss": 0.0213, "step": 90450 }, { "epoch": 57.65455704270236, "grad_norm": 0.4471738636493683, "learning_rate": 8.36775218367783e-05, "loss": 0.0207, "step": 90460 }, { "epoch": 57.660930528999366, "grad_norm": 0.24095755815505981, "learning_rate": 8.367344780627246e-05, "loss": 0.0192, "step": 90470 }, { "epoch": 57.66730401529637, "grad_norm": 0.28147220611572266, "learning_rate": 8.366937336660246e-05, "loss": 0.0213, "step": 90480 }, { "epoch": 57.673677501593374, "grad_norm": 0.380973756313324, "learning_rate": 8.366529851781776e-05, "loss": 0.0217, "step": 90490 }, { "epoch": 57.680050987890375, "grad_norm": 0.27357548475265503, "learning_rate": 8.36612232599679e-05, "loss": 0.0202, "step": 90500 }, { "epoch": 57.68642447418738, "grad_norm": 0.5490698218345642, "learning_rate": 8.36571475931024e-05, "loss": 0.0209, "step": 90510 }, { "epoch": 57.692797960484384, "grad_norm": 0.40168046951293945, "learning_rate": 8.365307151727078e-05, "loss": 0.0173, "step": 90520 }, { "epoch": 57.69917144678139, "grad_norm": 0.4302065670490265, "learning_rate": 8.364899503252257e-05, "loss": 0.0216, "step": 90530 }, { "epoch": 57.70554493307839, "grad_norm": 0.5215052962303162, "learning_rate": 8.36449181389073e-05, "loss": 0.0156, "step": 90540 }, { "epoch": 57.7119184193754, "grad_norm": 0.6719648838043213, "learning_rate": 8.36408408364745e-05, "loss": 0.0287, "step": 90550 }, { "epoch": 57.7182919056724, "grad_norm": 0.41499289870262146, "learning_rate": 8.363676312527372e-05, "loss": 0.024, "step": 90560 }, { "epoch": 57.72466539196941, "grad_norm": 0.37071362137794495, "learning_rate": 8.363268500535452e-05, "loss": 0.0201, "step": 90570 }, { "epoch": 57.73103887826641, "grad_norm": 0.49417492747306824, "learning_rate": 8.362860647676644e-05, "loss": 0.0162, "step": 90580 }, { "epoch": 57.73741236456342, "grad_norm": 0.520778238773346, "learning_rate": 8.362452753955904e-05, "loss": 0.0211, "step": 90590 }, { "epoch": 57.74378585086042, "grad_norm": 0.3248370587825775, "learning_rate": 8.362044819378188e-05, "loss": 0.0238, "step": 90600 }, { "epoch": 57.75015933715743, "grad_norm": 0.45907577872276306, "learning_rate": 8.361636843948454e-05, "loss": 0.0139, "step": 90610 }, { "epoch": 57.75653282345443, "grad_norm": 0.39414939284324646, "learning_rate": 8.361228827671657e-05, "loss": 0.0154, "step": 90620 }, { "epoch": 57.762906309751436, "grad_norm": 0.5685986876487732, "learning_rate": 8.360820770552757e-05, "loss": 0.0256, "step": 90630 }, { "epoch": 57.76927979604844, "grad_norm": 0.4703851342201233, "learning_rate": 8.360412672596712e-05, "loss": 0.0108, "step": 90640 }, { "epoch": 57.775653282345445, "grad_norm": 0.4228808879852295, "learning_rate": 8.36000453380848e-05, "loss": 0.0126, "step": 90650 }, { "epoch": 57.782026768642446, "grad_norm": 0.29923856258392334, "learning_rate": 8.35959635419302e-05, "loss": 0.0179, "step": 90660 }, { "epoch": 57.788400254939454, "grad_norm": 0.7327203154563904, "learning_rate": 8.359188133755294e-05, "loss": 0.0289, "step": 90670 }, { "epoch": 57.794773741236455, "grad_norm": 0.4385051131248474, "learning_rate": 8.358779872500259e-05, "loss": 0.0172, "step": 90680 }, { "epoch": 57.80114722753346, "grad_norm": 0.4585539400577545, "learning_rate": 8.358371570432877e-05, "loss": 0.0179, "step": 90690 }, { "epoch": 57.80752071383046, "grad_norm": 0.5885264873504639, "learning_rate": 8.35796322755811e-05, "loss": 0.0196, "step": 90700 }, { "epoch": 57.81389420012747, "grad_norm": 0.25167813897132874, "learning_rate": 8.357554843880919e-05, "loss": 0.0182, "step": 90710 }, { "epoch": 57.82026768642447, "grad_norm": 0.45056578516960144, "learning_rate": 8.357146419406268e-05, "loss": 0.0292, "step": 90720 }, { "epoch": 57.82664117272148, "grad_norm": 0.8499084115028381, "learning_rate": 8.356737954139118e-05, "loss": 0.0327, "step": 90730 }, { "epoch": 57.83301465901848, "grad_norm": 0.6569238305091858, "learning_rate": 8.356329448084432e-05, "loss": 0.0181, "step": 90740 }, { "epoch": 57.83938814531549, "grad_norm": 0.3645811676979065, "learning_rate": 8.355920901247174e-05, "loss": 0.0199, "step": 90750 }, { "epoch": 57.84576163161249, "grad_norm": 0.45569077134132385, "learning_rate": 8.355512313632309e-05, "loss": 0.0189, "step": 90760 }, { "epoch": 57.8521351179095, "grad_norm": 0.28670087456703186, "learning_rate": 8.355103685244801e-05, "loss": 0.0197, "step": 90770 }, { "epoch": 57.8585086042065, "grad_norm": 0.24539832770824432, "learning_rate": 8.354695016089616e-05, "loss": 0.0178, "step": 90780 }, { "epoch": 57.86488209050351, "grad_norm": 0.23846940696239471, "learning_rate": 8.35428630617172e-05, "loss": 0.0234, "step": 90790 }, { "epoch": 57.87125557680051, "grad_norm": 0.519631564617157, "learning_rate": 8.353877555496076e-05, "loss": 0.0182, "step": 90800 }, { "epoch": 57.877629063097515, "grad_norm": 0.6478245854377747, "learning_rate": 8.353468764067654e-05, "loss": 0.033, "step": 90810 }, { "epoch": 57.884002549394516, "grad_norm": 0.3000773787498474, "learning_rate": 8.353059931891422e-05, "loss": 0.0236, "step": 90820 }, { "epoch": 57.890376035691524, "grad_norm": 0.2870498597621918, "learning_rate": 8.352651058972344e-05, "loss": 0.024, "step": 90830 }, { "epoch": 57.896749521988525, "grad_norm": 0.5736768841743469, "learning_rate": 8.352242145315392e-05, "loss": 0.0234, "step": 90840 }, { "epoch": 57.90312300828553, "grad_norm": 0.2978558838367462, "learning_rate": 8.351833190925531e-05, "loss": 0.0143, "step": 90850 }, { "epoch": 57.909496494582534, "grad_norm": 0.23081423342227936, "learning_rate": 8.351424195807733e-05, "loss": 0.0191, "step": 90860 }, { "epoch": 57.91586998087954, "grad_norm": 0.5447297692298889, "learning_rate": 8.351015159966966e-05, "loss": 0.0159, "step": 90870 }, { "epoch": 57.92224346717654, "grad_norm": 0.3341819643974304, "learning_rate": 8.350606083408202e-05, "loss": 0.0215, "step": 90880 }, { "epoch": 57.92861695347355, "grad_norm": 0.36266979575157166, "learning_rate": 8.35019696613641e-05, "loss": 0.0179, "step": 90890 }, { "epoch": 57.93499043977055, "grad_norm": 0.2404865026473999, "learning_rate": 8.349787808156562e-05, "loss": 0.0194, "step": 90900 }, { "epoch": 57.94136392606756, "grad_norm": 0.3146337568759918, "learning_rate": 8.34937860947363e-05, "loss": 0.0232, "step": 90910 }, { "epoch": 57.94773741236456, "grad_norm": 0.42501819133758545, "learning_rate": 8.348969370092584e-05, "loss": 0.0312, "step": 90920 }, { "epoch": 57.95411089866157, "grad_norm": 0.3568401634693146, "learning_rate": 8.348560090018399e-05, "loss": 0.0129, "step": 90930 }, { "epoch": 57.96048438495857, "grad_norm": 0.5907946825027466, "learning_rate": 8.348150769256047e-05, "loss": 0.0222, "step": 90940 }, { "epoch": 57.96685787125558, "grad_norm": 0.41706475615501404, "learning_rate": 8.347741407810503e-05, "loss": 0.026, "step": 90950 }, { "epoch": 57.97323135755258, "grad_norm": 0.39935413002967834, "learning_rate": 8.34733200568674e-05, "loss": 0.0369, "step": 90960 }, { "epoch": 57.979604843849586, "grad_norm": 0.3769091069698334, "learning_rate": 8.346922562889732e-05, "loss": 0.0251, "step": 90970 }, { "epoch": 57.98597833014659, "grad_norm": 0.38300979137420654, "learning_rate": 8.346513079424455e-05, "loss": 0.0193, "step": 90980 }, { "epoch": 57.992351816443595, "grad_norm": 0.4498318135738373, "learning_rate": 8.346103555295883e-05, "loss": 0.0197, "step": 90990 }, { "epoch": 57.998725302740596, "grad_norm": 0.455406129360199, "learning_rate": 8.345693990508996e-05, "loss": 0.0227, "step": 91000 }, { "epoch": 58.0050987890376, "grad_norm": 0.5942676067352295, "learning_rate": 8.345284385068768e-05, "loss": 0.021, "step": 91010 }, { "epoch": 58.01147227533461, "grad_norm": 0.3158631920814514, "learning_rate": 8.344874738980174e-05, "loss": 0.0244, "step": 91020 }, { "epoch": 58.01784576163161, "grad_norm": 0.713674008846283, "learning_rate": 8.344465052248196e-05, "loss": 0.027, "step": 91030 }, { "epoch": 58.02421924792862, "grad_norm": 0.3068496584892273, "learning_rate": 8.34405532487781e-05, "loss": 0.0181, "step": 91040 }, { "epoch": 58.03059273422562, "grad_norm": 0.437433123588562, "learning_rate": 8.343645556873994e-05, "loss": 0.0132, "step": 91050 }, { "epoch": 58.03696622052263, "grad_norm": 0.5695892572402954, "learning_rate": 8.343235748241727e-05, "loss": 0.0277, "step": 91060 }, { "epoch": 58.04333970681963, "grad_norm": 0.4153307378292084, "learning_rate": 8.342825898985989e-05, "loss": 0.0277, "step": 91070 }, { "epoch": 58.04971319311664, "grad_norm": 0.4828537702560425, "learning_rate": 8.34241600911176e-05, "loss": 0.0169, "step": 91080 }, { "epoch": 58.05608667941364, "grad_norm": 0.5518837571144104, "learning_rate": 8.342006078624023e-05, "loss": 0.021, "step": 91090 }, { "epoch": 58.06246016571065, "grad_norm": 0.6673458814620972, "learning_rate": 8.341596107527753e-05, "loss": 0.0238, "step": 91100 }, { "epoch": 58.06883365200765, "grad_norm": 0.3064559996128082, "learning_rate": 8.341186095827939e-05, "loss": 0.024, "step": 91110 }, { "epoch": 58.075207138304656, "grad_norm": 0.6941390037536621, "learning_rate": 8.340776043529557e-05, "loss": 0.015, "step": 91120 }, { "epoch": 58.081580624601656, "grad_norm": 0.5841116905212402, "learning_rate": 8.340365950637592e-05, "loss": 0.0171, "step": 91130 }, { "epoch": 58.087954110898664, "grad_norm": 0.48552754521369934, "learning_rate": 8.339955817157028e-05, "loss": 0.0259, "step": 91140 }, { "epoch": 58.094327597195665, "grad_norm": 0.5929617285728455, "learning_rate": 8.339545643092848e-05, "loss": 0.03, "step": 91150 }, { "epoch": 58.10070108349267, "grad_norm": 0.49149858951568604, "learning_rate": 8.339135428450034e-05, "loss": 0.0371, "step": 91160 }, { "epoch": 58.107074569789674, "grad_norm": 0.4183204174041748, "learning_rate": 8.338725173233571e-05, "loss": 0.0172, "step": 91170 }, { "epoch": 58.11344805608668, "grad_norm": 0.27595144510269165, "learning_rate": 8.338314877448447e-05, "loss": 0.0189, "step": 91180 }, { "epoch": 58.11982154238368, "grad_norm": 0.7357737421989441, "learning_rate": 8.337904541099643e-05, "loss": 0.0199, "step": 91190 }, { "epoch": 58.12619502868069, "grad_norm": 0.40764376521110535, "learning_rate": 8.337494164192149e-05, "loss": 0.0294, "step": 91200 }, { "epoch": 58.13256851497769, "grad_norm": 0.3002212345600128, "learning_rate": 8.33708374673095e-05, "loss": 0.0204, "step": 91210 }, { "epoch": 58.1389420012747, "grad_norm": 0.28194454312324524, "learning_rate": 8.336673288721031e-05, "loss": 0.0213, "step": 91220 }, { "epoch": 58.1453154875717, "grad_norm": 0.35231664776802063, "learning_rate": 8.336262790167383e-05, "loss": 0.0237, "step": 91230 }, { "epoch": 58.15168897386871, "grad_norm": 0.48193374276161194, "learning_rate": 8.335852251074992e-05, "loss": 0.0218, "step": 91240 }, { "epoch": 58.15806246016571, "grad_norm": 0.3032952547073364, "learning_rate": 8.335441671448845e-05, "loss": 0.0151, "step": 91250 }, { "epoch": 58.16443594646272, "grad_norm": 0.7593832612037659, "learning_rate": 8.335031051293933e-05, "loss": 0.0346, "step": 91260 }, { "epoch": 58.17080943275972, "grad_norm": 0.2837711572647095, "learning_rate": 8.334620390615246e-05, "loss": 0.0188, "step": 91270 }, { "epoch": 58.177182919056726, "grad_norm": 0.272061824798584, "learning_rate": 8.334209689417773e-05, "loss": 0.0192, "step": 91280 }, { "epoch": 58.18355640535373, "grad_norm": 0.4461846351623535, "learning_rate": 8.333798947706504e-05, "loss": 0.022, "step": 91290 }, { "epoch": 58.189929891650735, "grad_norm": 0.6074658036231995, "learning_rate": 8.333388165486429e-05, "loss": 0.0239, "step": 91300 }, { "epoch": 58.196303377947736, "grad_norm": 0.43046948313713074, "learning_rate": 8.332977342762541e-05, "loss": 0.0206, "step": 91310 }, { "epoch": 58.202676864244744, "grad_norm": 0.19798019528388977, "learning_rate": 8.332566479539833e-05, "loss": 0.0127, "step": 91320 }, { "epoch": 58.209050350541744, "grad_norm": 0.3855135440826416, "learning_rate": 8.332155575823294e-05, "loss": 0.0193, "step": 91330 }, { "epoch": 58.21542383683875, "grad_norm": 0.32505425810813904, "learning_rate": 8.33174463161792e-05, "loss": 0.0181, "step": 91340 }, { "epoch": 58.22179732313575, "grad_norm": 0.46706852316856384, "learning_rate": 8.331333646928701e-05, "loss": 0.0196, "step": 91350 }, { "epoch": 58.22817080943276, "grad_norm": 0.3934280574321747, "learning_rate": 8.330922621760635e-05, "loss": 0.0152, "step": 91360 }, { "epoch": 58.23454429572976, "grad_norm": 0.3870113790035248, "learning_rate": 8.330511556118713e-05, "loss": 0.0173, "step": 91370 }, { "epoch": 58.24091778202677, "grad_norm": 0.42795488238334656, "learning_rate": 8.330100450007931e-05, "loss": 0.0246, "step": 91380 }, { "epoch": 58.24729126832377, "grad_norm": 0.379820853471756, "learning_rate": 8.329689303433285e-05, "loss": 0.0201, "step": 91390 }, { "epoch": 58.25366475462078, "grad_norm": 0.6734418869018555, "learning_rate": 8.32927811639977e-05, "loss": 0.0159, "step": 91400 }, { "epoch": 58.26003824091778, "grad_norm": 0.3046272397041321, "learning_rate": 8.328866888912384e-05, "loss": 0.0191, "step": 91410 }, { "epoch": 58.26641172721479, "grad_norm": 0.2194511443376541, "learning_rate": 8.32845562097612e-05, "loss": 0.0141, "step": 91420 }, { "epoch": 58.27278521351179, "grad_norm": 0.516089916229248, "learning_rate": 8.32804431259598e-05, "loss": 0.0178, "step": 91430 }, { "epoch": 58.2791586998088, "grad_norm": 0.49367237091064453, "learning_rate": 8.327632963776956e-05, "loss": 0.0154, "step": 91440 }, { "epoch": 58.2855321861058, "grad_norm": 0.23633047938346863, "learning_rate": 8.327221574524053e-05, "loss": 0.0247, "step": 91450 }, { "epoch": 58.291905672402805, "grad_norm": 0.46073925495147705, "learning_rate": 8.326810144842265e-05, "loss": 0.0183, "step": 91460 }, { "epoch": 58.298279158699806, "grad_norm": 0.3954058587551117, "learning_rate": 8.326398674736592e-05, "loss": 0.0147, "step": 91470 }, { "epoch": 58.304652644996814, "grad_norm": 0.8969352841377258, "learning_rate": 8.325987164212036e-05, "loss": 0.0198, "step": 91480 }, { "epoch": 58.311026131293815, "grad_norm": 0.38300052285194397, "learning_rate": 8.325575613273595e-05, "loss": 0.0166, "step": 91490 }, { "epoch": 58.31739961759082, "grad_norm": 0.3057568073272705, "learning_rate": 8.32516402192627e-05, "loss": 0.0188, "step": 91500 }, { "epoch": 58.323773103887824, "grad_norm": 0.1950339376926422, "learning_rate": 8.324752390175063e-05, "loss": 0.0263, "step": 91510 }, { "epoch": 58.33014659018483, "grad_norm": 0.5426521897315979, "learning_rate": 8.324340718024977e-05, "loss": 0.0177, "step": 91520 }, { "epoch": 58.33652007648183, "grad_norm": 0.37104398012161255, "learning_rate": 8.32392900548101e-05, "loss": 0.0202, "step": 91530 }, { "epoch": 58.34289356277884, "grad_norm": 0.3837517499923706, "learning_rate": 8.323517252548169e-05, "loss": 0.0184, "step": 91540 }, { "epoch": 58.34926704907584, "grad_norm": 0.4697130620479584, "learning_rate": 8.323105459231455e-05, "loss": 0.0128, "step": 91550 }, { "epoch": 58.35564053537285, "grad_norm": 0.46409332752227783, "learning_rate": 8.322693625535872e-05, "loss": 0.0206, "step": 91560 }, { "epoch": 58.36201402166985, "grad_norm": 0.30239129066467285, "learning_rate": 8.322281751466424e-05, "loss": 0.0316, "step": 91570 }, { "epoch": 58.36838750796686, "grad_norm": 0.6283408403396606, "learning_rate": 8.321869837028116e-05, "loss": 0.0232, "step": 91580 }, { "epoch": 58.37476099426386, "grad_norm": 0.601788341999054, "learning_rate": 8.321457882225952e-05, "loss": 0.0167, "step": 91590 }, { "epoch": 58.38113448056087, "grad_norm": 0.5789155960083008, "learning_rate": 8.32104588706494e-05, "loss": 0.023, "step": 91600 }, { "epoch": 58.38750796685787, "grad_norm": 0.4448530673980713, "learning_rate": 8.320633851550086e-05, "loss": 0.0204, "step": 91610 }, { "epoch": 58.393881453154876, "grad_norm": 0.6346665620803833, "learning_rate": 8.320221775686394e-05, "loss": 0.0155, "step": 91620 }, { "epoch": 58.400254939451884, "grad_norm": 0.5220521092414856, "learning_rate": 8.319809659478873e-05, "loss": 0.0264, "step": 91630 }, { "epoch": 58.406628425748885, "grad_norm": 0.2869974970817566, "learning_rate": 8.31939750293253e-05, "loss": 0.0216, "step": 91640 }, { "epoch": 58.41300191204589, "grad_norm": 0.3271142244338989, "learning_rate": 8.318985306052374e-05, "loss": 0.0255, "step": 91650 }, { "epoch": 58.41937539834289, "grad_norm": 0.3671657145023346, "learning_rate": 8.318573068843413e-05, "loss": 0.0257, "step": 91660 }, { "epoch": 58.4257488846399, "grad_norm": 0.2706374228000641, "learning_rate": 8.318160791310653e-05, "loss": 0.024, "step": 91670 }, { "epoch": 58.4321223709369, "grad_norm": 0.4009813666343689, "learning_rate": 8.317748473459111e-05, "loss": 0.0206, "step": 91680 }, { "epoch": 58.43849585723391, "grad_norm": 0.4837583899497986, "learning_rate": 8.31733611529379e-05, "loss": 0.0157, "step": 91690 }, { "epoch": 58.44486934353091, "grad_norm": 0.42084696888923645, "learning_rate": 8.316923716819704e-05, "loss": 0.0163, "step": 91700 }, { "epoch": 58.45124282982792, "grad_norm": 0.21271800994873047, "learning_rate": 8.316511278041862e-05, "loss": 0.0201, "step": 91710 }, { "epoch": 58.45761631612492, "grad_norm": 0.5939106941223145, "learning_rate": 8.316098798965279e-05, "loss": 0.0192, "step": 91720 }, { "epoch": 58.46398980242193, "grad_norm": 0.26054078340530396, "learning_rate": 8.315686279594964e-05, "loss": 0.0158, "step": 91730 }, { "epoch": 58.47036328871893, "grad_norm": 0.664340615272522, "learning_rate": 8.315273719935928e-05, "loss": 0.0216, "step": 91740 }, { "epoch": 58.47673677501594, "grad_norm": 0.6056423783302307, "learning_rate": 8.314861119993186e-05, "loss": 0.0214, "step": 91750 }, { "epoch": 58.48311026131294, "grad_norm": 0.42953068017959595, "learning_rate": 8.314448479771754e-05, "loss": 0.0194, "step": 91760 }, { "epoch": 58.489483747609945, "grad_norm": 0.5340674519538879, "learning_rate": 8.314035799276643e-05, "loss": 0.0172, "step": 91770 }, { "epoch": 58.495857233906946, "grad_norm": 0.26035353541374207, "learning_rate": 8.313623078512869e-05, "loss": 0.0131, "step": 91780 }, { "epoch": 58.502230720203954, "grad_norm": 0.4818899631500244, "learning_rate": 8.313210317485443e-05, "loss": 0.0184, "step": 91790 }, { "epoch": 58.508604206500955, "grad_norm": 0.586942195892334, "learning_rate": 8.312797516199387e-05, "loss": 0.0214, "step": 91800 }, { "epoch": 58.51497769279796, "grad_norm": 0.5176035165786743, "learning_rate": 8.31238467465971e-05, "loss": 0.0247, "step": 91810 }, { "epoch": 58.521351179094964, "grad_norm": 0.60249924659729, "learning_rate": 8.311971792871435e-05, "loss": 0.0157, "step": 91820 }, { "epoch": 58.52772466539197, "grad_norm": 0.45866042375564575, "learning_rate": 8.311558870839573e-05, "loss": 0.0284, "step": 91830 }, { "epoch": 58.53409815168897, "grad_norm": 0.5293552875518799, "learning_rate": 8.311145908569147e-05, "loss": 0.0156, "step": 91840 }, { "epoch": 58.54047163798598, "grad_norm": 0.4334186911582947, "learning_rate": 8.310732906065168e-05, "loss": 0.023, "step": 91850 }, { "epoch": 58.54684512428298, "grad_norm": 0.2653294801712036, "learning_rate": 8.310319863332662e-05, "loss": 0.0179, "step": 91860 }, { "epoch": 58.55321861057999, "grad_norm": 0.34615960717201233, "learning_rate": 8.309906780376642e-05, "loss": 0.0142, "step": 91870 }, { "epoch": 58.55959209687699, "grad_norm": 0.3851594924926758, "learning_rate": 8.309493657202131e-05, "loss": 0.0192, "step": 91880 }, { "epoch": 58.565965583174, "grad_norm": 0.3321509063243866, "learning_rate": 8.309080493814147e-05, "loss": 0.0181, "step": 91890 }, { "epoch": 58.572339069471, "grad_norm": 0.270266592502594, "learning_rate": 8.30866729021771e-05, "loss": 0.0215, "step": 91900 }, { "epoch": 58.57871255576801, "grad_norm": 0.42378073930740356, "learning_rate": 8.308254046417843e-05, "loss": 0.0215, "step": 91910 }, { "epoch": 58.58508604206501, "grad_norm": 0.6370362639427185, "learning_rate": 8.307840762419563e-05, "loss": 0.0271, "step": 91920 }, { "epoch": 58.591459528362016, "grad_norm": 0.5408931970596313, "learning_rate": 8.307427438227897e-05, "loss": 0.0211, "step": 91930 }, { "epoch": 58.59783301465902, "grad_norm": 0.35163483023643494, "learning_rate": 8.307014073847864e-05, "loss": 0.0255, "step": 91940 }, { "epoch": 58.604206500956025, "grad_norm": 0.2837541401386261, "learning_rate": 8.306600669284488e-05, "loss": 0.0174, "step": 91950 }, { "epoch": 58.610579987253026, "grad_norm": 0.28027379512786865, "learning_rate": 8.306187224542791e-05, "loss": 0.0138, "step": 91960 }, { "epoch": 58.61695347355003, "grad_norm": 0.35531988739967346, "learning_rate": 8.3057737396278e-05, "loss": 0.0257, "step": 91970 }, { "epoch": 58.623326959847034, "grad_norm": 0.47617557644844055, "learning_rate": 8.305360214544535e-05, "loss": 0.0162, "step": 91980 }, { "epoch": 58.62970044614404, "grad_norm": 0.41942328214645386, "learning_rate": 8.304946649298023e-05, "loss": 0.0132, "step": 91990 }, { "epoch": 58.63607393244104, "grad_norm": 0.33593735098838806, "learning_rate": 8.304533043893288e-05, "loss": 0.027, "step": 92000 }, { "epoch": 58.64244741873805, "grad_norm": 0.1853705793619156, "learning_rate": 8.304119398335357e-05, "loss": 0.0176, "step": 92010 }, { "epoch": 58.64882090503505, "grad_norm": 0.750265896320343, "learning_rate": 8.303705712629257e-05, "loss": 0.0208, "step": 92020 }, { "epoch": 58.65519439133206, "grad_norm": 0.4242609739303589, "learning_rate": 8.303291986780014e-05, "loss": 0.0176, "step": 92030 }, { "epoch": 58.66156787762906, "grad_norm": 0.4339333176612854, "learning_rate": 8.302878220792651e-05, "loss": 0.0248, "step": 92040 }, { "epoch": 58.66794136392607, "grad_norm": 0.38782384991645813, "learning_rate": 8.302464414672202e-05, "loss": 0.0296, "step": 92050 }, { "epoch": 58.67431485022307, "grad_norm": 0.648065447807312, "learning_rate": 8.302050568423692e-05, "loss": 0.02, "step": 92060 }, { "epoch": 58.68068833652008, "grad_norm": 0.532376766204834, "learning_rate": 8.301636682052149e-05, "loss": 0.0225, "step": 92070 }, { "epoch": 58.68706182281708, "grad_norm": 0.32371726632118225, "learning_rate": 8.301222755562603e-05, "loss": 0.0153, "step": 92080 }, { "epoch": 58.693435309114086, "grad_norm": 0.6140322089195251, "learning_rate": 8.300808788960083e-05, "loss": 0.0237, "step": 92090 }, { "epoch": 58.69980879541109, "grad_norm": 0.2801460921764374, "learning_rate": 8.300394782249621e-05, "loss": 0.0174, "step": 92100 }, { "epoch": 58.706182281708095, "grad_norm": 0.42450451850891113, "learning_rate": 8.299980735436248e-05, "loss": 0.0219, "step": 92110 }, { "epoch": 58.712555768005096, "grad_norm": 0.607071578502655, "learning_rate": 8.299566648524991e-05, "loss": 0.0201, "step": 92120 }, { "epoch": 58.718929254302104, "grad_norm": 0.5148131251335144, "learning_rate": 8.299152521520884e-05, "loss": 0.0262, "step": 92130 }, { "epoch": 58.725302740599105, "grad_norm": 0.3308785557746887, "learning_rate": 8.29873835442896e-05, "loss": 0.0263, "step": 92140 }, { "epoch": 58.73167622689611, "grad_norm": 0.7380141019821167, "learning_rate": 8.29832414725425e-05, "loss": 0.0126, "step": 92150 }, { "epoch": 58.738049713193114, "grad_norm": 0.3335932195186615, "learning_rate": 8.297909900001787e-05, "loss": 0.0165, "step": 92160 }, { "epoch": 58.74442319949012, "grad_norm": 0.27494731545448303, "learning_rate": 8.297495612676604e-05, "loss": 0.0314, "step": 92170 }, { "epoch": 58.75079668578712, "grad_norm": 0.3603711724281311, "learning_rate": 8.297081285283738e-05, "loss": 0.0217, "step": 92180 }, { "epoch": 58.75717017208413, "grad_norm": 0.3022996485233307, "learning_rate": 8.296666917828222e-05, "loss": 0.0264, "step": 92190 }, { "epoch": 58.76354365838113, "grad_norm": 0.3028657138347626, "learning_rate": 8.296252510315089e-05, "loss": 0.0246, "step": 92200 }, { "epoch": 58.76991714467814, "grad_norm": 0.7696290612220764, "learning_rate": 8.295838062749374e-05, "loss": 0.0245, "step": 92210 }, { "epoch": 58.77629063097514, "grad_norm": 0.3597107231616974, "learning_rate": 8.295423575136117e-05, "loss": 0.0446, "step": 92220 }, { "epoch": 58.78266411727215, "grad_norm": 0.7335372567176819, "learning_rate": 8.295009047480354e-05, "loss": 0.018, "step": 92230 }, { "epoch": 58.78903760356915, "grad_norm": 0.31487876176834106, "learning_rate": 8.294594479787116e-05, "loss": 0.0183, "step": 92240 }, { "epoch": 58.79541108986616, "grad_norm": 0.47309112548828125, "learning_rate": 8.294179872061447e-05, "loss": 0.0318, "step": 92250 }, { "epoch": 58.80178457616316, "grad_norm": 0.29446467757225037, "learning_rate": 8.293765224308381e-05, "loss": 0.0251, "step": 92260 }, { "epoch": 58.808158062460166, "grad_norm": 0.2540798783302307, "learning_rate": 8.29335053653296e-05, "loss": 0.0227, "step": 92270 }, { "epoch": 58.814531548757174, "grad_norm": 0.35325634479522705, "learning_rate": 8.292935808740218e-05, "loss": 0.0203, "step": 92280 }, { "epoch": 58.820905035054174, "grad_norm": 0.6192834973335266, "learning_rate": 8.292521040935197e-05, "loss": 0.0339, "step": 92290 }, { "epoch": 58.82727852135118, "grad_norm": 0.6190416812896729, "learning_rate": 8.292106233122939e-05, "loss": 0.0203, "step": 92300 }, { "epoch": 58.83365200764818, "grad_norm": 0.6363612413406372, "learning_rate": 8.29169138530848e-05, "loss": 0.0348, "step": 92310 }, { "epoch": 58.84002549394519, "grad_norm": 0.444632351398468, "learning_rate": 8.291276497496863e-05, "loss": 0.0178, "step": 92320 }, { "epoch": 58.84639898024219, "grad_norm": 0.779443085193634, "learning_rate": 8.290861569693129e-05, "loss": 0.028, "step": 92330 }, { "epoch": 58.8527724665392, "grad_norm": 0.6048526167869568, "learning_rate": 8.29044660190232e-05, "loss": 0.0277, "step": 92340 }, { "epoch": 58.8591459528362, "grad_norm": 0.6131560802459717, "learning_rate": 8.29003159412948e-05, "loss": 0.0212, "step": 92350 }, { "epoch": 58.86551943913321, "grad_norm": 0.5680294632911682, "learning_rate": 8.289616546379646e-05, "loss": 0.0187, "step": 92360 }, { "epoch": 58.87189292543021, "grad_norm": 0.5247848033905029, "learning_rate": 8.289201458657868e-05, "loss": 0.0245, "step": 92370 }, { "epoch": 58.87826641172722, "grad_norm": 0.48301443457603455, "learning_rate": 8.288786330969185e-05, "loss": 0.0238, "step": 92380 }, { "epoch": 58.88463989802422, "grad_norm": 0.6292096376419067, "learning_rate": 8.288371163318645e-05, "loss": 0.0221, "step": 92390 }, { "epoch": 58.89101338432123, "grad_norm": 0.32192355394363403, "learning_rate": 8.287955955711289e-05, "loss": 0.0156, "step": 92400 }, { "epoch": 58.89738687061823, "grad_norm": 0.8315854668617249, "learning_rate": 8.287540708152164e-05, "loss": 0.0359, "step": 92410 }, { "epoch": 58.903760356915235, "grad_norm": 0.6144301891326904, "learning_rate": 8.287125420646316e-05, "loss": 0.0268, "step": 92420 }, { "epoch": 58.910133843212236, "grad_norm": 0.4382031559944153, "learning_rate": 8.286710093198791e-05, "loss": 0.0191, "step": 92430 }, { "epoch": 58.916507329509244, "grad_norm": 0.3596999943256378, "learning_rate": 8.286294725814635e-05, "loss": 0.0235, "step": 92440 }, { "epoch": 58.922880815806245, "grad_norm": 0.4521946310997009, "learning_rate": 8.285879318498893e-05, "loss": 0.0202, "step": 92450 }, { "epoch": 58.92925430210325, "grad_norm": 0.20184192061424255, "learning_rate": 8.285463871256618e-05, "loss": 0.0174, "step": 92460 }, { "epoch": 58.935627788400254, "grad_norm": 0.2752222418785095, "learning_rate": 8.285048384092853e-05, "loss": 0.0261, "step": 92470 }, { "epoch": 58.94200127469726, "grad_norm": 0.38707879185676575, "learning_rate": 8.28463285701265e-05, "loss": 0.0176, "step": 92480 }, { "epoch": 58.94837476099426, "grad_norm": 1.079452633857727, "learning_rate": 8.284217290021055e-05, "loss": 0.0302, "step": 92490 }, { "epoch": 58.95474824729127, "grad_norm": 0.48543980717658997, "learning_rate": 8.28380168312312e-05, "loss": 0.0213, "step": 92500 }, { "epoch": 58.96112173358827, "grad_norm": 0.4242631494998932, "learning_rate": 8.283386036323894e-05, "loss": 0.018, "step": 92510 }, { "epoch": 58.96749521988528, "grad_norm": 0.3445139229297638, "learning_rate": 8.282970349628427e-05, "loss": 0.0142, "step": 92520 }, { "epoch": 58.97386870618228, "grad_norm": 0.1800592988729477, "learning_rate": 8.282554623041773e-05, "loss": 0.0181, "step": 92530 }, { "epoch": 58.98024219247929, "grad_norm": 0.5218713879585266, "learning_rate": 8.282138856568978e-05, "loss": 0.0254, "step": 92540 }, { "epoch": 58.98661567877629, "grad_norm": 0.5554972887039185, "learning_rate": 8.2817230502151e-05, "loss": 0.0182, "step": 92550 }, { "epoch": 58.9929891650733, "grad_norm": 0.3502090275287628, "learning_rate": 8.281307203985186e-05, "loss": 0.0198, "step": 92560 }, { "epoch": 58.9993626513703, "grad_norm": 0.34401580691337585, "learning_rate": 8.280891317884291e-05, "loss": 0.0207, "step": 92570 }, { "epoch": 59.005736137667306, "grad_norm": 0.3515547215938568, "learning_rate": 8.28047539191747e-05, "loss": 0.0266, "step": 92580 }, { "epoch": 59.01210962396431, "grad_norm": 0.7415618896484375, "learning_rate": 8.280059426089777e-05, "loss": 0.0276, "step": 92590 }, { "epoch": 59.018483110261315, "grad_norm": 0.4392441511154175, "learning_rate": 8.279643420406265e-05, "loss": 0.0197, "step": 92600 }, { "epoch": 59.024856596558315, "grad_norm": 0.4125823676586151, "learning_rate": 8.279227374871988e-05, "loss": 0.0219, "step": 92610 }, { "epoch": 59.03123008285532, "grad_norm": 0.2168254852294922, "learning_rate": 8.278811289492001e-05, "loss": 0.0163, "step": 92620 }, { "epoch": 59.037603569152324, "grad_norm": 0.5477132201194763, "learning_rate": 8.278395164271362e-05, "loss": 0.0287, "step": 92630 }, { "epoch": 59.04397705544933, "grad_norm": 0.5849064588546753, "learning_rate": 8.277978999215127e-05, "loss": 0.0297, "step": 92640 }, { "epoch": 59.05035054174633, "grad_norm": 0.34799671173095703, "learning_rate": 8.277562794328352e-05, "loss": 0.0212, "step": 92650 }, { "epoch": 59.05672402804334, "grad_norm": 0.21220850944519043, "learning_rate": 8.277146549616094e-05, "loss": 0.0225, "step": 92660 }, { "epoch": 59.06309751434034, "grad_norm": 0.34040048718452454, "learning_rate": 8.276730265083412e-05, "loss": 0.0235, "step": 92670 }, { "epoch": 59.06947100063735, "grad_norm": 0.4934118688106537, "learning_rate": 8.276313940735363e-05, "loss": 0.0213, "step": 92680 }, { "epoch": 59.07584448693435, "grad_norm": 0.6165369749069214, "learning_rate": 8.275897576577007e-05, "loss": 0.0237, "step": 92690 }, { "epoch": 59.08221797323136, "grad_norm": 0.5506377220153809, "learning_rate": 8.275481172613402e-05, "loss": 0.0158, "step": 92700 }, { "epoch": 59.08859145952836, "grad_norm": 0.4537414312362671, "learning_rate": 8.275064728849608e-05, "loss": 0.0205, "step": 92710 }, { "epoch": 59.09496494582537, "grad_norm": 0.38948196172714233, "learning_rate": 8.274648245290686e-05, "loss": 0.0168, "step": 92720 }, { "epoch": 59.10133843212237, "grad_norm": 0.26395899057388306, "learning_rate": 8.274231721941695e-05, "loss": 0.0272, "step": 92730 }, { "epoch": 59.107711918419376, "grad_norm": 0.41868460178375244, "learning_rate": 8.273815158807699e-05, "loss": 0.0225, "step": 92740 }, { "epoch": 59.11408540471638, "grad_norm": 0.6008462905883789, "learning_rate": 8.273398555893756e-05, "loss": 0.0235, "step": 92750 }, { "epoch": 59.120458891013385, "grad_norm": 0.5641804337501526, "learning_rate": 8.272981913204931e-05, "loss": 0.0249, "step": 92760 }, { "epoch": 59.126832377310386, "grad_norm": 0.43208417296409607, "learning_rate": 8.272565230746284e-05, "loss": 0.0145, "step": 92770 }, { "epoch": 59.133205863607394, "grad_norm": 0.24256716668605804, "learning_rate": 8.272148508522882e-05, "loss": 0.0112, "step": 92780 }, { "epoch": 59.139579349904395, "grad_norm": 0.5895251035690308, "learning_rate": 8.271731746539784e-05, "loss": 0.0206, "step": 92790 }, { "epoch": 59.1459528362014, "grad_norm": 0.46897920966148376, "learning_rate": 8.271314944802057e-05, "loss": 0.0236, "step": 92800 }, { "epoch": 59.1523263224984, "grad_norm": 0.47891977429389954, "learning_rate": 8.270898103314764e-05, "loss": 0.0237, "step": 92810 }, { "epoch": 59.15869980879541, "grad_norm": 0.5131921768188477, "learning_rate": 8.270481222082974e-05, "loss": 0.0201, "step": 92820 }, { "epoch": 59.16507329509241, "grad_norm": 0.2485666424036026, "learning_rate": 8.270064301111746e-05, "loss": 0.0203, "step": 92830 }, { "epoch": 59.17144678138942, "grad_norm": 0.31107446551322937, "learning_rate": 8.26964734040615e-05, "loss": 0.0275, "step": 92840 }, { "epoch": 59.17782026768642, "grad_norm": 0.8494387865066528, "learning_rate": 8.269230339971253e-05, "loss": 0.0294, "step": 92850 }, { "epoch": 59.18419375398343, "grad_norm": 0.30487751960754395, "learning_rate": 8.268813299812119e-05, "loss": 0.0295, "step": 92860 }, { "epoch": 59.19056724028043, "grad_norm": 0.46762609481811523, "learning_rate": 8.268396219933819e-05, "loss": 0.0214, "step": 92870 }, { "epoch": 59.19694072657744, "grad_norm": 0.33592015504837036, "learning_rate": 8.267979100341418e-05, "loss": 0.0236, "step": 92880 }, { "epoch": 59.203314212874446, "grad_norm": 0.41084024310112, "learning_rate": 8.267561941039985e-05, "loss": 0.0275, "step": 92890 }, { "epoch": 59.20968769917145, "grad_norm": 0.5496814250946045, "learning_rate": 8.26714474203459e-05, "loss": 0.0249, "step": 92900 }, { "epoch": 59.216061185468455, "grad_norm": 0.4792845845222473, "learning_rate": 8.266727503330302e-05, "loss": 0.0409, "step": 92910 }, { "epoch": 59.222434671765456, "grad_norm": 0.23644034564495087, "learning_rate": 8.26631022493219e-05, "loss": 0.0327, "step": 92920 }, { "epoch": 59.22880815806246, "grad_norm": 0.5033069252967834, "learning_rate": 8.265892906845325e-05, "loss": 0.0254, "step": 92930 }, { "epoch": 59.235181644359464, "grad_norm": 0.4442286491394043, "learning_rate": 8.265475549074777e-05, "loss": 0.03, "step": 92940 }, { "epoch": 59.24155513065647, "grad_norm": 0.3033333420753479, "learning_rate": 8.265058151625617e-05, "loss": 0.0321, "step": 92950 }, { "epoch": 59.24792861695347, "grad_norm": 0.40133604407310486, "learning_rate": 8.26464071450292e-05, "loss": 0.014, "step": 92960 }, { "epoch": 59.25430210325048, "grad_norm": 0.37757736444473267, "learning_rate": 8.264223237711756e-05, "loss": 0.0168, "step": 92970 }, { "epoch": 59.26067558954748, "grad_norm": 0.35297250747680664, "learning_rate": 8.263805721257198e-05, "loss": 0.022, "step": 92980 }, { "epoch": 59.26704907584449, "grad_norm": 0.3134089708328247, "learning_rate": 8.263388165144317e-05, "loss": 0.0258, "step": 92990 }, { "epoch": 59.27342256214149, "grad_norm": 0.4771571755409241, "learning_rate": 8.26297056937819e-05, "loss": 0.0268, "step": 93000 }, { "epoch": 59.2797960484385, "grad_norm": 0.6517699956893921, "learning_rate": 8.26255293396389e-05, "loss": 0.0221, "step": 93010 }, { "epoch": 59.2861695347355, "grad_norm": 0.6683975458145142, "learning_rate": 8.26213525890649e-05, "loss": 0.0334, "step": 93020 }, { "epoch": 59.29254302103251, "grad_norm": 0.599269688129425, "learning_rate": 8.261717544211067e-05, "loss": 0.021, "step": 93030 }, { "epoch": 59.29891650732951, "grad_norm": 0.32951679825782776, "learning_rate": 8.261299789882697e-05, "loss": 0.0149, "step": 93040 }, { "epoch": 59.305289993626516, "grad_norm": 0.4724511206150055, "learning_rate": 8.260881995926455e-05, "loss": 0.0319, "step": 93050 }, { "epoch": 59.31166347992352, "grad_norm": 0.7388535737991333, "learning_rate": 8.260464162347418e-05, "loss": 0.0304, "step": 93060 }, { "epoch": 59.318036966220525, "grad_norm": 0.37618768215179443, "learning_rate": 8.260046289150662e-05, "loss": 0.0234, "step": 93070 }, { "epoch": 59.324410452517526, "grad_norm": 0.6019701361656189, "learning_rate": 8.259628376341269e-05, "loss": 0.0216, "step": 93080 }, { "epoch": 59.330783938814534, "grad_norm": 0.8116775751113892, "learning_rate": 8.25921042392431e-05, "loss": 0.0241, "step": 93090 }, { "epoch": 59.337157425111535, "grad_norm": 0.2904490530490875, "learning_rate": 8.25879243190487e-05, "loss": 0.017, "step": 93100 }, { "epoch": 59.34353091140854, "grad_norm": 0.298493891954422, "learning_rate": 8.258374400288022e-05, "loss": 0.0197, "step": 93110 }, { "epoch": 59.349904397705544, "grad_norm": 0.5321019887924194, "learning_rate": 8.257956329078851e-05, "loss": 0.0321, "step": 93120 }, { "epoch": 59.35627788400255, "grad_norm": 0.37851524353027344, "learning_rate": 8.257538218282433e-05, "loss": 0.0158, "step": 93130 }, { "epoch": 59.36265137029955, "grad_norm": 0.4570450484752655, "learning_rate": 8.257120067903852e-05, "loss": 0.0284, "step": 93140 }, { "epoch": 59.36902485659656, "grad_norm": 0.44990336894989014, "learning_rate": 8.256701877948185e-05, "loss": 0.0339, "step": 93150 }, { "epoch": 59.37539834289356, "grad_norm": 0.621115505695343, "learning_rate": 8.256283648420516e-05, "loss": 0.0234, "step": 93160 }, { "epoch": 59.38177182919057, "grad_norm": 0.5486621856689453, "learning_rate": 8.255865379325927e-05, "loss": 0.0177, "step": 93170 }, { "epoch": 59.38814531548757, "grad_norm": 0.3188325762748718, "learning_rate": 8.2554470706695e-05, "loss": 0.0238, "step": 93180 }, { "epoch": 59.39451880178458, "grad_norm": 0.4178983271121979, "learning_rate": 8.255028722456314e-05, "loss": 0.0197, "step": 93190 }, { "epoch": 59.40089228808158, "grad_norm": 0.3587867021560669, "learning_rate": 8.25461033469146e-05, "loss": 0.0194, "step": 93200 }, { "epoch": 59.40726577437859, "grad_norm": 0.35675036907196045, "learning_rate": 8.254191907380016e-05, "loss": 0.0261, "step": 93210 }, { "epoch": 59.41363926067559, "grad_norm": 0.49231988191604614, "learning_rate": 8.253773440527066e-05, "loss": 0.0198, "step": 93220 }, { "epoch": 59.420012746972596, "grad_norm": 0.6578166484832764, "learning_rate": 8.2533549341377e-05, "loss": 0.0267, "step": 93230 }, { "epoch": 59.4263862332696, "grad_norm": 0.5459911823272705, "learning_rate": 8.252936388216998e-05, "loss": 0.0156, "step": 93240 }, { "epoch": 59.432759719566604, "grad_norm": 0.5205587148666382, "learning_rate": 8.252517802770048e-05, "loss": 0.0268, "step": 93250 }, { "epoch": 59.439133205863605, "grad_norm": 0.1915075182914734, "learning_rate": 8.252099177801935e-05, "loss": 0.0154, "step": 93260 }, { "epoch": 59.44550669216061, "grad_norm": 0.2727292478084564, "learning_rate": 8.251680513317746e-05, "loss": 0.0233, "step": 93270 }, { "epoch": 59.451880178457614, "grad_norm": 0.3489912450313568, "learning_rate": 8.25126180932257e-05, "loss": 0.0237, "step": 93280 }, { "epoch": 59.45825366475462, "grad_norm": 0.7254497408866882, "learning_rate": 8.250843065821494e-05, "loss": 0.0192, "step": 93290 }, { "epoch": 59.46462715105162, "grad_norm": 0.4784906208515167, "learning_rate": 8.250424282819605e-05, "loss": 0.0166, "step": 93300 }, { "epoch": 59.47100063734863, "grad_norm": 0.4103676974773407, "learning_rate": 8.25000546032199e-05, "loss": 0.0212, "step": 93310 }, { "epoch": 59.47737412364563, "grad_norm": 0.46184688806533813, "learning_rate": 8.249586598333742e-05, "loss": 0.0295, "step": 93320 }, { "epoch": 59.48374760994264, "grad_norm": 0.28949835896492004, "learning_rate": 8.24916769685995e-05, "loss": 0.0208, "step": 93330 }, { "epoch": 59.49012109623964, "grad_norm": 0.3224061131477356, "learning_rate": 8.2487487559057e-05, "loss": 0.0206, "step": 93340 }, { "epoch": 59.49649458253665, "grad_norm": 0.24052144587039948, "learning_rate": 8.248329775476087e-05, "loss": 0.0213, "step": 93350 }, { "epoch": 59.50286806883365, "grad_norm": 0.45123717188835144, "learning_rate": 8.2479107555762e-05, "loss": 0.0211, "step": 93360 }, { "epoch": 59.50924155513066, "grad_norm": 0.2407354861497879, "learning_rate": 8.247491696211129e-05, "loss": 0.0134, "step": 93370 }, { "epoch": 59.51561504142766, "grad_norm": 0.5685559511184692, "learning_rate": 8.24707259738597e-05, "loss": 0.0304, "step": 93380 }, { "epoch": 59.521988527724666, "grad_norm": 0.5383599996566772, "learning_rate": 8.246653459105813e-05, "loss": 0.0156, "step": 93390 }, { "epoch": 59.52836201402167, "grad_norm": 0.6316070556640625, "learning_rate": 8.246234281375751e-05, "loss": 0.0185, "step": 93400 }, { "epoch": 59.534735500318675, "grad_norm": 0.5343974828720093, "learning_rate": 8.245815064200877e-05, "loss": 0.022, "step": 93410 }, { "epoch": 59.541108986615676, "grad_norm": 0.6939537525177002, "learning_rate": 8.245395807586284e-05, "loss": 0.0265, "step": 93420 }, { "epoch": 59.547482472912684, "grad_norm": 0.3878512978553772, "learning_rate": 8.24497651153707e-05, "loss": 0.0145, "step": 93430 }, { "epoch": 59.553855959209685, "grad_norm": 0.33297035098075867, "learning_rate": 8.244557176058327e-05, "loss": 0.0177, "step": 93440 }, { "epoch": 59.56022944550669, "grad_norm": 0.3212027847766876, "learning_rate": 8.244137801155151e-05, "loss": 0.0156, "step": 93450 }, { "epoch": 59.56660293180369, "grad_norm": 0.28421279788017273, "learning_rate": 8.243718386832638e-05, "loss": 0.0104, "step": 93460 }, { "epoch": 59.5729764181007, "grad_norm": 0.34876295924186707, "learning_rate": 8.243298933095883e-05, "loss": 0.0198, "step": 93470 }, { "epoch": 59.5793499043977, "grad_norm": 0.439009428024292, "learning_rate": 8.242879439949984e-05, "loss": 0.0204, "step": 93480 }, { "epoch": 59.58572339069471, "grad_norm": 0.37273940443992615, "learning_rate": 8.242459907400039e-05, "loss": 0.0174, "step": 93490 }, { "epoch": 59.59209687699171, "grad_norm": 0.5547595620155334, "learning_rate": 8.242040335451144e-05, "loss": 0.0185, "step": 93500 }, { "epoch": 59.59847036328872, "grad_norm": 0.5012189149856567, "learning_rate": 8.241620724108399e-05, "loss": 0.0236, "step": 93510 }, { "epoch": 59.60484384958572, "grad_norm": 0.24134746193885803, "learning_rate": 8.2412010733769e-05, "loss": 0.0225, "step": 93520 }, { "epoch": 59.61121733588273, "grad_norm": 0.27147793769836426, "learning_rate": 8.240781383261749e-05, "loss": 0.0161, "step": 93530 }, { "epoch": 59.617590822179736, "grad_norm": 0.3964003026485443, "learning_rate": 8.240361653768042e-05, "loss": 0.0201, "step": 93540 }, { "epoch": 59.62396430847674, "grad_norm": 0.4895612597465515, "learning_rate": 8.239941884900884e-05, "loss": 0.0275, "step": 93550 }, { "epoch": 59.630337794773745, "grad_norm": 0.6507510542869568, "learning_rate": 8.239522076665374e-05, "loss": 0.0217, "step": 93560 }, { "epoch": 59.636711281070745, "grad_norm": 0.3782055675983429, "learning_rate": 8.23910222906661e-05, "loss": 0.0147, "step": 93570 }, { "epoch": 59.64308476736775, "grad_norm": 0.5713593363761902, "learning_rate": 8.238682342109695e-05, "loss": 0.0446, "step": 93580 }, { "epoch": 59.649458253664754, "grad_norm": 0.27633485198020935, "learning_rate": 8.238262415799734e-05, "loss": 0.0302, "step": 93590 }, { "epoch": 59.65583173996176, "grad_norm": 0.42386385798454285, "learning_rate": 8.237842450141826e-05, "loss": 0.0146, "step": 93600 }, { "epoch": 59.66220522625876, "grad_norm": 0.42912089824676514, "learning_rate": 8.237422445141075e-05, "loss": 0.0241, "step": 93610 }, { "epoch": 59.66857871255577, "grad_norm": 0.42248407006263733, "learning_rate": 8.237002400802586e-05, "loss": 0.0156, "step": 93620 }, { "epoch": 59.67495219885277, "grad_norm": 0.7035057544708252, "learning_rate": 8.23658231713146e-05, "loss": 0.0319, "step": 93630 }, { "epoch": 59.68132568514978, "grad_norm": 0.3953835368156433, "learning_rate": 8.236162194132803e-05, "loss": 0.023, "step": 93640 }, { "epoch": 59.68769917144678, "grad_norm": 0.3582005798816681, "learning_rate": 8.235742031811722e-05, "loss": 0.0207, "step": 93650 }, { "epoch": 59.69407265774379, "grad_norm": 0.5828365683555603, "learning_rate": 8.235321830173318e-05, "loss": 0.0181, "step": 93660 }, { "epoch": 59.70044614404079, "grad_norm": 0.2936023771762848, "learning_rate": 8.234901589222701e-05, "loss": 0.0238, "step": 93670 }, { "epoch": 59.7068196303378, "grad_norm": 0.2770211398601532, "learning_rate": 8.234481308964974e-05, "loss": 0.028, "step": 93680 }, { "epoch": 59.7131931166348, "grad_norm": 0.4537712335586548, "learning_rate": 8.234060989405246e-05, "loss": 0.0154, "step": 93690 }, { "epoch": 59.719566602931806, "grad_norm": 0.32808563113212585, "learning_rate": 8.233640630548622e-05, "loss": 0.0164, "step": 93700 }, { "epoch": 59.72594008922881, "grad_norm": 0.2303311675786972, "learning_rate": 8.233220232400214e-05, "loss": 0.0232, "step": 93710 }, { "epoch": 59.732313575525815, "grad_norm": 0.7458415031433105, "learning_rate": 8.232799794965125e-05, "loss": 0.0251, "step": 93720 }, { "epoch": 59.738687061822816, "grad_norm": 0.5359320640563965, "learning_rate": 8.232379318248466e-05, "loss": 0.0189, "step": 93730 }, { "epoch": 59.745060548119824, "grad_norm": 0.6634984016418457, "learning_rate": 8.231958802255348e-05, "loss": 0.0158, "step": 93740 }, { "epoch": 59.751434034416825, "grad_norm": 0.3584100306034088, "learning_rate": 8.23153824699088e-05, "loss": 0.0161, "step": 93750 }, { "epoch": 59.75780752071383, "grad_norm": 0.3296360373497009, "learning_rate": 8.231117652460169e-05, "loss": 0.0231, "step": 93760 }, { "epoch": 59.76418100701083, "grad_norm": 0.26982828974723816, "learning_rate": 8.230697018668329e-05, "loss": 0.0299, "step": 93770 }, { "epoch": 59.77055449330784, "grad_norm": 0.7162748575210571, "learning_rate": 8.23027634562047e-05, "loss": 0.0246, "step": 93780 }, { "epoch": 59.77692797960484, "grad_norm": 0.81397545337677, "learning_rate": 8.229855633321702e-05, "loss": 0.0352, "step": 93790 }, { "epoch": 59.78330146590185, "grad_norm": 0.6027111411094666, "learning_rate": 8.22943488177714e-05, "loss": 0.0218, "step": 93800 }, { "epoch": 59.78967495219885, "grad_norm": 0.3139319121837616, "learning_rate": 8.229014090991896e-05, "loss": 0.0207, "step": 93810 }, { "epoch": 59.79604843849586, "grad_norm": 0.43289288878440857, "learning_rate": 8.22859326097108e-05, "loss": 0.02, "step": 93820 }, { "epoch": 59.80242192479286, "grad_norm": 0.6288232207298279, "learning_rate": 8.228172391719809e-05, "loss": 0.0169, "step": 93830 }, { "epoch": 59.80879541108987, "grad_norm": 0.24853788316249847, "learning_rate": 8.227751483243195e-05, "loss": 0.0247, "step": 93840 }, { "epoch": 59.81516889738687, "grad_norm": 0.7029234766960144, "learning_rate": 8.227330535546353e-05, "loss": 0.0209, "step": 93850 }, { "epoch": 59.82154238368388, "grad_norm": 0.4341537058353424, "learning_rate": 8.226909548634398e-05, "loss": 0.0175, "step": 93860 }, { "epoch": 59.82791586998088, "grad_norm": 0.51505446434021, "learning_rate": 8.226488522512446e-05, "loss": 0.0181, "step": 93870 }, { "epoch": 59.834289356277885, "grad_norm": 0.2237633466720581, "learning_rate": 8.226067457185611e-05, "loss": 0.0193, "step": 93880 }, { "epoch": 59.840662842574886, "grad_norm": 0.5321182012557983, "learning_rate": 8.225646352659011e-05, "loss": 0.0201, "step": 93890 }, { "epoch": 59.847036328871894, "grad_norm": 0.446595162153244, "learning_rate": 8.225225208937761e-05, "loss": 0.0255, "step": 93900 }, { "epoch": 59.853409815168895, "grad_norm": 1.4309738874435425, "learning_rate": 8.224804026026981e-05, "loss": 0.0312, "step": 93910 }, { "epoch": 59.8597833014659, "grad_norm": 0.28338631987571716, "learning_rate": 8.224382803931788e-05, "loss": 0.0174, "step": 93920 }, { "epoch": 59.866156787762904, "grad_norm": 0.46626850962638855, "learning_rate": 8.223961542657297e-05, "loss": 0.0216, "step": 93930 }, { "epoch": 59.87253027405991, "grad_norm": 0.23753514885902405, "learning_rate": 8.223540242208631e-05, "loss": 0.0303, "step": 93940 }, { "epoch": 59.87890376035691, "grad_norm": 0.5911844372749329, "learning_rate": 8.223118902590908e-05, "loss": 0.022, "step": 93950 }, { "epoch": 59.88527724665392, "grad_norm": 0.6595186591148376, "learning_rate": 8.222697523809248e-05, "loss": 0.022, "step": 93960 }, { "epoch": 59.89165073295092, "grad_norm": 0.5199986696243286, "learning_rate": 8.22227610586877e-05, "loss": 0.0182, "step": 93970 }, { "epoch": 59.89802421924793, "grad_norm": 0.3174617290496826, "learning_rate": 8.221854648774592e-05, "loss": 0.0238, "step": 93980 }, { "epoch": 59.90439770554493, "grad_norm": 0.48729005455970764, "learning_rate": 8.221433152531841e-05, "loss": 0.0247, "step": 93990 }, { "epoch": 59.91077119184194, "grad_norm": 0.5597407817840576, "learning_rate": 8.221011617145635e-05, "loss": 0.0201, "step": 94000 }, { "epoch": 59.91714467813894, "grad_norm": 0.31119054555892944, "learning_rate": 8.220590042621097e-05, "loss": 0.0264, "step": 94010 }, { "epoch": 59.92351816443595, "grad_norm": 0.33432674407958984, "learning_rate": 8.220168428963348e-05, "loss": 0.0239, "step": 94020 }, { "epoch": 59.92989165073295, "grad_norm": 0.4152015447616577, "learning_rate": 8.219746776177513e-05, "loss": 0.0137, "step": 94030 }, { "epoch": 59.936265137029956, "grad_norm": 0.6544189453125, "learning_rate": 8.219325084268715e-05, "loss": 0.0247, "step": 94040 }, { "epoch": 59.94263862332696, "grad_norm": 0.4218325912952423, "learning_rate": 8.218903353242077e-05, "loss": 0.0158, "step": 94050 }, { "epoch": 59.949012109623965, "grad_norm": 0.4683205485343933, "learning_rate": 8.218481583102726e-05, "loss": 0.0165, "step": 94060 }, { "epoch": 59.955385595920966, "grad_norm": 0.4502445161342621, "learning_rate": 8.218059773855782e-05, "loss": 0.0192, "step": 94070 }, { "epoch": 59.961759082217974, "grad_norm": 0.4613311290740967, "learning_rate": 8.217637925506376e-05, "loss": 0.0283, "step": 94080 }, { "epoch": 59.968132568514974, "grad_norm": 0.39782169461250305, "learning_rate": 8.21721603805963e-05, "loss": 0.0239, "step": 94090 }, { "epoch": 59.97450605481198, "grad_norm": 0.574532151222229, "learning_rate": 8.216794111520671e-05, "loss": 0.0304, "step": 94100 }, { "epoch": 59.98087954110898, "grad_norm": 0.37155768275260925, "learning_rate": 8.216372145894627e-05, "loss": 0.0212, "step": 94110 }, { "epoch": 59.98725302740599, "grad_norm": 0.5361500382423401, "learning_rate": 8.215950141186626e-05, "loss": 0.0178, "step": 94120 }, { "epoch": 59.99362651370299, "grad_norm": 0.4008428752422333, "learning_rate": 8.215528097401793e-05, "loss": 0.0172, "step": 94130 }, { "epoch": 60.0, "grad_norm": 0.740408182144165, "learning_rate": 8.215106014545258e-05, "loss": 0.0118, "step": 94140 }, { "epoch": 60.00637348629701, "grad_norm": 0.36434316635131836, "learning_rate": 8.21468389262215e-05, "loss": 0.0196, "step": 94150 }, { "epoch": 60.01274697259401, "grad_norm": 0.2639552652835846, "learning_rate": 8.214261731637596e-05, "loss": 0.0246, "step": 94160 }, { "epoch": 60.01912045889102, "grad_norm": 0.33472633361816406, "learning_rate": 8.213839531596728e-05, "loss": 0.0197, "step": 94170 }, { "epoch": 60.02549394518802, "grad_norm": 0.6816985607147217, "learning_rate": 8.213417292504674e-05, "loss": 0.0255, "step": 94180 }, { "epoch": 60.031867431485026, "grad_norm": 0.5622140169143677, "learning_rate": 8.212995014366569e-05, "loss": 0.02, "step": 94190 }, { "epoch": 60.038240917782026, "grad_norm": 0.30110636353492737, "learning_rate": 8.212572697187539e-05, "loss": 0.0179, "step": 94200 }, { "epoch": 60.044614404079034, "grad_norm": 0.471817284822464, "learning_rate": 8.212150340972717e-05, "loss": 0.0159, "step": 94210 }, { "epoch": 60.050987890376035, "grad_norm": 0.5833938121795654, "learning_rate": 8.211727945727237e-05, "loss": 0.0132, "step": 94220 }, { "epoch": 60.05736137667304, "grad_norm": 0.7653937339782715, "learning_rate": 8.211305511456229e-05, "loss": 0.0212, "step": 94230 }, { "epoch": 60.063734862970044, "grad_norm": 0.3829895555973053, "learning_rate": 8.210883038164828e-05, "loss": 0.0199, "step": 94240 }, { "epoch": 60.07010834926705, "grad_norm": 0.21789944171905518, "learning_rate": 8.210460525858166e-05, "loss": 0.0219, "step": 94250 }, { "epoch": 60.07648183556405, "grad_norm": 0.47638124227523804, "learning_rate": 8.210037974541377e-05, "loss": 0.0187, "step": 94260 }, { "epoch": 60.08285532186106, "grad_norm": 0.469731867313385, "learning_rate": 8.209615384219596e-05, "loss": 0.023, "step": 94270 }, { "epoch": 60.08922880815806, "grad_norm": 0.6022607088088989, "learning_rate": 8.209192754897958e-05, "loss": 0.017, "step": 94280 }, { "epoch": 60.09560229445507, "grad_norm": 0.1777062714099884, "learning_rate": 8.2087700865816e-05, "loss": 0.0122, "step": 94290 }, { "epoch": 60.10197578075207, "grad_norm": 0.2868332862854004, "learning_rate": 8.208347379275654e-05, "loss": 0.0154, "step": 94300 }, { "epoch": 60.10834926704908, "grad_norm": 0.5057234168052673, "learning_rate": 8.207924632985258e-05, "loss": 0.0281, "step": 94310 }, { "epoch": 60.11472275334608, "grad_norm": 0.406100332736969, "learning_rate": 8.207501847715549e-05, "loss": 0.017, "step": 94320 }, { "epoch": 60.12109623964309, "grad_norm": 0.33227604627609253, "learning_rate": 8.207079023471664e-05, "loss": 0.0136, "step": 94330 }, { "epoch": 60.12746972594009, "grad_norm": 0.5410879254341125, "learning_rate": 8.206656160258742e-05, "loss": 0.0251, "step": 94340 }, { "epoch": 60.133843212237096, "grad_norm": 0.4204516112804413, "learning_rate": 8.206233258081918e-05, "loss": 0.0176, "step": 94350 }, { "epoch": 60.1402166985341, "grad_norm": 0.3921091556549072, "learning_rate": 8.205810316946334e-05, "loss": 0.0155, "step": 94360 }, { "epoch": 60.146590184831105, "grad_norm": 0.49312520027160645, "learning_rate": 8.205387336857128e-05, "loss": 0.0218, "step": 94370 }, { "epoch": 60.152963671128106, "grad_norm": 0.7163779735565186, "learning_rate": 8.204964317819439e-05, "loss": 0.0276, "step": 94380 }, { "epoch": 60.159337157425114, "grad_norm": 0.6908515095710754, "learning_rate": 8.204541259838408e-05, "loss": 0.0254, "step": 94390 }, { "epoch": 60.165710643722115, "grad_norm": 0.27434495091438293, "learning_rate": 8.204118162919174e-05, "loss": 0.0177, "step": 94400 }, { "epoch": 60.17208413001912, "grad_norm": 0.5171014666557312, "learning_rate": 8.20369502706688e-05, "loss": 0.0194, "step": 94410 }, { "epoch": 60.17845761631612, "grad_norm": 0.4232417047023773, "learning_rate": 8.203271852286667e-05, "loss": 0.0189, "step": 94420 }, { "epoch": 60.18483110261313, "grad_norm": 0.37033259868621826, "learning_rate": 8.202848638583675e-05, "loss": 0.0171, "step": 94430 }, { "epoch": 60.19120458891013, "grad_norm": 0.3788917660713196, "learning_rate": 8.20242538596305e-05, "loss": 0.0151, "step": 94440 }, { "epoch": 60.19757807520714, "grad_norm": 0.5438482165336609, "learning_rate": 8.202002094429933e-05, "loss": 0.0212, "step": 94450 }, { "epoch": 60.20395156150414, "grad_norm": 0.5093549489974976, "learning_rate": 8.201578763989466e-05, "loss": 0.0238, "step": 94460 }, { "epoch": 60.21032504780115, "grad_norm": 0.15039405226707458, "learning_rate": 8.201155394646796e-05, "loss": 0.0188, "step": 94470 }, { "epoch": 60.21669853409815, "grad_norm": 0.2884507477283478, "learning_rate": 8.200731986407064e-05, "loss": 0.0177, "step": 94480 }, { "epoch": 60.22307202039516, "grad_norm": 0.3501623570919037, "learning_rate": 8.200308539275417e-05, "loss": 0.0257, "step": 94490 }, { "epoch": 60.22944550669216, "grad_norm": 0.3756864666938782, "learning_rate": 8.199885053256999e-05, "loss": 0.0157, "step": 94500 }, { "epoch": 60.23581899298917, "grad_norm": 0.46834036707878113, "learning_rate": 8.199461528356958e-05, "loss": 0.0312, "step": 94510 }, { "epoch": 60.24219247928617, "grad_norm": 0.600739598274231, "learning_rate": 8.199037964580436e-05, "loss": 0.0217, "step": 94520 }, { "epoch": 60.248565965583175, "grad_norm": 0.4493928849697113, "learning_rate": 8.198614361932583e-05, "loss": 0.0144, "step": 94530 }, { "epoch": 60.254939451880176, "grad_norm": 0.38744163513183594, "learning_rate": 8.198190720418549e-05, "loss": 0.0223, "step": 94540 }, { "epoch": 60.261312938177184, "grad_norm": 0.47216832637786865, "learning_rate": 8.197767040043474e-05, "loss": 0.0169, "step": 94550 }, { "epoch": 60.267686424474185, "grad_norm": 0.538172721862793, "learning_rate": 8.197343320812512e-05, "loss": 0.0226, "step": 94560 }, { "epoch": 60.27405991077119, "grad_norm": 0.35447004437446594, "learning_rate": 8.19691956273081e-05, "loss": 0.0337, "step": 94570 }, { "epoch": 60.280433397068194, "grad_norm": 0.9600151777267456, "learning_rate": 8.196495765803515e-05, "loss": 0.034, "step": 94580 }, { "epoch": 60.2868068833652, "grad_norm": 0.297158420085907, "learning_rate": 8.19607193003578e-05, "loss": 0.0268, "step": 94590 }, { "epoch": 60.2931803696622, "grad_norm": 0.3200402557849884, "learning_rate": 8.195648055432755e-05, "loss": 0.0139, "step": 94600 }, { "epoch": 60.29955385595921, "grad_norm": 0.42138639092445374, "learning_rate": 8.195224141999587e-05, "loss": 0.0271, "step": 94610 }, { "epoch": 60.30592734225621, "grad_norm": 0.6901991963386536, "learning_rate": 8.19480018974143e-05, "loss": 0.0278, "step": 94620 }, { "epoch": 60.31230082855322, "grad_norm": 0.7307824492454529, "learning_rate": 8.194376198663434e-05, "loss": 0.0184, "step": 94630 }, { "epoch": 60.31867431485022, "grad_norm": 0.4220949411392212, "learning_rate": 8.19395216877075e-05, "loss": 0.0305, "step": 94640 }, { "epoch": 60.32504780114723, "grad_norm": 0.3536379933357239, "learning_rate": 8.193528100068532e-05, "loss": 0.0228, "step": 94650 }, { "epoch": 60.33142128744423, "grad_norm": 0.42400944232940674, "learning_rate": 8.193103992561933e-05, "loss": 0.0189, "step": 94660 }, { "epoch": 60.33779477374124, "grad_norm": 0.32926443219184875, "learning_rate": 8.192679846256106e-05, "loss": 0.0217, "step": 94670 }, { "epoch": 60.34416826003824, "grad_norm": 0.5759711861610413, "learning_rate": 8.192255661156202e-05, "loss": 0.026, "step": 94680 }, { "epoch": 60.350541746335246, "grad_norm": 0.679394006729126, "learning_rate": 8.19183143726738e-05, "loss": 0.0265, "step": 94690 }, { "epoch": 60.35691523263225, "grad_norm": 0.2350999414920807, "learning_rate": 8.191407174594794e-05, "loss": 0.0235, "step": 94700 }, { "epoch": 60.363288718929255, "grad_norm": 0.5690944790840149, "learning_rate": 8.190982873143594e-05, "loss": 0.0142, "step": 94710 }, { "epoch": 60.369662205226255, "grad_norm": 0.3832121193408966, "learning_rate": 8.190558532918941e-05, "loss": 0.023, "step": 94720 }, { "epoch": 60.37603569152326, "grad_norm": 0.8401877284049988, "learning_rate": 8.19013415392599e-05, "loss": 0.0252, "step": 94730 }, { "epoch": 60.382409177820264, "grad_norm": 0.7321261763572693, "learning_rate": 8.189709736169898e-05, "loss": 0.0229, "step": 94740 }, { "epoch": 60.38878266411727, "grad_norm": 0.38541319966316223, "learning_rate": 8.189285279655818e-05, "loss": 0.0292, "step": 94750 }, { "epoch": 60.39515615041427, "grad_norm": 0.3211003243923187, "learning_rate": 8.188860784388913e-05, "loss": 0.0222, "step": 94760 }, { "epoch": 60.40152963671128, "grad_norm": 0.41377875208854675, "learning_rate": 8.188436250374339e-05, "loss": 0.0205, "step": 94770 }, { "epoch": 60.40790312300829, "grad_norm": 0.6658848524093628, "learning_rate": 8.188011677617251e-05, "loss": 0.0272, "step": 94780 }, { "epoch": 60.41427660930529, "grad_norm": 0.46615052223205566, "learning_rate": 8.187587066122814e-05, "loss": 0.0324, "step": 94790 }, { "epoch": 60.4206500956023, "grad_norm": 0.6167742609977722, "learning_rate": 8.187162415896184e-05, "loss": 0.0189, "step": 94800 }, { "epoch": 60.4270235818993, "grad_norm": 0.2914963662624359, "learning_rate": 8.186737726942522e-05, "loss": 0.025, "step": 94810 }, { "epoch": 60.43339706819631, "grad_norm": 0.5335922837257385, "learning_rate": 8.186312999266987e-05, "loss": 0.0235, "step": 94820 }, { "epoch": 60.43977055449331, "grad_norm": 0.33137714862823486, "learning_rate": 8.18588823287474e-05, "loss": 0.0152, "step": 94830 }, { "epoch": 60.446144040790315, "grad_norm": 0.35201752185821533, "learning_rate": 8.185463427770944e-05, "loss": 0.0193, "step": 94840 }, { "epoch": 60.452517527087316, "grad_norm": 0.3196587562561035, "learning_rate": 8.18503858396076e-05, "loss": 0.0211, "step": 94850 }, { "epoch": 60.458891013384324, "grad_norm": 0.45294442772865295, "learning_rate": 8.184613701449349e-05, "loss": 0.0178, "step": 94860 }, { "epoch": 60.465264499681325, "grad_norm": 0.38281112909317017, "learning_rate": 8.184188780241876e-05, "loss": 0.0245, "step": 94870 }, { "epoch": 60.47163798597833, "grad_norm": 0.5284104943275452, "learning_rate": 8.183763820343501e-05, "loss": 0.0225, "step": 94880 }, { "epoch": 60.478011472275334, "grad_norm": 0.21307097375392914, "learning_rate": 8.183338821759392e-05, "loss": 0.03, "step": 94890 }, { "epoch": 60.48438495857234, "grad_norm": 0.3982771039009094, "learning_rate": 8.18291378449471e-05, "loss": 0.0224, "step": 94900 }, { "epoch": 60.49075844486934, "grad_norm": 0.26546910405158997, "learning_rate": 8.18248870855462e-05, "loss": 0.0385, "step": 94910 }, { "epoch": 60.49713193116635, "grad_norm": 0.44611799716949463, "learning_rate": 8.182063593944287e-05, "loss": 0.0232, "step": 94920 }, { "epoch": 60.50350541746335, "grad_norm": 0.3830655515193939, "learning_rate": 8.181638440668879e-05, "loss": 0.0224, "step": 94930 }, { "epoch": 60.50987890376036, "grad_norm": 0.31376492977142334, "learning_rate": 8.181213248733556e-05, "loss": 0.0161, "step": 94940 }, { "epoch": 60.51625239005736, "grad_norm": 0.3416409194469452, "learning_rate": 8.180788018143492e-05, "loss": 0.0198, "step": 94950 }, { "epoch": 60.52262587635437, "grad_norm": 0.30884602665901184, "learning_rate": 8.180362748903849e-05, "loss": 0.0134, "step": 94960 }, { "epoch": 60.52899936265137, "grad_norm": 0.3671185076236725, "learning_rate": 8.179937441019798e-05, "loss": 0.0161, "step": 94970 }, { "epoch": 60.53537284894838, "grad_norm": 0.3416841924190521, "learning_rate": 8.179512094496502e-05, "loss": 0.0193, "step": 94980 }, { "epoch": 60.54174633524538, "grad_norm": 0.4086098074913025, "learning_rate": 8.179086709339132e-05, "loss": 0.0108, "step": 94990 }, { "epoch": 60.548119821542386, "grad_norm": 0.5163726210594177, "learning_rate": 8.178661285552858e-05, "loss": 0.0162, "step": 95000 }, { "epoch": 60.55449330783939, "grad_norm": 0.7988197803497314, "learning_rate": 8.178235823142849e-05, "loss": 0.0258, "step": 95010 }, { "epoch": 60.560866794136395, "grad_norm": 0.5301536917686462, "learning_rate": 8.177810322114274e-05, "loss": 0.0298, "step": 95020 }, { "epoch": 60.567240280433396, "grad_norm": 0.4332931339740753, "learning_rate": 8.177384782472302e-05, "loss": 0.0248, "step": 95030 }, { "epoch": 60.5736137667304, "grad_norm": 0.3519963026046753, "learning_rate": 8.176959204222106e-05, "loss": 0.0177, "step": 95040 }, { "epoch": 60.579987253027404, "grad_norm": 0.4729333817958832, "learning_rate": 8.176533587368854e-05, "loss": 0.0268, "step": 95050 }, { "epoch": 60.58636073932441, "grad_norm": 0.27976396679878235, "learning_rate": 8.176107931917721e-05, "loss": 0.0241, "step": 95060 }, { "epoch": 60.59273422562141, "grad_norm": 0.2705378234386444, "learning_rate": 8.175682237873878e-05, "loss": 0.0124, "step": 95070 }, { "epoch": 60.59910771191842, "grad_norm": 0.4462827444076538, "learning_rate": 8.175256505242497e-05, "loss": 0.0239, "step": 95080 }, { "epoch": 60.60548119821542, "grad_norm": 0.3681188225746155, "learning_rate": 8.174830734028753e-05, "loss": 0.0267, "step": 95090 }, { "epoch": 60.61185468451243, "grad_norm": 0.4259365200996399, "learning_rate": 8.174404924237817e-05, "loss": 0.0162, "step": 95100 }, { "epoch": 60.61822817080943, "grad_norm": 0.30780887603759766, "learning_rate": 8.173979075874864e-05, "loss": 0.0214, "step": 95110 }, { "epoch": 60.62460165710644, "grad_norm": 1.293228268623352, "learning_rate": 8.173553188945069e-05, "loss": 0.016, "step": 95120 }, { "epoch": 60.63097514340344, "grad_norm": 0.38159847259521484, "learning_rate": 8.173127263453606e-05, "loss": 0.0331, "step": 95130 }, { "epoch": 60.63734862970045, "grad_norm": 0.4378938376903534, "learning_rate": 8.172701299405651e-05, "loss": 0.0287, "step": 95140 }, { "epoch": 60.64372211599745, "grad_norm": 0.3534691631793976, "learning_rate": 8.17227529680638e-05, "loss": 0.0193, "step": 95150 }, { "epoch": 60.650095602294456, "grad_norm": 0.5505616068840027, "learning_rate": 8.171849255660968e-05, "loss": 0.0343, "step": 95160 }, { "epoch": 60.65646908859146, "grad_norm": 0.1939314603805542, "learning_rate": 8.171423175974594e-05, "loss": 0.026, "step": 95170 }, { "epoch": 60.662842574888465, "grad_norm": 0.2661811411380768, "learning_rate": 8.170997057752434e-05, "loss": 0.0196, "step": 95180 }, { "epoch": 60.669216061185466, "grad_norm": 0.5986185669898987, "learning_rate": 8.170570900999665e-05, "loss": 0.0191, "step": 95190 }, { "epoch": 60.675589547482474, "grad_norm": 0.5243088006973267, "learning_rate": 8.170144705721465e-05, "loss": 0.0256, "step": 95200 }, { "epoch": 60.681963033779475, "grad_norm": 0.5933504104614258, "learning_rate": 8.169718471923016e-05, "loss": 0.0178, "step": 95210 }, { "epoch": 60.68833652007648, "grad_norm": 0.27748236060142517, "learning_rate": 8.169292199609494e-05, "loss": 0.0196, "step": 95220 }, { "epoch": 60.694710006373484, "grad_norm": 0.38703417778015137, "learning_rate": 8.16886588878608e-05, "loss": 0.0271, "step": 95230 }, { "epoch": 60.70108349267049, "grad_norm": 0.28786715865135193, "learning_rate": 8.168439539457953e-05, "loss": 0.0092, "step": 95240 }, { "epoch": 60.70745697896749, "grad_norm": 0.28245866298675537, "learning_rate": 8.168013151630294e-05, "loss": 0.0208, "step": 95250 }, { "epoch": 60.7138304652645, "grad_norm": 0.44695496559143066, "learning_rate": 8.167586725308283e-05, "loss": 0.0216, "step": 95260 }, { "epoch": 60.7202039515615, "grad_norm": 0.2618371844291687, "learning_rate": 8.167160260497105e-05, "loss": 0.0264, "step": 95270 }, { "epoch": 60.72657743785851, "grad_norm": 0.4371306598186493, "learning_rate": 8.166733757201937e-05, "loss": 0.0216, "step": 95280 }, { "epoch": 60.73295092415551, "grad_norm": 0.2725467383861542, "learning_rate": 8.166307215427967e-05, "loss": 0.0174, "step": 95290 }, { "epoch": 60.73932441045252, "grad_norm": 0.4796711802482605, "learning_rate": 8.165880635180372e-05, "loss": 0.0188, "step": 95300 }, { "epoch": 60.74569789674952, "grad_norm": 0.45764002203941345, "learning_rate": 8.16545401646434e-05, "loss": 0.0207, "step": 95310 }, { "epoch": 60.75207138304653, "grad_norm": 0.49266281723976135, "learning_rate": 8.165027359285052e-05, "loss": 0.0243, "step": 95320 }, { "epoch": 60.75844486934353, "grad_norm": 0.25585174560546875, "learning_rate": 8.164600663647693e-05, "loss": 0.0276, "step": 95330 }, { "epoch": 60.764818355640536, "grad_norm": 0.2526433765888214, "learning_rate": 8.164173929557449e-05, "loss": 0.0217, "step": 95340 }, { "epoch": 60.77119184193754, "grad_norm": 0.275379478931427, "learning_rate": 8.163747157019503e-05, "loss": 0.0248, "step": 95350 }, { "epoch": 60.777565328234544, "grad_norm": 0.39631468057632446, "learning_rate": 8.163320346039044e-05, "loss": 0.0214, "step": 95360 }, { "epoch": 60.783938814531545, "grad_norm": 0.31717073917388916, "learning_rate": 8.162893496621255e-05, "loss": 0.0138, "step": 95370 }, { "epoch": 60.79031230082855, "grad_norm": 0.23715782165527344, "learning_rate": 8.162466608771324e-05, "loss": 0.0127, "step": 95380 }, { "epoch": 60.796685787125554, "grad_norm": 0.42857521772384644, "learning_rate": 8.162039682494437e-05, "loss": 0.0321, "step": 95390 }, { "epoch": 60.80305927342256, "grad_norm": 0.6073705554008484, "learning_rate": 8.161612717795783e-05, "loss": 0.033, "step": 95400 }, { "epoch": 60.80943275971957, "grad_norm": 0.3969240188598633, "learning_rate": 8.161185714680551e-05, "loss": 0.0154, "step": 95410 }, { "epoch": 60.81580624601657, "grad_norm": 0.544790506362915, "learning_rate": 8.160758673153926e-05, "loss": 0.0267, "step": 95420 }, { "epoch": 60.82217973231358, "grad_norm": 0.622099757194519, "learning_rate": 8.160331593221099e-05, "loss": 0.0199, "step": 95430 }, { "epoch": 60.82855321861058, "grad_norm": 0.9002111554145813, "learning_rate": 8.159904474887259e-05, "loss": 0.0263, "step": 95440 }, { "epoch": 60.83492670490759, "grad_norm": 0.2974507808685303, "learning_rate": 8.159477318157598e-05, "loss": 0.0199, "step": 95450 }, { "epoch": 60.84130019120459, "grad_norm": 0.5293838381767273, "learning_rate": 8.159050123037301e-05, "loss": 0.0232, "step": 95460 }, { "epoch": 60.8476736775016, "grad_norm": 0.42628180980682373, "learning_rate": 8.158622889531567e-05, "loss": 0.0199, "step": 95470 }, { "epoch": 60.8540471637986, "grad_norm": 0.40943604707717896, "learning_rate": 8.158195617645579e-05, "loss": 0.0205, "step": 95480 }, { "epoch": 60.860420650095605, "grad_norm": 0.23170246183872223, "learning_rate": 8.157768307384535e-05, "loss": 0.0189, "step": 95490 }, { "epoch": 60.866794136392606, "grad_norm": 0.6541046500205994, "learning_rate": 8.157340958753624e-05, "loss": 0.0248, "step": 95500 }, { "epoch": 60.873167622689614, "grad_norm": 0.2681419253349304, "learning_rate": 8.156913571758037e-05, "loss": 0.027, "step": 95510 }, { "epoch": 60.879541108986615, "grad_norm": 0.6248667240142822, "learning_rate": 8.156486146402972e-05, "loss": 0.0228, "step": 95520 }, { "epoch": 60.88591459528362, "grad_norm": 0.4736101031303406, "learning_rate": 8.156058682693619e-05, "loss": 0.0173, "step": 95530 }, { "epoch": 60.892288081580624, "grad_norm": 0.58992999792099, "learning_rate": 8.155631180635172e-05, "loss": 0.0377, "step": 95540 }, { "epoch": 60.89866156787763, "grad_norm": 0.40210843086242676, "learning_rate": 8.155203640232829e-05, "loss": 0.0219, "step": 95550 }, { "epoch": 60.90503505417463, "grad_norm": 0.3092411160469055, "learning_rate": 8.154776061491782e-05, "loss": 0.0268, "step": 95560 }, { "epoch": 60.91140854047164, "grad_norm": 0.5102787613868713, "learning_rate": 8.154348444417227e-05, "loss": 0.0146, "step": 95570 }, { "epoch": 60.91778202676864, "grad_norm": 0.3227364122867584, "learning_rate": 8.15392078901436e-05, "loss": 0.0173, "step": 95580 }, { "epoch": 60.92415551306565, "grad_norm": 0.8906571865081787, "learning_rate": 8.153493095288379e-05, "loss": 0.0195, "step": 95590 }, { "epoch": 60.93052899936265, "grad_norm": 0.659977912902832, "learning_rate": 8.153065363244479e-05, "loss": 0.0151, "step": 95600 }, { "epoch": 60.93690248565966, "grad_norm": 0.4790312349796295, "learning_rate": 8.152637592887857e-05, "loss": 0.0181, "step": 95610 }, { "epoch": 60.94327597195666, "grad_norm": 0.5095083713531494, "learning_rate": 8.152209784223712e-05, "loss": 0.0203, "step": 95620 }, { "epoch": 60.94964945825367, "grad_norm": 0.29002413153648376, "learning_rate": 8.151781937257241e-05, "loss": 0.0219, "step": 95630 }, { "epoch": 60.95602294455067, "grad_norm": 0.5270630717277527, "learning_rate": 8.151354051993644e-05, "loss": 0.0148, "step": 95640 }, { "epoch": 60.962396430847676, "grad_norm": 0.9076797962188721, "learning_rate": 8.150926128438122e-05, "loss": 0.0198, "step": 95650 }, { "epoch": 60.96876991714468, "grad_norm": 0.3433191776275635, "learning_rate": 8.150498166595871e-05, "loss": 0.019, "step": 95660 }, { "epoch": 60.975143403441685, "grad_norm": 0.4772210717201233, "learning_rate": 8.150070166472093e-05, "loss": 0.0321, "step": 95670 }, { "epoch": 60.981516889738685, "grad_norm": 0.4406152367591858, "learning_rate": 8.149642128071988e-05, "loss": 0.0209, "step": 95680 }, { "epoch": 60.98789037603569, "grad_norm": 0.4365846514701843, "learning_rate": 8.149214051400757e-05, "loss": 0.0246, "step": 95690 }, { "epoch": 60.994263862332694, "grad_norm": 0.4946298897266388, "learning_rate": 8.148785936463603e-05, "loss": 0.0443, "step": 95700 }, { "epoch": 61.0006373486297, "grad_norm": 0.39964190125465393, "learning_rate": 8.148357783265727e-05, "loss": 0.0154, "step": 95710 }, { "epoch": 61.0070108349267, "grad_norm": 0.3821636140346527, "learning_rate": 8.147929591812331e-05, "loss": 0.0276, "step": 95720 }, { "epoch": 61.01338432122371, "grad_norm": 0.4688991904258728, "learning_rate": 8.147501362108617e-05, "loss": 0.0181, "step": 95730 }, { "epoch": 61.01975780752071, "grad_norm": 0.5446555018424988, "learning_rate": 8.147073094159792e-05, "loss": 0.026, "step": 95740 }, { "epoch": 61.02613129381772, "grad_norm": 0.2936679720878601, "learning_rate": 8.146644787971056e-05, "loss": 0.0228, "step": 95750 }, { "epoch": 61.03250478011472, "grad_norm": 0.26948612928390503, "learning_rate": 8.146216443547615e-05, "loss": 0.0166, "step": 95760 }, { "epoch": 61.03887826641173, "grad_norm": 0.4455488324165344, "learning_rate": 8.145788060894674e-05, "loss": 0.0205, "step": 95770 }, { "epoch": 61.04525175270873, "grad_norm": 0.36903873085975647, "learning_rate": 8.145359640017438e-05, "loss": 0.0213, "step": 95780 }, { "epoch": 61.05162523900574, "grad_norm": 0.4006199538707733, "learning_rate": 8.144931180921114e-05, "loss": 0.0128, "step": 95790 }, { "epoch": 61.05799872530274, "grad_norm": 0.4446595311164856, "learning_rate": 8.144502683610904e-05, "loss": 0.0197, "step": 95800 }, { "epoch": 61.064372211599746, "grad_norm": 0.33677929639816284, "learning_rate": 8.144074148092021e-05, "loss": 0.0223, "step": 95810 }, { "epoch": 61.07074569789675, "grad_norm": 0.3379022777080536, "learning_rate": 8.143645574369664e-05, "loss": 0.0229, "step": 95820 }, { "epoch": 61.077119184193755, "grad_norm": 0.4703719913959503, "learning_rate": 8.14321696244905e-05, "loss": 0.0165, "step": 95830 }, { "epoch": 61.083492670490756, "grad_norm": 0.3277742862701416, "learning_rate": 8.142788312335379e-05, "loss": 0.0152, "step": 95840 }, { "epoch": 61.089866156787764, "grad_norm": 0.6366003751754761, "learning_rate": 8.142359624033865e-05, "loss": 0.0293, "step": 95850 }, { "epoch": 61.096239643084765, "grad_norm": 0.25846773386001587, "learning_rate": 8.141930897549713e-05, "loss": 0.0182, "step": 95860 }, { "epoch": 61.10261312938177, "grad_norm": 0.32974714040756226, "learning_rate": 8.141502132888132e-05, "loss": 0.0213, "step": 95870 }, { "epoch": 61.10898661567877, "grad_norm": 0.6804724931716919, "learning_rate": 8.141073330054338e-05, "loss": 0.0265, "step": 95880 }, { "epoch": 61.11536010197578, "grad_norm": 0.2344520390033722, "learning_rate": 8.140644489053535e-05, "loss": 0.0201, "step": 95890 }, { "epoch": 61.12173358827278, "grad_norm": 0.3404674530029297, "learning_rate": 8.140215609890938e-05, "loss": 0.0144, "step": 95900 }, { "epoch": 61.12810707456979, "grad_norm": 0.4693468511104584, "learning_rate": 8.139786692571755e-05, "loss": 0.027, "step": 95910 }, { "epoch": 61.13448056086679, "grad_norm": 0.3311028480529785, "learning_rate": 8.139357737101201e-05, "loss": 0.0146, "step": 95920 }, { "epoch": 61.1408540471638, "grad_norm": 0.4409269094467163, "learning_rate": 8.138928743484484e-05, "loss": 0.0212, "step": 95930 }, { "epoch": 61.1472275334608, "grad_norm": 0.4106735587120056, "learning_rate": 8.138499711726821e-05, "loss": 0.02, "step": 95940 }, { "epoch": 61.15360101975781, "grad_norm": 0.43851763010025024, "learning_rate": 8.138070641833421e-05, "loss": 0.0216, "step": 95950 }, { "epoch": 61.15997450605481, "grad_norm": 0.8622010946273804, "learning_rate": 8.137641533809504e-05, "loss": 0.0175, "step": 95960 }, { "epoch": 61.16634799235182, "grad_norm": 0.3043321371078491, "learning_rate": 8.137212387660276e-05, "loss": 0.0422, "step": 95970 }, { "epoch": 61.17272147864882, "grad_norm": 0.36679545044898987, "learning_rate": 8.136783203390958e-05, "loss": 0.0204, "step": 95980 }, { "epoch": 61.179094964945826, "grad_norm": 0.33673644065856934, "learning_rate": 8.13635398100676e-05, "loss": 0.0195, "step": 95990 }, { "epoch": 61.185468451242826, "grad_norm": 0.42075634002685547, "learning_rate": 8.135924720512903e-05, "loss": 0.0265, "step": 96000 }, { "epoch": 61.191841937539834, "grad_norm": 0.3706428110599518, "learning_rate": 8.135495421914597e-05, "loss": 0.0179, "step": 96010 }, { "epoch": 61.198215423836835, "grad_norm": 0.9598209261894226, "learning_rate": 8.135066085217062e-05, "loss": 0.0294, "step": 96020 }, { "epoch": 61.20458891013384, "grad_norm": 0.3607963025569916, "learning_rate": 8.134636710425513e-05, "loss": 0.0183, "step": 96030 }, { "epoch": 61.21096239643085, "grad_norm": 0.40037673711776733, "learning_rate": 8.134207297545171e-05, "loss": 0.0241, "step": 96040 }, { "epoch": 61.21733588272785, "grad_norm": 0.3431549668312073, "learning_rate": 8.13377784658125e-05, "loss": 0.0204, "step": 96050 }, { "epoch": 61.22370936902486, "grad_norm": 0.3988195061683655, "learning_rate": 8.133348357538971e-05, "loss": 0.0201, "step": 96060 }, { "epoch": 61.23008285532186, "grad_norm": 0.29358696937561035, "learning_rate": 8.132918830423548e-05, "loss": 0.0178, "step": 96070 }, { "epoch": 61.23645634161887, "grad_norm": 0.2851429581642151, "learning_rate": 8.132489265240205e-05, "loss": 0.02, "step": 96080 }, { "epoch": 61.24282982791587, "grad_norm": 0.32349932193756104, "learning_rate": 8.13205966199416e-05, "loss": 0.0187, "step": 96090 }, { "epoch": 61.24920331421288, "grad_norm": 0.3963437080383301, "learning_rate": 8.131630020690632e-05, "loss": 0.0305, "step": 96100 }, { "epoch": 61.25557680050988, "grad_norm": 0.2823528051376343, "learning_rate": 8.131200341334843e-05, "loss": 0.0169, "step": 96110 }, { "epoch": 61.261950286806886, "grad_norm": 0.6109834909439087, "learning_rate": 8.130770623932013e-05, "loss": 0.0311, "step": 96120 }, { "epoch": 61.26832377310389, "grad_norm": 0.37428396940231323, "learning_rate": 8.130340868487365e-05, "loss": 0.0171, "step": 96130 }, { "epoch": 61.274697259400895, "grad_norm": 0.38158851861953735, "learning_rate": 8.12991107500612e-05, "loss": 0.0146, "step": 96140 }, { "epoch": 61.281070745697896, "grad_norm": 0.2713254392147064, "learning_rate": 8.1294812434935e-05, "loss": 0.0155, "step": 96150 }, { "epoch": 61.287444231994904, "grad_norm": 0.41676440834999084, "learning_rate": 8.129051373954728e-05, "loss": 0.0341, "step": 96160 }, { "epoch": 61.293817718291905, "grad_norm": 0.5367199182510376, "learning_rate": 8.128621466395028e-05, "loss": 0.0178, "step": 96170 }, { "epoch": 61.30019120458891, "grad_norm": 0.48037564754486084, "learning_rate": 8.128191520819621e-05, "loss": 0.0165, "step": 96180 }, { "epoch": 61.306564690885914, "grad_norm": 0.7687711119651794, "learning_rate": 8.127761537233736e-05, "loss": 0.0245, "step": 96190 }, { "epoch": 61.31293817718292, "grad_norm": 0.6043808460235596, "learning_rate": 8.127331515642595e-05, "loss": 0.0292, "step": 96200 }, { "epoch": 61.31931166347992, "grad_norm": 0.40612417459487915, "learning_rate": 8.126901456051421e-05, "loss": 0.0151, "step": 96210 }, { "epoch": 61.32568514977693, "grad_norm": 0.31163692474365234, "learning_rate": 8.126471358465445e-05, "loss": 0.016, "step": 96220 }, { "epoch": 61.33205863607393, "grad_norm": 0.5293201208114624, "learning_rate": 8.126041222889889e-05, "loss": 0.0235, "step": 96230 }, { "epoch": 61.33843212237094, "grad_norm": 0.4634445309638977, "learning_rate": 8.125611049329981e-05, "loss": 0.0243, "step": 96240 }, { "epoch": 61.34480560866794, "grad_norm": 0.49156299233436584, "learning_rate": 8.125180837790946e-05, "loss": 0.0178, "step": 96250 }, { "epoch": 61.35117909496495, "grad_norm": 0.38550353050231934, "learning_rate": 8.124750588278017e-05, "loss": 0.0186, "step": 96260 }, { "epoch": 61.35755258126195, "grad_norm": 0.3207261860370636, "learning_rate": 8.124320300796414e-05, "loss": 0.0138, "step": 96270 }, { "epoch": 61.36392606755896, "grad_norm": 0.2582796812057495, "learning_rate": 8.12388997535137e-05, "loss": 0.0128, "step": 96280 }, { "epoch": 61.37029955385596, "grad_norm": 0.8275569677352905, "learning_rate": 8.123459611948116e-05, "loss": 0.0201, "step": 96290 }, { "epoch": 61.376673040152966, "grad_norm": 0.46774613857269287, "learning_rate": 8.123029210591876e-05, "loss": 0.0345, "step": 96300 }, { "epoch": 61.38304652644997, "grad_norm": 0.1818963587284088, "learning_rate": 8.122598771287883e-05, "loss": 0.0216, "step": 96310 }, { "epoch": 61.389420012746974, "grad_norm": 0.5210548043251038, "learning_rate": 8.122168294041368e-05, "loss": 0.0197, "step": 96320 }, { "epoch": 61.395793499043975, "grad_norm": 0.306632936000824, "learning_rate": 8.12173777885756e-05, "loss": 0.0142, "step": 96330 }, { "epoch": 61.40216698534098, "grad_norm": 0.2799775302410126, "learning_rate": 8.12130722574169e-05, "loss": 0.0165, "step": 96340 }, { "epoch": 61.408540471637984, "grad_norm": 0.5510163307189941, "learning_rate": 8.120876634698991e-05, "loss": 0.0201, "step": 96350 }, { "epoch": 61.41491395793499, "grad_norm": 0.3885762393474579, "learning_rate": 8.120446005734694e-05, "loss": 0.041, "step": 96360 }, { "epoch": 61.42128744423199, "grad_norm": 0.3218502104282379, "learning_rate": 8.12001533885403e-05, "loss": 0.0202, "step": 96370 }, { "epoch": 61.427660930529, "grad_norm": 0.42520588636398315, "learning_rate": 8.119584634062237e-05, "loss": 0.0242, "step": 96380 }, { "epoch": 61.434034416826, "grad_norm": 0.3278224468231201, "learning_rate": 8.119153891364544e-05, "loss": 0.0167, "step": 96390 }, { "epoch": 61.44040790312301, "grad_norm": 0.40820711851119995, "learning_rate": 8.118723110766185e-05, "loss": 0.0179, "step": 96400 }, { "epoch": 61.44678138942001, "grad_norm": 0.441184937953949, "learning_rate": 8.118292292272399e-05, "loss": 0.0163, "step": 96410 }, { "epoch": 61.45315487571702, "grad_norm": 0.356674462556839, "learning_rate": 8.117861435888413e-05, "loss": 0.0137, "step": 96420 }, { "epoch": 61.45952836201402, "grad_norm": 0.6369248628616333, "learning_rate": 8.11743054161947e-05, "loss": 0.0276, "step": 96430 }, { "epoch": 61.46590184831103, "grad_norm": 0.42223960161209106, "learning_rate": 8.116999609470801e-05, "loss": 0.0137, "step": 96440 }, { "epoch": 61.47227533460803, "grad_norm": 0.3947093188762665, "learning_rate": 8.116568639447644e-05, "loss": 0.0202, "step": 96450 }, { "epoch": 61.478648820905036, "grad_norm": 0.3592786490917206, "learning_rate": 8.116137631555236e-05, "loss": 0.0216, "step": 96460 }, { "epoch": 61.48502230720204, "grad_norm": 0.3977213203907013, "learning_rate": 8.115706585798814e-05, "loss": 0.0188, "step": 96470 }, { "epoch": 61.491395793499045, "grad_norm": 0.3117183744907379, "learning_rate": 8.115275502183615e-05, "loss": 0.0137, "step": 96480 }, { "epoch": 61.497769279796046, "grad_norm": 0.5651962757110596, "learning_rate": 8.114844380714876e-05, "loss": 0.0336, "step": 96490 }, { "epoch": 61.504142766093054, "grad_norm": 0.286560595035553, "learning_rate": 8.114413221397838e-05, "loss": 0.0166, "step": 96500 }, { "epoch": 61.510516252390055, "grad_norm": 0.40250521898269653, "learning_rate": 8.11398202423774e-05, "loss": 0.0205, "step": 96510 }, { "epoch": 61.51688973868706, "grad_norm": 0.3812200129032135, "learning_rate": 8.113550789239818e-05, "loss": 0.0174, "step": 96520 }, { "epoch": 61.52326322498406, "grad_norm": 0.36307859420776367, "learning_rate": 8.113119516409316e-05, "loss": 0.0127, "step": 96530 }, { "epoch": 61.52963671128107, "grad_norm": 0.39085614681243896, "learning_rate": 8.112688205751472e-05, "loss": 0.0155, "step": 96540 }, { "epoch": 61.53601019757807, "grad_norm": 0.4594487249851227, "learning_rate": 8.112256857271528e-05, "loss": 0.0176, "step": 96550 }, { "epoch": 61.54238368387508, "grad_norm": 0.37602975964546204, "learning_rate": 8.111825470974724e-05, "loss": 0.016, "step": 96560 }, { "epoch": 61.54875717017208, "grad_norm": 0.2828028202056885, "learning_rate": 8.111394046866302e-05, "loss": 0.0161, "step": 96570 }, { "epoch": 61.55513065646909, "grad_norm": 0.32374995946884155, "learning_rate": 8.110962584951506e-05, "loss": 0.0196, "step": 96580 }, { "epoch": 61.56150414276609, "grad_norm": 0.607700765132904, "learning_rate": 8.110531085235578e-05, "loss": 0.0173, "step": 96590 }, { "epoch": 61.5678776290631, "grad_norm": 0.4178611636161804, "learning_rate": 8.110099547723757e-05, "loss": 0.0196, "step": 96600 }, { "epoch": 61.5742511153601, "grad_norm": 0.4291205108165741, "learning_rate": 8.109667972421293e-05, "loss": 0.0134, "step": 96610 }, { "epoch": 61.58062460165711, "grad_norm": 0.4629915654659271, "learning_rate": 8.109236359333426e-05, "loss": 0.0198, "step": 96620 }, { "epoch": 61.58699808795411, "grad_norm": 0.5891873836517334, "learning_rate": 8.108804708465403e-05, "loss": 0.0232, "step": 96630 }, { "epoch": 61.593371574251115, "grad_norm": 0.3979177474975586, "learning_rate": 8.108373019822466e-05, "loss": 0.0203, "step": 96640 }, { "epoch": 61.599745060548116, "grad_norm": 0.2961391508579254, "learning_rate": 8.107941293409863e-05, "loss": 0.0217, "step": 96650 }, { "epoch": 61.606118546845124, "grad_norm": 0.4202664792537689, "learning_rate": 8.107509529232838e-05, "loss": 0.0216, "step": 96660 }, { "epoch": 61.61249203314213, "grad_norm": 0.29675528407096863, "learning_rate": 8.107077727296639e-05, "loss": 0.0156, "step": 96670 }, { "epoch": 61.61886551943913, "grad_norm": 0.7197170257568359, "learning_rate": 8.106645887606512e-05, "loss": 0.0155, "step": 96680 }, { "epoch": 61.62523900573614, "grad_norm": 0.41031673550605774, "learning_rate": 8.106214010167704e-05, "loss": 0.0176, "step": 96690 }, { "epoch": 61.63161249203314, "grad_norm": 0.6684808135032654, "learning_rate": 8.105782094985464e-05, "loss": 0.0198, "step": 96700 }, { "epoch": 61.63798597833015, "grad_norm": 0.40795400738716125, "learning_rate": 8.105350142065037e-05, "loss": 0.0252, "step": 96710 }, { "epoch": 61.64435946462715, "grad_norm": 0.47454220056533813, "learning_rate": 8.104918151411676e-05, "loss": 0.0188, "step": 96720 }, { "epoch": 61.65073295092416, "grad_norm": 0.5199906229972839, "learning_rate": 8.104486123030627e-05, "loss": 0.0179, "step": 96730 }, { "epoch": 61.65710643722116, "grad_norm": 0.3327793478965759, "learning_rate": 8.104054056927139e-05, "loss": 0.0166, "step": 96740 }, { "epoch": 61.66347992351817, "grad_norm": 0.3099733293056488, "learning_rate": 8.103621953106466e-05, "loss": 0.0239, "step": 96750 }, { "epoch": 61.66985340981517, "grad_norm": 0.18611934781074524, "learning_rate": 8.103189811573855e-05, "loss": 0.0227, "step": 96760 }, { "epoch": 61.676226896112176, "grad_norm": 0.4122558534145355, "learning_rate": 8.102757632334558e-05, "loss": 0.0232, "step": 96770 }, { "epoch": 61.68260038240918, "grad_norm": 0.5746716260910034, "learning_rate": 8.102325415393826e-05, "loss": 0.0251, "step": 96780 }, { "epoch": 61.688973868706185, "grad_norm": 0.5040014982223511, "learning_rate": 8.101893160756913e-05, "loss": 0.0191, "step": 96790 }, { "epoch": 61.695347355003186, "grad_norm": 0.48117077350616455, "learning_rate": 8.101460868429067e-05, "loss": 0.0241, "step": 96800 }, { "epoch": 61.701720841300194, "grad_norm": 0.40665462613105774, "learning_rate": 8.101028538415543e-05, "loss": 0.0187, "step": 96810 }, { "epoch": 61.708094327597195, "grad_norm": 0.6190661191940308, "learning_rate": 8.100596170721597e-05, "loss": 0.0184, "step": 96820 }, { "epoch": 61.7144678138942, "grad_norm": 0.3660863935947418, "learning_rate": 8.100163765352478e-05, "loss": 0.0246, "step": 96830 }, { "epoch": 61.7208413001912, "grad_norm": 0.25920796394348145, "learning_rate": 8.099731322313442e-05, "loss": 0.0143, "step": 96840 }, { "epoch": 61.72721478648821, "grad_norm": 0.9363122582435608, "learning_rate": 8.099298841609745e-05, "loss": 0.0198, "step": 96850 }, { "epoch": 61.73358827278521, "grad_norm": 0.6316894292831421, "learning_rate": 8.098866323246642e-05, "loss": 0.023, "step": 96860 }, { "epoch": 61.73996175908222, "grad_norm": 0.603313148021698, "learning_rate": 8.098433767229384e-05, "loss": 0.0217, "step": 96870 }, { "epoch": 61.74633524537922, "grad_norm": 0.6552619934082031, "learning_rate": 8.098001173563232e-05, "loss": 0.0146, "step": 96880 }, { "epoch": 61.75270873167623, "grad_norm": 0.37630438804626465, "learning_rate": 8.097568542253441e-05, "loss": 0.0241, "step": 96890 }, { "epoch": 61.75908221797323, "grad_norm": 0.33536455035209656, "learning_rate": 8.097135873305267e-05, "loss": 0.016, "step": 96900 }, { "epoch": 61.76545570427024, "grad_norm": 0.4042932689189911, "learning_rate": 8.096703166723968e-05, "loss": 0.0192, "step": 96910 }, { "epoch": 61.77182919056724, "grad_norm": 0.6125134229660034, "learning_rate": 8.096270422514802e-05, "loss": 0.016, "step": 96920 }, { "epoch": 61.77820267686425, "grad_norm": 0.46857771277427673, "learning_rate": 8.095837640683029e-05, "loss": 0.0237, "step": 96930 }, { "epoch": 61.78457616316125, "grad_norm": 0.5416263341903687, "learning_rate": 8.095404821233903e-05, "loss": 0.0216, "step": 96940 }, { "epoch": 61.790949649458256, "grad_norm": 0.28832900524139404, "learning_rate": 8.094971964172687e-05, "loss": 0.0185, "step": 96950 }, { "epoch": 61.797323135755256, "grad_norm": 0.3597073554992676, "learning_rate": 8.094539069504639e-05, "loss": 0.0192, "step": 96960 }, { "epoch": 61.803696622052264, "grad_norm": 0.3857431411743164, "learning_rate": 8.094106137235019e-05, "loss": 0.0274, "step": 96970 }, { "epoch": 61.810070108349265, "grad_norm": 0.4957766830921173, "learning_rate": 8.093673167369089e-05, "loss": 0.0199, "step": 96980 }, { "epoch": 61.81644359464627, "grad_norm": 0.33757513761520386, "learning_rate": 8.093240159912108e-05, "loss": 0.0241, "step": 96990 }, { "epoch": 61.822817080943274, "grad_norm": 0.42400628328323364, "learning_rate": 8.09280711486934e-05, "loss": 0.0176, "step": 97000 }, { "epoch": 61.82919056724028, "grad_norm": 0.33330148458480835, "learning_rate": 8.092374032246045e-05, "loss": 0.0156, "step": 97010 }, { "epoch": 61.83556405353728, "grad_norm": 0.38017845153808594, "learning_rate": 8.091940912047486e-05, "loss": 0.0186, "step": 97020 }, { "epoch": 61.84193753983429, "grad_norm": 0.472946435213089, "learning_rate": 8.091507754278927e-05, "loss": 0.0275, "step": 97030 }, { "epoch": 61.84831102613129, "grad_norm": 0.1467377245426178, "learning_rate": 8.091074558945626e-05, "loss": 0.0155, "step": 97040 }, { "epoch": 61.8546845124283, "grad_norm": 0.758780837059021, "learning_rate": 8.090641326052854e-05, "loss": 0.0159, "step": 97050 }, { "epoch": 61.8610579987253, "grad_norm": 0.573049008846283, "learning_rate": 8.09020805560587e-05, "loss": 0.0263, "step": 97060 }, { "epoch": 61.86743148502231, "grad_norm": 0.38568615913391113, "learning_rate": 8.089774747609942e-05, "loss": 0.0118, "step": 97070 }, { "epoch": 61.87380497131931, "grad_norm": 0.2276342213153839, "learning_rate": 8.089341402070331e-05, "loss": 0.0124, "step": 97080 }, { "epoch": 61.88017845761632, "grad_norm": 0.8106245398521423, "learning_rate": 8.088908018992309e-05, "loss": 0.0238, "step": 97090 }, { "epoch": 61.88655194391332, "grad_norm": 0.39384767413139343, "learning_rate": 8.088474598381135e-05, "loss": 0.0239, "step": 97100 }, { "epoch": 61.892925430210326, "grad_norm": 0.428750216960907, "learning_rate": 8.088041140242081e-05, "loss": 0.016, "step": 97110 }, { "epoch": 61.89929891650733, "grad_norm": 0.5590060353279114, "learning_rate": 8.08760764458041e-05, "loss": 0.0223, "step": 97120 }, { "epoch": 61.905672402804335, "grad_norm": 0.3726309537887573, "learning_rate": 8.087174111401393e-05, "loss": 0.0233, "step": 97130 }, { "epoch": 61.912045889101336, "grad_norm": 0.5455565452575684, "learning_rate": 8.086740540710293e-05, "loss": 0.0222, "step": 97140 }, { "epoch": 61.918419375398344, "grad_norm": 0.5680257081985474, "learning_rate": 8.086306932512382e-05, "loss": 0.0167, "step": 97150 }, { "epoch": 61.924792861695344, "grad_norm": 0.4302905201911926, "learning_rate": 8.085873286812928e-05, "loss": 0.016, "step": 97160 }, { "epoch": 61.93116634799235, "grad_norm": 0.5440407991409302, "learning_rate": 8.085439603617199e-05, "loss": 0.0339, "step": 97170 }, { "epoch": 61.93753983428935, "grad_norm": 0.6203112602233887, "learning_rate": 8.085005882930468e-05, "loss": 0.0149, "step": 97180 }, { "epoch": 61.94391332058636, "grad_norm": 0.38591358065605164, "learning_rate": 8.084572124758001e-05, "loss": 0.0167, "step": 97190 }, { "epoch": 61.95028680688336, "grad_norm": 0.38126569986343384, "learning_rate": 8.084138329105071e-05, "loss": 0.0208, "step": 97200 }, { "epoch": 61.95666029318037, "grad_norm": 0.25042882561683655, "learning_rate": 8.08370449597695e-05, "loss": 0.0169, "step": 97210 }, { "epoch": 61.96303377947737, "grad_norm": 0.4394133687019348, "learning_rate": 8.083270625378905e-05, "loss": 0.0243, "step": 97220 }, { "epoch": 61.96940726577438, "grad_norm": 0.6699545979499817, "learning_rate": 8.08283671731621e-05, "loss": 0.0195, "step": 97230 }, { "epoch": 61.97578075207138, "grad_norm": 0.2697429060935974, "learning_rate": 8.082402771794141e-05, "loss": 0.0136, "step": 97240 }, { "epoch": 61.98215423836839, "grad_norm": 0.5341203808784485, "learning_rate": 8.081968788817967e-05, "loss": 0.023, "step": 97250 }, { "epoch": 61.98852772466539, "grad_norm": 0.29507946968078613, "learning_rate": 8.081534768392963e-05, "loss": 0.0167, "step": 97260 }, { "epoch": 61.9949012109624, "grad_norm": 0.3525209426879883, "learning_rate": 8.081100710524401e-05, "loss": 0.0208, "step": 97270 }, { "epoch": 62.001274697259404, "grad_norm": 0.5168564915657043, "learning_rate": 8.080666615217557e-05, "loss": 0.0248, "step": 97280 }, { "epoch": 62.007648183556405, "grad_norm": 0.5288978219032288, "learning_rate": 8.080232482477705e-05, "loss": 0.0354, "step": 97290 }, { "epoch": 62.01402166985341, "grad_norm": 0.38537314534187317, "learning_rate": 8.07979831231012e-05, "loss": 0.0167, "step": 97300 }, { "epoch": 62.020395156150414, "grad_norm": 0.38581278920173645, "learning_rate": 8.079364104720078e-05, "loss": 0.0182, "step": 97310 }, { "epoch": 62.02676864244742, "grad_norm": 0.3052176833152771, "learning_rate": 8.078929859712855e-05, "loss": 0.0297, "step": 97320 }, { "epoch": 62.03314212874442, "grad_norm": 0.3487243056297302, "learning_rate": 8.078495577293727e-05, "loss": 0.025, "step": 97330 }, { "epoch": 62.03951561504143, "grad_norm": 0.6181469559669495, "learning_rate": 8.07806125746797e-05, "loss": 0.0265, "step": 97340 }, { "epoch": 62.04588910133843, "grad_norm": 0.468288391828537, "learning_rate": 8.077626900240865e-05, "loss": 0.0239, "step": 97350 }, { "epoch": 62.05226258763544, "grad_norm": 0.6747252345085144, "learning_rate": 8.077192505617684e-05, "loss": 0.0163, "step": 97360 }, { "epoch": 62.05863607393244, "grad_norm": 0.26907429099082947, "learning_rate": 8.076758073603712e-05, "loss": 0.02, "step": 97370 }, { "epoch": 62.06500956022945, "grad_norm": 0.43044260144233704, "learning_rate": 8.076323604204222e-05, "loss": 0.0153, "step": 97380 }, { "epoch": 62.07138304652645, "grad_norm": 0.20085649192333221, "learning_rate": 8.075889097424496e-05, "loss": 0.0169, "step": 97390 }, { "epoch": 62.07775653282346, "grad_norm": 0.5214897990226746, "learning_rate": 8.075454553269815e-05, "loss": 0.0159, "step": 97400 }, { "epoch": 62.08413001912046, "grad_norm": 0.31792503595352173, "learning_rate": 8.075019971745457e-05, "loss": 0.0371, "step": 97410 }, { "epoch": 62.090503505417466, "grad_norm": 0.27119511365890503, "learning_rate": 8.074585352856701e-05, "loss": 0.0196, "step": 97420 }, { "epoch": 62.09687699171447, "grad_norm": 0.38794219493865967, "learning_rate": 8.074150696608832e-05, "loss": 0.0172, "step": 97430 }, { "epoch": 62.103250478011475, "grad_norm": 0.4796970784664154, "learning_rate": 8.07371600300713e-05, "loss": 0.0295, "step": 97440 }, { "epoch": 62.109623964308476, "grad_norm": 0.5864056348800659, "learning_rate": 8.073281272056874e-05, "loss": 0.0178, "step": 97450 }, { "epoch": 62.115997450605484, "grad_norm": 0.7065044045448303, "learning_rate": 8.072846503763351e-05, "loss": 0.0283, "step": 97460 }, { "epoch": 62.122370936902485, "grad_norm": 0.4096756875514984, "learning_rate": 8.072411698131841e-05, "loss": 0.0217, "step": 97470 }, { "epoch": 62.12874442319949, "grad_norm": 0.8006265163421631, "learning_rate": 8.071976855167629e-05, "loss": 0.0207, "step": 97480 }, { "epoch": 62.13511790949649, "grad_norm": 0.5989569425582886, "learning_rate": 8.071541974875996e-05, "loss": 0.0184, "step": 97490 }, { "epoch": 62.1414913957935, "grad_norm": 0.7030155062675476, "learning_rate": 8.071107057262231e-05, "loss": 0.0187, "step": 97500 }, { "epoch": 62.1478648820905, "grad_norm": 0.8609601259231567, "learning_rate": 8.070672102331614e-05, "loss": 0.0272, "step": 97510 }, { "epoch": 62.15423836838751, "grad_norm": 0.6227999925613403, "learning_rate": 8.07023711008943e-05, "loss": 0.0137, "step": 97520 }, { "epoch": 62.16061185468451, "grad_norm": 0.42242833971977234, "learning_rate": 8.069802080540969e-05, "loss": 0.0229, "step": 97530 }, { "epoch": 62.16698534098152, "grad_norm": 0.42195412516593933, "learning_rate": 8.069367013691512e-05, "loss": 0.0196, "step": 97540 }, { "epoch": 62.17335882727852, "grad_norm": 0.44220268726348877, "learning_rate": 8.068931909546349e-05, "loss": 0.0236, "step": 97550 }, { "epoch": 62.17973231357553, "grad_norm": 0.402554988861084, "learning_rate": 8.068496768110766e-05, "loss": 0.0186, "step": 97560 }, { "epoch": 62.18610579987253, "grad_norm": 0.467043399810791, "learning_rate": 8.06806158939005e-05, "loss": 0.0398, "step": 97570 }, { "epoch": 62.19247928616954, "grad_norm": 0.502907395362854, "learning_rate": 8.067626373389488e-05, "loss": 0.0282, "step": 97580 }, { "epoch": 62.19885277246654, "grad_norm": 0.4417386054992676, "learning_rate": 8.06719112011437e-05, "loss": 0.0159, "step": 97590 }, { "epoch": 62.205226258763545, "grad_norm": 0.6036456823348999, "learning_rate": 8.066755829569985e-05, "loss": 0.0267, "step": 97600 }, { "epoch": 62.211599745060546, "grad_norm": 0.7202828526496887, "learning_rate": 8.066320501761619e-05, "loss": 0.0262, "step": 97610 }, { "epoch": 62.217973231357554, "grad_norm": 0.3794562816619873, "learning_rate": 8.065885136694566e-05, "loss": 0.015, "step": 97620 }, { "epoch": 62.224346717654555, "grad_norm": 0.47899314761161804, "learning_rate": 8.065449734374113e-05, "loss": 0.0196, "step": 97630 }, { "epoch": 62.23072020395156, "grad_norm": 0.4210587739944458, "learning_rate": 8.065014294805551e-05, "loss": 0.0132, "step": 97640 }, { "epoch": 62.237093690248564, "grad_norm": 0.6893230080604553, "learning_rate": 8.064578817994172e-05, "loss": 0.017, "step": 97650 }, { "epoch": 62.24346717654557, "grad_norm": 0.4412880837917328, "learning_rate": 8.064143303945266e-05, "loss": 0.0178, "step": 97660 }, { "epoch": 62.24984066284257, "grad_norm": 0.32984164357185364, "learning_rate": 8.063707752664127e-05, "loss": 0.0191, "step": 97670 }, { "epoch": 62.25621414913958, "grad_norm": 0.20432505011558533, "learning_rate": 8.063272164156046e-05, "loss": 0.0245, "step": 97680 }, { "epoch": 62.26258763543658, "grad_norm": 0.5770242214202881, "learning_rate": 8.062836538426316e-05, "loss": 0.0224, "step": 97690 }, { "epoch": 62.26896112173359, "grad_norm": 0.3613213300704956, "learning_rate": 8.06240087548023e-05, "loss": 0.0351, "step": 97700 }, { "epoch": 62.27533460803059, "grad_norm": 0.3641296923160553, "learning_rate": 8.061965175323083e-05, "loss": 0.0317, "step": 97710 }, { "epoch": 62.2817080943276, "grad_norm": 0.3477620482444763, "learning_rate": 8.061529437960167e-05, "loss": 0.0228, "step": 97720 }, { "epoch": 62.2880815806246, "grad_norm": 0.587247371673584, "learning_rate": 8.061093663396778e-05, "loss": 0.016, "step": 97730 }, { "epoch": 62.29445506692161, "grad_norm": 0.3307460844516754, "learning_rate": 8.060657851638209e-05, "loss": 0.0137, "step": 97740 }, { "epoch": 62.30082855321861, "grad_norm": 0.33146101236343384, "learning_rate": 8.06022200268976e-05, "loss": 0.0194, "step": 97750 }, { "epoch": 62.307202039515616, "grad_norm": 0.348911315202713, "learning_rate": 8.059786116556724e-05, "loss": 0.0156, "step": 97760 }, { "epoch": 62.31357552581262, "grad_norm": 0.2472749948501587, "learning_rate": 8.059350193244396e-05, "loss": 0.0144, "step": 97770 }, { "epoch": 62.319949012109625, "grad_norm": 0.3716954290866852, "learning_rate": 8.058914232758078e-05, "loss": 0.0223, "step": 97780 }, { "epoch": 62.326322498406626, "grad_norm": 0.5336238741874695, "learning_rate": 8.05847823510306e-05, "loss": 0.0201, "step": 97790 }, { "epoch": 62.33269598470363, "grad_norm": 0.30397364497184753, "learning_rate": 8.058042200284645e-05, "loss": 0.0191, "step": 97800 }, { "epoch": 62.339069471000634, "grad_norm": 0.27326685190200806, "learning_rate": 8.057606128308132e-05, "loss": 0.0177, "step": 97810 }, { "epoch": 62.34544295729764, "grad_norm": 0.40142396092414856, "learning_rate": 8.057170019178817e-05, "loss": 0.0244, "step": 97820 }, { "epoch": 62.35181644359464, "grad_norm": 0.1699044406414032, "learning_rate": 8.056733872901997e-05, "loss": 0.0187, "step": 97830 }, { "epoch": 62.35818992989165, "grad_norm": 0.3968382477760315, "learning_rate": 8.056297689482978e-05, "loss": 0.0204, "step": 97840 }, { "epoch": 62.36456341618865, "grad_norm": 0.25534868240356445, "learning_rate": 8.055861468927055e-05, "loss": 0.025, "step": 97850 }, { "epoch": 62.37093690248566, "grad_norm": 0.5727357268333435, "learning_rate": 8.055425211239529e-05, "loss": 0.011, "step": 97860 }, { "epoch": 62.37731038878266, "grad_norm": 0.568152129650116, "learning_rate": 8.054988916425702e-05, "loss": 0.0245, "step": 97870 }, { "epoch": 62.38368387507967, "grad_norm": 1.2434951066970825, "learning_rate": 8.054552584490878e-05, "loss": 0.0205, "step": 97880 }, { "epoch": 62.39005736137667, "grad_norm": 0.45836806297302246, "learning_rate": 8.054116215440354e-05, "loss": 0.03, "step": 97890 }, { "epoch": 62.39643084767368, "grad_norm": 0.1967131495475769, "learning_rate": 8.053679809279434e-05, "loss": 0.0258, "step": 97900 }, { "epoch": 62.40280433397068, "grad_norm": 0.4425431191921234, "learning_rate": 8.053243366013422e-05, "loss": 0.0155, "step": 97910 }, { "epoch": 62.409177820267686, "grad_norm": 0.5517053604125977, "learning_rate": 8.052806885647621e-05, "loss": 0.0232, "step": 97920 }, { "epoch": 62.415551306564694, "grad_norm": 0.40846145153045654, "learning_rate": 8.052370368187332e-05, "loss": 0.023, "step": 97930 }, { "epoch": 62.421924792861695, "grad_norm": 0.46480873227119446, "learning_rate": 8.051933813637863e-05, "loss": 0.0199, "step": 97940 }, { "epoch": 62.4282982791587, "grad_norm": 0.18525362014770508, "learning_rate": 8.051497222004516e-05, "loss": 0.0187, "step": 97950 }, { "epoch": 62.434671765455704, "grad_norm": 0.3228902518749237, "learning_rate": 8.051060593292596e-05, "loss": 0.0221, "step": 97960 }, { "epoch": 62.44104525175271, "grad_norm": 0.316935658454895, "learning_rate": 8.05062392750741e-05, "loss": 0.0191, "step": 97970 }, { "epoch": 62.44741873804971, "grad_norm": 0.35347357392311096, "learning_rate": 8.050187224654263e-05, "loss": 0.0159, "step": 97980 }, { "epoch": 62.45379222434672, "grad_norm": 0.36566323041915894, "learning_rate": 8.049750484738461e-05, "loss": 0.0169, "step": 97990 }, { "epoch": 62.46016571064372, "grad_norm": 0.3364529311656952, "learning_rate": 8.049313707765312e-05, "loss": 0.0128, "step": 98000 }, { "epoch": 62.46653919694073, "grad_norm": 0.21278861165046692, "learning_rate": 8.048876893740121e-05, "loss": 0.0203, "step": 98010 }, { "epoch": 62.47291268323773, "grad_norm": 0.3901107907295227, "learning_rate": 8.048440042668198e-05, "loss": 0.0301, "step": 98020 }, { "epoch": 62.47928616953474, "grad_norm": 0.24456846714019775, "learning_rate": 8.048003154554851e-05, "loss": 0.0124, "step": 98030 }, { "epoch": 62.48565965583174, "grad_norm": 0.26253634691238403, "learning_rate": 8.047566229405386e-05, "loss": 0.0171, "step": 98040 }, { "epoch": 62.49203314212875, "grad_norm": 0.3108423054218292, "learning_rate": 8.047129267225115e-05, "loss": 0.0153, "step": 98050 }, { "epoch": 62.49840662842575, "grad_norm": 0.2625300884246826, "learning_rate": 8.046692268019345e-05, "loss": 0.0175, "step": 98060 }, { "epoch": 62.504780114722756, "grad_norm": 0.42729923129081726, "learning_rate": 8.046255231793389e-05, "loss": 0.0232, "step": 98070 }, { "epoch": 62.51115360101976, "grad_norm": 0.7011959552764893, "learning_rate": 8.045818158552555e-05, "loss": 0.0181, "step": 98080 }, { "epoch": 62.517527087316765, "grad_norm": 0.49554136395454407, "learning_rate": 8.045381048302157e-05, "loss": 0.0175, "step": 98090 }, { "epoch": 62.523900573613766, "grad_norm": 0.4114202857017517, "learning_rate": 8.044943901047502e-05, "loss": 0.0204, "step": 98100 }, { "epoch": 62.530274059910774, "grad_norm": 0.4409479796886444, "learning_rate": 8.044506716793904e-05, "loss": 0.0262, "step": 98110 }, { "epoch": 62.536647546207774, "grad_norm": 0.46096691489219666, "learning_rate": 8.044069495546674e-05, "loss": 0.0161, "step": 98120 }, { "epoch": 62.54302103250478, "grad_norm": 0.45059147477149963, "learning_rate": 8.043632237311127e-05, "loss": 0.0159, "step": 98130 }, { "epoch": 62.54939451880178, "grad_norm": 0.9806071519851685, "learning_rate": 8.043194942092574e-05, "loss": 0.0302, "step": 98140 }, { "epoch": 62.55576800509879, "grad_norm": 0.2080533653497696, "learning_rate": 8.042757609896327e-05, "loss": 0.0225, "step": 98150 }, { "epoch": 62.56214149139579, "grad_norm": 0.42297086119651794, "learning_rate": 8.042320240727705e-05, "loss": 0.0186, "step": 98160 }, { "epoch": 62.5685149776928, "grad_norm": 0.3181159496307373, "learning_rate": 8.04188283459202e-05, "loss": 0.0186, "step": 98170 }, { "epoch": 62.5748884639898, "grad_norm": 0.34635987877845764, "learning_rate": 8.041445391494584e-05, "loss": 0.0227, "step": 98180 }, { "epoch": 62.58126195028681, "grad_norm": 0.491744726896286, "learning_rate": 8.041007911440718e-05, "loss": 0.0199, "step": 98190 }, { "epoch": 62.58763543658381, "grad_norm": 0.4063069820404053, "learning_rate": 8.040570394435733e-05, "loss": 0.0224, "step": 98200 }, { "epoch": 62.59400892288082, "grad_norm": 0.39454934000968933, "learning_rate": 8.040132840484946e-05, "loss": 0.0172, "step": 98210 }, { "epoch": 62.60038240917782, "grad_norm": 0.2943691611289978, "learning_rate": 8.039695249593676e-05, "loss": 0.0172, "step": 98220 }, { "epoch": 62.606755895474826, "grad_norm": 0.478152334690094, "learning_rate": 8.039257621767238e-05, "loss": 0.0197, "step": 98230 }, { "epoch": 62.61312938177183, "grad_norm": 0.3306417763233185, "learning_rate": 8.038819957010952e-05, "loss": 0.0195, "step": 98240 }, { "epoch": 62.619502868068835, "grad_norm": 0.28508543968200684, "learning_rate": 8.038382255330133e-05, "loss": 0.0172, "step": 98250 }, { "epoch": 62.625876354365836, "grad_norm": 0.2696250379085541, "learning_rate": 8.037944516730101e-05, "loss": 0.0185, "step": 98260 }, { "epoch": 62.632249840662844, "grad_norm": 0.4812609851360321, "learning_rate": 8.037506741216174e-05, "loss": 0.0178, "step": 98270 }, { "epoch": 62.638623326959845, "grad_norm": 0.5007364749908447, "learning_rate": 8.037068928793674e-05, "loss": 0.0144, "step": 98280 }, { "epoch": 62.64499681325685, "grad_norm": 0.2383960783481598, "learning_rate": 8.036631079467917e-05, "loss": 0.0145, "step": 98290 }, { "epoch": 62.651370299553854, "grad_norm": 0.3832099437713623, "learning_rate": 8.036193193244228e-05, "loss": 0.0165, "step": 98300 }, { "epoch": 62.65774378585086, "grad_norm": 0.6132698655128479, "learning_rate": 8.035755270127922e-05, "loss": 0.0168, "step": 98310 }, { "epoch": 62.66411727214786, "grad_norm": 0.3570716679096222, "learning_rate": 8.035317310124327e-05, "loss": 0.0169, "step": 98320 }, { "epoch": 62.67049075844487, "grad_norm": 0.2383190095424652, "learning_rate": 8.034879313238758e-05, "loss": 0.0182, "step": 98330 }, { "epoch": 62.67686424474187, "grad_norm": 0.30801406502723694, "learning_rate": 8.03444127947654e-05, "loss": 0.0124, "step": 98340 }, { "epoch": 62.68323773103888, "grad_norm": 0.3085496723651886, "learning_rate": 8.034003208842997e-05, "loss": 0.0237, "step": 98350 }, { "epoch": 62.68961121733588, "grad_norm": 0.47724223136901855, "learning_rate": 8.033565101343448e-05, "loss": 0.0273, "step": 98360 }, { "epoch": 62.69598470363289, "grad_norm": 0.2284173220396042, "learning_rate": 8.033126956983221e-05, "loss": 0.0235, "step": 98370 }, { "epoch": 62.70235818992989, "grad_norm": 0.6044796109199524, "learning_rate": 8.032688775767637e-05, "loss": 0.0304, "step": 98380 }, { "epoch": 62.7087316762269, "grad_norm": 0.3556976020336151, "learning_rate": 8.032250557702021e-05, "loss": 0.017, "step": 98390 }, { "epoch": 62.7151051625239, "grad_norm": 0.36444106698036194, "learning_rate": 8.0318123027917e-05, "loss": 0.0263, "step": 98400 }, { "epoch": 62.721478648820906, "grad_norm": 0.33993998169898987, "learning_rate": 8.031374011041993e-05, "loss": 0.0179, "step": 98410 }, { "epoch": 62.72785213511791, "grad_norm": 1.283555269241333, "learning_rate": 8.030935682458231e-05, "loss": 0.0344, "step": 98420 }, { "epoch": 62.734225621414915, "grad_norm": 0.2985171675682068, "learning_rate": 8.03049731704574e-05, "loss": 0.0188, "step": 98430 }, { "epoch": 62.740599107711915, "grad_norm": 0.7376689314842224, "learning_rate": 8.030058914809844e-05, "loss": 0.018, "step": 98440 }, { "epoch": 62.74697259400892, "grad_norm": 0.5051977038383484, "learning_rate": 8.029620475755872e-05, "loss": 0.035, "step": 98450 }, { "epoch": 62.753346080305924, "grad_norm": 0.4614267945289612, "learning_rate": 8.029181999889152e-05, "loss": 0.0126, "step": 98460 }, { "epoch": 62.75971956660293, "grad_norm": 0.37160399556159973, "learning_rate": 8.028743487215009e-05, "loss": 0.0176, "step": 98470 }, { "epoch": 62.76609305289993, "grad_norm": 0.37515145540237427, "learning_rate": 8.028304937738773e-05, "loss": 0.02, "step": 98480 }, { "epoch": 62.77246653919694, "grad_norm": 0.33021432161331177, "learning_rate": 8.027866351465775e-05, "loss": 0.0314, "step": 98490 }, { "epoch": 62.77884002549394, "grad_norm": 0.47927749156951904, "learning_rate": 8.02742772840134e-05, "loss": 0.0155, "step": 98500 }, { "epoch": 62.78521351179095, "grad_norm": 0.6512821316719055, "learning_rate": 8.026989068550801e-05, "loss": 0.0283, "step": 98510 }, { "epoch": 62.79158699808795, "grad_norm": 0.29518041014671326, "learning_rate": 8.026550371919485e-05, "loss": 0.0199, "step": 98520 }, { "epoch": 62.79796048438496, "grad_norm": 0.48598840832710266, "learning_rate": 8.026111638512725e-05, "loss": 0.0169, "step": 98530 }, { "epoch": 62.80433397068197, "grad_norm": 0.48206278681755066, "learning_rate": 8.025672868335851e-05, "loss": 0.0309, "step": 98540 }, { "epoch": 62.81070745697897, "grad_norm": 0.26152753829956055, "learning_rate": 8.025234061394198e-05, "loss": 0.0129, "step": 98550 }, { "epoch": 62.817080943275975, "grad_norm": 0.5359047651290894, "learning_rate": 8.024795217693094e-05, "loss": 0.0144, "step": 98560 }, { "epoch": 62.823454429572976, "grad_norm": 0.4500282406806946, "learning_rate": 8.024356337237868e-05, "loss": 0.0197, "step": 98570 }, { "epoch": 62.829827915869984, "grad_norm": 0.3998371660709381, "learning_rate": 8.023917420033862e-05, "loss": 0.0245, "step": 98580 }, { "epoch": 62.836201402166985, "grad_norm": 0.4074002504348755, "learning_rate": 8.0234784660864e-05, "loss": 0.0194, "step": 98590 }, { "epoch": 62.84257488846399, "grad_norm": 0.33811888098716736, "learning_rate": 8.023039475400825e-05, "loss": 0.016, "step": 98600 }, { "epoch": 62.848948374760994, "grad_norm": 0.5253702402114868, "learning_rate": 8.022600447982462e-05, "loss": 0.0161, "step": 98610 }, { "epoch": 62.855321861058, "grad_norm": 0.668700098991394, "learning_rate": 8.022161383836652e-05, "loss": 0.0212, "step": 98620 }, { "epoch": 62.861695347355, "grad_norm": 0.7551515698432922, "learning_rate": 8.021722282968727e-05, "loss": 0.0218, "step": 98630 }, { "epoch": 62.86806883365201, "grad_norm": 0.4215162992477417, "learning_rate": 8.021283145384022e-05, "loss": 0.0188, "step": 98640 }, { "epoch": 62.87444231994901, "grad_norm": 0.5689658522605896, "learning_rate": 8.020843971087877e-05, "loss": 0.0194, "step": 98650 }, { "epoch": 62.88081580624602, "grad_norm": 0.5398138165473938, "learning_rate": 8.020404760085622e-05, "loss": 0.0191, "step": 98660 }, { "epoch": 62.88718929254302, "grad_norm": 0.4774570167064667, "learning_rate": 8.0199655123826e-05, "loss": 0.0193, "step": 98670 }, { "epoch": 62.89356277884003, "grad_norm": 0.3100110590457916, "learning_rate": 8.019526227984147e-05, "loss": 0.0209, "step": 98680 }, { "epoch": 62.89993626513703, "grad_norm": 0.5663490891456604, "learning_rate": 8.019086906895596e-05, "loss": 0.0256, "step": 98690 }, { "epoch": 62.90630975143404, "grad_norm": 0.5103445053100586, "learning_rate": 8.018647549122291e-05, "loss": 0.0267, "step": 98700 }, { "epoch": 62.91268323773104, "grad_norm": 0.4636227488517761, "learning_rate": 8.018208154669568e-05, "loss": 0.0198, "step": 98710 }, { "epoch": 62.919056724028046, "grad_norm": 0.283550500869751, "learning_rate": 8.017768723542765e-05, "loss": 0.0146, "step": 98720 }, { "epoch": 62.92543021032505, "grad_norm": 0.3818202614784241, "learning_rate": 8.017329255747223e-05, "loss": 0.0163, "step": 98730 }, { "epoch": 62.931803696622055, "grad_norm": 0.20966285467147827, "learning_rate": 8.016889751288283e-05, "loss": 0.02, "step": 98740 }, { "epoch": 62.938177182919055, "grad_norm": 0.35521215200424194, "learning_rate": 8.016450210171283e-05, "loss": 0.0242, "step": 98750 }, { "epoch": 62.94455066921606, "grad_norm": 0.6651269197463989, "learning_rate": 8.016010632401565e-05, "loss": 0.0247, "step": 98760 }, { "epoch": 62.950924155513064, "grad_norm": 0.539497971534729, "learning_rate": 8.015571017984471e-05, "loss": 0.0228, "step": 98770 }, { "epoch": 62.95729764181007, "grad_norm": 0.33614158630371094, "learning_rate": 8.015131366925341e-05, "loss": 0.019, "step": 98780 }, { "epoch": 62.96367112810707, "grad_norm": 0.4557303488254547, "learning_rate": 8.014691679229518e-05, "loss": 0.0319, "step": 98790 }, { "epoch": 62.97004461440408, "grad_norm": 0.2390744686126709, "learning_rate": 8.014251954902347e-05, "loss": 0.018, "step": 98800 }, { "epoch": 62.97641810070108, "grad_norm": 0.4323665201663971, "learning_rate": 8.013812193949166e-05, "loss": 0.0132, "step": 98810 }, { "epoch": 62.98279158699809, "grad_norm": 0.7057623863220215, "learning_rate": 8.013372396375323e-05, "loss": 0.0216, "step": 98820 }, { "epoch": 62.98916507329509, "grad_norm": 0.21151228249073029, "learning_rate": 8.012932562186159e-05, "loss": 0.02, "step": 98830 }, { "epoch": 62.9955385595921, "grad_norm": 0.37299472093582153, "learning_rate": 8.012492691387021e-05, "loss": 0.0147, "step": 98840 }, { "epoch": 63.0019120458891, "grad_norm": 0.5209152102470398, "learning_rate": 8.012052783983252e-05, "loss": 0.017, "step": 98850 }, { "epoch": 63.00828553218611, "grad_norm": 0.3561956286430359, "learning_rate": 8.011612839980198e-05, "loss": 0.0138, "step": 98860 }, { "epoch": 63.01465901848311, "grad_norm": 0.5387434363365173, "learning_rate": 8.011172859383204e-05, "loss": 0.0298, "step": 98870 }, { "epoch": 63.021032504780116, "grad_norm": 0.5193247199058533, "learning_rate": 8.010732842197616e-05, "loss": 0.0273, "step": 98880 }, { "epoch": 63.02740599107712, "grad_norm": 0.35978639125823975, "learning_rate": 8.010292788428784e-05, "loss": 0.0182, "step": 98890 }, { "epoch": 63.033779477374125, "grad_norm": 0.2811342775821686, "learning_rate": 8.00985269808205e-05, "loss": 0.0119, "step": 98900 }, { "epoch": 63.040152963671126, "grad_norm": 0.6008613705635071, "learning_rate": 8.009412571162765e-05, "loss": 0.029, "step": 98910 }, { "epoch": 63.046526449968134, "grad_norm": 0.4641428291797638, "learning_rate": 8.008972407676277e-05, "loss": 0.0209, "step": 98920 }, { "epoch": 63.052899936265135, "grad_norm": 0.3382489085197449, "learning_rate": 8.008532207627932e-05, "loss": 0.0183, "step": 98930 }, { "epoch": 63.05927342256214, "grad_norm": 0.5956593155860901, "learning_rate": 8.008091971023079e-05, "loss": 0.0147, "step": 98940 }, { "epoch": 63.065646908859144, "grad_norm": 0.32215505838394165, "learning_rate": 8.007651697867069e-05, "loss": 0.0107, "step": 98950 }, { "epoch": 63.07202039515615, "grad_norm": 0.2750360667705536, "learning_rate": 8.007211388165252e-05, "loss": 0.013, "step": 98960 }, { "epoch": 63.07839388145315, "grad_norm": 0.2838267982006073, "learning_rate": 8.006771041922978e-05, "loss": 0.0184, "step": 98970 }, { "epoch": 63.08476736775016, "grad_norm": 0.6140388250350952, "learning_rate": 8.006330659145595e-05, "loss": 0.0156, "step": 98980 }, { "epoch": 63.09114085404716, "grad_norm": 0.5899274349212646, "learning_rate": 8.005890239838458e-05, "loss": 0.0213, "step": 98990 }, { "epoch": 63.09751434034417, "grad_norm": 0.4996708035469055, "learning_rate": 8.005449784006917e-05, "loss": 0.0225, "step": 99000 }, { "epoch": 63.10388782664117, "grad_norm": 0.23874390125274658, "learning_rate": 8.005009291656322e-05, "loss": 0.0137, "step": 99010 }, { "epoch": 63.11026131293818, "grad_norm": 0.33274585008621216, "learning_rate": 8.004568762792026e-05, "loss": 0.0258, "step": 99020 }, { "epoch": 63.11663479923518, "grad_norm": 0.4330150783061981, "learning_rate": 8.004128197419385e-05, "loss": 0.0185, "step": 99030 }, { "epoch": 63.12300828553219, "grad_norm": 0.41974076628685, "learning_rate": 8.003687595543749e-05, "loss": 0.0105, "step": 99040 }, { "epoch": 63.12938177182919, "grad_norm": 0.40363675355911255, "learning_rate": 8.003246957170473e-05, "loss": 0.0189, "step": 99050 }, { "epoch": 63.135755258126196, "grad_norm": 0.3172776997089386, "learning_rate": 8.00280628230491e-05, "loss": 0.0163, "step": 99060 }, { "epoch": 63.1421287444232, "grad_norm": 0.4893060624599457, "learning_rate": 8.002365570952417e-05, "loss": 0.0122, "step": 99070 }, { "epoch": 63.148502230720204, "grad_norm": 0.3942698836326599, "learning_rate": 8.001924823118346e-05, "loss": 0.0153, "step": 99080 }, { "epoch": 63.154875717017205, "grad_norm": 0.3238273561000824, "learning_rate": 8.001484038808056e-05, "loss": 0.0171, "step": 99090 }, { "epoch": 63.16124920331421, "grad_norm": 0.36308565735816956, "learning_rate": 8.0010432180269e-05, "loss": 0.0141, "step": 99100 }, { "epoch": 63.167622689611214, "grad_norm": 0.5347263813018799, "learning_rate": 8.000602360780234e-05, "loss": 0.0193, "step": 99110 }, { "epoch": 63.17399617590822, "grad_norm": 0.5116556882858276, "learning_rate": 8.000161467073419e-05, "loss": 0.019, "step": 99120 }, { "epoch": 63.18036966220522, "grad_norm": 0.6092588305473328, "learning_rate": 7.999720536911807e-05, "loss": 0.0237, "step": 99130 }, { "epoch": 63.18674314850223, "grad_norm": 0.4888933300971985, "learning_rate": 7.999279570300761e-05, "loss": 0.0199, "step": 99140 }, { "epoch": 63.19311663479923, "grad_norm": 0.4203897714614868, "learning_rate": 7.998838567245634e-05, "loss": 0.0189, "step": 99150 }, { "epoch": 63.19949012109624, "grad_norm": 0.621531069278717, "learning_rate": 7.998397527751788e-05, "loss": 0.0195, "step": 99160 }, { "epoch": 63.20586360739325, "grad_norm": 0.4950043261051178, "learning_rate": 7.997956451824582e-05, "loss": 0.0244, "step": 99170 }, { "epoch": 63.21223709369025, "grad_norm": 0.4597020745277405, "learning_rate": 7.997515339469372e-05, "loss": 0.0235, "step": 99180 }, { "epoch": 63.218610579987256, "grad_norm": 0.24398472905158997, "learning_rate": 7.997074190691522e-05, "loss": 0.0187, "step": 99190 }, { "epoch": 63.22498406628426, "grad_norm": 0.3873862326145172, "learning_rate": 7.996633005496392e-05, "loss": 0.0253, "step": 99200 }, { "epoch": 63.231357552581265, "grad_norm": 0.6634920835494995, "learning_rate": 7.99619178388934e-05, "loss": 0.0181, "step": 99210 }, { "epoch": 63.237731038878266, "grad_norm": 0.3712419271469116, "learning_rate": 7.995750525875731e-05, "loss": 0.0163, "step": 99220 }, { "epoch": 63.244104525175274, "grad_norm": 0.5725627541542053, "learning_rate": 7.995309231460923e-05, "loss": 0.0172, "step": 99230 }, { "epoch": 63.250478011472275, "grad_norm": 0.45920196175575256, "learning_rate": 7.99486790065028e-05, "loss": 0.0146, "step": 99240 }, { "epoch": 63.25685149776928, "grad_norm": 0.24594923853874207, "learning_rate": 7.994426533449165e-05, "loss": 0.0244, "step": 99250 }, { "epoch": 63.263224984066284, "grad_norm": 0.2665552794933319, "learning_rate": 7.99398512986294e-05, "loss": 0.0129, "step": 99260 }, { "epoch": 63.26959847036329, "grad_norm": 0.5905922651290894, "learning_rate": 7.993543689896968e-05, "loss": 0.017, "step": 99270 }, { "epoch": 63.27597195666029, "grad_norm": 0.36916956305503845, "learning_rate": 7.993102213556616e-05, "loss": 0.0166, "step": 99280 }, { "epoch": 63.2823454429573, "grad_norm": 0.3216252326965332, "learning_rate": 7.992660700847244e-05, "loss": 0.0172, "step": 99290 }, { "epoch": 63.2887189292543, "grad_norm": 0.4990364909172058, "learning_rate": 7.99221915177422e-05, "loss": 0.0181, "step": 99300 }, { "epoch": 63.29509241555131, "grad_norm": 0.3280979096889496, "learning_rate": 7.991777566342909e-05, "loss": 0.0189, "step": 99310 }, { "epoch": 63.30146590184831, "grad_norm": 0.43353983759880066, "learning_rate": 7.991335944558673e-05, "loss": 0.0117, "step": 99320 }, { "epoch": 63.30783938814532, "grad_norm": 0.26578187942504883, "learning_rate": 7.990894286426883e-05, "loss": 0.0234, "step": 99330 }, { "epoch": 63.31421287444232, "grad_norm": 0.25533702969551086, "learning_rate": 7.990452591952902e-05, "loss": 0.0236, "step": 99340 }, { "epoch": 63.32058636073933, "grad_norm": 0.36629024147987366, "learning_rate": 7.990010861142101e-05, "loss": 0.0175, "step": 99350 }, { "epoch": 63.32695984703633, "grad_norm": 0.28898096084594727, "learning_rate": 7.989569093999843e-05, "loss": 0.0261, "step": 99360 }, { "epoch": 63.333333333333336, "grad_norm": 0.28239476680755615, "learning_rate": 7.989127290531499e-05, "loss": 0.0125, "step": 99370 }, { "epoch": 63.33970681963034, "grad_norm": 0.43497779965400696, "learning_rate": 7.988685450742437e-05, "loss": 0.0204, "step": 99380 }, { "epoch": 63.346080305927345, "grad_norm": 0.29697126150131226, "learning_rate": 7.988243574638024e-05, "loss": 0.0133, "step": 99390 }, { "epoch": 63.352453792224345, "grad_norm": 0.5886693000793457, "learning_rate": 7.987801662223631e-05, "loss": 0.0305, "step": 99400 }, { "epoch": 63.35882727852135, "grad_norm": 0.4977903664112091, "learning_rate": 7.987359713504626e-05, "loss": 0.0218, "step": 99410 }, { "epoch": 63.365200764818354, "grad_norm": 0.751430869102478, "learning_rate": 7.986917728486378e-05, "loss": 0.0186, "step": 99420 }, { "epoch": 63.37157425111536, "grad_norm": 0.8391770720481873, "learning_rate": 7.986475707174261e-05, "loss": 0.0256, "step": 99430 }, { "epoch": 63.37794773741236, "grad_norm": 0.6434748768806458, "learning_rate": 7.986033649573645e-05, "loss": 0.0219, "step": 99440 }, { "epoch": 63.38432122370937, "grad_norm": 0.4687001407146454, "learning_rate": 7.985591555689901e-05, "loss": 0.0301, "step": 99450 }, { "epoch": 63.39069471000637, "grad_norm": 0.539650559425354, "learning_rate": 7.9851494255284e-05, "loss": 0.0215, "step": 99460 }, { "epoch": 63.39706819630338, "grad_norm": 0.37941938638687134, "learning_rate": 7.984707259094517e-05, "loss": 0.0122, "step": 99470 }, { "epoch": 63.40344168260038, "grad_norm": 0.39016401767730713, "learning_rate": 7.98426505639362e-05, "loss": 0.0163, "step": 99480 }, { "epoch": 63.40981516889739, "grad_norm": 0.5059176683425903, "learning_rate": 7.983822817431088e-05, "loss": 0.0141, "step": 99490 }, { "epoch": 63.41618865519439, "grad_norm": 0.7039914131164551, "learning_rate": 7.983380542212288e-05, "loss": 0.0174, "step": 99500 }, { "epoch": 63.4225621414914, "grad_norm": 0.30721616744995117, "learning_rate": 7.9829382307426e-05, "loss": 0.013, "step": 99510 }, { "epoch": 63.4289356277884, "grad_norm": 0.7046541571617126, "learning_rate": 7.982495883027395e-05, "loss": 0.0256, "step": 99520 }, { "epoch": 63.435309114085406, "grad_norm": 0.2535940408706665, "learning_rate": 7.982053499072051e-05, "loss": 0.0213, "step": 99530 }, { "epoch": 63.44168260038241, "grad_norm": 0.27109014987945557, "learning_rate": 7.98161107888194e-05, "loss": 0.0245, "step": 99540 }, { "epoch": 63.448056086679415, "grad_norm": 0.4865108132362366, "learning_rate": 7.981168622462439e-05, "loss": 0.0164, "step": 99550 }, { "epoch": 63.454429572976416, "grad_norm": 0.35810279846191406, "learning_rate": 7.980726129818925e-05, "loss": 0.0121, "step": 99560 }, { "epoch": 63.460803059273424, "grad_norm": 0.5062312483787537, "learning_rate": 7.980283600956774e-05, "loss": 0.0192, "step": 99570 }, { "epoch": 63.467176545570425, "grad_norm": 0.6737575531005859, "learning_rate": 7.979841035881363e-05, "loss": 0.0176, "step": 99580 }, { "epoch": 63.47355003186743, "grad_norm": 0.5870066285133362, "learning_rate": 7.979398434598072e-05, "loss": 0.0218, "step": 99590 }, { "epoch": 63.47992351816443, "grad_norm": 0.37848570942878723, "learning_rate": 7.978955797112275e-05, "loss": 0.0142, "step": 99600 }, { "epoch": 63.48629700446144, "grad_norm": 0.36420461535453796, "learning_rate": 7.978513123429353e-05, "loss": 0.0171, "step": 99610 }, { "epoch": 63.49267049075844, "grad_norm": 0.46157515048980713, "learning_rate": 7.978070413554684e-05, "loss": 0.0135, "step": 99620 }, { "epoch": 63.49904397705545, "grad_norm": 0.536571741104126, "learning_rate": 7.977627667493648e-05, "loss": 0.019, "step": 99630 }, { "epoch": 63.50541746335245, "grad_norm": 0.25162723660469055, "learning_rate": 7.977184885251625e-05, "loss": 0.0154, "step": 99640 }, { "epoch": 63.51179094964946, "grad_norm": 0.2925228476524353, "learning_rate": 7.976742066833995e-05, "loss": 0.0156, "step": 99650 }, { "epoch": 63.51816443594646, "grad_norm": 0.35020291805267334, "learning_rate": 7.976299212246136e-05, "loss": 0.0202, "step": 99660 }, { "epoch": 63.52453792224347, "grad_norm": 0.683990478515625, "learning_rate": 7.975856321493434e-05, "loss": 0.015, "step": 99670 }, { "epoch": 63.53091140854047, "grad_norm": 0.3023894429206848, "learning_rate": 7.975413394581268e-05, "loss": 0.0173, "step": 99680 }, { "epoch": 63.53728489483748, "grad_norm": 0.5436844825744629, "learning_rate": 7.974970431515019e-05, "loss": 0.0155, "step": 99690 }, { "epoch": 63.54365838113448, "grad_norm": 0.260058730840683, "learning_rate": 7.97452743230007e-05, "loss": 0.0136, "step": 99700 }, { "epoch": 63.550031867431485, "grad_norm": 0.6079813838005066, "learning_rate": 7.974084396941805e-05, "loss": 0.0234, "step": 99710 }, { "epoch": 63.556405353728486, "grad_norm": 0.41311898827552795, "learning_rate": 7.973641325445607e-05, "loss": 0.0182, "step": 99720 }, { "epoch": 63.562778840025494, "grad_norm": 0.5203515291213989, "learning_rate": 7.973198217816858e-05, "loss": 0.0272, "step": 99730 }, { "epoch": 63.569152326322495, "grad_norm": 0.33635422587394714, "learning_rate": 7.972755074060945e-05, "loss": 0.0154, "step": 99740 }, { "epoch": 63.5755258126195, "grad_norm": 0.41816583275794983, "learning_rate": 7.97231189418325e-05, "loss": 0.032, "step": 99750 }, { "epoch": 63.581899298916504, "grad_norm": 0.40470996499061584, "learning_rate": 7.971868678189161e-05, "loss": 0.0172, "step": 99760 }, { "epoch": 63.58827278521351, "grad_norm": 0.25811755657196045, "learning_rate": 7.971425426084059e-05, "loss": 0.0237, "step": 99770 }, { "epoch": 63.59464627151051, "grad_norm": 0.3381985127925873, "learning_rate": 7.970982137873333e-05, "loss": 0.0323, "step": 99780 }, { "epoch": 63.60101975780752, "grad_norm": 0.42816609144210815, "learning_rate": 7.970538813562371e-05, "loss": 0.0125, "step": 99790 }, { "epoch": 63.60739324410453, "grad_norm": 0.3816288411617279, "learning_rate": 7.970095453156556e-05, "loss": 0.0145, "step": 99800 }, { "epoch": 63.61376673040153, "grad_norm": 0.6040284633636475, "learning_rate": 7.969652056661277e-05, "loss": 0.0282, "step": 99810 }, { "epoch": 63.62014021669854, "grad_norm": 0.20247100293636322, "learning_rate": 7.969208624081924e-05, "loss": 0.0132, "step": 99820 }, { "epoch": 63.62651370299554, "grad_norm": 0.4193212687969208, "learning_rate": 7.96876515542388e-05, "loss": 0.0199, "step": 99830 }, { "epoch": 63.632887189292546, "grad_norm": 0.5176677703857422, "learning_rate": 7.968321650692536e-05, "loss": 0.0242, "step": 99840 }, { "epoch": 63.63926067558955, "grad_norm": 0.18144778907299042, "learning_rate": 7.967878109893283e-05, "loss": 0.0171, "step": 99850 }, { "epoch": 63.645634161886555, "grad_norm": 0.5400227308273315, "learning_rate": 7.96743453303151e-05, "loss": 0.0151, "step": 99860 }, { "epoch": 63.652007648183556, "grad_norm": 0.7711611390113831, "learning_rate": 7.966990920112603e-05, "loss": 0.0211, "step": 99870 }, { "epoch": 63.658381134480564, "grad_norm": 0.36018404364585876, "learning_rate": 7.966547271141957e-05, "loss": 0.0309, "step": 99880 }, { "epoch": 63.664754620777565, "grad_norm": 0.475949227809906, "learning_rate": 7.966103586124959e-05, "loss": 0.0221, "step": 99890 }, { "epoch": 63.67112810707457, "grad_norm": 0.5908346176147461, "learning_rate": 7.965659865067004e-05, "loss": 0.0445, "step": 99900 }, { "epoch": 63.677501593371574, "grad_norm": 0.43027475476264954, "learning_rate": 7.965216107973481e-05, "loss": 0.0177, "step": 99910 }, { "epoch": 63.68387507966858, "grad_norm": 0.442116379737854, "learning_rate": 7.964772314849782e-05, "loss": 0.0193, "step": 99920 }, { "epoch": 63.69024856596558, "grad_norm": 0.4515080749988556, "learning_rate": 7.9643284857013e-05, "loss": 0.0275, "step": 99930 }, { "epoch": 63.69662205226259, "grad_norm": 0.39885076880455017, "learning_rate": 7.963884620533428e-05, "loss": 0.0178, "step": 99940 }, { "epoch": 63.70299553855959, "grad_norm": 0.4135817885398865, "learning_rate": 7.96344071935156e-05, "loss": 0.019, "step": 99950 }, { "epoch": 63.7093690248566, "grad_norm": 0.19746170938014984, "learning_rate": 7.962996782161089e-05, "loss": 0.0175, "step": 99960 }, { "epoch": 63.7157425111536, "grad_norm": 0.43266063928604126, "learning_rate": 7.962552808967412e-05, "loss": 0.0239, "step": 99970 }, { "epoch": 63.72211599745061, "grad_norm": 0.2739451229572296, "learning_rate": 7.962108799775917e-05, "loss": 0.0205, "step": 99980 }, { "epoch": 63.72848948374761, "grad_norm": 0.5102591514587402, "learning_rate": 7.961664754592006e-05, "loss": 0.0211, "step": 99990 }, { "epoch": 63.73486297004462, "grad_norm": 0.29910773038864136, "learning_rate": 7.961220673421072e-05, "loss": 0.0199, "step": 100000 }, { "epoch": 63.74123645634162, "grad_norm": 0.504512369632721, "learning_rate": 7.960776556268512e-05, "loss": 0.0189, "step": 100010 }, { "epoch": 63.747609942638626, "grad_norm": 0.5625543594360352, "learning_rate": 7.960332403139719e-05, "loss": 0.0142, "step": 100020 }, { "epoch": 63.753983428935626, "grad_norm": 0.7551673650741577, "learning_rate": 7.959888214040093e-05, "loss": 0.0304, "step": 100030 }, { "epoch": 63.760356915232634, "grad_norm": 0.6405827403068542, "learning_rate": 7.95944398897503e-05, "loss": 0.0251, "step": 100040 }, { "epoch": 63.766730401529635, "grad_norm": 0.23830997943878174, "learning_rate": 7.95899972794993e-05, "loss": 0.0221, "step": 100050 }, { "epoch": 63.77310388782664, "grad_norm": 0.25130775570869446, "learning_rate": 7.958555430970189e-05, "loss": 0.0162, "step": 100060 }, { "epoch": 63.779477374123644, "grad_norm": 0.6759980320930481, "learning_rate": 7.958111098041206e-05, "loss": 0.0246, "step": 100070 }, { "epoch": 63.78585086042065, "grad_norm": 0.41152045130729675, "learning_rate": 7.957666729168381e-05, "loss": 0.0166, "step": 100080 }, { "epoch": 63.79222434671765, "grad_norm": 0.48469844460487366, "learning_rate": 7.957222324357111e-05, "loss": 0.0169, "step": 100090 }, { "epoch": 63.79859783301466, "grad_norm": 0.4016260504722595, "learning_rate": 7.956777883612799e-05, "loss": 0.0263, "step": 100100 }, { "epoch": 63.80497131931166, "grad_norm": 0.38035067915916443, "learning_rate": 7.956333406940843e-05, "loss": 0.012, "step": 100110 }, { "epoch": 63.81134480560867, "grad_norm": 0.5274645090103149, "learning_rate": 7.955888894346646e-05, "loss": 0.0156, "step": 100120 }, { "epoch": 63.81771829190567, "grad_norm": 0.3349745571613312, "learning_rate": 7.955444345835607e-05, "loss": 0.0127, "step": 100130 }, { "epoch": 63.82409177820268, "grad_norm": 0.6604814529418945, "learning_rate": 7.954999761413128e-05, "loss": 0.0181, "step": 100140 }, { "epoch": 63.83046526449968, "grad_norm": 0.7396057844161987, "learning_rate": 7.954555141084614e-05, "loss": 0.0221, "step": 100150 }, { "epoch": 63.83683875079669, "grad_norm": 0.4282257556915283, "learning_rate": 7.954110484855465e-05, "loss": 0.0195, "step": 100160 }, { "epoch": 63.84321223709369, "grad_norm": 0.35531067848205566, "learning_rate": 7.953665792731084e-05, "loss": 0.0155, "step": 100170 }, { "epoch": 63.849585723390696, "grad_norm": 0.23789873719215393, "learning_rate": 7.953221064716876e-05, "loss": 0.0163, "step": 100180 }, { "epoch": 63.8559592096877, "grad_norm": 0.48693209886550903, "learning_rate": 7.952776300818243e-05, "loss": 0.0138, "step": 100190 }, { "epoch": 63.862332695984705, "grad_norm": 0.7536044716835022, "learning_rate": 7.95233150104059e-05, "loss": 0.0282, "step": 100200 }, { "epoch": 63.868706182281706, "grad_norm": 0.4548039436340332, "learning_rate": 7.951886665389322e-05, "loss": 0.0139, "step": 100210 }, { "epoch": 63.875079668578714, "grad_norm": 0.3638702929019928, "learning_rate": 7.951441793869843e-05, "loss": 0.0278, "step": 100220 }, { "epoch": 63.881453154875715, "grad_norm": 0.5413004159927368, "learning_rate": 7.95099688648756e-05, "loss": 0.0191, "step": 100230 }, { "epoch": 63.88782664117272, "grad_norm": 0.345094233751297, "learning_rate": 7.950551943247879e-05, "loss": 0.0136, "step": 100240 }, { "epoch": 63.89420012746972, "grad_norm": 0.42283883690834045, "learning_rate": 7.950106964156206e-05, "loss": 0.0205, "step": 100250 }, { "epoch": 63.90057361376673, "grad_norm": 0.3873358964920044, "learning_rate": 7.949661949217948e-05, "loss": 0.0157, "step": 100260 }, { "epoch": 63.90694710006373, "grad_norm": 0.8664463758468628, "learning_rate": 7.949216898438512e-05, "loss": 0.0285, "step": 100270 }, { "epoch": 63.91332058636074, "grad_norm": 0.3149782717227936, "learning_rate": 7.948771811823307e-05, "loss": 0.018, "step": 100280 }, { "epoch": 63.91969407265774, "grad_norm": 0.33077171444892883, "learning_rate": 7.94832668937774e-05, "loss": 0.0177, "step": 100290 }, { "epoch": 63.92606755895475, "grad_norm": 0.6482410430908203, "learning_rate": 7.94788153110722e-05, "loss": 0.0197, "step": 100300 }, { "epoch": 63.93244104525175, "grad_norm": 0.5424817800521851, "learning_rate": 7.947436337017155e-05, "loss": 0.0169, "step": 100310 }, { "epoch": 63.93881453154876, "grad_norm": 0.38601818680763245, "learning_rate": 7.946991107112957e-05, "loss": 0.0182, "step": 100320 }, { "epoch": 63.94518801784576, "grad_norm": 0.2997458875179291, "learning_rate": 7.946545841400035e-05, "loss": 0.0162, "step": 100330 }, { "epoch": 63.95156150414277, "grad_norm": 0.3400866985321045, "learning_rate": 7.946100539883798e-05, "loss": 0.0113, "step": 100340 }, { "epoch": 63.95793499043977, "grad_norm": 0.18522697687149048, "learning_rate": 7.945655202569659e-05, "loss": 0.0207, "step": 100350 }, { "epoch": 63.964308476736775, "grad_norm": 0.394366592168808, "learning_rate": 7.945209829463027e-05, "loss": 0.0119, "step": 100360 }, { "epoch": 63.970681963033776, "grad_norm": 0.4166446328163147, "learning_rate": 7.944764420569314e-05, "loss": 0.0182, "step": 100370 }, { "epoch": 63.977055449330784, "grad_norm": 0.1872115433216095, "learning_rate": 7.944318975893936e-05, "loss": 0.0209, "step": 100380 }, { "epoch": 63.983428935627785, "grad_norm": 0.3367049992084503, "learning_rate": 7.9438734954423e-05, "loss": 0.0131, "step": 100390 }, { "epoch": 63.98980242192479, "grad_norm": 0.5766119360923767, "learning_rate": 7.943427979219822e-05, "loss": 0.0222, "step": 100400 }, { "epoch": 63.9961759082218, "grad_norm": 0.3367582857608795, "learning_rate": 7.942982427231916e-05, "loss": 0.0244, "step": 100410 }, { "epoch": 64.00254939451881, "grad_norm": 0.3722878098487854, "learning_rate": 7.942536839483995e-05, "loss": 0.0149, "step": 100420 }, { "epoch": 64.0089228808158, "grad_norm": 0.3097875714302063, "learning_rate": 7.942091215981473e-05, "loss": 0.0168, "step": 100430 }, { "epoch": 64.01529636711281, "grad_norm": 0.6235575675964355, "learning_rate": 7.941645556729763e-05, "loss": 0.033, "step": 100440 }, { "epoch": 64.02166985340982, "grad_norm": 0.49927809834480286, "learning_rate": 7.941199861734285e-05, "loss": 0.0226, "step": 100450 }, { "epoch": 64.02804333970683, "grad_norm": 0.4408954381942749, "learning_rate": 7.940754131000449e-05, "loss": 0.0275, "step": 100460 }, { "epoch": 64.03441682600382, "grad_norm": 0.5775007009506226, "learning_rate": 7.940308364533675e-05, "loss": 0.0435, "step": 100470 }, { "epoch": 64.04079031230083, "grad_norm": 0.3866294026374817, "learning_rate": 7.939862562339378e-05, "loss": 0.0167, "step": 100480 }, { "epoch": 64.04716379859784, "grad_norm": 1.362955093383789, "learning_rate": 7.939416724422976e-05, "loss": 0.0205, "step": 100490 }, { "epoch": 64.05353728489484, "grad_norm": 0.4709736704826355, "learning_rate": 7.938970850789886e-05, "loss": 0.0217, "step": 100500 }, { "epoch": 64.05991077119184, "grad_norm": 0.27373307943344116, "learning_rate": 7.938524941445522e-05, "loss": 0.0219, "step": 100510 }, { "epoch": 64.06628425748885, "grad_norm": 0.7043091058731079, "learning_rate": 7.938078996395309e-05, "loss": 0.0257, "step": 100520 }, { "epoch": 64.07265774378585, "grad_norm": 0.6914076805114746, "learning_rate": 7.937633015644661e-05, "loss": 0.0213, "step": 100530 }, { "epoch": 64.07903123008286, "grad_norm": 0.7368538975715637, "learning_rate": 7.937186999198997e-05, "loss": 0.0145, "step": 100540 }, { "epoch": 64.08540471637986, "grad_norm": 1.0276824235916138, "learning_rate": 7.936740947063738e-05, "loss": 0.0231, "step": 100550 }, { "epoch": 64.09177820267686, "grad_norm": 0.5121086835861206, "learning_rate": 7.936294859244307e-05, "loss": 0.0189, "step": 100560 }, { "epoch": 64.09815168897387, "grad_norm": 0.6267324090003967, "learning_rate": 7.935848735746119e-05, "loss": 0.0241, "step": 100570 }, { "epoch": 64.10452517527088, "grad_norm": 0.4705946743488312, "learning_rate": 7.935402576574597e-05, "loss": 0.0189, "step": 100580 }, { "epoch": 64.11089866156787, "grad_norm": 0.4142254889011383, "learning_rate": 7.934956381735161e-05, "loss": 0.0307, "step": 100590 }, { "epoch": 64.11727214786488, "grad_norm": 0.5279407501220703, "learning_rate": 7.934510151233236e-05, "loss": 0.0301, "step": 100600 }, { "epoch": 64.12364563416189, "grad_norm": 0.8567071557044983, "learning_rate": 7.93406388507424e-05, "loss": 0.0205, "step": 100610 }, { "epoch": 64.1300191204589, "grad_norm": 0.5787873864173889, "learning_rate": 7.9336175832636e-05, "loss": 0.0329, "step": 100620 }, { "epoch": 64.13639260675589, "grad_norm": 0.9041554927825928, "learning_rate": 7.933171245806734e-05, "loss": 0.0231, "step": 100630 }, { "epoch": 64.1427660930529, "grad_norm": 0.4500020444393158, "learning_rate": 7.93272487270907e-05, "loss": 0.0143, "step": 100640 }, { "epoch": 64.1491395793499, "grad_norm": 0.5958545804023743, "learning_rate": 7.932278463976028e-05, "loss": 0.0299, "step": 100650 }, { "epoch": 64.15551306564691, "grad_norm": 0.48936063051223755, "learning_rate": 7.931832019613035e-05, "loss": 0.0169, "step": 100660 }, { "epoch": 64.16188655194391, "grad_norm": 0.4239228367805481, "learning_rate": 7.931385539625515e-05, "loss": 0.0167, "step": 100670 }, { "epoch": 64.16826003824092, "grad_norm": 0.3314990997314453, "learning_rate": 7.930939024018893e-05, "loss": 0.0156, "step": 100680 }, { "epoch": 64.17463352453792, "grad_norm": 0.37809112668037415, "learning_rate": 7.930492472798594e-05, "loss": 0.0251, "step": 100690 }, { "epoch": 64.18100701083493, "grad_norm": 0.3460645377635956, "learning_rate": 7.930045885970047e-05, "loss": 0.0175, "step": 100700 }, { "epoch": 64.18738049713193, "grad_norm": 0.3932236135005951, "learning_rate": 7.929599263538673e-05, "loss": 0.0145, "step": 100710 }, { "epoch": 64.19375398342893, "grad_norm": 0.30228695273399353, "learning_rate": 7.929152605509904e-05, "loss": 0.0166, "step": 100720 }, { "epoch": 64.20012746972594, "grad_norm": 0.538913905620575, "learning_rate": 7.928705911889164e-05, "loss": 0.024, "step": 100730 }, { "epoch": 64.20650095602295, "grad_norm": 0.4455079734325409, "learning_rate": 7.928259182681883e-05, "loss": 0.0201, "step": 100740 }, { "epoch": 64.21287444231994, "grad_norm": 0.38340333104133606, "learning_rate": 7.927812417893487e-05, "loss": 0.0096, "step": 100750 }, { "epoch": 64.21924792861695, "grad_norm": 0.37559008598327637, "learning_rate": 7.927365617529407e-05, "loss": 0.0203, "step": 100760 }, { "epoch": 64.22562141491396, "grad_norm": 0.1946892887353897, "learning_rate": 7.92691878159507e-05, "loss": 0.0134, "step": 100770 }, { "epoch": 64.23199490121097, "grad_norm": 0.5607951879501343, "learning_rate": 7.926471910095907e-05, "loss": 0.0196, "step": 100780 }, { "epoch": 64.23836838750796, "grad_norm": 0.8088657855987549, "learning_rate": 7.926025003037348e-05, "loss": 0.0192, "step": 100790 }, { "epoch": 64.24474187380497, "grad_norm": 0.2770121693611145, "learning_rate": 7.925578060424819e-05, "loss": 0.0158, "step": 100800 }, { "epoch": 64.25111536010198, "grad_norm": 0.2666962146759033, "learning_rate": 7.925131082263758e-05, "loss": 0.0184, "step": 100810 }, { "epoch": 64.25748884639899, "grad_norm": 0.2115778625011444, "learning_rate": 7.924684068559592e-05, "loss": 0.0238, "step": 100820 }, { "epoch": 64.26386233269598, "grad_norm": 0.19178056716918945, "learning_rate": 7.924237019317753e-05, "loss": 0.0189, "step": 100830 }, { "epoch": 64.27023581899299, "grad_norm": 0.2301441878080368, "learning_rate": 7.923789934543673e-05, "loss": 0.0144, "step": 100840 }, { "epoch": 64.27660930529, "grad_norm": 0.2912479043006897, "learning_rate": 7.923342814242785e-05, "loss": 0.0086, "step": 100850 }, { "epoch": 64.282982791587, "grad_norm": 0.1716517210006714, "learning_rate": 7.92289565842052e-05, "loss": 0.0168, "step": 100860 }, { "epoch": 64.289356277884, "grad_norm": 0.4291152358055115, "learning_rate": 7.922448467082315e-05, "loss": 0.0168, "step": 100870 }, { "epoch": 64.295729764181, "grad_norm": 0.455714613199234, "learning_rate": 7.922001240233601e-05, "loss": 0.018, "step": 100880 }, { "epoch": 64.30210325047801, "grad_norm": 0.46090081334114075, "learning_rate": 7.921553977879815e-05, "loss": 0.0178, "step": 100890 }, { "epoch": 64.30847673677502, "grad_norm": 0.3887077271938324, "learning_rate": 7.921106680026387e-05, "loss": 0.0182, "step": 100900 }, { "epoch": 64.31485022307201, "grad_norm": 0.5262818336486816, "learning_rate": 7.920659346678755e-05, "loss": 0.0223, "step": 100910 }, { "epoch": 64.32122370936902, "grad_norm": 0.21705761551856995, "learning_rate": 7.920211977842357e-05, "loss": 0.0183, "step": 100920 }, { "epoch": 64.32759719566603, "grad_norm": 0.37105482816696167, "learning_rate": 7.919764573522624e-05, "loss": 0.0198, "step": 100930 }, { "epoch": 64.33397068196304, "grad_norm": 0.4694282114505768, "learning_rate": 7.919317133724996e-05, "loss": 0.016, "step": 100940 }, { "epoch": 64.34034416826003, "grad_norm": 0.3237643539905548, "learning_rate": 7.918869658454908e-05, "loss": 0.0146, "step": 100950 }, { "epoch": 64.34671765455704, "grad_norm": 0.2847658693790436, "learning_rate": 7.918422147717798e-05, "loss": 0.0168, "step": 100960 }, { "epoch": 64.35309114085405, "grad_norm": 0.28392845392227173, "learning_rate": 7.917974601519102e-05, "loss": 0.0178, "step": 100970 }, { "epoch": 64.35946462715106, "grad_norm": 0.35983288288116455, "learning_rate": 7.917527019864262e-05, "loss": 0.0131, "step": 100980 }, { "epoch": 64.36583811344805, "grad_norm": 0.5620695352554321, "learning_rate": 7.917079402758713e-05, "loss": 0.0226, "step": 100990 }, { "epoch": 64.37221159974506, "grad_norm": 0.3042221665382385, "learning_rate": 7.916631750207894e-05, "loss": 0.0253, "step": 101000 }, { "epoch": 64.37858508604207, "grad_norm": 0.3776514232158661, "learning_rate": 7.916184062217247e-05, "loss": 0.0188, "step": 101010 }, { "epoch": 64.38495857233907, "grad_norm": 0.37415656447410583, "learning_rate": 7.91573633879221e-05, "loss": 0.0183, "step": 101020 }, { "epoch": 64.39133205863607, "grad_norm": 0.3529919981956482, "learning_rate": 7.915288579938224e-05, "loss": 0.0204, "step": 101030 }, { "epoch": 64.39770554493307, "grad_norm": 0.4255618453025818, "learning_rate": 7.914840785660729e-05, "loss": 0.021, "step": 101040 }, { "epoch": 64.40407903123008, "grad_norm": 0.2587633430957794, "learning_rate": 7.914392955965165e-05, "loss": 0.0178, "step": 101050 }, { "epoch": 64.41045251752709, "grad_norm": 0.46349823474884033, "learning_rate": 7.913945090856976e-05, "loss": 0.0191, "step": 101060 }, { "epoch": 64.4168260038241, "grad_norm": 0.3097410500049591, "learning_rate": 7.913497190341603e-05, "loss": 0.0239, "step": 101070 }, { "epoch": 64.42319949012109, "grad_norm": 0.29241394996643066, "learning_rate": 7.913049254424489e-05, "loss": 0.017, "step": 101080 }, { "epoch": 64.4295729764181, "grad_norm": 0.23196350038051605, "learning_rate": 7.912601283111077e-05, "loss": 0.0288, "step": 101090 }, { "epoch": 64.43594646271511, "grad_norm": 0.32502835988998413, "learning_rate": 7.912153276406807e-05, "loss": 0.0215, "step": 101100 }, { "epoch": 64.44231994901212, "grad_norm": 0.5920019745826721, "learning_rate": 7.911705234317126e-05, "loss": 0.0226, "step": 101110 }, { "epoch": 64.44869343530911, "grad_norm": 0.395711213350296, "learning_rate": 7.911257156847476e-05, "loss": 0.018, "step": 101120 }, { "epoch": 64.45506692160612, "grad_norm": 0.5862853527069092, "learning_rate": 7.910809044003304e-05, "loss": 0.0205, "step": 101130 }, { "epoch": 64.46144040790313, "grad_norm": 0.4178614914417267, "learning_rate": 7.910360895790055e-05, "loss": 0.0191, "step": 101140 }, { "epoch": 64.46781389420013, "grad_norm": 0.3227098286151886, "learning_rate": 7.909912712213172e-05, "loss": 0.0193, "step": 101150 }, { "epoch": 64.47418738049713, "grad_norm": 0.25548678636550903, "learning_rate": 7.909464493278103e-05, "loss": 0.0131, "step": 101160 }, { "epoch": 64.48056086679414, "grad_norm": 0.46040797233581543, "learning_rate": 7.909016238990292e-05, "loss": 0.0175, "step": 101170 }, { "epoch": 64.48693435309114, "grad_norm": 0.32197993993759155, "learning_rate": 7.90856794935519e-05, "loss": 0.0169, "step": 101180 }, { "epoch": 64.49330783938815, "grad_norm": 0.213952898979187, "learning_rate": 7.908119624378238e-05, "loss": 0.0172, "step": 101190 }, { "epoch": 64.49968132568515, "grad_norm": 0.4883028566837311, "learning_rate": 7.907671264064888e-05, "loss": 0.0224, "step": 101200 }, { "epoch": 64.50605481198215, "grad_norm": 0.7371419072151184, "learning_rate": 7.907222868420587e-05, "loss": 0.0208, "step": 101210 }, { "epoch": 64.51242829827916, "grad_norm": 0.18632501363754272, "learning_rate": 7.906774437450784e-05, "loss": 0.0103, "step": 101220 }, { "epoch": 64.51880178457617, "grad_norm": 0.5247831344604492, "learning_rate": 7.906325971160927e-05, "loss": 0.0281, "step": 101230 }, { "epoch": 64.52517527087316, "grad_norm": 0.5225400924682617, "learning_rate": 7.905877469556464e-05, "loss": 0.0236, "step": 101240 }, { "epoch": 64.53154875717017, "grad_norm": 0.2231888324022293, "learning_rate": 7.905428932642848e-05, "loss": 0.0123, "step": 101250 }, { "epoch": 64.53792224346718, "grad_norm": 0.2814962863922119, "learning_rate": 7.904980360425525e-05, "loss": 0.0167, "step": 101260 }, { "epoch": 64.54429572976419, "grad_norm": 0.3209356367588043, "learning_rate": 7.904531752909949e-05, "loss": 0.02, "step": 101270 }, { "epoch": 64.55066921606118, "grad_norm": 0.2787807881832123, "learning_rate": 7.90408311010157e-05, "loss": 0.0235, "step": 101280 }, { "epoch": 64.55704270235819, "grad_norm": 0.340372234582901, "learning_rate": 7.90363443200584e-05, "loss": 0.0218, "step": 101290 }, { "epoch": 64.5634161886552, "grad_norm": 0.4209139049053192, "learning_rate": 7.90318571862821e-05, "loss": 0.0182, "step": 101300 }, { "epoch": 64.5697896749522, "grad_norm": 0.32832300662994385, "learning_rate": 7.902736969974132e-05, "loss": 0.0211, "step": 101310 }, { "epoch": 64.5761631612492, "grad_norm": 0.49134567379951477, "learning_rate": 7.902288186049059e-05, "loss": 0.0311, "step": 101320 }, { "epoch": 64.5825366475462, "grad_norm": 0.2537376284599304, "learning_rate": 7.901839366858443e-05, "loss": 0.0281, "step": 101330 }, { "epoch": 64.58891013384321, "grad_norm": 0.30038487911224365, "learning_rate": 7.901390512407741e-05, "loss": 0.0153, "step": 101340 }, { "epoch": 64.59528362014022, "grad_norm": 0.2778017520904541, "learning_rate": 7.900941622702404e-05, "loss": 0.0164, "step": 101350 }, { "epoch": 64.60165710643722, "grad_norm": 0.3513677716255188, "learning_rate": 7.900492697747887e-05, "loss": 0.0171, "step": 101360 }, { "epoch": 64.60803059273422, "grad_norm": 0.4721972644329071, "learning_rate": 7.900043737549647e-05, "loss": 0.0239, "step": 101370 }, { "epoch": 64.61440407903123, "grad_norm": 0.2930135726928711, "learning_rate": 7.899594742113136e-05, "loss": 0.0127, "step": 101380 }, { "epoch": 64.62077756532824, "grad_norm": 0.3552321493625641, "learning_rate": 7.899145711443812e-05, "loss": 0.025, "step": 101390 }, { "epoch": 64.62715105162523, "grad_norm": 0.5242018699645996, "learning_rate": 7.89869664554713e-05, "loss": 0.0211, "step": 101400 }, { "epoch": 64.63352453792224, "grad_norm": 0.4109097719192505, "learning_rate": 7.898247544428547e-05, "loss": 0.0217, "step": 101410 }, { "epoch": 64.63989802421925, "grad_norm": 0.22159899771213531, "learning_rate": 7.89779840809352e-05, "loss": 0.0193, "step": 101420 }, { "epoch": 64.64627151051626, "grad_norm": 0.5730328559875488, "learning_rate": 7.897349236547506e-05, "loss": 0.0151, "step": 101430 }, { "epoch": 64.65264499681325, "grad_norm": 0.49782100319862366, "learning_rate": 7.896900029795961e-05, "loss": 0.0186, "step": 101440 }, { "epoch": 64.65901848311026, "grad_norm": 0.2766443192958832, "learning_rate": 7.89645078784435e-05, "loss": 0.0112, "step": 101450 }, { "epoch": 64.66539196940727, "grad_norm": 0.693105161190033, "learning_rate": 7.896001510698125e-05, "loss": 0.0254, "step": 101460 }, { "epoch": 64.67176545570427, "grad_norm": 0.5014377236366272, "learning_rate": 7.895552198362748e-05, "loss": 0.0131, "step": 101470 }, { "epoch": 64.67813894200127, "grad_norm": 0.3709430694580078, "learning_rate": 7.895102850843676e-05, "loss": 0.0215, "step": 101480 }, { "epoch": 64.68451242829828, "grad_norm": 0.39387303590774536, "learning_rate": 7.894653468146374e-05, "loss": 0.024, "step": 101490 }, { "epoch": 64.69088591459528, "grad_norm": 0.4278917908668518, "learning_rate": 7.894204050276297e-05, "loss": 0.0332, "step": 101500 }, { "epoch": 64.69725940089229, "grad_norm": 0.46081745624542236, "learning_rate": 7.893754597238908e-05, "loss": 0.0138, "step": 101510 }, { "epoch": 64.70363288718929, "grad_norm": 0.5743962526321411, "learning_rate": 7.893305109039671e-05, "loss": 0.0201, "step": 101520 }, { "epoch": 64.7100063734863, "grad_norm": 0.3479151427745819, "learning_rate": 7.892855585684043e-05, "loss": 0.0132, "step": 101530 }, { "epoch": 64.7163798597833, "grad_norm": 0.5232152342796326, "learning_rate": 7.892406027177488e-05, "loss": 0.024, "step": 101540 }, { "epoch": 64.72275334608031, "grad_norm": 0.3444971442222595, "learning_rate": 7.891956433525471e-05, "loss": 0.0358, "step": 101550 }, { "epoch": 64.7291268323773, "grad_norm": 0.41029149293899536, "learning_rate": 7.89150680473345e-05, "loss": 0.0149, "step": 101560 }, { "epoch": 64.73550031867431, "grad_norm": 0.23681877553462982, "learning_rate": 7.891057140806895e-05, "loss": 0.0191, "step": 101570 }, { "epoch": 64.74187380497132, "grad_norm": 0.41550537943840027, "learning_rate": 7.890607441751263e-05, "loss": 0.0186, "step": 101580 }, { "epoch": 64.74824729126833, "grad_norm": 0.35216400027275085, "learning_rate": 7.890157707572023e-05, "loss": 0.0112, "step": 101590 }, { "epoch": 64.75462077756532, "grad_norm": 0.49337026476860046, "learning_rate": 7.889707938274636e-05, "loss": 0.0135, "step": 101600 }, { "epoch": 64.76099426386233, "grad_norm": 0.3101850152015686, "learning_rate": 7.889258133864572e-05, "loss": 0.0167, "step": 101610 }, { "epoch": 64.76736775015934, "grad_norm": 0.3876095414161682, "learning_rate": 7.888808294347291e-05, "loss": 0.0175, "step": 101620 }, { "epoch": 64.77374123645635, "grad_norm": 0.563385546207428, "learning_rate": 7.888358419728263e-05, "loss": 0.0159, "step": 101630 }, { "epoch": 64.78011472275334, "grad_norm": 0.3106054365634918, "learning_rate": 7.887908510012954e-05, "loss": 0.0104, "step": 101640 }, { "epoch": 64.78648820905035, "grad_norm": 0.5170860886573792, "learning_rate": 7.88745856520683e-05, "loss": 0.0154, "step": 101650 }, { "epoch": 64.79286169534736, "grad_norm": 0.5176358222961426, "learning_rate": 7.887008585315357e-05, "loss": 0.0191, "step": 101660 }, { "epoch": 64.79923518164436, "grad_norm": 0.2890208959579468, "learning_rate": 7.886558570344005e-05, "loss": 0.0148, "step": 101670 }, { "epoch": 64.80560866794136, "grad_norm": 0.7593268752098083, "learning_rate": 7.886108520298241e-05, "loss": 0.0264, "step": 101680 }, { "epoch": 64.81198215423836, "grad_norm": 0.29270845651626587, "learning_rate": 7.885658435183532e-05, "loss": 0.0211, "step": 101690 }, { "epoch": 64.81835564053537, "grad_norm": 0.29428961873054504, "learning_rate": 7.88520831500535e-05, "loss": 0.0235, "step": 101700 }, { "epoch": 64.82472912683238, "grad_norm": 0.21894976496696472, "learning_rate": 7.884758159769164e-05, "loss": 0.0135, "step": 101710 }, { "epoch": 64.83110261312939, "grad_norm": 0.24917179346084595, "learning_rate": 7.884307969480439e-05, "loss": 0.0254, "step": 101720 }, { "epoch": 64.83747609942638, "grad_norm": 0.4067525863647461, "learning_rate": 7.883857744144651e-05, "loss": 0.0149, "step": 101730 }, { "epoch": 64.84384958572339, "grad_norm": 0.36483076214790344, "learning_rate": 7.883407483767269e-05, "loss": 0.0187, "step": 101740 }, { "epoch": 64.8502230720204, "grad_norm": 0.45491382479667664, "learning_rate": 7.882957188353764e-05, "loss": 0.0183, "step": 101750 }, { "epoch": 64.8565965583174, "grad_norm": 0.6092327833175659, "learning_rate": 7.882506857909606e-05, "loss": 0.0176, "step": 101760 }, { "epoch": 64.8629700446144, "grad_norm": 0.49361664056777954, "learning_rate": 7.882056492440269e-05, "loss": 0.0476, "step": 101770 }, { "epoch": 64.86934353091141, "grad_norm": 0.33111509680747986, "learning_rate": 7.881606091951223e-05, "loss": 0.0133, "step": 101780 }, { "epoch": 64.87571701720842, "grad_norm": 0.3197232484817505, "learning_rate": 7.881155656447944e-05, "loss": 0.0164, "step": 101790 }, { "epoch": 64.88209050350542, "grad_norm": 0.41550832986831665, "learning_rate": 7.880705185935904e-05, "loss": 0.0274, "step": 101800 }, { "epoch": 64.88846398980242, "grad_norm": 0.3176764249801636, "learning_rate": 7.880254680420574e-05, "loss": 0.0189, "step": 101810 }, { "epoch": 64.89483747609943, "grad_norm": 0.31465575098991394, "learning_rate": 7.879804139907431e-05, "loss": 0.021, "step": 101820 }, { "epoch": 64.90121096239643, "grad_norm": 0.39449888467788696, "learning_rate": 7.87935356440195e-05, "loss": 0.0213, "step": 101830 }, { "epoch": 64.90758444869344, "grad_norm": 0.4956305921077728, "learning_rate": 7.878902953909604e-05, "loss": 0.0375, "step": 101840 }, { "epoch": 64.91395793499044, "grad_norm": 0.3838893473148346, "learning_rate": 7.878452308435867e-05, "loss": 0.0263, "step": 101850 }, { "epoch": 64.92033142128744, "grad_norm": 0.39091402292251587, "learning_rate": 7.87800162798622e-05, "loss": 0.0175, "step": 101860 }, { "epoch": 64.92670490758445, "grad_norm": 0.3478544056415558, "learning_rate": 7.877550912566134e-05, "loss": 0.0086, "step": 101870 }, { "epoch": 64.93307839388146, "grad_norm": 0.4270361065864563, "learning_rate": 7.877100162181088e-05, "loss": 0.0164, "step": 101880 }, { "epoch": 64.93945188017845, "grad_norm": 0.3549550175666809, "learning_rate": 7.876649376836558e-05, "loss": 0.0211, "step": 101890 }, { "epoch": 64.94582536647546, "grad_norm": 0.3081165850162506, "learning_rate": 7.876198556538023e-05, "loss": 0.0119, "step": 101900 }, { "epoch": 64.95219885277247, "grad_norm": 0.8338729739189148, "learning_rate": 7.875747701290959e-05, "loss": 0.0133, "step": 101910 }, { "epoch": 64.95857233906948, "grad_norm": 0.3619285821914673, "learning_rate": 7.875296811100845e-05, "loss": 0.0106, "step": 101920 }, { "epoch": 64.96494582536647, "grad_norm": 0.3653522729873657, "learning_rate": 7.874845885973162e-05, "loss": 0.0226, "step": 101930 }, { "epoch": 64.97131931166348, "grad_norm": 0.3748699426651001, "learning_rate": 7.874394925913385e-05, "loss": 0.0183, "step": 101940 }, { "epoch": 64.97769279796049, "grad_norm": 0.6844147443771362, "learning_rate": 7.873943930926997e-05, "loss": 0.0212, "step": 101950 }, { "epoch": 64.9840662842575, "grad_norm": 0.2633829712867737, "learning_rate": 7.873492901019476e-05, "loss": 0.0184, "step": 101960 }, { "epoch": 64.99043977055449, "grad_norm": 0.42903703451156616, "learning_rate": 7.873041836196305e-05, "loss": 0.0195, "step": 101970 }, { "epoch": 64.9968132568515, "grad_norm": 0.4704042375087738, "learning_rate": 7.872590736462962e-05, "loss": 0.0136, "step": 101980 }, { "epoch": 65.0031867431485, "grad_norm": 0.20374086499214172, "learning_rate": 7.872139601824929e-05, "loss": 0.0311, "step": 101990 }, { "epoch": 65.00956022944551, "grad_norm": 0.5119425058364868, "learning_rate": 7.87168843228769e-05, "loss": 0.0437, "step": 102000 }, { "epoch": 65.0159337157425, "grad_norm": 0.31094542145729065, "learning_rate": 7.871237227856721e-05, "loss": 0.0227, "step": 102010 }, { "epoch": 65.02230720203951, "grad_norm": 0.26588964462280273, "learning_rate": 7.870785988537512e-05, "loss": 0.0239, "step": 102020 }, { "epoch": 65.02868068833652, "grad_norm": 0.3512263894081116, "learning_rate": 7.870334714335544e-05, "loss": 0.0283, "step": 102030 }, { "epoch": 65.03505417463353, "grad_norm": 0.3284313678741455, "learning_rate": 7.869883405256295e-05, "loss": 0.0172, "step": 102040 }, { "epoch": 65.04142766093052, "grad_norm": 0.42981940507888794, "learning_rate": 7.869432061305255e-05, "loss": 0.026, "step": 102050 }, { "epoch": 65.04780114722753, "grad_norm": 0.2298647165298462, "learning_rate": 7.868980682487907e-05, "loss": 0.0201, "step": 102060 }, { "epoch": 65.05417463352454, "grad_norm": 0.6069732308387756, "learning_rate": 7.868529268809735e-05, "loss": 0.0252, "step": 102070 }, { "epoch": 65.06054811982155, "grad_norm": 0.4655470848083496, "learning_rate": 7.868077820276223e-05, "loss": 0.0286, "step": 102080 }, { "epoch": 65.06692160611854, "grad_norm": 0.34539562463760376, "learning_rate": 7.867626336892858e-05, "loss": 0.0219, "step": 102090 }, { "epoch": 65.07329509241555, "grad_norm": 0.6529168486595154, "learning_rate": 7.867174818665126e-05, "loss": 0.0153, "step": 102100 }, { "epoch": 65.07966857871256, "grad_norm": 0.509239137172699, "learning_rate": 7.866723265598512e-05, "loss": 0.015, "step": 102110 }, { "epoch": 65.08604206500956, "grad_norm": 0.36583852767944336, "learning_rate": 7.866271677698504e-05, "loss": 0.0258, "step": 102120 }, { "epoch": 65.09241555130656, "grad_norm": 0.48289310932159424, "learning_rate": 7.86582005497059e-05, "loss": 0.0182, "step": 102130 }, { "epoch": 65.09878903760357, "grad_norm": 1.14234459400177, "learning_rate": 7.865368397420256e-05, "loss": 0.0324, "step": 102140 }, { "epoch": 65.10516252390057, "grad_norm": 0.47853198647499084, "learning_rate": 7.864916705052989e-05, "loss": 0.0196, "step": 102150 }, { "epoch": 65.11153601019758, "grad_norm": 0.3444884419441223, "learning_rate": 7.864464977874282e-05, "loss": 0.0194, "step": 102160 }, { "epoch": 65.11790949649458, "grad_norm": 0.3368479907512665, "learning_rate": 7.864013215889618e-05, "loss": 0.019, "step": 102170 }, { "epoch": 65.12428298279158, "grad_norm": 0.4835352897644043, "learning_rate": 7.863561419104491e-05, "loss": 0.0161, "step": 102180 }, { "epoch": 65.13065646908859, "grad_norm": 0.38151365518569946, "learning_rate": 7.863109587524389e-05, "loss": 0.0221, "step": 102190 }, { "epoch": 65.1370299553856, "grad_norm": 0.28853705525398254, "learning_rate": 7.862657721154802e-05, "loss": 0.0138, "step": 102200 }, { "epoch": 65.1434034416826, "grad_norm": 0.531181812286377, "learning_rate": 7.862205820001221e-05, "loss": 0.0143, "step": 102210 }, { "epoch": 65.1497769279796, "grad_norm": 0.5171051025390625, "learning_rate": 7.861753884069137e-05, "loss": 0.0161, "step": 102220 }, { "epoch": 65.15615041427661, "grad_norm": 0.47504138946533203, "learning_rate": 7.861301913364042e-05, "loss": 0.0233, "step": 102230 }, { "epoch": 65.16252390057362, "grad_norm": 0.4146222770214081, "learning_rate": 7.860849907891425e-05, "loss": 0.019, "step": 102240 }, { "epoch": 65.16889738687061, "grad_norm": 0.29952529072761536, "learning_rate": 7.860397867656784e-05, "loss": 0.0158, "step": 102250 }, { "epoch": 65.17527087316762, "grad_norm": 0.5573179125785828, "learning_rate": 7.859945792665605e-05, "loss": 0.0251, "step": 102260 }, { "epoch": 65.18164435946463, "grad_norm": 0.6797312498092651, "learning_rate": 7.859493682923387e-05, "loss": 0.0255, "step": 102270 }, { "epoch": 65.18801784576164, "grad_norm": 0.759575366973877, "learning_rate": 7.859041538435619e-05, "loss": 0.0262, "step": 102280 }, { "epoch": 65.19439133205863, "grad_norm": 0.6905320286750793, "learning_rate": 7.858589359207797e-05, "loss": 0.0268, "step": 102290 }, { "epoch": 65.20076481835564, "grad_norm": 0.23406296968460083, "learning_rate": 7.858137145245416e-05, "loss": 0.0173, "step": 102300 }, { "epoch": 65.20713830465264, "grad_norm": 0.30583304166793823, "learning_rate": 7.85768489655397e-05, "loss": 0.0194, "step": 102310 }, { "epoch": 65.21351179094965, "grad_norm": 0.44565093517303467, "learning_rate": 7.857232613138955e-05, "loss": 0.0297, "step": 102320 }, { "epoch": 65.21988527724666, "grad_norm": 0.4663126766681671, "learning_rate": 7.856780295005865e-05, "loss": 0.0253, "step": 102330 }, { "epoch": 65.22625876354365, "grad_norm": 0.43653979897499084, "learning_rate": 7.856327942160197e-05, "loss": 0.0131, "step": 102340 }, { "epoch": 65.23263224984066, "grad_norm": 0.5796127915382385, "learning_rate": 7.85587555460745e-05, "loss": 0.0215, "step": 102350 }, { "epoch": 65.23900573613767, "grad_norm": 0.5583221912384033, "learning_rate": 7.855423132353117e-05, "loss": 0.0191, "step": 102360 }, { "epoch": 65.24537922243468, "grad_norm": 0.5176437497138977, "learning_rate": 7.854970675402696e-05, "loss": 0.0215, "step": 102370 }, { "epoch": 65.25175270873167, "grad_norm": 0.49771931767463684, "learning_rate": 7.854518183761688e-05, "loss": 0.0131, "step": 102380 }, { "epoch": 65.25812619502868, "grad_norm": 0.4438225328922272, "learning_rate": 7.854065657435588e-05, "loss": 0.0395, "step": 102390 }, { "epoch": 65.26449968132569, "grad_norm": 0.7916486859321594, "learning_rate": 7.853613096429894e-05, "loss": 0.0188, "step": 102400 }, { "epoch": 65.2708731676227, "grad_norm": 0.37696781754493713, "learning_rate": 7.853160500750107e-05, "loss": 0.0146, "step": 102410 }, { "epoch": 65.27724665391969, "grad_norm": 0.2348046600818634, "learning_rate": 7.852707870401728e-05, "loss": 0.0212, "step": 102420 }, { "epoch": 65.2836201402167, "grad_norm": 0.25182101130485535, "learning_rate": 7.852255205390253e-05, "loss": 0.0163, "step": 102430 }, { "epoch": 65.2899936265137, "grad_norm": 0.16943129897117615, "learning_rate": 7.851802505721186e-05, "loss": 0.0223, "step": 102440 }, { "epoch": 65.29636711281071, "grad_norm": 0.3626856207847595, "learning_rate": 7.851349771400025e-05, "loss": 0.0194, "step": 102450 }, { "epoch": 65.30274059910771, "grad_norm": 0.3006613850593567, "learning_rate": 7.850897002432274e-05, "loss": 0.016, "step": 102460 }, { "epoch": 65.30911408540472, "grad_norm": 0.3690333664417267, "learning_rate": 7.850444198823431e-05, "loss": 0.0146, "step": 102470 }, { "epoch": 65.31548757170172, "grad_norm": 0.2569896876811981, "learning_rate": 7.849991360578999e-05, "loss": 0.0272, "step": 102480 }, { "epoch": 65.32186105799873, "grad_norm": 0.6101712584495544, "learning_rate": 7.849538487704483e-05, "loss": 0.0196, "step": 102490 }, { "epoch": 65.32823454429573, "grad_norm": 0.2981135845184326, "learning_rate": 7.849085580205382e-05, "loss": 0.0195, "step": 102500 }, { "epoch": 65.33460803059273, "grad_norm": 0.4539845287799835, "learning_rate": 7.848632638087202e-05, "loss": 0.021, "step": 102510 }, { "epoch": 65.34098151688974, "grad_norm": 0.25157296657562256, "learning_rate": 7.848179661355446e-05, "loss": 0.0261, "step": 102520 }, { "epoch": 65.34735500318675, "grad_norm": 0.3460191786289215, "learning_rate": 7.847726650015619e-05, "loss": 0.0248, "step": 102530 }, { "epoch": 65.35372848948374, "grad_norm": 0.37448257207870483, "learning_rate": 7.847273604073222e-05, "loss": 0.0184, "step": 102540 }, { "epoch": 65.36010197578075, "grad_norm": 0.48398613929748535, "learning_rate": 7.846820523533764e-05, "loss": 0.0268, "step": 102550 }, { "epoch": 65.36647546207776, "grad_norm": 0.41866376996040344, "learning_rate": 7.846367408402749e-05, "loss": 0.0241, "step": 102560 }, { "epoch": 65.37284894837477, "grad_norm": 0.43720099329948425, "learning_rate": 7.845914258685682e-05, "loss": 0.0171, "step": 102570 }, { "epoch": 65.37922243467176, "grad_norm": 0.1811823844909668, "learning_rate": 7.84546107438807e-05, "loss": 0.0181, "step": 102580 }, { "epoch": 65.38559592096877, "grad_norm": 0.35956311225891113, "learning_rate": 7.84500785551542e-05, "loss": 0.0163, "step": 102590 }, { "epoch": 65.39196940726578, "grad_norm": 0.3005768954753876, "learning_rate": 7.844554602073238e-05, "loss": 0.012, "step": 102600 }, { "epoch": 65.39834289356278, "grad_norm": 0.529806911945343, "learning_rate": 7.844101314067032e-05, "loss": 0.0276, "step": 102610 }, { "epoch": 65.40471637985978, "grad_norm": 0.7234546542167664, "learning_rate": 7.843647991502309e-05, "loss": 0.0185, "step": 102620 }, { "epoch": 65.41108986615679, "grad_norm": 0.676247239112854, "learning_rate": 7.843194634384578e-05, "loss": 0.0211, "step": 102630 }, { "epoch": 65.4174633524538, "grad_norm": 0.5337989330291748, "learning_rate": 7.84274124271935e-05, "loss": 0.0226, "step": 102640 }, { "epoch": 65.4238368387508, "grad_norm": 0.3495458960533142, "learning_rate": 7.84228781651213e-05, "loss": 0.0216, "step": 102650 }, { "epoch": 65.4302103250478, "grad_norm": 0.2972962558269501, "learning_rate": 7.841834355768429e-05, "loss": 0.0201, "step": 102660 }, { "epoch": 65.4365838113448, "grad_norm": 0.73860102891922, "learning_rate": 7.841380860493759e-05, "loss": 0.0193, "step": 102670 }, { "epoch": 65.44295729764181, "grad_norm": 0.5754537582397461, "learning_rate": 7.840927330693629e-05, "loss": 0.0127, "step": 102680 }, { "epoch": 65.44933078393882, "grad_norm": 1.4802402257919312, "learning_rate": 7.840473766373548e-05, "loss": 0.0342, "step": 102690 }, { "epoch": 65.45570427023581, "grad_norm": 0.335588663816452, "learning_rate": 7.840020167539028e-05, "loss": 0.0184, "step": 102700 }, { "epoch": 65.46207775653282, "grad_norm": 0.6408677697181702, "learning_rate": 7.839566534195584e-05, "loss": 0.015, "step": 102710 }, { "epoch": 65.46845124282983, "grad_norm": 0.4173913896083832, "learning_rate": 7.839112866348724e-05, "loss": 0.0187, "step": 102720 }, { "epoch": 65.47482472912684, "grad_norm": 0.3144106864929199, "learning_rate": 7.838659164003963e-05, "loss": 0.0181, "step": 102730 }, { "epoch": 65.48119821542383, "grad_norm": 0.2749316096305847, "learning_rate": 7.838205427166814e-05, "loss": 0.031, "step": 102740 }, { "epoch": 65.48757170172084, "grad_norm": 0.2788013517856598, "learning_rate": 7.837751655842788e-05, "loss": 0.0169, "step": 102750 }, { "epoch": 65.49394518801785, "grad_norm": 0.4586140513420105, "learning_rate": 7.837297850037399e-05, "loss": 0.0181, "step": 102760 }, { "epoch": 65.50031867431485, "grad_norm": 0.4848775863647461, "learning_rate": 7.836844009756164e-05, "loss": 0.0214, "step": 102770 }, { "epoch": 65.50669216061185, "grad_norm": 0.42470505833625793, "learning_rate": 7.836390135004596e-05, "loss": 0.0204, "step": 102780 }, { "epoch": 65.51306564690886, "grad_norm": 0.3330576717853546, "learning_rate": 7.83593622578821e-05, "loss": 0.0182, "step": 102790 }, { "epoch": 65.51943913320586, "grad_norm": 0.3691929876804352, "learning_rate": 7.83548228211252e-05, "loss": 0.02, "step": 102800 }, { "epoch": 65.52581261950287, "grad_norm": 0.6243945956230164, "learning_rate": 7.835028303983043e-05, "loss": 0.0318, "step": 102810 }, { "epoch": 65.53218610579987, "grad_norm": 0.9150430560112, "learning_rate": 7.834574291405297e-05, "loss": 0.0233, "step": 102820 }, { "epoch": 65.53855959209687, "grad_norm": 0.38269171118736267, "learning_rate": 7.834120244384795e-05, "loss": 0.0232, "step": 102830 }, { "epoch": 65.54493307839388, "grad_norm": 0.4879479706287384, "learning_rate": 7.833666162927058e-05, "loss": 0.0186, "step": 102840 }, { "epoch": 65.55130656469089, "grad_norm": 0.4995993673801422, "learning_rate": 7.833212047037602e-05, "loss": 0.0203, "step": 102850 }, { "epoch": 65.55768005098788, "grad_norm": 0.6419657468795776, "learning_rate": 7.832757896721943e-05, "loss": 0.0196, "step": 102860 }, { "epoch": 65.56405353728489, "grad_norm": 0.48423707485198975, "learning_rate": 7.832303711985602e-05, "loss": 0.0204, "step": 102870 }, { "epoch": 65.5704270235819, "grad_norm": 0.3181213140487671, "learning_rate": 7.831849492834096e-05, "loss": 0.023, "step": 102880 }, { "epoch": 65.57680050987891, "grad_norm": 0.4334322214126587, "learning_rate": 7.831395239272945e-05, "loss": 0.0199, "step": 102890 }, { "epoch": 65.5831739961759, "grad_norm": 0.4734525680541992, "learning_rate": 7.830940951307669e-05, "loss": 0.0179, "step": 102900 }, { "epoch": 65.58954748247291, "grad_norm": 0.3188893795013428, "learning_rate": 7.830486628943788e-05, "loss": 0.0252, "step": 102910 }, { "epoch": 65.59592096876992, "grad_norm": 0.36049970984458923, "learning_rate": 7.830032272186822e-05, "loss": 0.0275, "step": 102920 }, { "epoch": 65.60229445506693, "grad_norm": 0.4017672836780548, "learning_rate": 7.829577881042291e-05, "loss": 0.0203, "step": 102930 }, { "epoch": 65.60866794136393, "grad_norm": 0.7531338334083557, "learning_rate": 7.829123455515718e-05, "loss": 0.0281, "step": 102940 }, { "epoch": 65.61504142766093, "grad_norm": 0.28667938709259033, "learning_rate": 7.828668995612622e-05, "loss": 0.0245, "step": 102950 }, { "epoch": 65.62141491395793, "grad_norm": 0.4678466320037842, "learning_rate": 7.82821450133853e-05, "loss": 0.0271, "step": 102960 }, { "epoch": 65.62778840025494, "grad_norm": 0.5310764908790588, "learning_rate": 7.82775997269896e-05, "loss": 0.0188, "step": 102970 }, { "epoch": 65.63416188655195, "grad_norm": 0.5667780637741089, "learning_rate": 7.827305409699437e-05, "loss": 0.0163, "step": 102980 }, { "epoch": 65.64053537284894, "grad_norm": 0.2960823178291321, "learning_rate": 7.826850812345483e-05, "loss": 0.0131, "step": 102990 }, { "epoch": 65.64690885914595, "grad_norm": 0.28298553824424744, "learning_rate": 7.826396180642623e-05, "loss": 0.0151, "step": 103000 }, { "epoch": 65.65328234544296, "grad_norm": 0.3857429325580597, "learning_rate": 7.825941514596381e-05, "loss": 0.0146, "step": 103010 }, { "epoch": 65.65965583173997, "grad_norm": 0.24088533222675323, "learning_rate": 7.825486814212282e-05, "loss": 0.0162, "step": 103020 }, { "epoch": 65.66602931803696, "grad_norm": 0.3241817057132721, "learning_rate": 7.825032079495849e-05, "loss": 0.0283, "step": 103030 }, { "epoch": 65.67240280433397, "grad_norm": 0.371898353099823, "learning_rate": 7.82457731045261e-05, "loss": 0.018, "step": 103040 }, { "epoch": 65.67877629063098, "grad_norm": 0.6257537603378296, "learning_rate": 7.82412250708809e-05, "loss": 0.0195, "step": 103050 }, { "epoch": 65.68514977692799, "grad_norm": 0.533287763595581, "learning_rate": 7.823667669407815e-05, "loss": 0.0169, "step": 103060 }, { "epoch": 65.69152326322498, "grad_norm": 0.34694167971611023, "learning_rate": 7.823212797417311e-05, "loss": 0.0193, "step": 103070 }, { "epoch": 65.69789674952199, "grad_norm": 0.30658912658691406, "learning_rate": 7.822757891122107e-05, "loss": 0.0129, "step": 103080 }, { "epoch": 65.704270235819, "grad_norm": 0.3185724914073944, "learning_rate": 7.822302950527729e-05, "loss": 0.0192, "step": 103090 }, { "epoch": 65.710643722116, "grad_norm": 0.7462679743766785, "learning_rate": 7.821847975639705e-05, "loss": 0.018, "step": 103100 }, { "epoch": 65.717017208413, "grad_norm": 0.29510313272476196, "learning_rate": 7.821392966463564e-05, "loss": 0.0135, "step": 103110 }, { "epoch": 65.72339069471, "grad_norm": 0.32646939158439636, "learning_rate": 7.820937923004835e-05, "loss": 0.0176, "step": 103120 }, { "epoch": 65.72976418100701, "grad_norm": 0.5003933906555176, "learning_rate": 7.820482845269047e-05, "loss": 0.0204, "step": 103130 }, { "epoch": 65.73613766730402, "grad_norm": 0.5240833163261414, "learning_rate": 7.82002773326173e-05, "loss": 0.0407, "step": 103140 }, { "epoch": 65.74251115360101, "grad_norm": 0.349303275346756, "learning_rate": 7.819572586988413e-05, "loss": 0.0262, "step": 103150 }, { "epoch": 65.74888463989802, "grad_norm": 0.27445346117019653, "learning_rate": 7.819117406454626e-05, "loss": 0.0268, "step": 103160 }, { "epoch": 65.75525812619503, "grad_norm": 0.5163291692733765, "learning_rate": 7.818662191665901e-05, "loss": 0.0201, "step": 103170 }, { "epoch": 65.76163161249204, "grad_norm": 0.5932561159133911, "learning_rate": 7.81820694262777e-05, "loss": 0.0396, "step": 103180 }, { "epoch": 65.76800509878903, "grad_norm": 0.3425849974155426, "learning_rate": 7.817751659345765e-05, "loss": 0.0215, "step": 103190 }, { "epoch": 65.77437858508604, "grad_norm": 0.29944124817848206, "learning_rate": 7.817296341825413e-05, "loss": 0.0173, "step": 103200 }, { "epoch": 65.78075207138305, "grad_norm": 0.3647506535053253, "learning_rate": 7.816840990072254e-05, "loss": 0.0172, "step": 103210 }, { "epoch": 65.78712555768006, "grad_norm": 0.34036916494369507, "learning_rate": 7.816385604091815e-05, "loss": 0.0197, "step": 103220 }, { "epoch": 65.79349904397705, "grad_norm": 0.5521475672721863, "learning_rate": 7.815930183889635e-05, "loss": 0.0252, "step": 103230 }, { "epoch": 65.79987253027406, "grad_norm": 0.6016308665275574, "learning_rate": 7.815474729471241e-05, "loss": 0.0193, "step": 103240 }, { "epoch": 65.80624601657107, "grad_norm": 0.5104973316192627, "learning_rate": 7.815019240842174e-05, "loss": 0.0156, "step": 103250 }, { "epoch": 65.81261950286807, "grad_norm": 0.34524697065353394, "learning_rate": 7.814563718007963e-05, "loss": 0.0161, "step": 103260 }, { "epoch": 65.81899298916507, "grad_norm": 0.43747833371162415, "learning_rate": 7.814108160974148e-05, "loss": 0.0166, "step": 103270 }, { "epoch": 65.82536647546208, "grad_norm": 0.30830204486846924, "learning_rate": 7.81365256974626e-05, "loss": 0.017, "step": 103280 }, { "epoch": 65.83173996175908, "grad_norm": 0.4012058973312378, "learning_rate": 7.813196944329836e-05, "loss": 0.0144, "step": 103290 }, { "epoch": 65.83811344805609, "grad_norm": 0.3441618084907532, "learning_rate": 7.812741284730413e-05, "loss": 0.0174, "step": 103300 }, { "epoch": 65.84448693435309, "grad_norm": 0.4442175626754761, "learning_rate": 7.81228559095353e-05, "loss": 0.0178, "step": 103310 }, { "epoch": 65.8508604206501, "grad_norm": 0.35823187232017517, "learning_rate": 7.81182986300472e-05, "loss": 0.0166, "step": 103320 }, { "epoch": 65.8572339069471, "grad_norm": 0.522636353969574, "learning_rate": 7.811374100889524e-05, "loss": 0.0134, "step": 103330 }, { "epoch": 65.86360739324411, "grad_norm": 0.355129212141037, "learning_rate": 7.810918304613478e-05, "loss": 0.0223, "step": 103340 }, { "epoch": 65.8699808795411, "grad_norm": 0.3701978921890259, "learning_rate": 7.810462474182119e-05, "loss": 0.0246, "step": 103350 }, { "epoch": 65.87635436583811, "grad_norm": 0.6545206308364868, "learning_rate": 7.810006609600988e-05, "loss": 0.0212, "step": 103360 }, { "epoch": 65.88272785213512, "grad_norm": 0.4972892701625824, "learning_rate": 7.809550710875624e-05, "loss": 0.0277, "step": 103370 }, { "epoch": 65.88910133843213, "grad_norm": 0.5639822483062744, "learning_rate": 7.809094778011565e-05, "loss": 0.0231, "step": 103380 }, { "epoch": 65.89547482472912, "grad_norm": 0.44901537895202637, "learning_rate": 7.808638811014354e-05, "loss": 0.0316, "step": 103390 }, { "epoch": 65.90184831102613, "grad_norm": 0.46642786264419556, "learning_rate": 7.80818280988953e-05, "loss": 0.0197, "step": 103400 }, { "epoch": 65.90822179732314, "grad_norm": 0.30104172229766846, "learning_rate": 7.807726774642631e-05, "loss": 0.0192, "step": 103410 }, { "epoch": 65.91459528362014, "grad_norm": 0.6449616551399231, "learning_rate": 7.807270705279204e-05, "loss": 0.0157, "step": 103420 }, { "epoch": 65.92096876991714, "grad_norm": 0.2343538999557495, "learning_rate": 7.806814601804784e-05, "loss": 0.0211, "step": 103430 }, { "epoch": 65.92734225621415, "grad_norm": 0.6807414293289185, "learning_rate": 7.806358464224919e-05, "loss": 0.034, "step": 103440 }, { "epoch": 65.93371574251115, "grad_norm": 0.24237528443336487, "learning_rate": 7.805902292545147e-05, "loss": 0.0125, "step": 103450 }, { "epoch": 65.94008922880816, "grad_norm": 0.6697959899902344, "learning_rate": 7.805446086771016e-05, "loss": 0.0189, "step": 103460 }, { "epoch": 65.94646271510516, "grad_norm": 0.46799135208129883, "learning_rate": 7.804989846908063e-05, "loss": 0.017, "step": 103470 }, { "epoch": 65.95283620140216, "grad_norm": 0.268690288066864, "learning_rate": 7.804533572961836e-05, "loss": 0.0162, "step": 103480 }, { "epoch": 65.95920968769917, "grad_norm": 0.32006779313087463, "learning_rate": 7.80407726493788e-05, "loss": 0.0176, "step": 103490 }, { "epoch": 65.96558317399618, "grad_norm": 0.5648157596588135, "learning_rate": 7.803620922841736e-05, "loss": 0.0271, "step": 103500 }, { "epoch": 65.97195666029317, "grad_norm": 0.3454248309135437, "learning_rate": 7.80316454667895e-05, "loss": 0.0123, "step": 103510 }, { "epoch": 65.97833014659018, "grad_norm": 0.7566028833389282, "learning_rate": 7.802708136455071e-05, "loss": 0.0166, "step": 103520 }, { "epoch": 65.98470363288719, "grad_norm": 0.24555304646492004, "learning_rate": 7.802251692175638e-05, "loss": 0.0254, "step": 103530 }, { "epoch": 65.9910771191842, "grad_norm": 0.7986894845962524, "learning_rate": 7.801795213846204e-05, "loss": 0.0191, "step": 103540 }, { "epoch": 65.99745060548119, "grad_norm": 0.38679730892181396, "learning_rate": 7.801338701472311e-05, "loss": 0.0176, "step": 103550 }, { "epoch": 66.0038240917782, "grad_norm": 0.9482989311218262, "learning_rate": 7.800882155059511e-05, "loss": 0.0281, "step": 103560 }, { "epoch": 66.0101975780752, "grad_norm": 0.27796244621276855, "learning_rate": 7.800425574613345e-05, "loss": 0.015, "step": 103570 }, { "epoch": 66.01657106437222, "grad_norm": 0.2370956540107727, "learning_rate": 7.799968960139366e-05, "loss": 0.0151, "step": 103580 }, { "epoch": 66.02294455066922, "grad_norm": 0.3358617126941681, "learning_rate": 7.79951231164312e-05, "loss": 0.0269, "step": 103590 }, { "epoch": 66.02931803696622, "grad_norm": 0.5765673518180847, "learning_rate": 7.799055629130156e-05, "loss": 0.0155, "step": 103600 }, { "epoch": 66.03569152326322, "grad_norm": 0.4407922923564911, "learning_rate": 7.798598912606024e-05, "loss": 0.0221, "step": 103610 }, { "epoch": 66.04206500956023, "grad_norm": 0.325094074010849, "learning_rate": 7.798142162076273e-05, "loss": 0.0192, "step": 103620 }, { "epoch": 66.04843849585724, "grad_norm": 0.45933645963668823, "learning_rate": 7.797685377546451e-05, "loss": 0.0167, "step": 103630 }, { "epoch": 66.05481198215423, "grad_norm": 0.25464367866516113, "learning_rate": 7.797228559022113e-05, "loss": 0.0132, "step": 103640 }, { "epoch": 66.06118546845124, "grad_norm": 0.5071162581443787, "learning_rate": 7.796771706508805e-05, "loss": 0.0378, "step": 103650 }, { "epoch": 66.06755895474825, "grad_norm": 0.3177618384361267, "learning_rate": 7.796314820012081e-05, "loss": 0.0204, "step": 103660 }, { "epoch": 66.07393244104526, "grad_norm": 0.2850005328655243, "learning_rate": 7.795857899537492e-05, "loss": 0.0181, "step": 103670 }, { "epoch": 66.08030592734225, "grad_norm": 0.5369991660118103, "learning_rate": 7.79540094509059e-05, "loss": 0.0162, "step": 103680 }, { "epoch": 66.08667941363926, "grad_norm": 0.3731865882873535, "learning_rate": 7.794943956676925e-05, "loss": 0.0142, "step": 103690 }, { "epoch": 66.09305289993627, "grad_norm": 0.4750994145870209, "learning_rate": 7.794486934302054e-05, "loss": 0.0215, "step": 103700 }, { "epoch": 66.09942638623328, "grad_norm": 0.26797664165496826, "learning_rate": 7.794029877971529e-05, "loss": 0.0121, "step": 103710 }, { "epoch": 66.10579987253027, "grad_norm": 0.4954093098640442, "learning_rate": 7.793572787690902e-05, "loss": 0.0186, "step": 103720 }, { "epoch": 66.11217335882728, "grad_norm": 0.6190300583839417, "learning_rate": 7.793115663465727e-05, "loss": 0.0271, "step": 103730 }, { "epoch": 66.11854684512429, "grad_norm": 0.7718958258628845, "learning_rate": 7.792658505301561e-05, "loss": 0.0153, "step": 103740 }, { "epoch": 66.1249203314213, "grad_norm": 0.637861430644989, "learning_rate": 7.792201313203957e-05, "loss": 0.0153, "step": 103750 }, { "epoch": 66.13129381771829, "grad_norm": 0.39405620098114014, "learning_rate": 7.791744087178471e-05, "loss": 0.0174, "step": 103760 }, { "epoch": 66.1376673040153, "grad_norm": 0.2939327657222748, "learning_rate": 7.791286827230657e-05, "loss": 0.0184, "step": 103770 }, { "epoch": 66.1440407903123, "grad_norm": 0.3212132751941681, "learning_rate": 7.790829533366075e-05, "loss": 0.018, "step": 103780 }, { "epoch": 66.15041427660931, "grad_norm": 0.30288732051849365, "learning_rate": 7.790372205590277e-05, "loss": 0.0174, "step": 103790 }, { "epoch": 66.1567877629063, "grad_norm": 0.30808648467063904, "learning_rate": 7.789914843908824e-05, "loss": 0.02, "step": 103800 }, { "epoch": 66.16316124920331, "grad_norm": 0.21060657501220703, "learning_rate": 7.78945744832727e-05, "loss": 0.0325, "step": 103810 }, { "epoch": 66.16953473550032, "grad_norm": 0.48340728878974915, "learning_rate": 7.789000018851176e-05, "loss": 0.0148, "step": 103820 }, { "epoch": 66.17590822179733, "grad_norm": 0.30996447801589966, "learning_rate": 7.788542555486098e-05, "loss": 0.0184, "step": 103830 }, { "epoch": 66.18228170809432, "grad_norm": 0.4478999078273773, "learning_rate": 7.788085058237594e-05, "loss": 0.0191, "step": 103840 }, { "epoch": 66.18865519439133, "grad_norm": 0.6804331541061401, "learning_rate": 7.787627527111226e-05, "loss": 0.0189, "step": 103850 }, { "epoch": 66.19502868068834, "grad_norm": 0.8778570890426636, "learning_rate": 7.78716996211255e-05, "loss": 0.0174, "step": 103860 }, { "epoch": 66.20140216698535, "grad_norm": 0.680188775062561, "learning_rate": 7.786712363247129e-05, "loss": 0.0235, "step": 103870 }, { "epoch": 66.20777565328234, "grad_norm": 0.24531130492687225, "learning_rate": 7.786254730520519e-05, "loss": 0.0239, "step": 103880 }, { "epoch": 66.21414913957935, "grad_norm": 0.3847886919975281, "learning_rate": 7.785797063938284e-05, "loss": 0.0148, "step": 103890 }, { "epoch": 66.22052262587636, "grad_norm": 0.6601198315620422, "learning_rate": 7.785339363505987e-05, "loss": 0.0264, "step": 103900 }, { "epoch": 66.22689611217336, "grad_norm": 0.5291872620582581, "learning_rate": 7.784881629229185e-05, "loss": 0.0168, "step": 103910 }, { "epoch": 66.23326959847036, "grad_norm": 0.5817005634307861, "learning_rate": 7.784423861113443e-05, "loss": 0.0205, "step": 103920 }, { "epoch": 66.23964308476737, "grad_norm": 0.468478798866272, "learning_rate": 7.78396605916432e-05, "loss": 0.017, "step": 103930 }, { "epoch": 66.24601657106437, "grad_norm": 0.3667430579662323, "learning_rate": 7.783508223387383e-05, "loss": 0.0132, "step": 103940 }, { "epoch": 66.25239005736138, "grad_norm": 0.40649276971817017, "learning_rate": 7.783050353788193e-05, "loss": 0.0198, "step": 103950 }, { "epoch": 66.25876354365838, "grad_norm": 0.42306339740753174, "learning_rate": 7.782592450372312e-05, "loss": 0.0313, "step": 103960 }, { "epoch": 66.26513702995538, "grad_norm": 0.5502456426620483, "learning_rate": 7.782134513145307e-05, "loss": 0.0148, "step": 103970 }, { "epoch": 66.27151051625239, "grad_norm": 0.4431155323982239, "learning_rate": 7.781676542112739e-05, "loss": 0.0303, "step": 103980 }, { "epoch": 66.2778840025494, "grad_norm": 0.3459029197692871, "learning_rate": 7.781218537280178e-05, "loss": 0.0165, "step": 103990 }, { "epoch": 66.2842574888464, "grad_norm": 0.38782626390457153, "learning_rate": 7.780760498653183e-05, "loss": 0.0175, "step": 104000 }, { "epoch": 66.2906309751434, "grad_norm": 0.2680876851081848, "learning_rate": 7.780302426237323e-05, "loss": 0.0226, "step": 104010 }, { "epoch": 66.29700446144041, "grad_norm": 0.326782763004303, "learning_rate": 7.779844320038163e-05, "loss": 0.0203, "step": 104020 }, { "epoch": 66.30337794773742, "grad_norm": 0.7530038356781006, "learning_rate": 7.779386180061272e-05, "loss": 0.0271, "step": 104030 }, { "epoch": 66.30975143403441, "grad_norm": 0.2513943314552307, "learning_rate": 7.778928006312212e-05, "loss": 0.0174, "step": 104040 }, { "epoch": 66.31612492033142, "grad_norm": 0.35788676142692566, "learning_rate": 7.778469798796555e-05, "loss": 0.0274, "step": 104050 }, { "epoch": 66.32249840662843, "grad_norm": 0.290964275598526, "learning_rate": 7.778011557519867e-05, "loss": 0.0112, "step": 104060 }, { "epoch": 66.32887189292543, "grad_norm": 0.389392614364624, "learning_rate": 7.777553282487714e-05, "loss": 0.0259, "step": 104070 }, { "epoch": 66.33524537922243, "grad_norm": 0.32394707202911377, "learning_rate": 7.777094973705668e-05, "loss": 0.0219, "step": 104080 }, { "epoch": 66.34161886551944, "grad_norm": 0.5177465677261353, "learning_rate": 7.776636631179292e-05, "loss": 0.0188, "step": 104090 }, { "epoch": 66.34799235181644, "grad_norm": 0.6246376633644104, "learning_rate": 7.776178254914163e-05, "loss": 0.0192, "step": 104100 }, { "epoch": 66.35436583811345, "grad_norm": 0.2061404436826706, "learning_rate": 7.775719844915848e-05, "loss": 0.0175, "step": 104110 }, { "epoch": 66.36073932441045, "grad_norm": 0.5537052154541016, "learning_rate": 7.775261401189914e-05, "loss": 0.0189, "step": 104120 }, { "epoch": 66.36711281070745, "grad_norm": 0.38510414958000183, "learning_rate": 7.774802923741934e-05, "loss": 0.0372, "step": 104130 }, { "epoch": 66.37348629700446, "grad_norm": 0.30153512954711914, "learning_rate": 7.774344412577479e-05, "loss": 0.0149, "step": 104140 }, { "epoch": 66.37985978330147, "grad_norm": 0.5645362138748169, "learning_rate": 7.77388586770212e-05, "loss": 0.0263, "step": 104150 }, { "epoch": 66.38623326959846, "grad_norm": 0.2406090497970581, "learning_rate": 7.773427289121429e-05, "loss": 0.0126, "step": 104160 }, { "epoch": 66.39260675589547, "grad_norm": 0.29537901282310486, "learning_rate": 7.772968676840977e-05, "loss": 0.0162, "step": 104170 }, { "epoch": 66.39898024219248, "grad_norm": 0.33754897117614746, "learning_rate": 7.772510030866339e-05, "loss": 0.0227, "step": 104180 }, { "epoch": 66.40535372848949, "grad_norm": 0.6791728734970093, "learning_rate": 7.772051351203085e-05, "loss": 0.0172, "step": 104190 }, { "epoch": 66.4117272147865, "grad_norm": 0.4084722101688385, "learning_rate": 7.771592637856789e-05, "loss": 0.0123, "step": 104200 }, { "epoch": 66.41810070108349, "grad_norm": 0.31572774052619934, "learning_rate": 7.771133890833026e-05, "loss": 0.017, "step": 104210 }, { "epoch": 66.4244741873805, "grad_norm": 0.3071862459182739, "learning_rate": 7.770675110137371e-05, "loss": 0.0148, "step": 104220 }, { "epoch": 66.4308476736775, "grad_norm": 0.2591749131679535, "learning_rate": 7.770216295775398e-05, "loss": 0.0138, "step": 104230 }, { "epoch": 66.43722115997451, "grad_norm": 0.2720262408256531, "learning_rate": 7.76975744775268e-05, "loss": 0.0157, "step": 104240 }, { "epoch": 66.4435946462715, "grad_norm": 0.2967155873775482, "learning_rate": 7.769298566074794e-05, "loss": 0.0108, "step": 104250 }, { "epoch": 66.44996813256851, "grad_norm": 0.4368475377559662, "learning_rate": 7.768839650747317e-05, "loss": 0.0129, "step": 104260 }, { "epoch": 66.45634161886552, "grad_norm": 0.3891008496284485, "learning_rate": 7.768380701775822e-05, "loss": 0.0301, "step": 104270 }, { "epoch": 66.46271510516253, "grad_norm": 0.5583856701850891, "learning_rate": 7.767921719165889e-05, "loss": 0.0162, "step": 104280 }, { "epoch": 66.46908859145952, "grad_norm": 0.37496843934059143, "learning_rate": 7.767462702923095e-05, "loss": 0.0138, "step": 104290 }, { "epoch": 66.47546207775653, "grad_norm": 0.5572828650474548, "learning_rate": 7.767003653053015e-05, "loss": 0.0148, "step": 104300 }, { "epoch": 66.48183556405354, "grad_norm": 0.43109455704689026, "learning_rate": 7.766544569561228e-05, "loss": 0.0178, "step": 104310 }, { "epoch": 66.48820905035055, "grad_norm": 0.42166683077812195, "learning_rate": 7.766085452453312e-05, "loss": 0.0221, "step": 104320 }, { "epoch": 66.49458253664754, "grad_norm": 0.4453985095024109, "learning_rate": 7.765626301734848e-05, "loss": 0.0188, "step": 104330 }, { "epoch": 66.50095602294455, "grad_norm": 0.28248581290245056, "learning_rate": 7.765167117411412e-05, "loss": 0.0296, "step": 104340 }, { "epoch": 66.50732950924156, "grad_norm": 0.29273292422294617, "learning_rate": 7.764707899488584e-05, "loss": 0.0175, "step": 104350 }, { "epoch": 66.51370299553857, "grad_norm": 0.5694992542266846, "learning_rate": 7.764248647971945e-05, "loss": 0.0215, "step": 104360 }, { "epoch": 66.52007648183556, "grad_norm": 0.4754321873188019, "learning_rate": 7.763789362867075e-05, "loss": 0.0118, "step": 104370 }, { "epoch": 66.52644996813257, "grad_norm": 0.35538235306739807, "learning_rate": 7.763330044179557e-05, "loss": 0.0204, "step": 104380 }, { "epoch": 66.53282345442958, "grad_norm": 0.44605621695518494, "learning_rate": 7.762870691914968e-05, "loss": 0.0183, "step": 104390 }, { "epoch": 66.53919694072658, "grad_norm": 0.3377116322517395, "learning_rate": 7.762411306078892e-05, "loss": 0.0176, "step": 104400 }, { "epoch": 66.54557042702358, "grad_norm": 0.3596092462539673, "learning_rate": 7.761951886676908e-05, "loss": 0.0264, "step": 104410 }, { "epoch": 66.55194391332059, "grad_norm": 0.5701419115066528, "learning_rate": 7.761492433714604e-05, "loss": 0.0244, "step": 104420 }, { "epoch": 66.5583173996176, "grad_norm": 0.45920702815055847, "learning_rate": 7.761032947197558e-05, "loss": 0.0265, "step": 104430 }, { "epoch": 66.5646908859146, "grad_norm": 0.34056341648101807, "learning_rate": 7.760573427131356e-05, "loss": 0.0145, "step": 104440 }, { "epoch": 66.5710643722116, "grad_norm": 0.48435258865356445, "learning_rate": 7.760113873521581e-05, "loss": 0.0253, "step": 104450 }, { "epoch": 66.5774378585086, "grad_norm": 0.32668203115463257, "learning_rate": 7.759654286373814e-05, "loss": 0.0199, "step": 104460 }, { "epoch": 66.58381134480561, "grad_norm": 0.2782791256904602, "learning_rate": 7.759194665693646e-05, "loss": 0.0306, "step": 104470 }, { "epoch": 66.59018483110262, "grad_norm": 0.31007471680641174, "learning_rate": 7.758735011486653e-05, "loss": 0.0252, "step": 104480 }, { "epoch": 66.59655831739961, "grad_norm": 0.5312166810035706, "learning_rate": 7.758275323758429e-05, "loss": 0.011, "step": 104490 }, { "epoch": 66.60293180369662, "grad_norm": 0.33997178077697754, "learning_rate": 7.757815602514552e-05, "loss": 0.0185, "step": 104500 }, { "epoch": 66.60930528999363, "grad_norm": 0.30017730593681335, "learning_rate": 7.757355847760613e-05, "loss": 0.025, "step": 104510 }, { "epoch": 66.61567877629064, "grad_norm": 0.24849814176559448, "learning_rate": 7.756896059502197e-05, "loss": 0.0157, "step": 104520 }, { "epoch": 66.62205226258763, "grad_norm": 0.4428296387195587, "learning_rate": 7.756436237744891e-05, "loss": 0.0236, "step": 104530 }, { "epoch": 66.62842574888464, "grad_norm": 0.3234385848045349, "learning_rate": 7.755976382494284e-05, "loss": 0.0163, "step": 104540 }, { "epoch": 66.63479923518165, "grad_norm": 0.5087635517120361, "learning_rate": 7.75551649375596e-05, "loss": 0.0135, "step": 104550 }, { "epoch": 66.64117272147865, "grad_norm": 0.3576057553291321, "learning_rate": 7.755056571535508e-05, "loss": 0.0141, "step": 104560 }, { "epoch": 66.64754620777565, "grad_norm": 0.2613471448421478, "learning_rate": 7.754596615838518e-05, "loss": 0.0157, "step": 104570 }, { "epoch": 66.65391969407266, "grad_norm": 0.5001068711280823, "learning_rate": 7.754136626670579e-05, "loss": 0.0193, "step": 104580 }, { "epoch": 66.66029318036966, "grad_norm": 0.3789062798023224, "learning_rate": 7.753676604037282e-05, "loss": 0.0181, "step": 104590 }, { "epoch": 66.66666666666667, "grad_norm": 0.20813186466693878, "learning_rate": 7.753216547944211e-05, "loss": 0.031, "step": 104600 }, { "epoch": 66.67304015296367, "grad_norm": 0.35968807339668274, "learning_rate": 7.75275645839696e-05, "loss": 0.0227, "step": 104610 }, { "epoch": 66.67941363926067, "grad_norm": 0.5767723321914673, "learning_rate": 7.75229633540112e-05, "loss": 0.0244, "step": 104620 }, { "epoch": 66.68578712555768, "grad_norm": 0.6653258204460144, "learning_rate": 7.75183617896228e-05, "loss": 0.0301, "step": 104630 }, { "epoch": 66.69216061185469, "grad_norm": 0.28615257143974304, "learning_rate": 7.751375989086032e-05, "loss": 0.0228, "step": 104640 }, { "epoch": 66.69853409815168, "grad_norm": 0.4136192500591278, "learning_rate": 7.75091576577797e-05, "loss": 0.0121, "step": 104650 }, { "epoch": 66.70490758444869, "grad_norm": 0.567590594291687, "learning_rate": 7.750455509043681e-05, "loss": 0.029, "step": 104660 }, { "epoch": 66.7112810707457, "grad_norm": 0.3617996573448181, "learning_rate": 7.749995218888763e-05, "loss": 0.0262, "step": 104670 }, { "epoch": 66.7176545570427, "grad_norm": 0.5951144099235535, "learning_rate": 7.749534895318806e-05, "loss": 0.0099, "step": 104680 }, { "epoch": 66.7240280433397, "grad_norm": 0.501395583152771, "learning_rate": 7.749074538339405e-05, "loss": 0.0183, "step": 104690 }, { "epoch": 66.73040152963671, "grad_norm": 1.0098766088485718, "learning_rate": 7.748614147956153e-05, "loss": 0.0276, "step": 104700 }, { "epoch": 66.73677501593372, "grad_norm": 0.49463239312171936, "learning_rate": 7.748153724174642e-05, "loss": 0.0254, "step": 104710 }, { "epoch": 66.74314850223072, "grad_norm": 0.8672518134117126, "learning_rate": 7.747693267000469e-05, "loss": 0.0266, "step": 104720 }, { "epoch": 66.74952198852772, "grad_norm": 0.3697878420352936, "learning_rate": 7.74723277643923e-05, "loss": 0.0198, "step": 104730 }, { "epoch": 66.75589547482473, "grad_norm": 0.32793936133384705, "learning_rate": 7.746772252496519e-05, "loss": 0.0231, "step": 104740 }, { "epoch": 66.76226896112173, "grad_norm": 0.37291669845581055, "learning_rate": 7.74631169517793e-05, "loss": 0.0228, "step": 104750 }, { "epoch": 66.76864244741874, "grad_norm": 0.3412865996360779, "learning_rate": 7.745851104489061e-05, "loss": 0.0189, "step": 104760 }, { "epoch": 66.77501593371574, "grad_norm": 0.5117918252944946, "learning_rate": 7.74539048043551e-05, "loss": 0.0203, "step": 104770 }, { "epoch": 66.78138942001274, "grad_norm": 0.3399236798286438, "learning_rate": 7.744929823022872e-05, "loss": 0.0282, "step": 104780 }, { "epoch": 66.78776290630975, "grad_norm": 3.5026469230651855, "learning_rate": 7.744469132256746e-05, "loss": 0.0174, "step": 104790 }, { "epoch": 66.79413639260676, "grad_norm": 0.5342695713043213, "learning_rate": 7.744008408142726e-05, "loss": 0.0174, "step": 104800 }, { "epoch": 66.80050987890377, "grad_norm": 0.6316962242126465, "learning_rate": 7.743547650686415e-05, "loss": 0.0184, "step": 104810 }, { "epoch": 66.80688336520076, "grad_norm": 30.871097564697266, "learning_rate": 7.74308685989341e-05, "loss": 0.0254, "step": 104820 }, { "epoch": 66.81325685149777, "grad_norm": 0.3209657669067383, "learning_rate": 7.742626035769309e-05, "loss": 0.0261, "step": 104830 }, { "epoch": 66.81963033779478, "grad_norm": 0.3342027962207794, "learning_rate": 7.742165178319712e-05, "loss": 0.0222, "step": 104840 }, { "epoch": 66.82600382409179, "grad_norm": 0.35139378905296326, "learning_rate": 7.74170428755022e-05, "loss": 0.0142, "step": 104850 }, { "epoch": 66.83237731038878, "grad_norm": 0.923721969127655, "learning_rate": 7.741243363466433e-05, "loss": 0.0298, "step": 104860 }, { "epoch": 66.83875079668579, "grad_norm": 0.3711704909801483, "learning_rate": 7.74078240607395e-05, "loss": 0.0191, "step": 104870 }, { "epoch": 66.8451242829828, "grad_norm": 0.5540165305137634, "learning_rate": 7.740321415378373e-05, "loss": 0.0203, "step": 104880 }, { "epoch": 66.8514977692798, "grad_norm": 0.3774334192276001, "learning_rate": 7.739860391385303e-05, "loss": 0.0194, "step": 104890 }, { "epoch": 66.8578712555768, "grad_norm": 0.7168303728103638, "learning_rate": 7.739399334100342e-05, "loss": 0.0217, "step": 104900 }, { "epoch": 66.8642447418738, "grad_norm": 0.46840304136276245, "learning_rate": 7.738938243529094e-05, "loss": 0.0158, "step": 104910 }, { "epoch": 66.87061822817081, "grad_norm": 0.2594188451766968, "learning_rate": 7.73847711967716e-05, "loss": 0.0231, "step": 104920 }, { "epoch": 66.87699171446782, "grad_norm": 0.3350170850753784, "learning_rate": 7.738015962550142e-05, "loss": 0.0178, "step": 104930 }, { "epoch": 66.88336520076481, "grad_norm": 0.3165241777896881, "learning_rate": 7.737554772153647e-05, "loss": 0.0276, "step": 104940 }, { "epoch": 66.88973868706182, "grad_norm": 0.30383965373039246, "learning_rate": 7.737093548493276e-05, "loss": 0.0156, "step": 104950 }, { "epoch": 66.89611217335883, "grad_norm": 0.3588404059410095, "learning_rate": 7.736632291574634e-05, "loss": 0.0155, "step": 104960 }, { "epoch": 66.90248565965584, "grad_norm": 0.49292972683906555, "learning_rate": 7.736171001403326e-05, "loss": 0.0133, "step": 104970 }, { "epoch": 66.90885914595283, "grad_norm": 0.45548707246780396, "learning_rate": 7.735709677984957e-05, "loss": 0.0228, "step": 104980 }, { "epoch": 66.91523263224984, "grad_norm": 0.6077223420143127, "learning_rate": 7.735248321325133e-05, "loss": 0.0173, "step": 104990 }, { "epoch": 66.92160611854685, "grad_norm": 0.5410031676292419, "learning_rate": 7.734786931429457e-05, "loss": 0.0192, "step": 105000 }, { "epoch": 66.92797960484386, "grad_norm": 0.4797418713569641, "learning_rate": 7.73432550830354e-05, "loss": 0.0168, "step": 105010 }, { "epoch": 66.93435309114085, "grad_norm": 0.29916077852249146, "learning_rate": 7.733864051952987e-05, "loss": 0.0263, "step": 105020 }, { "epoch": 66.94072657743786, "grad_norm": 0.30439293384552, "learning_rate": 7.733402562383403e-05, "loss": 0.0115, "step": 105030 }, { "epoch": 66.94710006373487, "grad_norm": 0.3199256658554077, "learning_rate": 7.732941039600398e-05, "loss": 0.0187, "step": 105040 }, { "epoch": 66.95347355003187, "grad_norm": 0.8677883744239807, "learning_rate": 7.732479483609578e-05, "loss": 0.0318, "step": 105050 }, { "epoch": 66.95984703632887, "grad_norm": 0.7674688696861267, "learning_rate": 7.732017894416554e-05, "loss": 0.017, "step": 105060 }, { "epoch": 66.96622052262587, "grad_norm": 0.3315852880477905, "learning_rate": 7.731556272026932e-05, "loss": 0.0234, "step": 105070 }, { "epoch": 66.97259400892288, "grad_norm": 0.2697911858558655, "learning_rate": 7.731094616446323e-05, "loss": 0.0116, "step": 105080 }, { "epoch": 66.97896749521989, "grad_norm": 0.46022412180900574, "learning_rate": 7.730632927680334e-05, "loss": 0.0226, "step": 105090 }, { "epoch": 66.98534098151688, "grad_norm": 0.2678773105144501, "learning_rate": 7.730171205734578e-05, "loss": 0.0132, "step": 105100 }, { "epoch": 66.99171446781389, "grad_norm": 0.3796941936016083, "learning_rate": 7.729709450614666e-05, "loss": 0.0236, "step": 105110 }, { "epoch": 66.9980879541109, "grad_norm": 0.34052908420562744, "learning_rate": 7.729247662326205e-05, "loss": 0.0213, "step": 105120 }, { "epoch": 67.00446144040791, "grad_norm": 0.5408557057380676, "learning_rate": 7.728785840874808e-05, "loss": 0.0156, "step": 105130 }, { "epoch": 67.0108349267049, "grad_norm": 0.2174934446811676, "learning_rate": 7.728323986266086e-05, "loss": 0.018, "step": 105140 }, { "epoch": 67.01720841300191, "grad_norm": 0.40602123737335205, "learning_rate": 7.727862098505654e-05, "loss": 0.0144, "step": 105150 }, { "epoch": 67.02358189929892, "grad_norm": 0.37545037269592285, "learning_rate": 7.727400177599118e-05, "loss": 0.0215, "step": 105160 }, { "epoch": 67.02995538559593, "grad_norm": 0.533374547958374, "learning_rate": 7.726938223552098e-05, "loss": 0.033, "step": 105170 }, { "epoch": 67.03632887189292, "grad_norm": 0.4033690094947815, "learning_rate": 7.726476236370203e-05, "loss": 0.0146, "step": 105180 }, { "epoch": 67.04270235818993, "grad_norm": 0.3791067898273468, "learning_rate": 7.726014216059048e-05, "loss": 0.0163, "step": 105190 }, { "epoch": 67.04907584448694, "grad_norm": 0.5514476895332336, "learning_rate": 7.725552162624248e-05, "loss": 0.0161, "step": 105200 }, { "epoch": 67.05544933078394, "grad_norm": 0.43224847316741943, "learning_rate": 7.725090076071413e-05, "loss": 0.0245, "step": 105210 }, { "epoch": 67.06182281708094, "grad_norm": 0.2212410718202591, "learning_rate": 7.724627956406162e-05, "loss": 0.0129, "step": 105220 }, { "epoch": 67.06819630337795, "grad_norm": 0.19712330400943756, "learning_rate": 7.724165803634108e-05, "loss": 0.0141, "step": 105230 }, { "epoch": 67.07456978967495, "grad_norm": 0.39485466480255127, "learning_rate": 7.723703617760869e-05, "loss": 0.0274, "step": 105240 }, { "epoch": 67.08094327597196, "grad_norm": 0.21389883756637573, "learning_rate": 7.723241398792057e-05, "loss": 0.0133, "step": 105250 }, { "epoch": 67.08731676226896, "grad_norm": 0.4741755425930023, "learning_rate": 7.722779146733293e-05, "loss": 0.0207, "step": 105260 }, { "epoch": 67.09369024856596, "grad_norm": 0.41173914074897766, "learning_rate": 7.722316861590191e-05, "loss": 0.0188, "step": 105270 }, { "epoch": 67.10006373486297, "grad_norm": 0.2696644365787506, "learning_rate": 7.721854543368367e-05, "loss": 0.0113, "step": 105280 }, { "epoch": 67.10643722115998, "grad_norm": 0.30941760540008545, "learning_rate": 7.721392192073444e-05, "loss": 0.0109, "step": 105290 }, { "epoch": 67.11281070745697, "grad_norm": 0.21298426389694214, "learning_rate": 7.720929807711034e-05, "loss": 0.0185, "step": 105300 }, { "epoch": 67.11918419375398, "grad_norm": 0.7119069695472717, "learning_rate": 7.720467390286758e-05, "loss": 0.0263, "step": 105310 }, { "epoch": 67.12555768005099, "grad_norm": 0.6277884840965271, "learning_rate": 7.720004939806236e-05, "loss": 0.0153, "step": 105320 }, { "epoch": 67.131931166348, "grad_norm": 0.6795727610588074, "learning_rate": 7.719542456275084e-05, "loss": 0.0194, "step": 105330 }, { "epoch": 67.13830465264499, "grad_norm": 0.2808575928211212, "learning_rate": 7.719079939698925e-05, "loss": 0.017, "step": 105340 }, { "epoch": 67.144678138942, "grad_norm": 0.7953891158103943, "learning_rate": 7.718617390083376e-05, "loss": 0.0157, "step": 105350 }, { "epoch": 67.151051625239, "grad_norm": 0.2603432834148407, "learning_rate": 7.718154807434061e-05, "loss": 0.0196, "step": 105360 }, { "epoch": 67.15742511153601, "grad_norm": 0.5800109505653381, "learning_rate": 7.717692191756596e-05, "loss": 0.0231, "step": 105370 }, { "epoch": 67.16379859783301, "grad_norm": 0.36354097723960876, "learning_rate": 7.717229543056608e-05, "loss": 0.0169, "step": 105380 }, { "epoch": 67.17017208413002, "grad_norm": 0.3106566071510315, "learning_rate": 7.716766861339714e-05, "loss": 0.0164, "step": 105390 }, { "epoch": 67.17654557042702, "grad_norm": 0.5032699704170227, "learning_rate": 7.716304146611536e-05, "loss": 0.0243, "step": 105400 }, { "epoch": 67.18291905672403, "grad_norm": 0.4553208649158478, "learning_rate": 7.715841398877701e-05, "loss": 0.022, "step": 105410 }, { "epoch": 67.18929254302103, "grad_norm": 0.4614121615886688, "learning_rate": 7.715378618143827e-05, "loss": 0.0189, "step": 105420 }, { "epoch": 67.19566602931803, "grad_norm": 0.4059658944606781, "learning_rate": 7.714915804415542e-05, "loss": 0.0234, "step": 105430 }, { "epoch": 67.20203951561504, "grad_norm": 0.33004340529441833, "learning_rate": 7.714452957698462e-05, "loss": 0.0176, "step": 105440 }, { "epoch": 67.20841300191205, "grad_norm": 0.6053465604782104, "learning_rate": 7.713990077998219e-05, "loss": 0.0266, "step": 105450 }, { "epoch": 67.21478648820906, "grad_norm": 0.3490292429924011, "learning_rate": 7.713527165320432e-05, "loss": 0.0148, "step": 105460 }, { "epoch": 67.22115997450605, "grad_norm": 0.3693540394306183, "learning_rate": 7.713064219670729e-05, "loss": 0.0188, "step": 105470 }, { "epoch": 67.22753346080306, "grad_norm": 0.27278390526771545, "learning_rate": 7.712601241054734e-05, "loss": 0.0218, "step": 105480 }, { "epoch": 67.23390694710007, "grad_norm": 0.4609919786453247, "learning_rate": 7.712138229478075e-05, "loss": 0.0199, "step": 105490 }, { "epoch": 67.24028043339707, "grad_norm": 0.4214707911014557, "learning_rate": 7.711675184946373e-05, "loss": 0.0182, "step": 105500 }, { "epoch": 67.24665391969407, "grad_norm": 0.44127461314201355, "learning_rate": 7.711212107465257e-05, "loss": 0.0315, "step": 105510 }, { "epoch": 67.25302740599108, "grad_norm": 0.6578241586685181, "learning_rate": 7.710748997040356e-05, "loss": 0.0129, "step": 105520 }, { "epoch": 67.25940089228808, "grad_norm": 0.2958971858024597, "learning_rate": 7.710285853677295e-05, "loss": 0.0183, "step": 105530 }, { "epoch": 67.26577437858509, "grad_norm": 0.4070180356502533, "learning_rate": 7.709822677381702e-05, "loss": 0.0234, "step": 105540 }, { "epoch": 67.27214786488209, "grad_norm": 0.3588690161705017, "learning_rate": 7.709359468159203e-05, "loss": 0.0222, "step": 105550 }, { "epoch": 67.2785213511791, "grad_norm": 0.32245951890945435, "learning_rate": 7.70889622601543e-05, "loss": 0.011, "step": 105560 }, { "epoch": 67.2848948374761, "grad_norm": 0.2726690471172333, "learning_rate": 7.708432950956011e-05, "loss": 0.0109, "step": 105570 }, { "epoch": 67.29126832377311, "grad_norm": 0.25237929821014404, "learning_rate": 7.707969642986572e-05, "loss": 0.015, "step": 105580 }, { "epoch": 67.2976418100701, "grad_norm": 0.23396292328834534, "learning_rate": 7.707506302112746e-05, "loss": 0.0145, "step": 105590 }, { "epoch": 67.30401529636711, "grad_norm": 0.49927106499671936, "learning_rate": 7.707042928340164e-05, "loss": 0.0201, "step": 105600 }, { "epoch": 67.31038878266412, "grad_norm": 0.3887627124786377, "learning_rate": 7.706579521674453e-05, "loss": 0.0164, "step": 105610 }, { "epoch": 67.31676226896113, "grad_norm": 0.3990075886249542, "learning_rate": 7.706116082121245e-05, "loss": 0.0168, "step": 105620 }, { "epoch": 67.32313575525812, "grad_norm": 0.5167376399040222, "learning_rate": 7.705652609686173e-05, "loss": 0.0245, "step": 105630 }, { "epoch": 67.32950924155513, "grad_norm": 0.6399632096290588, "learning_rate": 7.705189104374866e-05, "loss": 0.0161, "step": 105640 }, { "epoch": 67.33588272785214, "grad_norm": 0.5228751301765442, "learning_rate": 7.704725566192958e-05, "loss": 0.0181, "step": 105650 }, { "epoch": 67.34225621414915, "grad_norm": 0.40531012415885925, "learning_rate": 7.70426199514608e-05, "loss": 0.0188, "step": 105660 }, { "epoch": 67.34862970044614, "grad_norm": 0.2802581191062927, "learning_rate": 7.703798391239866e-05, "loss": 0.0185, "step": 105670 }, { "epoch": 67.35500318674315, "grad_norm": 0.2423659712076187, "learning_rate": 7.703334754479949e-05, "loss": 0.0206, "step": 105680 }, { "epoch": 67.36137667304016, "grad_norm": 0.28774094581604004, "learning_rate": 7.70287108487196e-05, "loss": 0.0143, "step": 105690 }, { "epoch": 67.36775015933716, "grad_norm": 0.3289378881454468, "learning_rate": 7.702407382421539e-05, "loss": 0.0139, "step": 105700 }, { "epoch": 67.37412364563416, "grad_norm": 0.48083844780921936, "learning_rate": 7.701943647134315e-05, "loss": 0.0175, "step": 105710 }, { "epoch": 67.38049713193116, "grad_norm": 0.3376317620277405, "learning_rate": 7.701479879015925e-05, "loss": 0.0186, "step": 105720 }, { "epoch": 67.38687061822817, "grad_norm": 0.32823094725608826, "learning_rate": 7.701016078072005e-05, "loss": 0.0235, "step": 105730 }, { "epoch": 67.39324410452518, "grad_norm": 0.27656641602516174, "learning_rate": 7.700552244308188e-05, "loss": 0.015, "step": 105740 }, { "epoch": 67.39961759082217, "grad_norm": 0.23919177055358887, "learning_rate": 7.700088377730111e-05, "loss": 0.0099, "step": 105750 }, { "epoch": 67.40599107711918, "grad_norm": 0.29493051767349243, "learning_rate": 7.699624478343413e-05, "loss": 0.0167, "step": 105760 }, { "epoch": 67.41236456341619, "grad_norm": 0.32968592643737793, "learning_rate": 7.699160546153727e-05, "loss": 0.0254, "step": 105770 }, { "epoch": 67.4187380497132, "grad_norm": 0.5708123445510864, "learning_rate": 7.698696581166691e-05, "loss": 0.0184, "step": 105780 }, { "epoch": 67.42511153601019, "grad_norm": 0.7756767868995667, "learning_rate": 7.698232583387945e-05, "loss": 0.0206, "step": 105790 }, { "epoch": 67.4314850223072, "grad_norm": 0.44351449608802795, "learning_rate": 7.697768552823126e-05, "loss": 0.0275, "step": 105800 }, { "epoch": 67.43785850860421, "grad_norm": 0.2911220192909241, "learning_rate": 7.69730448947787e-05, "loss": 0.0174, "step": 105810 }, { "epoch": 67.44423199490122, "grad_norm": 0.38815680146217346, "learning_rate": 7.69684039335782e-05, "loss": 0.0162, "step": 105820 }, { "epoch": 67.45060548119821, "grad_norm": 0.33517536520957947, "learning_rate": 7.696376264468612e-05, "loss": 0.0242, "step": 105830 }, { "epoch": 67.45697896749522, "grad_norm": 0.5730772018432617, "learning_rate": 7.695912102815886e-05, "loss": 0.0216, "step": 105840 }, { "epoch": 67.46335245379223, "grad_norm": 0.3504476249217987, "learning_rate": 7.695447908405283e-05, "loss": 0.0141, "step": 105850 }, { "epoch": 67.46972594008923, "grad_norm": 0.23968160152435303, "learning_rate": 7.694983681242442e-05, "loss": 0.0269, "step": 105860 }, { "epoch": 67.47609942638623, "grad_norm": 0.4239415228366852, "learning_rate": 7.694519421333007e-05, "loss": 0.0118, "step": 105870 }, { "epoch": 67.48247291268324, "grad_norm": 0.24590325355529785, "learning_rate": 7.694055128682616e-05, "loss": 0.01, "step": 105880 }, { "epoch": 67.48884639898024, "grad_norm": 0.525265097618103, "learning_rate": 7.693590803296912e-05, "loss": 0.0266, "step": 105890 }, { "epoch": 67.49521988527725, "grad_norm": 0.2504498064517975, "learning_rate": 7.693126445181536e-05, "loss": 0.0164, "step": 105900 }, { "epoch": 67.50159337157424, "grad_norm": 0.20599915087223053, "learning_rate": 7.692662054342131e-05, "loss": 0.0138, "step": 105910 }, { "epoch": 67.50796685787125, "grad_norm": 0.8464505672454834, "learning_rate": 7.692197630784339e-05, "loss": 0.0178, "step": 105920 }, { "epoch": 67.51434034416826, "grad_norm": 0.29683056473731995, "learning_rate": 7.691733174513806e-05, "loss": 0.0156, "step": 105930 }, { "epoch": 67.52071383046527, "grad_norm": 0.45821741223335266, "learning_rate": 7.691268685536172e-05, "loss": 0.0204, "step": 105940 }, { "epoch": 67.52708731676226, "grad_norm": 0.3899914622306824, "learning_rate": 7.690804163857082e-05, "loss": 0.0177, "step": 105950 }, { "epoch": 67.53346080305927, "grad_norm": 0.39054110646247864, "learning_rate": 7.690339609482182e-05, "loss": 0.0133, "step": 105960 }, { "epoch": 67.53983428935628, "grad_norm": 0.32774627208709717, "learning_rate": 7.689875022417115e-05, "loss": 0.0182, "step": 105970 }, { "epoch": 67.54620777565329, "grad_norm": 0.4880833029747009, "learning_rate": 7.689410402667527e-05, "loss": 0.0126, "step": 105980 }, { "epoch": 67.55258126195028, "grad_norm": 0.3543986976146698, "learning_rate": 7.688945750239064e-05, "loss": 0.0185, "step": 105990 }, { "epoch": 67.55895474824729, "grad_norm": 0.6037893891334534, "learning_rate": 7.688481065137371e-05, "loss": 0.0233, "step": 106000 }, { "epoch": 67.5653282345443, "grad_norm": 0.34684044122695923, "learning_rate": 7.688016347368094e-05, "loss": 0.0172, "step": 106010 }, { "epoch": 67.5717017208413, "grad_norm": 0.41661015152931213, "learning_rate": 7.687551596936882e-05, "loss": 0.0197, "step": 106020 }, { "epoch": 67.5780752071383, "grad_norm": 0.2711317241191864, "learning_rate": 7.687086813849378e-05, "loss": 0.0128, "step": 106030 }, { "epoch": 67.5844486934353, "grad_norm": 0.602989137172699, "learning_rate": 7.686621998111234e-05, "loss": 0.0178, "step": 106040 }, { "epoch": 67.59082217973231, "grad_norm": 0.3907238245010376, "learning_rate": 7.686157149728096e-05, "loss": 0.0195, "step": 106050 }, { "epoch": 67.59719566602932, "grad_norm": 0.35256439447402954, "learning_rate": 7.685692268705614e-05, "loss": 0.0229, "step": 106060 }, { "epoch": 67.60356915232632, "grad_norm": 0.23378022015094757, "learning_rate": 7.685227355049434e-05, "loss": 0.0168, "step": 106070 }, { "epoch": 67.60994263862332, "grad_norm": 0.3681199848651886, "learning_rate": 7.684762408765205e-05, "loss": 0.0179, "step": 106080 }, { "epoch": 67.61631612492033, "grad_norm": 0.5050526261329651, "learning_rate": 7.684297429858579e-05, "loss": 0.0298, "step": 106090 }, { "epoch": 67.62268961121734, "grad_norm": 0.5170756578445435, "learning_rate": 7.683832418335204e-05, "loss": 0.0311, "step": 106100 }, { "epoch": 67.62906309751435, "grad_norm": 0.18287505209445953, "learning_rate": 7.683367374200732e-05, "loss": 0.0133, "step": 106110 }, { "epoch": 67.63543658381134, "grad_norm": 0.44249165058135986, "learning_rate": 7.682902297460812e-05, "loss": 0.0188, "step": 106120 }, { "epoch": 67.64181007010835, "grad_norm": 0.35246211290359497, "learning_rate": 7.682437188121097e-05, "loss": 0.0201, "step": 106130 }, { "epoch": 67.64818355640536, "grad_norm": 0.2629941999912262, "learning_rate": 7.681972046187238e-05, "loss": 0.0165, "step": 106140 }, { "epoch": 67.65455704270236, "grad_norm": 0.40199899673461914, "learning_rate": 7.681506871664886e-05, "loss": 0.0171, "step": 106150 }, { "epoch": 67.66093052899936, "grad_norm": 0.6481414437294006, "learning_rate": 7.681041664559693e-05, "loss": 0.033, "step": 106160 }, { "epoch": 67.66730401529637, "grad_norm": 0.23511670529842377, "learning_rate": 7.680576424877312e-05, "loss": 0.0162, "step": 106170 }, { "epoch": 67.67367750159337, "grad_norm": 0.34975284337997437, "learning_rate": 7.680111152623399e-05, "loss": 0.0142, "step": 106180 }, { "epoch": 67.68005098789038, "grad_norm": 0.55955970287323, "learning_rate": 7.679645847803603e-05, "loss": 0.0174, "step": 106190 }, { "epoch": 67.68642447418738, "grad_norm": 0.2405116707086563, "learning_rate": 7.67918051042358e-05, "loss": 0.0155, "step": 106200 }, { "epoch": 67.69279796048438, "grad_norm": 0.3025439381599426, "learning_rate": 7.678715140488985e-05, "loss": 0.0182, "step": 106210 }, { "epoch": 67.69917144678139, "grad_norm": 0.44550079107284546, "learning_rate": 7.678249738005472e-05, "loss": 0.0305, "step": 106220 }, { "epoch": 67.7055449330784, "grad_norm": 0.24607625603675842, "learning_rate": 7.677784302978697e-05, "loss": 0.0215, "step": 106230 }, { "epoch": 67.7119184193754, "grad_norm": 0.3581153452396393, "learning_rate": 7.67731883541431e-05, "loss": 0.0169, "step": 106240 }, { "epoch": 67.7182919056724, "grad_norm": 0.3534605801105499, "learning_rate": 7.676853335317977e-05, "loss": 0.0178, "step": 106250 }, { "epoch": 67.72466539196941, "grad_norm": 0.3339270055294037, "learning_rate": 7.676387802695345e-05, "loss": 0.0226, "step": 106260 }, { "epoch": 67.73103887826642, "grad_norm": 0.2533717453479767, "learning_rate": 7.675922237552075e-05, "loss": 0.0122, "step": 106270 }, { "epoch": 67.73741236456341, "grad_norm": 0.39450234174728394, "learning_rate": 7.675456639893823e-05, "loss": 0.0192, "step": 106280 }, { "epoch": 67.74378585086042, "grad_norm": 0.45683565735816956, "learning_rate": 7.674991009726249e-05, "loss": 0.0238, "step": 106290 }, { "epoch": 67.75015933715743, "grad_norm": 0.31965792179107666, "learning_rate": 7.674525347055006e-05, "loss": 0.0215, "step": 106300 }, { "epoch": 67.75653282345444, "grad_norm": 0.35750317573547363, "learning_rate": 7.674059651885757e-05, "loss": 0.0178, "step": 106310 }, { "epoch": 67.76290630975143, "grad_norm": 0.8767184615135193, "learning_rate": 7.673593924224158e-05, "loss": 0.0216, "step": 106320 }, { "epoch": 67.76927979604844, "grad_norm": 0.4145925045013428, "learning_rate": 7.673128164075867e-05, "loss": 0.039, "step": 106330 }, { "epoch": 67.77565328234544, "grad_norm": 0.19461078941822052, "learning_rate": 7.672662371446546e-05, "loss": 0.0192, "step": 106340 }, { "epoch": 67.78202676864245, "grad_norm": 0.35224634408950806, "learning_rate": 7.672196546341853e-05, "loss": 0.0137, "step": 106350 }, { "epoch": 67.78840025493945, "grad_norm": 0.38458341360092163, "learning_rate": 7.671730688767451e-05, "loss": 0.0165, "step": 106360 }, { "epoch": 67.79477374123645, "grad_norm": 0.5220276117324829, "learning_rate": 7.671264798728997e-05, "loss": 0.0221, "step": 106370 }, { "epoch": 67.80114722753346, "grad_norm": 0.5423200726509094, "learning_rate": 7.670798876232154e-05, "loss": 0.0224, "step": 106380 }, { "epoch": 67.80752071383047, "grad_norm": 0.3010280728340149, "learning_rate": 7.670332921282582e-05, "loss": 0.0173, "step": 106390 }, { "epoch": 67.81389420012746, "grad_norm": 0.46610260009765625, "learning_rate": 7.669866933885945e-05, "loss": 0.0142, "step": 106400 }, { "epoch": 67.82026768642447, "grad_norm": 0.4742877185344696, "learning_rate": 7.669400914047904e-05, "loss": 0.0138, "step": 106410 }, { "epoch": 67.82664117272148, "grad_norm": 0.28387489914894104, "learning_rate": 7.668934861774121e-05, "loss": 0.0342, "step": 106420 }, { "epoch": 67.83301465901849, "grad_norm": 0.4412648379802704, "learning_rate": 7.668468777070259e-05, "loss": 0.0219, "step": 106430 }, { "epoch": 67.83938814531548, "grad_norm": 0.7620427012443542, "learning_rate": 7.668002659941983e-05, "loss": 0.0224, "step": 106440 }, { "epoch": 67.84576163161249, "grad_norm": 0.43268439173698425, "learning_rate": 7.667536510394956e-05, "loss": 0.0157, "step": 106450 }, { "epoch": 67.8521351179095, "grad_norm": 0.4897032678127289, "learning_rate": 7.66707032843484e-05, "loss": 0.0234, "step": 106460 }, { "epoch": 67.8585086042065, "grad_norm": 0.7679625749588013, "learning_rate": 7.666604114067304e-05, "loss": 0.0152, "step": 106470 }, { "epoch": 67.8648820905035, "grad_norm": 0.35588929057121277, "learning_rate": 7.666137867298009e-05, "loss": 0.0182, "step": 106480 }, { "epoch": 67.87125557680051, "grad_norm": 0.7372584939002991, "learning_rate": 7.66567158813262e-05, "loss": 0.0282, "step": 106490 }, { "epoch": 67.87762906309752, "grad_norm": 0.4810059070587158, "learning_rate": 7.665205276576807e-05, "loss": 0.0164, "step": 106500 }, { "epoch": 67.88400254939452, "grad_norm": 0.5692694187164307, "learning_rate": 7.664738932636231e-05, "loss": 0.0143, "step": 106510 }, { "epoch": 67.89037603569152, "grad_norm": 0.278939813375473, "learning_rate": 7.664272556316564e-05, "loss": 0.0153, "step": 106520 }, { "epoch": 67.89674952198853, "grad_norm": 0.3903173804283142, "learning_rate": 7.663806147623467e-05, "loss": 0.0225, "step": 106530 }, { "epoch": 67.90312300828553, "grad_norm": 0.35728293657302856, "learning_rate": 7.663339706562612e-05, "loss": 0.0186, "step": 106540 }, { "epoch": 67.90949649458254, "grad_norm": 0.3167421817779541, "learning_rate": 7.662873233139664e-05, "loss": 0.0302, "step": 106550 }, { "epoch": 67.91586998087953, "grad_norm": 0.554246187210083, "learning_rate": 7.662406727360294e-05, "loss": 0.0266, "step": 106560 }, { "epoch": 67.92224346717654, "grad_norm": 0.5290642976760864, "learning_rate": 7.661940189230166e-05, "loss": 0.0132, "step": 106570 }, { "epoch": 67.92861695347355, "grad_norm": 0.4577910006046295, "learning_rate": 7.661473618754953e-05, "loss": 0.0331, "step": 106580 }, { "epoch": 67.93499043977056, "grad_norm": 0.518340528011322, "learning_rate": 7.661007015940323e-05, "loss": 0.0378, "step": 106590 }, { "epoch": 67.94136392606755, "grad_norm": 0.5243750810623169, "learning_rate": 7.660540380791942e-05, "loss": 0.016, "step": 106600 }, { "epoch": 67.94773741236456, "grad_norm": 0.5740789771080017, "learning_rate": 7.660073713315488e-05, "loss": 0.027, "step": 106610 }, { "epoch": 67.95411089866157, "grad_norm": 0.6874484419822693, "learning_rate": 7.659607013516624e-05, "loss": 0.023, "step": 106620 }, { "epoch": 67.96048438495858, "grad_norm": 0.6075157523155212, "learning_rate": 7.659140281401026e-05, "loss": 0.0208, "step": 106630 }, { "epoch": 67.96685787125557, "grad_norm": 0.44594675302505493, "learning_rate": 7.658673516974362e-05, "loss": 0.0186, "step": 106640 }, { "epoch": 67.97323135755258, "grad_norm": 0.39324626326560974, "learning_rate": 7.658206720242302e-05, "loss": 0.0151, "step": 106650 }, { "epoch": 67.97960484384959, "grad_norm": 0.6100729703903198, "learning_rate": 7.657739891210523e-05, "loss": 0.0249, "step": 106660 }, { "epoch": 67.9859783301466, "grad_norm": 0.3282598853111267, "learning_rate": 7.657273029884695e-05, "loss": 0.0146, "step": 106670 }, { "epoch": 67.99235181644359, "grad_norm": 0.337340772151947, "learning_rate": 7.656806136270492e-05, "loss": 0.0188, "step": 106680 }, { "epoch": 67.9987253027406, "grad_norm": 0.4167628884315491, "learning_rate": 7.656339210373584e-05, "loss": 0.0231, "step": 106690 }, { "epoch": 68.0050987890376, "grad_norm": 0.29945069551467896, "learning_rate": 7.655872252199646e-05, "loss": 0.0131, "step": 106700 }, { "epoch": 68.01147227533461, "grad_norm": 0.44897034764289856, "learning_rate": 7.655405261754354e-05, "loss": 0.0164, "step": 106710 }, { "epoch": 68.01784576163162, "grad_norm": 0.5033071637153625, "learning_rate": 7.654938239043381e-05, "loss": 0.0119, "step": 106720 }, { "epoch": 68.02421924792861, "grad_norm": 0.3997918367385864, "learning_rate": 7.654471184072401e-05, "loss": 0.0153, "step": 106730 }, { "epoch": 68.03059273422562, "grad_norm": 0.38981032371520996, "learning_rate": 7.65400409684709e-05, "loss": 0.0168, "step": 106740 }, { "epoch": 68.03696622052263, "grad_norm": 0.3457668125629425, "learning_rate": 7.653536977373123e-05, "loss": 0.047, "step": 106750 }, { "epoch": 68.04333970681964, "grad_norm": 0.4665244221687317, "learning_rate": 7.653069825656178e-05, "loss": 0.0136, "step": 106760 }, { "epoch": 68.04971319311663, "grad_norm": 0.2908143997192383, "learning_rate": 7.65260264170193e-05, "loss": 0.0145, "step": 106770 }, { "epoch": 68.05608667941364, "grad_norm": 0.48534539341926575, "learning_rate": 7.652135425516053e-05, "loss": 0.0164, "step": 106780 }, { "epoch": 68.06246016571065, "grad_norm": 0.5216912031173706, "learning_rate": 7.651668177104227e-05, "loss": 0.0166, "step": 106790 }, { "epoch": 68.06883365200765, "grad_norm": 0.44820600748062134, "learning_rate": 7.65120089647213e-05, "loss": 0.0192, "step": 106800 }, { "epoch": 68.07520713830465, "grad_norm": 0.4308840334415436, "learning_rate": 7.650733583625439e-05, "loss": 0.0154, "step": 106810 }, { "epoch": 68.08158062460166, "grad_norm": 0.6498591303825378, "learning_rate": 7.650266238569831e-05, "loss": 0.0294, "step": 106820 }, { "epoch": 68.08795411089866, "grad_norm": 0.7375837564468384, "learning_rate": 7.649798861310988e-05, "loss": 0.0161, "step": 106830 }, { "epoch": 68.09432759719567, "grad_norm": 0.36340686678886414, "learning_rate": 7.649331451854587e-05, "loss": 0.0187, "step": 106840 }, { "epoch": 68.10070108349267, "grad_norm": 0.39983057975769043, "learning_rate": 7.648864010206305e-05, "loss": 0.0205, "step": 106850 }, { "epoch": 68.10707456978967, "grad_norm": 0.33313754200935364, "learning_rate": 7.648396536371826e-05, "loss": 0.032, "step": 106860 }, { "epoch": 68.11344805608668, "grad_norm": 0.4528173804283142, "learning_rate": 7.647929030356829e-05, "loss": 0.0178, "step": 106870 }, { "epoch": 68.11982154238369, "grad_norm": 0.5261549949645996, "learning_rate": 7.647461492166993e-05, "loss": 0.0302, "step": 106880 }, { "epoch": 68.12619502868068, "grad_norm": 0.8389148712158203, "learning_rate": 7.646993921808001e-05, "loss": 0.0223, "step": 106890 }, { "epoch": 68.13256851497769, "grad_norm": 0.30468299984931946, "learning_rate": 7.646526319285533e-05, "loss": 0.0258, "step": 106900 }, { "epoch": 68.1389420012747, "grad_norm": 0.3896545469760895, "learning_rate": 7.646058684605273e-05, "loss": 0.0136, "step": 106910 }, { "epoch": 68.14531548757171, "grad_norm": 0.47277501225471497, "learning_rate": 7.6455910177729e-05, "loss": 0.023, "step": 106920 }, { "epoch": 68.1516889738687, "grad_norm": 0.30315983295440674, "learning_rate": 7.645123318794098e-05, "loss": 0.0126, "step": 106930 }, { "epoch": 68.15806246016571, "grad_norm": 0.2271626889705658, "learning_rate": 7.644655587674552e-05, "loss": 0.0191, "step": 106940 }, { "epoch": 68.16443594646272, "grad_norm": 0.21386992931365967, "learning_rate": 7.644187824419943e-05, "loss": 0.0265, "step": 106950 }, { "epoch": 68.17080943275973, "grad_norm": 0.7005563974380493, "learning_rate": 7.643720029035956e-05, "loss": 0.0203, "step": 106960 }, { "epoch": 68.17718291905672, "grad_norm": 0.3439531922340393, "learning_rate": 7.643252201528273e-05, "loss": 0.0155, "step": 106970 }, { "epoch": 68.18355640535373, "grad_norm": 0.5040580630302429, "learning_rate": 7.64278434190258e-05, "loss": 0.0208, "step": 106980 }, { "epoch": 68.18992989165073, "grad_norm": 0.6920354962348938, "learning_rate": 7.642316450164563e-05, "loss": 0.0168, "step": 106990 }, { "epoch": 68.19630337794774, "grad_norm": 0.30823206901550293, "learning_rate": 7.641848526319908e-05, "loss": 0.019, "step": 107000 }, { "epoch": 68.20267686424474, "grad_norm": 0.43885335326194763, "learning_rate": 7.641380570374296e-05, "loss": 0.0206, "step": 107010 }, { "epoch": 68.20905035054174, "grad_norm": 0.20622175931930542, "learning_rate": 7.640912582333418e-05, "loss": 0.0213, "step": 107020 }, { "epoch": 68.21542383683875, "grad_norm": 0.4821596145629883, "learning_rate": 7.640444562202959e-05, "loss": 0.0187, "step": 107030 }, { "epoch": 68.22179732313576, "grad_norm": 0.17340785264968872, "learning_rate": 7.639976509988605e-05, "loss": 0.0259, "step": 107040 }, { "epoch": 68.22817080943275, "grad_norm": 0.3724033236503601, "learning_rate": 7.639508425696042e-05, "loss": 0.0159, "step": 107050 }, { "epoch": 68.23454429572976, "grad_norm": 0.5715773701667786, "learning_rate": 7.639040309330963e-05, "loss": 0.0191, "step": 107060 }, { "epoch": 68.24091778202677, "grad_norm": 0.4976310133934021, "learning_rate": 7.63857216089905e-05, "loss": 0.0236, "step": 107070 }, { "epoch": 68.24729126832378, "grad_norm": 0.4499194920063019, "learning_rate": 7.638103980405995e-05, "loss": 0.0133, "step": 107080 }, { "epoch": 68.25366475462077, "grad_norm": 0.23533770442008972, "learning_rate": 7.637635767857486e-05, "loss": 0.0126, "step": 107090 }, { "epoch": 68.26003824091778, "grad_norm": 0.2602512538433075, "learning_rate": 7.637167523259211e-05, "loss": 0.0213, "step": 107100 }, { "epoch": 68.26641172721479, "grad_norm": 0.3403356075286865, "learning_rate": 7.636699246616862e-05, "loss": 0.0149, "step": 107110 }, { "epoch": 68.2727852135118, "grad_norm": 0.36458438634872437, "learning_rate": 7.636230937936128e-05, "loss": 0.0138, "step": 107120 }, { "epoch": 68.27915869980879, "grad_norm": 0.34950578212738037, "learning_rate": 7.635762597222698e-05, "loss": 0.0162, "step": 107130 }, { "epoch": 68.2855321861058, "grad_norm": 0.35765716433525085, "learning_rate": 7.635294224482263e-05, "loss": 0.0221, "step": 107140 }, { "epoch": 68.2919056724028, "grad_norm": 0.34105348587036133, "learning_rate": 7.634825819720517e-05, "loss": 0.0137, "step": 107150 }, { "epoch": 68.29827915869981, "grad_norm": 0.40952396392822266, "learning_rate": 7.634357382943149e-05, "loss": 0.0267, "step": 107160 }, { "epoch": 68.3046526449968, "grad_norm": 0.38946276903152466, "learning_rate": 7.63388891415585e-05, "loss": 0.0179, "step": 107170 }, { "epoch": 68.31102613129381, "grad_norm": 0.3432058095932007, "learning_rate": 7.633420413364317e-05, "loss": 0.0105, "step": 107180 }, { "epoch": 68.31739961759082, "grad_norm": 0.3013158142566681, "learning_rate": 7.632951880574237e-05, "loss": 0.0142, "step": 107190 }, { "epoch": 68.32377310388783, "grad_norm": 0.783240020275116, "learning_rate": 7.632483315791306e-05, "loss": 0.0302, "step": 107200 }, { "epoch": 68.33014659018482, "grad_norm": 0.3759140074253082, "learning_rate": 7.632014719021217e-05, "loss": 0.0152, "step": 107210 }, { "epoch": 68.33652007648183, "grad_norm": 0.5865796208381653, "learning_rate": 7.631546090269665e-05, "loss": 0.0271, "step": 107220 }, { "epoch": 68.34289356277884, "grad_norm": 0.3275274634361267, "learning_rate": 7.631077429542344e-05, "loss": 0.0258, "step": 107230 }, { "epoch": 68.34926704907585, "grad_norm": 0.2704906761646271, "learning_rate": 7.630608736844947e-05, "loss": 0.0224, "step": 107240 }, { "epoch": 68.35564053537284, "grad_norm": 0.3460225760936737, "learning_rate": 7.63014001218317e-05, "loss": 0.0171, "step": 107250 }, { "epoch": 68.36201402166985, "grad_norm": 0.39124923944473267, "learning_rate": 7.62967125556271e-05, "loss": 0.0124, "step": 107260 }, { "epoch": 68.36838750796686, "grad_norm": 0.3099464774131775, "learning_rate": 7.62920246698926e-05, "loss": 0.019, "step": 107270 }, { "epoch": 68.37476099426387, "grad_norm": 0.317867249250412, "learning_rate": 7.628733646468517e-05, "loss": 0.0173, "step": 107280 }, { "epoch": 68.38113448056086, "grad_norm": 0.2427472174167633, "learning_rate": 7.628264794006182e-05, "loss": 0.0151, "step": 107290 }, { "epoch": 68.38750796685787, "grad_norm": 0.4088838994503021, "learning_rate": 7.627795909607944e-05, "loss": 0.0209, "step": 107300 }, { "epoch": 68.39388145315488, "grad_norm": 0.5233646631240845, "learning_rate": 7.627326993279506e-05, "loss": 0.0253, "step": 107310 }, { "epoch": 68.40025493945188, "grad_norm": 0.19374558329582214, "learning_rate": 7.626858045026566e-05, "loss": 0.02, "step": 107320 }, { "epoch": 68.40662842574889, "grad_norm": 0.2931341230869293, "learning_rate": 7.626389064854818e-05, "loss": 0.0232, "step": 107330 }, { "epoch": 68.41300191204589, "grad_norm": 0.7147870659828186, "learning_rate": 7.625920052769966e-05, "loss": 0.0178, "step": 107340 }, { "epoch": 68.4193753983429, "grad_norm": 0.3726840913295746, "learning_rate": 7.625451008777705e-05, "loss": 0.0206, "step": 107350 }, { "epoch": 68.4257488846399, "grad_norm": 0.7866557240486145, "learning_rate": 7.624981932883734e-05, "loss": 0.0354, "step": 107360 }, { "epoch": 68.43212237093691, "grad_norm": 0.5845111608505249, "learning_rate": 7.624512825093755e-05, "loss": 0.0217, "step": 107370 }, { "epoch": 68.4384958572339, "grad_norm": 0.31185394525527954, "learning_rate": 7.624043685413467e-05, "loss": 0.0202, "step": 107380 }, { "epoch": 68.44486934353091, "grad_norm": 0.6676550507545471, "learning_rate": 7.623574513848572e-05, "loss": 0.0176, "step": 107390 }, { "epoch": 68.45124282982792, "grad_norm": 0.333631694316864, "learning_rate": 7.623105310404768e-05, "loss": 0.0189, "step": 107400 }, { "epoch": 68.45761631612493, "grad_norm": 0.5050011277198792, "learning_rate": 7.622636075087759e-05, "loss": 0.022, "step": 107410 }, { "epoch": 68.46398980242192, "grad_norm": 0.6961150169372559, "learning_rate": 7.622166807903243e-05, "loss": 0.0209, "step": 107420 }, { "epoch": 68.47036328871893, "grad_norm": 1.3149696588516235, "learning_rate": 7.621697508856929e-05, "loss": 0.024, "step": 107430 }, { "epoch": 68.47673677501594, "grad_norm": 0.33559754490852356, "learning_rate": 7.621228177954513e-05, "loss": 0.0138, "step": 107440 }, { "epoch": 68.48311026131294, "grad_norm": 0.4553198218345642, "learning_rate": 7.620758815201698e-05, "loss": 0.0292, "step": 107450 }, { "epoch": 68.48948374760994, "grad_norm": 0.40704768896102905, "learning_rate": 7.620289420604192e-05, "loss": 0.0331, "step": 107460 }, { "epoch": 68.49585723390695, "grad_norm": 0.6535449028015137, "learning_rate": 7.619819994167692e-05, "loss": 0.019, "step": 107470 }, { "epoch": 68.50223072020395, "grad_norm": 0.5713701844215393, "learning_rate": 7.619350535897909e-05, "loss": 0.0222, "step": 107480 }, { "epoch": 68.50860420650096, "grad_norm": 0.6810839772224426, "learning_rate": 7.61888104580054e-05, "loss": 0.023, "step": 107490 }, { "epoch": 68.51497769279796, "grad_norm": 0.41756436228752136, "learning_rate": 7.618411523881297e-05, "loss": 0.0209, "step": 107500 }, { "epoch": 68.52135117909496, "grad_norm": 0.5397517681121826, "learning_rate": 7.617941970145881e-05, "loss": 0.0206, "step": 107510 }, { "epoch": 68.52772466539197, "grad_norm": 0.3397981524467468, "learning_rate": 7.617472384599999e-05, "loss": 0.0148, "step": 107520 }, { "epoch": 68.53409815168898, "grad_norm": 0.5578193068504333, "learning_rate": 7.617002767249355e-05, "loss": 0.0302, "step": 107530 }, { "epoch": 68.54047163798597, "grad_norm": 0.4449341595172882, "learning_rate": 7.616533118099657e-05, "loss": 0.0246, "step": 107540 }, { "epoch": 68.54684512428298, "grad_norm": 0.4021769165992737, "learning_rate": 7.61606343715661e-05, "loss": 0.0124, "step": 107550 }, { "epoch": 68.55321861057999, "grad_norm": 0.3518792688846588, "learning_rate": 7.615593724425925e-05, "loss": 0.0145, "step": 107560 }, { "epoch": 68.559592096877, "grad_norm": 0.2632792592048645, "learning_rate": 7.615123979913305e-05, "loss": 0.0151, "step": 107570 }, { "epoch": 68.56596558317399, "grad_norm": 0.3523036241531372, "learning_rate": 7.614654203624458e-05, "loss": 0.0128, "step": 107580 }, { "epoch": 68.572339069471, "grad_norm": 0.30193230509757996, "learning_rate": 7.614184395565096e-05, "loss": 0.01, "step": 107590 }, { "epoch": 68.578712555768, "grad_norm": 0.6037343144416809, "learning_rate": 7.613714555740924e-05, "loss": 0.0228, "step": 107600 }, { "epoch": 68.58508604206502, "grad_norm": 0.37733855843544006, "learning_rate": 7.613244684157654e-05, "loss": 0.021, "step": 107610 }, { "epoch": 68.59145952836201, "grad_norm": 0.6305876970291138, "learning_rate": 7.612774780820994e-05, "loss": 0.0224, "step": 107620 }, { "epoch": 68.59783301465902, "grad_norm": 0.47082987427711487, "learning_rate": 7.612304845736651e-05, "loss": 0.0177, "step": 107630 }, { "epoch": 68.60420650095602, "grad_norm": 0.25820809602737427, "learning_rate": 7.61183487891034e-05, "loss": 0.0226, "step": 107640 }, { "epoch": 68.61057998725303, "grad_norm": 0.21953780949115753, "learning_rate": 7.61136488034777e-05, "loss": 0.0208, "step": 107650 }, { "epoch": 68.61695347355003, "grad_norm": 0.4183875024318695, "learning_rate": 7.610894850054648e-05, "loss": 0.0114, "step": 107660 }, { "epoch": 68.62332695984703, "grad_norm": 0.5300595164299011, "learning_rate": 7.61042478803669e-05, "loss": 0.0242, "step": 107670 }, { "epoch": 68.62970044614404, "grad_norm": 0.409984827041626, "learning_rate": 7.609954694299607e-05, "loss": 0.0119, "step": 107680 }, { "epoch": 68.63607393244105, "grad_norm": 0.3815591633319855, "learning_rate": 7.60948456884911e-05, "loss": 0.0137, "step": 107690 }, { "epoch": 68.64244741873804, "grad_norm": 0.35697662830352783, "learning_rate": 7.60901441169091e-05, "loss": 0.0241, "step": 107700 }, { "epoch": 68.64882090503505, "grad_norm": 0.39731234312057495, "learning_rate": 7.608544222830724e-05, "loss": 0.0103, "step": 107710 }, { "epoch": 68.65519439133206, "grad_norm": 0.3648849427700043, "learning_rate": 7.608074002274262e-05, "loss": 0.0175, "step": 107720 }, { "epoch": 68.66156787762907, "grad_norm": 0.42588314414024353, "learning_rate": 7.607603750027237e-05, "loss": 0.0156, "step": 107730 }, { "epoch": 68.66794136392606, "grad_norm": 0.3402612507343292, "learning_rate": 7.607133466095365e-05, "loss": 0.0112, "step": 107740 }, { "epoch": 68.67431485022307, "grad_norm": 0.47347623109817505, "learning_rate": 7.606663150484362e-05, "loss": 0.0276, "step": 107750 }, { "epoch": 68.68068833652008, "grad_norm": 0.35613930225372314, "learning_rate": 7.606192803199938e-05, "loss": 0.0126, "step": 107760 }, { "epoch": 68.68706182281709, "grad_norm": 0.4113629162311554, "learning_rate": 7.605722424247812e-05, "loss": 0.0334, "step": 107770 }, { "epoch": 68.69343530911408, "grad_norm": 0.2720666825771332, "learning_rate": 7.605252013633699e-05, "loss": 0.0282, "step": 107780 }, { "epoch": 68.69980879541109, "grad_norm": 0.38510674238204956, "learning_rate": 7.604781571363313e-05, "loss": 0.0233, "step": 107790 }, { "epoch": 68.7061822817081, "grad_norm": 0.5955936908721924, "learning_rate": 7.604311097442372e-05, "loss": 0.0402, "step": 107800 }, { "epoch": 68.7125557680051, "grad_norm": 0.5176739692687988, "learning_rate": 7.603840591876592e-05, "loss": 0.0202, "step": 107810 }, { "epoch": 68.7189292543021, "grad_norm": 0.20203040540218353, "learning_rate": 7.60337005467169e-05, "loss": 0.0114, "step": 107820 }, { "epoch": 68.7253027405991, "grad_norm": 0.34093043208122253, "learning_rate": 7.602899485833384e-05, "loss": 0.0186, "step": 107830 }, { "epoch": 68.73167622689611, "grad_norm": 0.3843778371810913, "learning_rate": 7.602428885367395e-05, "loss": 0.022, "step": 107840 }, { "epoch": 68.73804971319312, "grad_norm": 0.4147146940231323, "learning_rate": 7.601958253279435e-05, "loss": 0.0179, "step": 107850 }, { "epoch": 68.74442319949011, "grad_norm": 0.44987133145332336, "learning_rate": 7.601487589575224e-05, "loss": 0.0188, "step": 107860 }, { "epoch": 68.75079668578712, "grad_norm": 0.1732083559036255, "learning_rate": 7.601016894260485e-05, "loss": 0.0137, "step": 107870 }, { "epoch": 68.75717017208413, "grad_norm": 0.5704591274261475, "learning_rate": 7.600546167340934e-05, "loss": 0.015, "step": 107880 }, { "epoch": 68.76354365838114, "grad_norm": 0.4458199143409729, "learning_rate": 7.600075408822291e-05, "loss": 0.0103, "step": 107890 }, { "epoch": 68.76991714467813, "grad_norm": 0.1648261696100235, "learning_rate": 7.599604618710279e-05, "loss": 0.0181, "step": 107900 }, { "epoch": 68.77629063097514, "grad_norm": 0.2667195498943329, "learning_rate": 7.599133797010616e-05, "loss": 0.0199, "step": 107910 }, { "epoch": 68.78266411727215, "grad_norm": 0.23429815471172333, "learning_rate": 7.598662943729021e-05, "loss": 0.014, "step": 107920 }, { "epoch": 68.78903760356916, "grad_norm": 0.262127548456192, "learning_rate": 7.598192058871221e-05, "loss": 0.0189, "step": 107930 }, { "epoch": 68.79541108986615, "grad_norm": 0.22058309614658356, "learning_rate": 7.597721142442932e-05, "loss": 0.0218, "step": 107940 }, { "epoch": 68.80178457616316, "grad_norm": 0.36248114705085754, "learning_rate": 7.597250194449878e-05, "loss": 0.0168, "step": 107950 }, { "epoch": 68.80815806246017, "grad_norm": 0.2068217694759369, "learning_rate": 7.596779214897784e-05, "loss": 0.0184, "step": 107960 }, { "epoch": 68.81453154875717, "grad_norm": 0.6522926688194275, "learning_rate": 7.59630820379237e-05, "loss": 0.0313, "step": 107970 }, { "epoch": 68.82090503505418, "grad_norm": 0.6307727098464966, "learning_rate": 7.59583716113936e-05, "loss": 0.0157, "step": 107980 }, { "epoch": 68.82727852135118, "grad_norm": 0.3160719573497772, "learning_rate": 7.595366086944478e-05, "loss": 0.0201, "step": 107990 }, { "epoch": 68.83365200764818, "grad_norm": 0.5050654411315918, "learning_rate": 7.594894981213447e-05, "loss": 0.0183, "step": 108000 }, { "epoch": 68.84002549394519, "grad_norm": 0.4164694845676422, "learning_rate": 7.594423843951992e-05, "loss": 0.0133, "step": 108010 }, { "epoch": 68.8463989802422, "grad_norm": 0.32257139682769775, "learning_rate": 7.593952675165837e-05, "loss": 0.0145, "step": 108020 }, { "epoch": 68.8527724665392, "grad_norm": 0.4874633550643921, "learning_rate": 7.593481474860709e-05, "loss": 0.0304, "step": 108030 }, { "epoch": 68.8591459528362, "grad_norm": 0.7851148843765259, "learning_rate": 7.593010243042333e-05, "loss": 0.0179, "step": 108040 }, { "epoch": 68.86551943913321, "grad_norm": 0.29480835795402527, "learning_rate": 7.592538979716434e-05, "loss": 0.0212, "step": 108050 }, { "epoch": 68.87189292543022, "grad_norm": 0.6148595809936523, "learning_rate": 7.592067684888737e-05, "loss": 0.0208, "step": 108060 }, { "epoch": 68.87826641172721, "grad_norm": 0.4410087764263153, "learning_rate": 7.591596358564971e-05, "loss": 0.0172, "step": 108070 }, { "epoch": 68.88463989802422, "grad_norm": 0.5094189047813416, "learning_rate": 7.591125000750864e-05, "loss": 0.014, "step": 108080 }, { "epoch": 68.89101338432123, "grad_norm": 0.4877133369445801, "learning_rate": 7.59065361145214e-05, "loss": 0.0246, "step": 108090 }, { "epoch": 68.89738687061823, "grad_norm": 0.5226429104804993, "learning_rate": 7.59018219067453e-05, "loss": 0.0185, "step": 108100 }, { "epoch": 68.90376035691523, "grad_norm": 0.3255905210971832, "learning_rate": 7.58971073842376e-05, "loss": 0.0124, "step": 108110 }, { "epoch": 68.91013384321224, "grad_norm": 0.3183947801589966, "learning_rate": 7.58923925470556e-05, "loss": 0.0176, "step": 108120 }, { "epoch": 68.91650732950924, "grad_norm": 0.4008593261241913, "learning_rate": 7.588767739525659e-05, "loss": 0.0165, "step": 108130 }, { "epoch": 68.92288081580625, "grad_norm": 0.9146546721458435, "learning_rate": 7.588296192889784e-05, "loss": 0.0239, "step": 108140 }, { "epoch": 68.92925430210325, "grad_norm": 0.4135487675666809, "learning_rate": 7.587824614803667e-05, "loss": 0.0156, "step": 108150 }, { "epoch": 68.93562778840025, "grad_norm": 0.3348453938961029, "learning_rate": 7.58735300527304e-05, "loss": 0.0105, "step": 108160 }, { "epoch": 68.94200127469726, "grad_norm": 0.2534012496471405, "learning_rate": 7.586881364303629e-05, "loss": 0.0132, "step": 108170 }, { "epoch": 68.94837476099427, "grad_norm": 0.28328579664230347, "learning_rate": 7.586409691901167e-05, "loss": 0.0158, "step": 108180 }, { "epoch": 68.95474824729126, "grad_norm": 0.45651426911354065, "learning_rate": 7.585937988071386e-05, "loss": 0.0212, "step": 108190 }, { "epoch": 68.96112173358827, "grad_norm": 0.43294382095336914, "learning_rate": 7.585466252820014e-05, "loss": 0.0117, "step": 108200 }, { "epoch": 68.96749521988528, "grad_norm": 0.31954821944236755, "learning_rate": 7.584994486152788e-05, "loss": 0.0146, "step": 108210 }, { "epoch": 68.97386870618229, "grad_norm": 0.7792256474494934, "learning_rate": 7.58452268807544e-05, "loss": 0.0169, "step": 108220 }, { "epoch": 68.98024219247928, "grad_norm": 0.3245939612388611, "learning_rate": 7.584050858593699e-05, "loss": 0.0245, "step": 108230 }, { "epoch": 68.98661567877629, "grad_norm": 0.3179319500923157, "learning_rate": 7.5835789977133e-05, "loss": 0.0183, "step": 108240 }, { "epoch": 68.9929891650733, "grad_norm": 0.5070083141326904, "learning_rate": 7.583107105439979e-05, "loss": 0.0175, "step": 108250 }, { "epoch": 68.9993626513703, "grad_norm": 0.47172659635543823, "learning_rate": 7.582635181779467e-05, "loss": 0.0156, "step": 108260 }, { "epoch": 69.0057361376673, "grad_norm": 0.4728893041610718, "learning_rate": 7.582163226737499e-05, "loss": 0.0183, "step": 108270 }, { "epoch": 69.0121096239643, "grad_norm": 0.7240013480186462, "learning_rate": 7.58169124031981e-05, "loss": 0.0297, "step": 108280 }, { "epoch": 69.01848311026131, "grad_norm": 0.5990631580352783, "learning_rate": 7.581219222532134e-05, "loss": 0.0227, "step": 108290 }, { "epoch": 69.02485659655832, "grad_norm": 0.49733075499534607, "learning_rate": 7.580747173380209e-05, "loss": 0.0138, "step": 108300 }, { "epoch": 69.03123008285532, "grad_norm": 0.5253685712814331, "learning_rate": 7.580275092869766e-05, "loss": 0.0278, "step": 108310 }, { "epoch": 69.03760356915232, "grad_norm": 0.26072439551353455, "learning_rate": 7.579802981006548e-05, "loss": 0.0121, "step": 108320 }, { "epoch": 69.04397705544933, "grad_norm": 0.38928183913230896, "learning_rate": 7.579330837796287e-05, "loss": 0.0133, "step": 108330 }, { "epoch": 69.05035054174634, "grad_norm": 0.3050217032432556, "learning_rate": 7.57885866324472e-05, "loss": 0.0218, "step": 108340 }, { "epoch": 69.05672402804333, "grad_norm": 0.4450750946998596, "learning_rate": 7.578386457357587e-05, "loss": 0.0128, "step": 108350 }, { "epoch": 69.06309751434034, "grad_norm": 0.6043259501457214, "learning_rate": 7.577914220140623e-05, "loss": 0.0171, "step": 108360 }, { "epoch": 69.06947100063735, "grad_norm": 0.17940351366996765, "learning_rate": 7.577441951599567e-05, "loss": 0.0161, "step": 108370 }, { "epoch": 69.07584448693436, "grad_norm": 0.4216276705265045, "learning_rate": 7.576969651740157e-05, "loss": 0.0303, "step": 108380 }, { "epoch": 69.08221797323135, "grad_norm": 0.5170851349830627, "learning_rate": 7.576497320568135e-05, "loss": 0.0216, "step": 108390 }, { "epoch": 69.08859145952836, "grad_norm": 0.6065660715103149, "learning_rate": 7.576024958089236e-05, "loss": 0.0216, "step": 108400 }, { "epoch": 69.09496494582537, "grad_norm": 0.36588621139526367, "learning_rate": 7.575552564309202e-05, "loss": 0.0265, "step": 108410 }, { "epoch": 69.10133843212238, "grad_norm": 0.800003170967102, "learning_rate": 7.575080139233774e-05, "loss": 0.0162, "step": 108420 }, { "epoch": 69.10771191841937, "grad_norm": 0.4325380027294159, "learning_rate": 7.57460768286869e-05, "loss": 0.0171, "step": 108430 }, { "epoch": 69.11408540471638, "grad_norm": 0.40484583377838135, "learning_rate": 7.574135195219694e-05, "loss": 0.0234, "step": 108440 }, { "epoch": 69.12045889101339, "grad_norm": 0.407765656709671, "learning_rate": 7.57366267629252e-05, "loss": 0.0215, "step": 108450 }, { "epoch": 69.1268323773104, "grad_norm": 0.2727501094341278, "learning_rate": 7.573190126092919e-05, "loss": 0.0227, "step": 108460 }, { "epoch": 69.13320586360739, "grad_norm": 0.8473920822143555, "learning_rate": 7.572717544626627e-05, "loss": 0.0173, "step": 108470 }, { "epoch": 69.1395793499044, "grad_norm": 0.2807987332344055, "learning_rate": 7.572244931899388e-05, "loss": 0.0114, "step": 108480 }, { "epoch": 69.1459528362014, "grad_norm": 0.38958507776260376, "learning_rate": 7.571772287916945e-05, "loss": 0.0258, "step": 108490 }, { "epoch": 69.15232632249841, "grad_norm": 0.3884463906288147, "learning_rate": 7.57129961268504e-05, "loss": 0.0124, "step": 108500 }, { "epoch": 69.1586998087954, "grad_norm": 0.5014525055885315, "learning_rate": 7.570826906209417e-05, "loss": 0.0175, "step": 108510 }, { "epoch": 69.16507329509241, "grad_norm": 0.2555745840072632, "learning_rate": 7.570354168495818e-05, "loss": 0.0176, "step": 108520 }, { "epoch": 69.17144678138942, "grad_norm": 0.4079521596431732, "learning_rate": 7.569881399549992e-05, "loss": 0.0223, "step": 108530 }, { "epoch": 69.17782026768643, "grad_norm": 0.48922199010849, "learning_rate": 7.569408599377678e-05, "loss": 0.0259, "step": 108540 }, { "epoch": 69.18419375398342, "grad_norm": 0.36028140783309937, "learning_rate": 7.568935767984624e-05, "loss": 0.0127, "step": 108550 }, { "epoch": 69.19056724028043, "grad_norm": 0.3056938946247101, "learning_rate": 7.568462905376577e-05, "loss": 0.0171, "step": 108560 }, { "epoch": 69.19694072657744, "grad_norm": 0.48632609844207764, "learning_rate": 7.567990011559278e-05, "loss": 0.0143, "step": 108570 }, { "epoch": 69.20331421287445, "grad_norm": 0.2394338846206665, "learning_rate": 7.567517086538478e-05, "loss": 0.0207, "step": 108580 }, { "epoch": 69.20968769917145, "grad_norm": 0.325132817029953, "learning_rate": 7.56704413031992e-05, "loss": 0.0191, "step": 108590 }, { "epoch": 69.21606118546845, "grad_norm": 0.5196437239646912, "learning_rate": 7.566571142909353e-05, "loss": 0.0325, "step": 108600 }, { "epoch": 69.22243467176546, "grad_norm": 0.5635237693786621, "learning_rate": 7.566098124312524e-05, "loss": 0.0212, "step": 108610 }, { "epoch": 69.22880815806246, "grad_norm": 0.2639165222644806, "learning_rate": 7.565625074535179e-05, "loss": 0.0235, "step": 108620 }, { "epoch": 69.23518164435947, "grad_norm": 0.3383626341819763, "learning_rate": 7.565151993583066e-05, "loss": 0.0266, "step": 108630 }, { "epoch": 69.24155513065647, "grad_norm": 0.4056667685508728, "learning_rate": 7.564678881461935e-05, "loss": 0.0169, "step": 108640 }, { "epoch": 69.24792861695347, "grad_norm": 0.2835894823074341, "learning_rate": 7.564205738177535e-05, "loss": 0.0233, "step": 108650 }, { "epoch": 69.25430210325048, "grad_norm": 0.2543014585971832, "learning_rate": 7.563732563735613e-05, "loss": 0.0147, "step": 108660 }, { "epoch": 69.26067558954749, "grad_norm": 0.2068164050579071, "learning_rate": 7.56325935814192e-05, "loss": 0.016, "step": 108670 }, { "epoch": 69.26704907584448, "grad_norm": 0.19365648925304413, "learning_rate": 7.562786121402208e-05, "loss": 0.0157, "step": 108680 }, { "epoch": 69.27342256214149, "grad_norm": 0.27341821789741516, "learning_rate": 7.562312853522222e-05, "loss": 0.0164, "step": 108690 }, { "epoch": 69.2797960484385, "grad_norm": 0.3066481351852417, "learning_rate": 7.561839554507717e-05, "loss": 0.0234, "step": 108700 }, { "epoch": 69.2861695347355, "grad_norm": 0.363720178604126, "learning_rate": 7.561366224364443e-05, "loss": 0.021, "step": 108710 }, { "epoch": 69.2925430210325, "grad_norm": 0.5038548111915588, "learning_rate": 7.56089286309815e-05, "loss": 0.0335, "step": 108720 }, { "epoch": 69.29891650732951, "grad_norm": 0.5308862328529358, "learning_rate": 7.56041947071459e-05, "loss": 0.0248, "step": 108730 }, { "epoch": 69.30528999362652, "grad_norm": 0.759466826915741, "learning_rate": 7.559946047219518e-05, "loss": 0.0298, "step": 108740 }, { "epoch": 69.31166347992352, "grad_norm": 0.7935208082199097, "learning_rate": 7.559472592618683e-05, "loss": 0.0258, "step": 108750 }, { "epoch": 69.31803696622052, "grad_norm": 0.26888805627822876, "learning_rate": 7.558999106917841e-05, "loss": 0.0127, "step": 108760 }, { "epoch": 69.32441045251753, "grad_norm": 0.4722280204296112, "learning_rate": 7.558525590122743e-05, "loss": 0.0262, "step": 108770 }, { "epoch": 69.33078393881453, "grad_norm": 0.7265624403953552, "learning_rate": 7.558052042239144e-05, "loss": 0.0133, "step": 108780 }, { "epoch": 69.33715742511154, "grad_norm": 0.26435384154319763, "learning_rate": 7.557578463272797e-05, "loss": 0.0194, "step": 108790 }, { "epoch": 69.34353091140854, "grad_norm": 0.5723888278007507, "learning_rate": 7.557104853229457e-05, "loss": 0.0175, "step": 108800 }, { "epoch": 69.34990439770554, "grad_norm": 0.3668348789215088, "learning_rate": 7.556631212114878e-05, "loss": 0.0215, "step": 108810 }, { "epoch": 69.35627788400255, "grad_norm": 0.2692742943763733, "learning_rate": 7.556157539934817e-05, "loss": 0.0209, "step": 108820 }, { "epoch": 69.36265137029956, "grad_norm": 0.3940919041633606, "learning_rate": 7.555683836695028e-05, "loss": 0.0154, "step": 108830 }, { "epoch": 69.36902485659655, "grad_norm": 0.24596382677555084, "learning_rate": 7.555210102401267e-05, "loss": 0.0142, "step": 108840 }, { "epoch": 69.37539834289356, "grad_norm": 0.289396196603775, "learning_rate": 7.554736337059292e-05, "loss": 0.0173, "step": 108850 }, { "epoch": 69.38177182919057, "grad_norm": 0.40313586592674255, "learning_rate": 7.554262540674857e-05, "loss": 0.0171, "step": 108860 }, { "epoch": 69.38814531548758, "grad_norm": 0.6086888909339905, "learning_rate": 7.55378871325372e-05, "loss": 0.0266, "step": 108870 }, { "epoch": 69.39451880178457, "grad_norm": 0.2553968131542206, "learning_rate": 7.553314854801641e-05, "loss": 0.0145, "step": 108880 }, { "epoch": 69.40089228808158, "grad_norm": 0.46494343876838684, "learning_rate": 7.552840965324375e-05, "loss": 0.0171, "step": 108890 }, { "epoch": 69.40726577437859, "grad_norm": 0.45463627576828003, "learning_rate": 7.552367044827681e-05, "loss": 0.0158, "step": 108900 }, { "epoch": 69.4136392606756, "grad_norm": 0.26690274477005005, "learning_rate": 7.551893093317318e-05, "loss": 0.0175, "step": 108910 }, { "epoch": 69.42001274697259, "grad_norm": 0.6234889626502991, "learning_rate": 7.551419110799043e-05, "loss": 0.0161, "step": 108920 }, { "epoch": 69.4263862332696, "grad_norm": 0.40611496567726135, "learning_rate": 7.550945097278617e-05, "loss": 0.015, "step": 108930 }, { "epoch": 69.4327597195666, "grad_norm": 0.39796438813209534, "learning_rate": 7.550471052761801e-05, "loss": 0.0288, "step": 108940 }, { "epoch": 69.43913320586361, "grad_norm": 0.7154423594474792, "learning_rate": 7.549996977254352e-05, "loss": 0.018, "step": 108950 }, { "epoch": 69.4455066921606, "grad_norm": 0.39691969752311707, "learning_rate": 7.549522870762034e-05, "loss": 0.0169, "step": 108960 }, { "epoch": 69.45188017845761, "grad_norm": 0.2973479628562927, "learning_rate": 7.549048733290604e-05, "loss": 0.0137, "step": 108970 }, { "epoch": 69.45825366475462, "grad_norm": 0.23621079325675964, "learning_rate": 7.548574564845827e-05, "loss": 0.0218, "step": 108980 }, { "epoch": 69.46462715105163, "grad_norm": 0.5880181789398193, "learning_rate": 7.548100365433462e-05, "loss": 0.0235, "step": 108990 }, { "epoch": 69.47100063734862, "grad_norm": 0.28229820728302, "learning_rate": 7.547626135059272e-05, "loss": 0.0146, "step": 109000 }, { "epoch": 69.47737412364563, "grad_norm": 0.3866879343986511, "learning_rate": 7.547151873729019e-05, "loss": 0.0173, "step": 109010 }, { "epoch": 69.48374760994264, "grad_norm": 0.6951055526733398, "learning_rate": 7.546677581448466e-05, "loss": 0.0226, "step": 109020 }, { "epoch": 69.49012109623965, "grad_norm": 0.5561010837554932, "learning_rate": 7.546203258223375e-05, "loss": 0.0148, "step": 109030 }, { "epoch": 69.49649458253664, "grad_norm": 0.31488171219825745, "learning_rate": 7.545728904059511e-05, "loss": 0.0196, "step": 109040 }, { "epoch": 69.50286806883365, "grad_norm": 0.4457908868789673, "learning_rate": 7.545254518962638e-05, "loss": 0.0219, "step": 109050 }, { "epoch": 69.50924155513066, "grad_norm": 0.5393133163452148, "learning_rate": 7.54478010293852e-05, "loss": 0.0241, "step": 109060 }, { "epoch": 69.51561504142767, "grad_norm": 0.5015327334403992, "learning_rate": 7.544305655992919e-05, "loss": 0.0129, "step": 109070 }, { "epoch": 69.52198852772466, "grad_norm": 0.35094714164733887, "learning_rate": 7.543831178131603e-05, "loss": 0.0088, "step": 109080 }, { "epoch": 69.52836201402167, "grad_norm": 0.3180698752403259, "learning_rate": 7.543356669360335e-05, "loss": 0.0208, "step": 109090 }, { "epoch": 69.53473550031867, "grad_norm": 0.6049396395683289, "learning_rate": 7.542882129684885e-05, "loss": 0.0232, "step": 109100 }, { "epoch": 69.54110898661568, "grad_norm": 0.4450114965438843, "learning_rate": 7.542407559111013e-05, "loss": 0.0227, "step": 109110 }, { "epoch": 69.54748247291268, "grad_norm": 0.5628753304481506, "learning_rate": 7.541932957644492e-05, "loss": 0.0188, "step": 109120 }, { "epoch": 69.55385595920968, "grad_norm": 0.27134501934051514, "learning_rate": 7.541458325291085e-05, "loss": 0.0147, "step": 109130 }, { "epoch": 69.56022944550669, "grad_norm": 0.6742274761199951, "learning_rate": 7.540983662056557e-05, "loss": 0.0184, "step": 109140 }, { "epoch": 69.5666029318037, "grad_norm": 0.6480090022087097, "learning_rate": 7.540508967946682e-05, "loss": 0.0258, "step": 109150 }, { "epoch": 69.5729764181007, "grad_norm": 0.3208577036857605, "learning_rate": 7.540034242967223e-05, "loss": 0.0202, "step": 109160 }, { "epoch": 69.5793499043977, "grad_norm": 0.2661128640174866, "learning_rate": 7.539559487123949e-05, "loss": 0.0148, "step": 109170 }, { "epoch": 69.58572339069471, "grad_norm": 0.31732413172721863, "learning_rate": 7.53908470042263e-05, "loss": 0.0133, "step": 109180 }, { "epoch": 69.59209687699172, "grad_norm": 0.5046302676200867, "learning_rate": 7.538609882869035e-05, "loss": 0.0183, "step": 109190 }, { "epoch": 69.59847036328873, "grad_norm": 0.25919511914253235, "learning_rate": 7.538135034468931e-05, "loss": 0.0132, "step": 109200 }, { "epoch": 69.60484384958572, "grad_norm": 0.4193447530269623, "learning_rate": 7.537660155228092e-05, "loss": 0.0229, "step": 109210 }, { "epoch": 69.61121733588273, "grad_norm": 0.30501729249954224, "learning_rate": 7.537185245152286e-05, "loss": 0.018, "step": 109220 }, { "epoch": 69.61759082217974, "grad_norm": 0.40610194206237793, "learning_rate": 7.536710304247283e-05, "loss": 0.023, "step": 109230 }, { "epoch": 69.62396430847674, "grad_norm": 0.24277259409427643, "learning_rate": 7.536235332518855e-05, "loss": 0.0158, "step": 109240 }, { "epoch": 69.63033779477374, "grad_norm": 0.532320499420166, "learning_rate": 7.535760329972772e-05, "loss": 0.0176, "step": 109250 }, { "epoch": 69.63671128107075, "grad_norm": 0.2230072319507599, "learning_rate": 7.535285296614809e-05, "loss": 0.017, "step": 109260 }, { "epoch": 69.64308476736775, "grad_norm": 0.16715764999389648, "learning_rate": 7.534810232450736e-05, "loss": 0.0126, "step": 109270 }, { "epoch": 69.64945825366476, "grad_norm": 0.2856670022010803, "learning_rate": 7.534335137486323e-05, "loss": 0.0146, "step": 109280 }, { "epoch": 69.65583173996176, "grad_norm": 0.3535715341567993, "learning_rate": 7.533860011727348e-05, "loss": 0.0298, "step": 109290 }, { "epoch": 69.66220522625876, "grad_norm": 0.5474178791046143, "learning_rate": 7.533384855179581e-05, "loss": 0.0207, "step": 109300 }, { "epoch": 69.66857871255577, "grad_norm": 0.3114345967769623, "learning_rate": 7.532909667848795e-05, "loss": 0.0204, "step": 109310 }, { "epoch": 69.67495219885278, "grad_norm": 0.2399299591779709, "learning_rate": 7.532434449740765e-05, "loss": 0.0181, "step": 109320 }, { "epoch": 69.68132568514977, "grad_norm": 0.4175562858581543, "learning_rate": 7.531959200861267e-05, "loss": 0.0199, "step": 109330 }, { "epoch": 69.68769917144678, "grad_norm": 0.3727715015411377, "learning_rate": 7.531483921216072e-05, "loss": 0.0183, "step": 109340 }, { "epoch": 69.69407265774379, "grad_norm": 0.6036545634269714, "learning_rate": 7.531008610810958e-05, "loss": 0.017, "step": 109350 }, { "epoch": 69.7004461440408, "grad_norm": 0.314609169960022, "learning_rate": 7.530533269651701e-05, "loss": 0.0206, "step": 109360 }, { "epoch": 69.70681963033779, "grad_norm": 0.4172629714012146, "learning_rate": 7.530057897744074e-05, "loss": 0.014, "step": 109370 }, { "epoch": 69.7131931166348, "grad_norm": 0.2995571792125702, "learning_rate": 7.529582495093854e-05, "loss": 0.0187, "step": 109380 }, { "epoch": 69.7195666029318, "grad_norm": 0.4165508449077606, "learning_rate": 7.529107061706822e-05, "loss": 0.0277, "step": 109390 }, { "epoch": 69.72594008922881, "grad_norm": 0.21330003440380096, "learning_rate": 7.528631597588746e-05, "loss": 0.022, "step": 109400 }, { "epoch": 69.73231357552581, "grad_norm": 0.5039772391319275, "learning_rate": 7.528156102745413e-05, "loss": 0.0279, "step": 109410 }, { "epoch": 69.73868706182282, "grad_norm": 0.6918461322784424, "learning_rate": 7.527680577182595e-05, "loss": 0.0268, "step": 109420 }, { "epoch": 69.74506054811982, "grad_norm": 0.4251154959201813, "learning_rate": 7.527205020906071e-05, "loss": 0.0176, "step": 109430 }, { "epoch": 69.75143403441683, "grad_norm": 0.6789569854736328, "learning_rate": 7.52672943392162e-05, "loss": 0.0278, "step": 109440 }, { "epoch": 69.75780752071383, "grad_norm": 0.3526040315628052, "learning_rate": 7.526253816235023e-05, "loss": 0.0194, "step": 109450 }, { "epoch": 69.76418100701083, "grad_norm": 0.26495787501335144, "learning_rate": 7.525778167852054e-05, "loss": 0.0209, "step": 109460 }, { "epoch": 69.77055449330784, "grad_norm": 0.21037988364696503, "learning_rate": 7.525302488778497e-05, "loss": 0.0273, "step": 109470 }, { "epoch": 69.77692797960485, "grad_norm": 0.6247171759605408, "learning_rate": 7.52482677902013e-05, "loss": 0.0299, "step": 109480 }, { "epoch": 69.78330146590184, "grad_norm": 0.46315404772758484, "learning_rate": 7.524351038582735e-05, "loss": 0.0164, "step": 109490 }, { "epoch": 69.78967495219885, "grad_norm": 0.635569155216217, "learning_rate": 7.523875267472089e-05, "loss": 0.0328, "step": 109500 }, { "epoch": 69.79604843849586, "grad_norm": 0.41595327854156494, "learning_rate": 7.523399465693977e-05, "loss": 0.0159, "step": 109510 }, { "epoch": 69.80242192479287, "grad_norm": 0.28905194997787476, "learning_rate": 7.52292363325418e-05, "loss": 0.0188, "step": 109520 }, { "epoch": 69.80879541108986, "grad_norm": 0.2691091299057007, "learning_rate": 7.522447770158477e-05, "loss": 0.0149, "step": 109530 }, { "epoch": 69.81516889738687, "grad_norm": 0.4444141685962677, "learning_rate": 7.521971876412653e-05, "loss": 0.0189, "step": 109540 }, { "epoch": 69.82154238368388, "grad_norm": 0.44206503033638, "learning_rate": 7.521495952022489e-05, "loss": 0.0224, "step": 109550 }, { "epoch": 69.82791586998088, "grad_norm": 0.5703757405281067, "learning_rate": 7.52101999699377e-05, "loss": 0.0225, "step": 109560 }, { "epoch": 69.83428935627788, "grad_norm": 0.3079952299594879, "learning_rate": 7.520544011332276e-05, "loss": 0.0195, "step": 109570 }, { "epoch": 69.84066284257489, "grad_norm": 0.6065131425857544, "learning_rate": 7.520067995043794e-05, "loss": 0.0222, "step": 109580 }, { "epoch": 69.8470363288719, "grad_norm": 0.4130432903766632, "learning_rate": 7.519591948134104e-05, "loss": 0.0132, "step": 109590 }, { "epoch": 69.8534098151689, "grad_norm": 0.609714150428772, "learning_rate": 7.519115870608996e-05, "loss": 0.0415, "step": 109600 }, { "epoch": 69.8597833014659, "grad_norm": 0.5145106911659241, "learning_rate": 7.51863976247425e-05, "loss": 0.0207, "step": 109610 }, { "epoch": 69.8661567877629, "grad_norm": 0.2117522656917572, "learning_rate": 7.518163623735654e-05, "loss": 0.0203, "step": 109620 }, { "epoch": 69.87253027405991, "grad_norm": 0.25884926319122314, "learning_rate": 7.517687454398992e-05, "loss": 0.0157, "step": 109630 }, { "epoch": 69.87890376035692, "grad_norm": 0.31358516216278076, "learning_rate": 7.517211254470049e-05, "loss": 0.0198, "step": 109640 }, { "epoch": 69.88527724665391, "grad_norm": 0.6769780516624451, "learning_rate": 7.516735023954614e-05, "loss": 0.0202, "step": 109650 }, { "epoch": 69.89165073295092, "grad_norm": 0.28477713465690613, "learning_rate": 7.516258762858472e-05, "loss": 0.016, "step": 109660 }, { "epoch": 69.89802421924793, "grad_norm": 0.432976633310318, "learning_rate": 7.51578247118741e-05, "loss": 0.0181, "step": 109670 }, { "epoch": 69.90439770554494, "grad_norm": 0.6001273989677429, "learning_rate": 7.515306148947215e-05, "loss": 0.0184, "step": 109680 }, { "epoch": 69.91077119184193, "grad_norm": 0.4394828975200653, "learning_rate": 7.514829796143677e-05, "loss": 0.0155, "step": 109690 }, { "epoch": 69.91714467813894, "grad_norm": 0.6605792045593262, "learning_rate": 7.514353412782583e-05, "loss": 0.0301, "step": 109700 }, { "epoch": 69.92351816443595, "grad_norm": 0.20071323215961456, "learning_rate": 7.513876998869719e-05, "loss": 0.0128, "step": 109710 }, { "epoch": 69.92989165073296, "grad_norm": 0.456286758184433, "learning_rate": 7.513400554410877e-05, "loss": 0.0173, "step": 109720 }, { "epoch": 69.93626513702995, "grad_norm": 0.2944415807723999, "learning_rate": 7.512924079411846e-05, "loss": 0.0204, "step": 109730 }, { "epoch": 69.94263862332696, "grad_norm": 0.5973456501960754, "learning_rate": 7.512447573878414e-05, "loss": 0.0167, "step": 109740 }, { "epoch": 69.94901210962396, "grad_norm": 0.4693790078163147, "learning_rate": 7.51197103781637e-05, "loss": 0.0211, "step": 109750 }, { "epoch": 69.95538559592097, "grad_norm": 0.2627420723438263, "learning_rate": 7.511494471231509e-05, "loss": 0.0192, "step": 109760 }, { "epoch": 69.96175908221797, "grad_norm": 0.32957616448402405, "learning_rate": 7.511017874129618e-05, "loss": 0.0138, "step": 109770 }, { "epoch": 69.96813256851497, "grad_norm": 0.23858226835727692, "learning_rate": 7.51054124651649e-05, "loss": 0.0147, "step": 109780 }, { "epoch": 69.97450605481198, "grad_norm": 0.516528308391571, "learning_rate": 7.510064588397914e-05, "loss": 0.0176, "step": 109790 }, { "epoch": 69.98087954110899, "grad_norm": 0.45180732011795044, "learning_rate": 7.509587899779684e-05, "loss": 0.0185, "step": 109800 }, { "epoch": 69.98725302740598, "grad_norm": 0.11954141408205032, "learning_rate": 7.509111180667591e-05, "loss": 0.0189, "step": 109810 }, { "epoch": 69.99362651370299, "grad_norm": 0.4547651708126068, "learning_rate": 7.50863443106743e-05, "loss": 0.0124, "step": 109820 }, { "epoch": 70.0, "grad_norm": 0.34960684180259705, "learning_rate": 7.50815765098499e-05, "loss": 0.0145, "step": 109830 }, { "epoch": 70.00637348629701, "grad_norm": 0.2908296585083008, "learning_rate": 7.507680840426067e-05, "loss": 0.021, "step": 109840 }, { "epoch": 70.01274697259402, "grad_norm": 0.6088809370994568, "learning_rate": 7.507203999396455e-05, "loss": 0.0323, "step": 109850 }, { "epoch": 70.01912045889101, "grad_norm": 0.6494719982147217, "learning_rate": 7.506727127901946e-05, "loss": 0.0407, "step": 109860 }, { "epoch": 70.02549394518802, "grad_norm": 0.5126445293426514, "learning_rate": 7.506250225948336e-05, "loss": 0.0187, "step": 109870 }, { "epoch": 70.03186743148503, "grad_norm": 0.5078320503234863, "learning_rate": 7.505773293541421e-05, "loss": 0.0208, "step": 109880 }, { "epoch": 70.03824091778203, "grad_norm": 0.3333047926425934, "learning_rate": 7.505296330686993e-05, "loss": 0.0144, "step": 109890 }, { "epoch": 70.04461440407903, "grad_norm": 0.6727513074874878, "learning_rate": 7.50481933739085e-05, "loss": 0.0267, "step": 109900 }, { "epoch": 70.05098789037604, "grad_norm": 0.36027681827545166, "learning_rate": 7.504342313658785e-05, "loss": 0.0218, "step": 109910 }, { "epoch": 70.05736137667304, "grad_norm": 0.2214171439409256, "learning_rate": 7.503865259496599e-05, "loss": 0.013, "step": 109920 }, { "epoch": 70.06373486297005, "grad_norm": 0.40567103028297424, "learning_rate": 7.503388174910084e-05, "loss": 0.0172, "step": 109930 }, { "epoch": 70.07010834926704, "grad_norm": 0.785518229007721, "learning_rate": 7.50291105990504e-05, "loss": 0.0171, "step": 109940 }, { "epoch": 70.07648183556405, "grad_norm": 0.39964479207992554, "learning_rate": 7.502433914487265e-05, "loss": 0.02, "step": 109950 }, { "epoch": 70.08285532186106, "grad_norm": 0.5685421228408813, "learning_rate": 7.501956738662554e-05, "loss": 0.0305, "step": 109960 }, { "epoch": 70.08922880815807, "grad_norm": 0.9127949476242065, "learning_rate": 7.501479532436706e-05, "loss": 0.0193, "step": 109970 }, { "epoch": 70.09560229445506, "grad_norm": 0.5881022810935974, "learning_rate": 7.501002295815521e-05, "loss": 0.0175, "step": 109980 }, { "epoch": 70.10197578075207, "grad_norm": 0.5793927907943726, "learning_rate": 7.500525028804794e-05, "loss": 0.0283, "step": 109990 }, { "epoch": 70.10834926704908, "grad_norm": 0.33323097229003906, "learning_rate": 7.500047731410329e-05, "loss": 0.0244, "step": 110000 }, { "epoch": 70.11472275334609, "grad_norm": 0.3606049716472626, "learning_rate": 7.499570403637924e-05, "loss": 0.0187, "step": 110010 }, { "epoch": 70.12109623964308, "grad_norm": 0.5225717425346375, "learning_rate": 7.499093045493379e-05, "loss": 0.0197, "step": 110020 }, { "epoch": 70.12746972594009, "grad_norm": 0.35697054862976074, "learning_rate": 7.498615656982493e-05, "loss": 0.0227, "step": 110030 }, { "epoch": 70.1338432122371, "grad_norm": 0.4346321225166321, "learning_rate": 7.498138238111068e-05, "loss": 0.0158, "step": 110040 }, { "epoch": 70.1402166985341, "grad_norm": 0.5753293633460999, "learning_rate": 7.497660788884903e-05, "loss": 0.0237, "step": 110050 }, { "epoch": 70.1465901848311, "grad_norm": 0.507948637008667, "learning_rate": 7.497183309309804e-05, "loss": 0.0175, "step": 110060 }, { "epoch": 70.1529636711281, "grad_norm": 0.279093861579895, "learning_rate": 7.496705799391569e-05, "loss": 0.0144, "step": 110070 }, { "epoch": 70.15933715742511, "grad_norm": 0.26793280243873596, "learning_rate": 7.496228259136001e-05, "loss": 0.0209, "step": 110080 }, { "epoch": 70.16571064372212, "grad_norm": 0.666017472743988, "learning_rate": 7.495750688548902e-05, "loss": 0.0298, "step": 110090 }, { "epoch": 70.17208413001912, "grad_norm": 0.45712095499038696, "learning_rate": 7.495273087636077e-05, "loss": 0.0151, "step": 110100 }, { "epoch": 70.17845761631612, "grad_norm": 0.4813659191131592, "learning_rate": 7.494795456403329e-05, "loss": 0.0254, "step": 110110 }, { "epoch": 70.18483110261313, "grad_norm": 0.29053524136543274, "learning_rate": 7.494317794856459e-05, "loss": 0.0155, "step": 110120 }, { "epoch": 70.19120458891014, "grad_norm": 0.3293651044368744, "learning_rate": 7.493840103001273e-05, "loss": 0.0147, "step": 110130 }, { "epoch": 70.19757807520713, "grad_norm": 0.38272738456726074, "learning_rate": 7.493362380843576e-05, "loss": 0.0179, "step": 110140 }, { "epoch": 70.20395156150414, "grad_norm": 0.3908238410949707, "learning_rate": 7.49288462838917e-05, "loss": 0.0198, "step": 110150 }, { "epoch": 70.21032504780115, "grad_norm": 0.3286839425563812, "learning_rate": 7.492406845643864e-05, "loss": 0.0201, "step": 110160 }, { "epoch": 70.21669853409816, "grad_norm": 0.5370028614997864, "learning_rate": 7.491929032613461e-05, "loss": 0.0188, "step": 110170 }, { "epoch": 70.22307202039515, "grad_norm": 0.32326292991638184, "learning_rate": 7.491451189303769e-05, "loss": 0.0152, "step": 110180 }, { "epoch": 70.22944550669216, "grad_norm": 0.4833873510360718, "learning_rate": 7.49097331572059e-05, "loss": 0.0156, "step": 110190 }, { "epoch": 70.23581899298917, "grad_norm": 0.27931270003318787, "learning_rate": 7.490495411869735e-05, "loss": 0.0113, "step": 110200 }, { "epoch": 70.24219247928617, "grad_norm": 0.3658488094806671, "learning_rate": 7.490017477757009e-05, "loss": 0.0123, "step": 110210 }, { "epoch": 70.24856596558317, "grad_norm": 0.21749505400657654, "learning_rate": 7.489539513388219e-05, "loss": 0.0202, "step": 110220 }, { "epoch": 70.25493945188018, "grad_norm": 0.4358943700790405, "learning_rate": 7.489061518769173e-05, "loss": 0.0254, "step": 110230 }, { "epoch": 70.26131293817718, "grad_norm": 0.6098313927650452, "learning_rate": 7.488583493905682e-05, "loss": 0.0276, "step": 110240 }, { "epoch": 70.26768642447419, "grad_norm": 0.4728843867778778, "learning_rate": 7.48810543880355e-05, "loss": 0.0167, "step": 110250 }, { "epoch": 70.27405991077119, "grad_norm": 0.5907619595527649, "learning_rate": 7.487627353468589e-05, "loss": 0.0203, "step": 110260 }, { "epoch": 70.2804333970682, "grad_norm": 0.33721596002578735, "learning_rate": 7.487149237906606e-05, "loss": 0.01, "step": 110270 }, { "epoch": 70.2868068833652, "grad_norm": 0.6818324327468872, "learning_rate": 7.486671092123412e-05, "loss": 0.0251, "step": 110280 }, { "epoch": 70.29318036966221, "grad_norm": 0.20555658638477325, "learning_rate": 7.486192916124817e-05, "loss": 0.0165, "step": 110290 }, { "epoch": 70.2995538559592, "grad_norm": 0.28444740176200867, "learning_rate": 7.48571470991663e-05, "loss": 0.0104, "step": 110300 }, { "epoch": 70.30592734225621, "grad_norm": 0.6349973678588867, "learning_rate": 7.485236473504662e-05, "loss": 0.0223, "step": 110310 }, { "epoch": 70.31230082855322, "grad_norm": 0.49866020679473877, "learning_rate": 7.484758206894724e-05, "loss": 0.0116, "step": 110320 }, { "epoch": 70.31867431485023, "grad_norm": 0.8094007968902588, "learning_rate": 7.484279910092628e-05, "loss": 0.0165, "step": 110330 }, { "epoch": 70.32504780114722, "grad_norm": 0.5938400626182556, "learning_rate": 7.483801583104186e-05, "loss": 0.0144, "step": 110340 }, { "epoch": 70.33142128744423, "grad_norm": 0.24431605637073517, "learning_rate": 7.48332322593521e-05, "loss": 0.0213, "step": 110350 }, { "epoch": 70.33779477374124, "grad_norm": 0.4685380756855011, "learning_rate": 7.482844838591512e-05, "loss": 0.0224, "step": 110360 }, { "epoch": 70.34416826003824, "grad_norm": 0.25866037607192993, "learning_rate": 7.482366421078903e-05, "loss": 0.0151, "step": 110370 }, { "epoch": 70.35054174633524, "grad_norm": 0.5931476950645447, "learning_rate": 7.481887973403202e-05, "loss": 0.0152, "step": 110380 }, { "epoch": 70.35691523263225, "grad_norm": 0.24416254460811615, "learning_rate": 7.481409495570215e-05, "loss": 0.0204, "step": 110390 }, { "epoch": 70.36328871892925, "grad_norm": 0.3839692175388336, "learning_rate": 7.480930987585762e-05, "loss": 0.0256, "step": 110400 }, { "epoch": 70.36966220522626, "grad_norm": 0.6211572289466858, "learning_rate": 7.480452449455655e-05, "loss": 0.0167, "step": 110410 }, { "epoch": 70.37603569152326, "grad_norm": 0.42414870858192444, "learning_rate": 7.479973881185708e-05, "loss": 0.0171, "step": 110420 }, { "epoch": 70.38240917782026, "grad_norm": 0.40094777941703796, "learning_rate": 7.47949528278174e-05, "loss": 0.0142, "step": 110430 }, { "epoch": 70.38878266411727, "grad_norm": 0.1959487795829773, "learning_rate": 7.479016654249558e-05, "loss": 0.0159, "step": 110440 }, { "epoch": 70.39515615041428, "grad_norm": 0.2864038944244385, "learning_rate": 7.478537995594988e-05, "loss": 0.0219, "step": 110450 }, { "epoch": 70.40152963671127, "grad_norm": 0.5128370523452759, "learning_rate": 7.478059306823838e-05, "loss": 0.0203, "step": 110460 }, { "epoch": 70.40790312300828, "grad_norm": 0.4851933717727661, "learning_rate": 7.47758058794193e-05, "loss": 0.019, "step": 110470 }, { "epoch": 70.41427660930529, "grad_norm": 0.5393998026847839, "learning_rate": 7.477101838955077e-05, "loss": 0.022, "step": 110480 }, { "epoch": 70.4206500956023, "grad_norm": 0.5386297702789307, "learning_rate": 7.476623059869098e-05, "loss": 0.0224, "step": 110490 }, { "epoch": 70.4270235818993, "grad_norm": 0.34283939003944397, "learning_rate": 7.47614425068981e-05, "loss": 0.0121, "step": 110500 }, { "epoch": 70.4333970681963, "grad_norm": 0.3141005337238312, "learning_rate": 7.475665411423032e-05, "loss": 0.0131, "step": 110510 }, { "epoch": 70.43977055449331, "grad_norm": 0.37007036805152893, "learning_rate": 7.475186542074584e-05, "loss": 0.0216, "step": 110520 }, { "epoch": 70.44614404079032, "grad_norm": 0.2906089127063751, "learning_rate": 7.474707642650279e-05, "loss": 0.0144, "step": 110530 }, { "epoch": 70.45251752708732, "grad_norm": 0.7727821469306946, "learning_rate": 7.474228713155942e-05, "loss": 0.0292, "step": 110540 }, { "epoch": 70.45889101338432, "grad_norm": 0.6066847443580627, "learning_rate": 7.473749753597389e-05, "loss": 0.0239, "step": 110550 }, { "epoch": 70.46526449968133, "grad_norm": 0.6106574535369873, "learning_rate": 7.473270763980441e-05, "loss": 0.0205, "step": 110560 }, { "epoch": 70.47163798597833, "grad_norm": 0.7122835516929626, "learning_rate": 7.472791744310919e-05, "loss": 0.0174, "step": 110570 }, { "epoch": 70.47801147227534, "grad_norm": 0.32336845993995667, "learning_rate": 7.472312694594641e-05, "loss": 0.017, "step": 110580 }, { "epoch": 70.48438495857233, "grad_norm": 0.22056859731674194, "learning_rate": 7.471833614837431e-05, "loss": 0.0201, "step": 110590 }, { "epoch": 70.49075844486934, "grad_norm": 0.47530868649482727, "learning_rate": 7.471354505045108e-05, "loss": 0.0194, "step": 110600 }, { "epoch": 70.49713193116635, "grad_norm": 0.3424436151981354, "learning_rate": 7.470875365223496e-05, "loss": 0.0251, "step": 110610 }, { "epoch": 70.50350541746336, "grad_norm": 0.29361146688461304, "learning_rate": 7.470396195378413e-05, "loss": 0.0253, "step": 110620 }, { "epoch": 70.50987890376035, "grad_norm": 0.33092179894447327, "learning_rate": 7.469916995515684e-05, "loss": 0.0257, "step": 110630 }, { "epoch": 70.51625239005736, "grad_norm": 0.4215983748435974, "learning_rate": 7.469437765641132e-05, "loss": 0.0188, "step": 110640 }, { "epoch": 70.52262587635437, "grad_norm": 0.3229232430458069, "learning_rate": 7.468958505760581e-05, "loss": 0.0198, "step": 110650 }, { "epoch": 70.52899936265138, "grad_norm": 0.37529292702674866, "learning_rate": 7.468479215879852e-05, "loss": 0.0212, "step": 110660 }, { "epoch": 70.53537284894837, "grad_norm": 0.35298141837120056, "learning_rate": 7.467999896004769e-05, "loss": 0.0209, "step": 110670 }, { "epoch": 70.54174633524538, "grad_norm": 0.3835197985172272, "learning_rate": 7.46752054614116e-05, "loss": 0.0244, "step": 110680 }, { "epoch": 70.54811982154239, "grad_norm": 0.4667966663837433, "learning_rate": 7.467041166294844e-05, "loss": 0.0165, "step": 110690 }, { "epoch": 70.5544933078394, "grad_norm": 0.3145020008087158, "learning_rate": 7.466561756471648e-05, "loss": 0.0251, "step": 110700 }, { "epoch": 70.56086679413639, "grad_norm": 0.4987790286540985, "learning_rate": 7.4660823166774e-05, "loss": 0.0207, "step": 110710 }, { "epoch": 70.5672402804334, "grad_norm": 0.5521416664123535, "learning_rate": 7.465602846917923e-05, "loss": 0.0234, "step": 110720 }, { "epoch": 70.5736137667304, "grad_norm": 0.630126953125, "learning_rate": 7.465123347199041e-05, "loss": 0.0212, "step": 110730 }, { "epoch": 70.57998725302741, "grad_norm": 0.6182052493095398, "learning_rate": 7.464643817526586e-05, "loss": 0.014, "step": 110740 }, { "epoch": 70.5863607393244, "grad_norm": 0.7427298426628113, "learning_rate": 7.46416425790638e-05, "loss": 0.0189, "step": 110750 }, { "epoch": 70.59273422562141, "grad_norm": 0.3367733061313629, "learning_rate": 7.463684668344254e-05, "loss": 0.0135, "step": 110760 }, { "epoch": 70.59910771191842, "grad_norm": 0.23937629163265228, "learning_rate": 7.463205048846031e-05, "loss": 0.0125, "step": 110770 }, { "epoch": 70.60548119821543, "grad_norm": 0.4011872410774231, "learning_rate": 7.462725399417541e-05, "loss": 0.0182, "step": 110780 }, { "epoch": 70.61185468451242, "grad_norm": 0.3864860236644745, "learning_rate": 7.462245720064612e-05, "loss": 0.0178, "step": 110790 }, { "epoch": 70.61822817080943, "grad_norm": 0.3954959809780121, "learning_rate": 7.461766010793073e-05, "loss": 0.0253, "step": 110800 }, { "epoch": 70.62460165710644, "grad_norm": 0.3908807635307312, "learning_rate": 7.461286271608754e-05, "loss": 0.0139, "step": 110810 }, { "epoch": 70.63097514340345, "grad_norm": 0.5494047403335571, "learning_rate": 7.460806502517481e-05, "loss": 0.017, "step": 110820 }, { "epoch": 70.63734862970044, "grad_norm": 0.24690695106983185, "learning_rate": 7.460326703525086e-05, "loss": 0.0141, "step": 110830 }, { "epoch": 70.64372211599745, "grad_norm": 0.21273037791252136, "learning_rate": 7.4598468746374e-05, "loss": 0.0109, "step": 110840 }, { "epoch": 70.65009560229446, "grad_norm": 0.28744566440582275, "learning_rate": 7.459367015860252e-05, "loss": 0.018, "step": 110850 }, { "epoch": 70.65646908859146, "grad_norm": 1.0926573276519775, "learning_rate": 7.458887127199471e-05, "loss": 0.0216, "step": 110860 }, { "epoch": 70.66284257488846, "grad_norm": 0.7990713715553284, "learning_rate": 7.458407208660892e-05, "loss": 0.0178, "step": 110870 }, { "epoch": 70.66921606118547, "grad_norm": 0.38773682713508606, "learning_rate": 7.457927260250342e-05, "loss": 0.0142, "step": 110880 }, { "epoch": 70.67558954748247, "grad_norm": 0.36491456627845764, "learning_rate": 7.457447281973656e-05, "loss": 0.0179, "step": 110890 }, { "epoch": 70.68196303377948, "grad_norm": 0.2779029607772827, "learning_rate": 7.456967273836665e-05, "loss": 0.0162, "step": 110900 }, { "epoch": 70.68833652007648, "grad_norm": 0.22199025750160217, "learning_rate": 7.456487235845204e-05, "loss": 0.011, "step": 110910 }, { "epoch": 70.69471000637348, "grad_norm": 0.44108274579048157, "learning_rate": 7.4560071680051e-05, "loss": 0.0145, "step": 110920 }, { "epoch": 70.70108349267049, "grad_norm": 0.28117650747299194, "learning_rate": 7.455527070322193e-05, "loss": 0.0179, "step": 110930 }, { "epoch": 70.7074569789675, "grad_norm": 0.32963407039642334, "learning_rate": 7.455046942802313e-05, "loss": 0.0251, "step": 110940 }, { "epoch": 70.7138304652645, "grad_norm": 0.6334715485572815, "learning_rate": 7.454566785451294e-05, "loss": 0.0206, "step": 110950 }, { "epoch": 70.7202039515615, "grad_norm": 0.387898713350296, "learning_rate": 7.454086598274972e-05, "loss": 0.0151, "step": 110960 }, { "epoch": 70.72657743785851, "grad_norm": 0.5127922296524048, "learning_rate": 7.45360638127918e-05, "loss": 0.0228, "step": 110970 }, { "epoch": 70.73295092415552, "grad_norm": 0.32181069254875183, "learning_rate": 7.453126134469755e-05, "loss": 0.0193, "step": 110980 }, { "epoch": 70.73932441045251, "grad_norm": 0.31377163529396057, "learning_rate": 7.45264585785253e-05, "loss": 0.0154, "step": 110990 }, { "epoch": 70.74569789674952, "grad_norm": 0.4714060127735138, "learning_rate": 7.452165551433343e-05, "loss": 0.02, "step": 111000 }, { "epoch": 70.75207138304653, "grad_norm": 0.39887291193008423, "learning_rate": 7.45168521521803e-05, "loss": 0.0209, "step": 111010 }, { "epoch": 70.75844486934353, "grad_norm": 0.1710035353899002, "learning_rate": 7.451204849212425e-05, "loss": 0.0147, "step": 111020 }, { "epoch": 70.76481835564053, "grad_norm": 0.36765360832214355, "learning_rate": 7.450724453422368e-05, "loss": 0.0233, "step": 111030 }, { "epoch": 70.77119184193754, "grad_norm": 0.28101223707199097, "learning_rate": 7.450244027853695e-05, "loss": 0.015, "step": 111040 }, { "epoch": 70.77756532823454, "grad_norm": 0.46950262784957886, "learning_rate": 7.449763572512242e-05, "loss": 0.0141, "step": 111050 }, { "epoch": 70.78393881453155, "grad_norm": 0.24851472675800323, "learning_rate": 7.449283087403851e-05, "loss": 0.0187, "step": 111060 }, { "epoch": 70.79031230082855, "grad_norm": 0.5380865931510925, "learning_rate": 7.448802572534356e-05, "loss": 0.0202, "step": 111070 }, { "epoch": 70.79668578712555, "grad_norm": 0.2650468945503235, "learning_rate": 7.4483220279096e-05, "loss": 0.02, "step": 111080 }, { "epoch": 70.80305927342256, "grad_norm": 0.4117903411388397, "learning_rate": 7.447841453535418e-05, "loss": 0.016, "step": 111090 }, { "epoch": 70.80943275971957, "grad_norm": 0.704867959022522, "learning_rate": 7.447360849417652e-05, "loss": 0.0208, "step": 111100 }, { "epoch": 70.81580624601658, "grad_norm": 0.25766023993492126, "learning_rate": 7.44688021556214e-05, "loss": 0.0195, "step": 111110 }, { "epoch": 70.82217973231357, "grad_norm": 0.3519233465194702, "learning_rate": 7.446399551974723e-05, "loss": 0.0178, "step": 111120 }, { "epoch": 70.82855321861058, "grad_norm": 0.6770761013031006, "learning_rate": 7.445918858661244e-05, "loss": 0.0306, "step": 111130 }, { "epoch": 70.83492670490759, "grad_norm": 0.35930171608924866, "learning_rate": 7.445438135627539e-05, "loss": 0.0149, "step": 111140 }, { "epoch": 70.8413001912046, "grad_norm": 0.48433917760849, "learning_rate": 7.444957382879454e-05, "loss": 0.0162, "step": 111150 }, { "epoch": 70.84767367750159, "grad_norm": 0.4752691090106964, "learning_rate": 7.444476600422828e-05, "loss": 0.0197, "step": 111160 }, { "epoch": 70.8540471637986, "grad_norm": 0.6118513345718384, "learning_rate": 7.4439957882635e-05, "loss": 0.0125, "step": 111170 }, { "epoch": 70.8604206500956, "grad_norm": 0.2735186815261841, "learning_rate": 7.44351494640732e-05, "loss": 0.0182, "step": 111180 }, { "epoch": 70.86679413639261, "grad_norm": 0.6872962713241577, "learning_rate": 7.443034074860124e-05, "loss": 0.0247, "step": 111190 }, { "epoch": 70.8731676226896, "grad_norm": 0.37924760580062866, "learning_rate": 7.442553173627757e-05, "loss": 0.0126, "step": 111200 }, { "epoch": 70.87954110898661, "grad_norm": 0.425611287355423, "learning_rate": 7.442072242716064e-05, "loss": 0.0141, "step": 111210 }, { "epoch": 70.88591459528362, "grad_norm": 0.3564702570438385, "learning_rate": 7.441591282130888e-05, "loss": 0.0121, "step": 111220 }, { "epoch": 70.89228808158063, "grad_norm": 0.286515474319458, "learning_rate": 7.441110291878072e-05, "loss": 0.0138, "step": 111230 }, { "epoch": 70.89866156787762, "grad_norm": 0.43566083908081055, "learning_rate": 7.44062927196346e-05, "loss": 0.0251, "step": 111240 }, { "epoch": 70.90503505417463, "grad_norm": 0.5162158608436584, "learning_rate": 7.4401482223929e-05, "loss": 0.0336, "step": 111250 }, { "epoch": 70.91140854047164, "grad_norm": 0.5080259442329407, "learning_rate": 7.439667143172232e-05, "loss": 0.0196, "step": 111260 }, { "epoch": 70.91778202676865, "grad_norm": 0.29871058464050293, "learning_rate": 7.439186034307308e-05, "loss": 0.0148, "step": 111270 }, { "epoch": 70.92415551306564, "grad_norm": 0.4005752205848694, "learning_rate": 7.43870489580397e-05, "loss": 0.0179, "step": 111280 }, { "epoch": 70.93052899936265, "grad_norm": 0.2640437185764313, "learning_rate": 7.438223727668065e-05, "loss": 0.0268, "step": 111290 }, { "epoch": 70.93690248565966, "grad_norm": 0.5144799947738647, "learning_rate": 7.437742529905438e-05, "loss": 0.0174, "step": 111300 }, { "epoch": 70.94327597195667, "grad_norm": 0.4153822958469391, "learning_rate": 7.437261302521941e-05, "loss": 0.0184, "step": 111310 }, { "epoch": 70.94964945825366, "grad_norm": 0.30273035168647766, "learning_rate": 7.436780045523416e-05, "loss": 0.0225, "step": 111320 }, { "epoch": 70.95602294455067, "grad_norm": 0.347936749458313, "learning_rate": 7.436298758915712e-05, "loss": 0.0202, "step": 111330 }, { "epoch": 70.96239643084768, "grad_norm": 0.2888943552970886, "learning_rate": 7.435817442704679e-05, "loss": 0.0142, "step": 111340 }, { "epoch": 70.96876991714468, "grad_norm": 0.45582571625709534, "learning_rate": 7.435336096896164e-05, "loss": 0.0134, "step": 111350 }, { "epoch": 70.97514340344168, "grad_norm": 0.17989078164100647, "learning_rate": 7.434854721496017e-05, "loss": 0.0197, "step": 111360 }, { "epoch": 70.98151688973869, "grad_norm": 0.37364208698272705, "learning_rate": 7.434373316510083e-05, "loss": 0.0234, "step": 111370 }, { "epoch": 70.9878903760357, "grad_norm": 0.4869678318500519, "learning_rate": 7.433891881944218e-05, "loss": 0.0251, "step": 111380 }, { "epoch": 70.9942638623327, "grad_norm": 0.4083375036716461, "learning_rate": 7.433410417804268e-05, "loss": 0.0129, "step": 111390 }, { "epoch": 71.0006373486297, "grad_norm": 0.2743567228317261, "learning_rate": 7.432928924096083e-05, "loss": 0.0174, "step": 111400 }, { "epoch": 71.0070108349267, "grad_norm": 0.23553141951560974, "learning_rate": 7.432447400825515e-05, "loss": 0.0187, "step": 111410 }, { "epoch": 71.01338432122371, "grad_norm": 0.2498624324798584, "learning_rate": 7.431965847998414e-05, "loss": 0.0186, "step": 111420 }, { "epoch": 71.01975780752072, "grad_norm": 0.2529947757720947, "learning_rate": 7.431484265620633e-05, "loss": 0.0166, "step": 111430 }, { "epoch": 71.02613129381771, "grad_norm": 0.49505311250686646, "learning_rate": 7.431002653698021e-05, "loss": 0.0107, "step": 111440 }, { "epoch": 71.03250478011472, "grad_norm": 0.5335766673088074, "learning_rate": 7.430521012236433e-05, "loss": 0.0133, "step": 111450 }, { "epoch": 71.03887826641173, "grad_norm": 0.3321877717971802, "learning_rate": 7.430039341241719e-05, "loss": 0.0098, "step": 111460 }, { "epoch": 71.04525175270874, "grad_norm": 0.6093344688415527, "learning_rate": 7.429557640719733e-05, "loss": 0.0173, "step": 111470 }, { "epoch": 71.05162523900573, "grad_norm": 0.20159131288528442, "learning_rate": 7.429075910676328e-05, "loss": 0.0136, "step": 111480 }, { "epoch": 71.05799872530274, "grad_norm": 0.2729185223579407, "learning_rate": 7.428594151117357e-05, "loss": 0.0156, "step": 111490 }, { "epoch": 71.06437221159975, "grad_norm": 0.6778607964515686, "learning_rate": 7.428112362048675e-05, "loss": 0.0227, "step": 111500 }, { "epoch": 71.07074569789675, "grad_norm": 0.3649970591068268, "learning_rate": 7.427630543476134e-05, "loss": 0.0135, "step": 111510 }, { "epoch": 71.07711918419375, "grad_norm": 0.2877497971057892, "learning_rate": 7.42714869540559e-05, "loss": 0.0143, "step": 111520 }, { "epoch": 71.08349267049076, "grad_norm": 0.28144481778144836, "learning_rate": 7.426666817842898e-05, "loss": 0.0102, "step": 111530 }, { "epoch": 71.08986615678776, "grad_norm": 0.5119896531105042, "learning_rate": 7.426184910793912e-05, "loss": 0.017, "step": 111540 }, { "epoch": 71.09623964308477, "grad_norm": 0.24396082758903503, "learning_rate": 7.42570297426449e-05, "loss": 0.0243, "step": 111550 }, { "epoch": 71.10261312938177, "grad_norm": 0.25271913409233093, "learning_rate": 7.425221008260487e-05, "loss": 0.0187, "step": 111560 }, { "epoch": 71.10898661567877, "grad_norm": 0.574061393737793, "learning_rate": 7.424739012787758e-05, "loss": 0.0241, "step": 111570 }, { "epoch": 71.11536010197578, "grad_norm": 0.42652013897895813, "learning_rate": 7.424256987852161e-05, "loss": 0.0201, "step": 111580 }, { "epoch": 71.12173358827279, "grad_norm": 0.2635291516780853, "learning_rate": 7.423774933459551e-05, "loss": 0.0142, "step": 111590 }, { "epoch": 71.12810707456978, "grad_norm": 0.5189371109008789, "learning_rate": 7.423292849615789e-05, "loss": 0.0197, "step": 111600 }, { "epoch": 71.13448056086679, "grad_norm": 0.36636531352996826, "learning_rate": 7.42281073632673e-05, "loss": 0.0179, "step": 111610 }, { "epoch": 71.1408540471638, "grad_norm": 0.2982560694217682, "learning_rate": 7.422328593598234e-05, "loss": 0.0129, "step": 111620 }, { "epoch": 71.1472275334608, "grad_norm": 0.1597938984632492, "learning_rate": 7.421846421436158e-05, "loss": 0.0117, "step": 111630 }, { "epoch": 71.1536010197578, "grad_norm": 0.6215270757675171, "learning_rate": 7.421364219846361e-05, "loss": 0.0206, "step": 111640 }, { "epoch": 71.15997450605481, "grad_norm": 0.6858156323432922, "learning_rate": 7.420881988834704e-05, "loss": 0.0189, "step": 111650 }, { "epoch": 71.16634799235182, "grad_norm": 0.5712109208106995, "learning_rate": 7.420399728407044e-05, "loss": 0.0288, "step": 111660 }, { "epoch": 71.17272147864882, "grad_norm": 0.6001536250114441, "learning_rate": 7.419917438569242e-05, "loss": 0.0292, "step": 111670 }, { "epoch": 71.17909496494582, "grad_norm": 0.27020469307899475, "learning_rate": 7.419435119327158e-05, "loss": 0.0193, "step": 111680 }, { "epoch": 71.18546845124283, "grad_norm": 0.5612889528274536, "learning_rate": 7.418952770686653e-05, "loss": 0.0212, "step": 111690 }, { "epoch": 71.19184193753983, "grad_norm": 0.5953080058097839, "learning_rate": 7.41847039265359e-05, "loss": 0.0181, "step": 111700 }, { "epoch": 71.19821542383684, "grad_norm": 0.5372666120529175, "learning_rate": 7.417987985233828e-05, "loss": 0.0191, "step": 111710 }, { "epoch": 71.20458891013385, "grad_norm": 0.3364112377166748, "learning_rate": 7.417505548433227e-05, "loss": 0.0337, "step": 111720 }, { "epoch": 71.21096239643084, "grad_norm": 0.26506349444389343, "learning_rate": 7.417023082257652e-05, "loss": 0.0254, "step": 111730 }, { "epoch": 71.21733588272785, "grad_norm": 0.48038312792778015, "learning_rate": 7.416540586712964e-05, "loss": 0.015, "step": 111740 }, { "epoch": 71.22370936902486, "grad_norm": 0.4602726995944977, "learning_rate": 7.416058061805027e-05, "loss": 0.0244, "step": 111750 }, { "epoch": 71.23008285532187, "grad_norm": 0.322104811668396, "learning_rate": 7.415575507539702e-05, "loss": 0.0159, "step": 111760 }, { "epoch": 71.23645634161886, "grad_norm": 0.3177740275859833, "learning_rate": 7.415092923922856e-05, "loss": 0.0204, "step": 111770 }, { "epoch": 71.24282982791587, "grad_norm": 0.38239848613739014, "learning_rate": 7.414610310960349e-05, "loss": 0.0285, "step": 111780 }, { "epoch": 71.24920331421288, "grad_norm": 0.4733118414878845, "learning_rate": 7.414127668658049e-05, "loss": 0.0139, "step": 111790 }, { "epoch": 71.25557680050989, "grad_norm": 0.4304279685020447, "learning_rate": 7.413644997021819e-05, "loss": 0.0256, "step": 111800 }, { "epoch": 71.26195028680688, "grad_norm": 0.3381074368953705, "learning_rate": 7.41316229605752e-05, "loss": 0.0216, "step": 111810 }, { "epoch": 71.26832377310389, "grad_norm": 0.32345151901245117, "learning_rate": 7.412679565771022e-05, "loss": 0.0156, "step": 111820 }, { "epoch": 71.2746972594009, "grad_norm": 0.2584652304649353, "learning_rate": 7.41219680616819e-05, "loss": 0.0138, "step": 111830 }, { "epoch": 71.2810707456979, "grad_norm": 0.3220323324203491, "learning_rate": 7.41171401725489e-05, "loss": 0.0105, "step": 111840 }, { "epoch": 71.2874442319949, "grad_norm": 0.33952227234840393, "learning_rate": 7.411231199036986e-05, "loss": 0.0253, "step": 111850 }, { "epoch": 71.2938177182919, "grad_norm": 0.48257675766944885, "learning_rate": 7.410748351520346e-05, "loss": 0.0195, "step": 111860 }, { "epoch": 71.30019120458891, "grad_norm": 0.4043757915496826, "learning_rate": 7.41026547471084e-05, "loss": 0.0175, "step": 111870 }, { "epoch": 71.30656469088592, "grad_norm": 0.5093423128128052, "learning_rate": 7.409782568614331e-05, "loss": 0.0159, "step": 111880 }, { "epoch": 71.31293817718291, "grad_norm": 0.49829819798469543, "learning_rate": 7.409299633236689e-05, "loss": 0.0156, "step": 111890 }, { "epoch": 71.31931166347992, "grad_norm": 0.3054064214229584, "learning_rate": 7.408816668583781e-05, "loss": 0.0147, "step": 111900 }, { "epoch": 71.32568514977693, "grad_norm": 0.2895759642124176, "learning_rate": 7.408333674661477e-05, "loss": 0.0212, "step": 111910 }, { "epoch": 71.33205863607394, "grad_norm": 0.6152570247650146, "learning_rate": 7.407850651475645e-05, "loss": 0.0175, "step": 111920 }, { "epoch": 71.33843212237093, "grad_norm": 0.4498957097530365, "learning_rate": 7.407367599032153e-05, "loss": 0.0145, "step": 111930 }, { "epoch": 71.34480560866794, "grad_norm": 0.5411485433578491, "learning_rate": 7.406884517336873e-05, "loss": 0.0206, "step": 111940 }, { "epoch": 71.35117909496495, "grad_norm": 0.2691161334514618, "learning_rate": 7.406401406395673e-05, "loss": 0.0259, "step": 111950 }, { "epoch": 71.35755258126196, "grad_norm": 0.2796146273612976, "learning_rate": 7.405918266214426e-05, "loss": 0.0125, "step": 111960 }, { "epoch": 71.36392606755895, "grad_norm": 0.58985835313797, "learning_rate": 7.405435096798999e-05, "loss": 0.0196, "step": 111970 }, { "epoch": 71.37029955385596, "grad_norm": 0.46045950055122375, "learning_rate": 7.404951898155263e-05, "loss": 0.0093, "step": 111980 }, { "epoch": 71.37667304015297, "grad_norm": 0.30071139335632324, "learning_rate": 7.404468670289092e-05, "loss": 0.0086, "step": 111990 }, { "epoch": 71.38304652644997, "grad_norm": 0.4210141599178314, "learning_rate": 7.403985413206357e-05, "loss": 0.019, "step": 112000 }, { "epoch": 71.38942001274697, "grad_norm": 0.43543553352355957, "learning_rate": 7.403502126912929e-05, "loss": 0.0186, "step": 112010 }, { "epoch": 71.39579349904398, "grad_norm": 0.476346492767334, "learning_rate": 7.40301881141468e-05, "loss": 0.0164, "step": 112020 }, { "epoch": 71.40216698534098, "grad_norm": 0.1843661665916443, "learning_rate": 7.402535466717484e-05, "loss": 0.0161, "step": 112030 }, { "epoch": 71.40854047163799, "grad_norm": 0.5556214451789856, "learning_rate": 7.402052092827214e-05, "loss": 0.0203, "step": 112040 }, { "epoch": 71.41491395793498, "grad_norm": 0.3086717426776886, "learning_rate": 7.401568689749745e-05, "loss": 0.0147, "step": 112050 }, { "epoch": 71.421287444232, "grad_norm": 0.26795196533203125, "learning_rate": 7.401085257490947e-05, "loss": 0.0142, "step": 112060 }, { "epoch": 71.427660930529, "grad_norm": 0.22148117423057556, "learning_rate": 7.400601796056696e-05, "loss": 0.0161, "step": 112070 }, { "epoch": 71.43403441682601, "grad_norm": 0.649986743927002, "learning_rate": 7.400118305452867e-05, "loss": 0.0333, "step": 112080 }, { "epoch": 71.440407903123, "grad_norm": 0.40546172857284546, "learning_rate": 7.399634785685335e-05, "loss": 0.0192, "step": 112090 }, { "epoch": 71.44678138942001, "grad_norm": 0.4664226174354553, "learning_rate": 7.399151236759974e-05, "loss": 0.0154, "step": 112100 }, { "epoch": 71.45315487571702, "grad_norm": 0.2864699363708496, "learning_rate": 7.39866765868266e-05, "loss": 0.029, "step": 112110 }, { "epoch": 71.45952836201403, "grad_norm": 0.3281095325946808, "learning_rate": 7.398184051459271e-05, "loss": 0.0112, "step": 112120 }, { "epoch": 71.46590184831102, "grad_norm": 0.3200412690639496, "learning_rate": 7.397700415095679e-05, "loss": 0.0314, "step": 112130 }, { "epoch": 71.47227533460803, "grad_norm": 0.5504675507545471, "learning_rate": 7.397216749597765e-05, "loss": 0.0178, "step": 112140 }, { "epoch": 71.47864882090504, "grad_norm": 0.38527223467826843, "learning_rate": 7.396733054971403e-05, "loss": 0.0168, "step": 112150 }, { "epoch": 71.48502230720204, "grad_norm": 0.43689268827438354, "learning_rate": 7.396249331222472e-05, "loss": 0.0135, "step": 112160 }, { "epoch": 71.49139579349904, "grad_norm": 0.529055655002594, "learning_rate": 7.395765578356849e-05, "loss": 0.021, "step": 112170 }, { "epoch": 71.49776927979605, "grad_norm": 0.3452892005443573, "learning_rate": 7.395281796380414e-05, "loss": 0.0256, "step": 112180 }, { "epoch": 71.50414276609305, "grad_norm": 0.3483910858631134, "learning_rate": 7.39479798529904e-05, "loss": 0.0117, "step": 112190 }, { "epoch": 71.51051625239006, "grad_norm": 0.36640042066574097, "learning_rate": 7.394314145118613e-05, "loss": 0.0142, "step": 112200 }, { "epoch": 71.51688973868706, "grad_norm": 0.5262042284011841, "learning_rate": 7.393830275845007e-05, "loss": 0.0204, "step": 112210 }, { "epoch": 71.52326322498406, "grad_norm": 0.16639140248298645, "learning_rate": 7.393346377484103e-05, "loss": 0.017, "step": 112220 }, { "epoch": 71.52963671128107, "grad_norm": 0.3318454921245575, "learning_rate": 7.392862450041781e-05, "loss": 0.0164, "step": 112230 }, { "epoch": 71.53601019757808, "grad_norm": 0.28175681829452515, "learning_rate": 7.39237849352392e-05, "loss": 0.0151, "step": 112240 }, { "epoch": 71.54238368387507, "grad_norm": 0.5442802309989929, "learning_rate": 7.391894507936403e-05, "loss": 0.0161, "step": 112250 }, { "epoch": 71.54875717017208, "grad_norm": 0.3890407085418701, "learning_rate": 7.391410493285109e-05, "loss": 0.0215, "step": 112260 }, { "epoch": 71.55513065646909, "grad_norm": 0.25443580746650696, "learning_rate": 7.39092644957592e-05, "loss": 0.007, "step": 112270 }, { "epoch": 71.5615041427661, "grad_norm": 0.35709139704704285, "learning_rate": 7.390442376814716e-05, "loss": 0.017, "step": 112280 }, { "epoch": 71.56787762906309, "grad_norm": 0.31479161977767944, "learning_rate": 7.389958275007381e-05, "loss": 0.0156, "step": 112290 }, { "epoch": 71.5742511153601, "grad_norm": 0.35649216175079346, "learning_rate": 7.389474144159796e-05, "loss": 0.0175, "step": 112300 }, { "epoch": 71.5806246016571, "grad_norm": 0.43991512060165405, "learning_rate": 7.388989984277845e-05, "loss": 0.0118, "step": 112310 }, { "epoch": 71.58699808795411, "grad_norm": 0.28414320945739746, "learning_rate": 7.388505795367409e-05, "loss": 0.0221, "step": 112320 }, { "epoch": 71.59337157425111, "grad_norm": 0.37139230966567993, "learning_rate": 7.388021577434374e-05, "loss": 0.0212, "step": 112330 }, { "epoch": 71.59974506054812, "grad_norm": 0.48078304529190063, "learning_rate": 7.38753733048462e-05, "loss": 0.0123, "step": 112340 }, { "epoch": 71.60611854684512, "grad_norm": 0.4004465341567993, "learning_rate": 7.387053054524035e-05, "loss": 0.0256, "step": 112350 }, { "epoch": 71.61249203314213, "grad_norm": 0.47982630133628845, "learning_rate": 7.3865687495585e-05, "loss": 0.017, "step": 112360 }, { "epoch": 71.61886551943914, "grad_norm": 0.3756134808063507, "learning_rate": 7.386084415593904e-05, "loss": 0.0177, "step": 112370 }, { "epoch": 71.62523900573613, "grad_norm": 0.2892858386039734, "learning_rate": 7.385600052636127e-05, "loss": 0.0203, "step": 112380 }, { "epoch": 71.63161249203314, "grad_norm": 0.28233975172042847, "learning_rate": 7.385115660691058e-05, "loss": 0.0196, "step": 112390 }, { "epoch": 71.63798597833015, "grad_norm": 0.41853615641593933, "learning_rate": 7.384631239764581e-05, "loss": 0.0161, "step": 112400 }, { "epoch": 71.64435946462716, "grad_norm": 0.5323163866996765, "learning_rate": 7.384146789862585e-05, "loss": 0.0231, "step": 112410 }, { "epoch": 71.65073295092415, "grad_norm": 0.7301079034805298, "learning_rate": 7.383662310990952e-05, "loss": 0.0243, "step": 112420 }, { "epoch": 71.65710643722116, "grad_norm": 0.26666516065597534, "learning_rate": 7.383177803155573e-05, "loss": 0.0198, "step": 112430 }, { "epoch": 71.66347992351817, "grad_norm": 0.5283469557762146, "learning_rate": 7.382693266362334e-05, "loss": 0.0161, "step": 112440 }, { "epoch": 71.66985340981518, "grad_norm": 0.32770204544067383, "learning_rate": 7.382208700617121e-05, "loss": 0.0154, "step": 112450 }, { "epoch": 71.67622689611217, "grad_norm": 0.4272671937942505, "learning_rate": 7.381724105925824e-05, "loss": 0.0189, "step": 112460 }, { "epoch": 71.68260038240918, "grad_norm": 0.3193916380405426, "learning_rate": 7.381239482294329e-05, "loss": 0.0122, "step": 112470 }, { "epoch": 71.68897386870619, "grad_norm": 0.3643152117729187, "learning_rate": 7.380754829728528e-05, "loss": 0.0233, "step": 112480 }, { "epoch": 71.6953473550032, "grad_norm": 0.46829432249069214, "learning_rate": 7.380270148234306e-05, "loss": 0.0193, "step": 112490 }, { "epoch": 71.70172084130019, "grad_norm": 0.5239201784133911, "learning_rate": 7.379785437817556e-05, "loss": 0.0222, "step": 112500 }, { "epoch": 71.7080943275972, "grad_norm": 0.24278642237186432, "learning_rate": 7.379300698484164e-05, "loss": 0.0121, "step": 112510 }, { "epoch": 71.7144678138942, "grad_norm": 0.28867438435554504, "learning_rate": 7.378815930240024e-05, "loss": 0.0146, "step": 112520 }, { "epoch": 71.72084130019121, "grad_norm": 0.40474647283554077, "learning_rate": 7.378331133091024e-05, "loss": 0.0161, "step": 112530 }, { "epoch": 71.7272147864882, "grad_norm": 0.442464679479599, "learning_rate": 7.377846307043055e-05, "loss": 0.0132, "step": 112540 }, { "epoch": 71.73358827278521, "grad_norm": 0.19204314053058624, "learning_rate": 7.377361452102009e-05, "loss": 0.0105, "step": 112550 }, { "epoch": 71.73996175908222, "grad_norm": 0.3332919180393219, "learning_rate": 7.376876568273776e-05, "loss": 0.0104, "step": 112560 }, { "epoch": 71.74633524537923, "grad_norm": 0.44868332147598267, "learning_rate": 7.376391655564249e-05, "loss": 0.018, "step": 112570 }, { "epoch": 71.75270873167622, "grad_norm": 0.3221135437488556, "learning_rate": 7.375906713979318e-05, "loss": 0.0108, "step": 112580 }, { "epoch": 71.75908221797323, "grad_norm": 0.7078045606613159, "learning_rate": 7.37542174352488e-05, "loss": 0.0109, "step": 112590 }, { "epoch": 71.76545570427024, "grad_norm": 0.43876132369041443, "learning_rate": 7.374936744206824e-05, "loss": 0.0194, "step": 112600 }, { "epoch": 71.77182919056725, "grad_norm": 0.566847026348114, "learning_rate": 7.374451716031043e-05, "loss": 0.0154, "step": 112610 }, { "epoch": 71.77820267686424, "grad_norm": 0.4566158950328827, "learning_rate": 7.373966659003433e-05, "loss": 0.0121, "step": 112620 }, { "epoch": 71.78457616316125, "grad_norm": 0.2289484441280365, "learning_rate": 7.373481573129886e-05, "loss": 0.0106, "step": 112630 }, { "epoch": 71.79094964945826, "grad_norm": 0.14812730252742767, "learning_rate": 7.372996458416298e-05, "loss": 0.0112, "step": 112640 }, { "epoch": 71.79732313575526, "grad_norm": 0.2204786241054535, "learning_rate": 7.372511314868561e-05, "loss": 0.0196, "step": 112650 }, { "epoch": 71.80369662205226, "grad_norm": 0.4022834300994873, "learning_rate": 7.372026142492573e-05, "loss": 0.013, "step": 112660 }, { "epoch": 71.81007010834927, "grad_norm": 0.328564316034317, "learning_rate": 7.371540941294226e-05, "loss": 0.0137, "step": 112670 }, { "epoch": 71.81644359464627, "grad_norm": 0.9040935635566711, "learning_rate": 7.37105571127942e-05, "loss": 0.0193, "step": 112680 }, { "epoch": 71.82281708094328, "grad_norm": 0.6369282007217407, "learning_rate": 7.370570452454047e-05, "loss": 0.0218, "step": 112690 }, { "epoch": 71.82919056724027, "grad_norm": 0.35568514466285706, "learning_rate": 7.370085164824004e-05, "loss": 0.0149, "step": 112700 }, { "epoch": 71.83556405353728, "grad_norm": 0.2949157953262329, "learning_rate": 7.36959984839519e-05, "loss": 0.0124, "step": 112710 }, { "epoch": 71.84193753983429, "grad_norm": 0.46655571460723877, "learning_rate": 7.3691145031735e-05, "loss": 0.0124, "step": 112720 }, { "epoch": 71.8483110261313, "grad_norm": 0.44947993755340576, "learning_rate": 7.368629129164833e-05, "loss": 0.0141, "step": 112730 }, { "epoch": 71.85468451242829, "grad_norm": 0.38359084725379944, "learning_rate": 7.368143726375083e-05, "loss": 0.0198, "step": 112740 }, { "epoch": 71.8610579987253, "grad_norm": 0.3058016896247864, "learning_rate": 7.367658294810154e-05, "loss": 0.0137, "step": 112750 }, { "epoch": 71.86743148502231, "grad_norm": 0.29920321702957153, "learning_rate": 7.367172834475939e-05, "loss": 0.0132, "step": 112760 }, { "epoch": 71.87380497131932, "grad_norm": 0.4284398555755615, "learning_rate": 7.36668734537834e-05, "loss": 0.0132, "step": 112770 }, { "epoch": 71.88017845761631, "grad_norm": 0.6181581616401672, "learning_rate": 7.366201827523256e-05, "loss": 0.0176, "step": 112780 }, { "epoch": 71.88655194391332, "grad_norm": 0.265926718711853, "learning_rate": 7.365716280916584e-05, "loss": 0.0105, "step": 112790 }, { "epoch": 71.89292543021033, "grad_norm": 0.472455769777298, "learning_rate": 7.365230705564227e-05, "loss": 0.0217, "step": 112800 }, { "epoch": 71.89929891650733, "grad_norm": 0.6640032529830933, "learning_rate": 7.364745101472083e-05, "loss": 0.0191, "step": 112810 }, { "epoch": 71.90567240280433, "grad_norm": 0.36768969893455505, "learning_rate": 7.364259468646053e-05, "loss": 0.0156, "step": 112820 }, { "epoch": 71.91204588910134, "grad_norm": 0.4640871286392212, "learning_rate": 7.363773807092038e-05, "loss": 0.0232, "step": 112830 }, { "epoch": 71.91841937539834, "grad_norm": 0.41178372502326965, "learning_rate": 7.363288116815941e-05, "loss": 0.0185, "step": 112840 }, { "epoch": 71.92479286169535, "grad_norm": 0.36267292499542236, "learning_rate": 7.362802397823663e-05, "loss": 0.0121, "step": 112850 }, { "epoch": 71.93116634799235, "grad_norm": 0.2497832328081131, "learning_rate": 7.362316650121102e-05, "loss": 0.0166, "step": 112860 }, { "epoch": 71.93753983428935, "grad_norm": 0.5851818323135376, "learning_rate": 7.361830873714165e-05, "loss": 0.0234, "step": 112870 }, { "epoch": 71.94391332058636, "grad_norm": 0.20047226548194885, "learning_rate": 7.361345068608752e-05, "loss": 0.0122, "step": 112880 }, { "epoch": 71.95028680688337, "grad_norm": 0.5246102213859558, "learning_rate": 7.360859234810767e-05, "loss": 0.0175, "step": 112890 }, { "epoch": 71.95666029318036, "grad_norm": 0.43953388929367065, "learning_rate": 7.360373372326114e-05, "loss": 0.0235, "step": 112900 }, { "epoch": 71.96303377947737, "grad_norm": 0.305100679397583, "learning_rate": 7.359887481160697e-05, "loss": 0.0265, "step": 112910 }, { "epoch": 71.96940726577438, "grad_norm": 0.6466211080551147, "learning_rate": 7.359401561320417e-05, "loss": 0.0159, "step": 112920 }, { "epoch": 71.97578075207139, "grad_norm": 0.7118947505950928, "learning_rate": 7.358915612811181e-05, "loss": 0.0292, "step": 112930 }, { "epoch": 71.98215423836838, "grad_norm": 0.23722565174102783, "learning_rate": 7.358429635638894e-05, "loss": 0.0227, "step": 112940 }, { "epoch": 71.98852772466539, "grad_norm": 0.3455184996128082, "learning_rate": 7.357943629809458e-05, "loss": 0.0166, "step": 112950 }, { "epoch": 71.9949012109624, "grad_norm": 0.40463805198669434, "learning_rate": 7.357457595328783e-05, "loss": 0.0155, "step": 112960 }, { "epoch": 72.0012746972594, "grad_norm": 0.5718831419944763, "learning_rate": 7.356971532202771e-05, "loss": 0.0149, "step": 112970 }, { "epoch": 72.00764818355641, "grad_norm": 0.3989647626876831, "learning_rate": 7.35648544043733e-05, "loss": 0.0307, "step": 112980 }, { "epoch": 72.0140216698534, "grad_norm": 0.702725350856781, "learning_rate": 7.355999320038366e-05, "loss": 0.0203, "step": 112990 }, { "epoch": 72.02039515615041, "grad_norm": 0.39064455032348633, "learning_rate": 7.355513171011787e-05, "loss": 0.0207, "step": 113000 }, { "epoch": 72.02676864244742, "grad_norm": 0.3058459758758545, "learning_rate": 7.355026993363496e-05, "loss": 0.0252, "step": 113010 }, { "epoch": 72.03314212874443, "grad_norm": 0.3144052028656006, "learning_rate": 7.354540787099406e-05, "loss": 0.0142, "step": 113020 }, { "epoch": 72.03951561504142, "grad_norm": 0.6099256277084351, "learning_rate": 7.354054552225422e-05, "loss": 0.02, "step": 113030 }, { "epoch": 72.04588910133843, "grad_norm": 0.46430057287216187, "learning_rate": 7.353568288747452e-05, "loss": 0.0183, "step": 113040 }, { "epoch": 72.05226258763544, "grad_norm": 0.2648930251598358, "learning_rate": 7.353081996671404e-05, "loss": 0.0229, "step": 113050 }, { "epoch": 72.05863607393245, "grad_norm": 0.23276522755622864, "learning_rate": 7.352595676003192e-05, "loss": 0.0132, "step": 113060 }, { "epoch": 72.06500956022944, "grad_norm": 0.4818512797355652, "learning_rate": 7.352109326748717e-05, "loss": 0.0209, "step": 113070 }, { "epoch": 72.07138304652645, "grad_norm": 0.575624942779541, "learning_rate": 7.351622948913896e-05, "loss": 0.0164, "step": 113080 }, { "epoch": 72.07775653282346, "grad_norm": 0.5992916822433472, "learning_rate": 7.351136542504634e-05, "loss": 0.0194, "step": 113090 }, { "epoch": 72.08413001912047, "grad_norm": 0.7430914640426636, "learning_rate": 7.350650107526845e-05, "loss": 0.0183, "step": 113100 }, { "epoch": 72.09050350541746, "grad_norm": 0.36094510555267334, "learning_rate": 7.350163643986435e-05, "loss": 0.0139, "step": 113110 }, { "epoch": 72.09687699171447, "grad_norm": 0.27551546692848206, "learning_rate": 7.349677151889321e-05, "loss": 0.0236, "step": 113120 }, { "epoch": 72.10325047801147, "grad_norm": 0.37647610902786255, "learning_rate": 7.349190631241408e-05, "loss": 0.0115, "step": 113130 }, { "epoch": 72.10962396430848, "grad_norm": 0.3141384422779083, "learning_rate": 7.348704082048613e-05, "loss": 0.0127, "step": 113140 }, { "epoch": 72.11599745060548, "grad_norm": 0.3731916844844818, "learning_rate": 7.348217504316845e-05, "loss": 0.021, "step": 113150 }, { "epoch": 72.12237093690248, "grad_norm": 0.5014616250991821, "learning_rate": 7.347730898052018e-05, "loss": 0.0146, "step": 113160 }, { "epoch": 72.12874442319949, "grad_norm": 0.7035441994667053, "learning_rate": 7.347244263260043e-05, "loss": 0.0181, "step": 113170 }, { "epoch": 72.1351179094965, "grad_norm": 0.22204901278018951, "learning_rate": 7.346757599946834e-05, "loss": 0.014, "step": 113180 }, { "epoch": 72.1414913957935, "grad_norm": 0.3503003716468811, "learning_rate": 7.346270908118306e-05, "loss": 0.0164, "step": 113190 }, { "epoch": 72.1478648820905, "grad_norm": 0.5039528012275696, "learning_rate": 7.34578418778037e-05, "loss": 0.0154, "step": 113200 }, { "epoch": 72.15423836838751, "grad_norm": 0.3733975887298584, "learning_rate": 7.345297438938942e-05, "loss": 0.0085, "step": 113210 }, { "epoch": 72.16061185468452, "grad_norm": 0.34849977493286133, "learning_rate": 7.344810661599935e-05, "loss": 0.0216, "step": 113220 }, { "epoch": 72.16698534098151, "grad_norm": 0.5941341519355774, "learning_rate": 7.344323855769265e-05, "loss": 0.0192, "step": 113230 }, { "epoch": 72.17335882727852, "grad_norm": 0.44508278369903564, "learning_rate": 7.343837021452846e-05, "loss": 0.0155, "step": 113240 }, { "epoch": 72.17973231357553, "grad_norm": 0.43203622102737427, "learning_rate": 7.343350158656596e-05, "loss": 0.0196, "step": 113250 }, { "epoch": 72.18610579987254, "grad_norm": 0.21516522765159607, "learning_rate": 7.342863267386428e-05, "loss": 0.0116, "step": 113260 }, { "epoch": 72.19247928616953, "grad_norm": 0.42442163825035095, "learning_rate": 7.342376347648259e-05, "loss": 0.0211, "step": 113270 }, { "epoch": 72.19885277246654, "grad_norm": 0.5309235453605652, "learning_rate": 7.341889399448006e-05, "loss": 0.025, "step": 113280 }, { "epoch": 72.20522625876355, "grad_norm": 0.19177095592021942, "learning_rate": 7.341402422791586e-05, "loss": 0.0135, "step": 113290 }, { "epoch": 72.21159974506055, "grad_norm": 0.35628488659858704, "learning_rate": 7.340915417684915e-05, "loss": 0.0123, "step": 113300 }, { "epoch": 72.21797323135755, "grad_norm": 0.3526313602924347, "learning_rate": 7.340428384133912e-05, "loss": 0.0162, "step": 113310 }, { "epoch": 72.22434671765456, "grad_norm": 0.2816762924194336, "learning_rate": 7.339941322144494e-05, "loss": 0.0237, "step": 113320 }, { "epoch": 72.23072020395156, "grad_norm": 0.32740554213523865, "learning_rate": 7.339454231722582e-05, "loss": 0.0412, "step": 113330 }, { "epoch": 72.23709369024857, "grad_norm": 0.24438093602657318, "learning_rate": 7.338967112874089e-05, "loss": 0.0158, "step": 113340 }, { "epoch": 72.24346717654556, "grad_norm": 0.5879038572311401, "learning_rate": 7.33847996560494e-05, "loss": 0.0132, "step": 113350 }, { "epoch": 72.24984066284257, "grad_norm": 0.21791058778762817, "learning_rate": 7.337992789921049e-05, "loss": 0.0125, "step": 113360 }, { "epoch": 72.25621414913958, "grad_norm": 0.42603424191474915, "learning_rate": 7.33750558582834e-05, "loss": 0.0183, "step": 113370 }, { "epoch": 72.26258763543659, "grad_norm": 0.34699559211730957, "learning_rate": 7.33701835333273e-05, "loss": 0.016, "step": 113380 }, { "epoch": 72.26896112173358, "grad_norm": 0.41082480549812317, "learning_rate": 7.336531092440142e-05, "loss": 0.0139, "step": 113390 }, { "epoch": 72.27533460803059, "grad_norm": 0.22218084335327148, "learning_rate": 7.336043803156493e-05, "loss": 0.0199, "step": 113400 }, { "epoch": 72.2817080943276, "grad_norm": 0.31930333375930786, "learning_rate": 7.335556485487708e-05, "loss": 0.0166, "step": 113410 }, { "epoch": 72.2880815806246, "grad_norm": 0.36845988035202026, "learning_rate": 7.335069139439706e-05, "loss": 0.0234, "step": 113420 }, { "epoch": 72.2944550669216, "grad_norm": 0.39159321784973145, "learning_rate": 7.334581765018409e-05, "loss": 0.0167, "step": 113430 }, { "epoch": 72.30082855321861, "grad_norm": 0.45930665731430054, "learning_rate": 7.33409436222974e-05, "loss": 0.0226, "step": 113440 }, { "epoch": 72.30720203951562, "grad_norm": 0.34750086069107056, "learning_rate": 7.33360693107962e-05, "loss": 0.02, "step": 113450 }, { "epoch": 72.31357552581262, "grad_norm": 0.4266999661922455, "learning_rate": 7.333119471573974e-05, "loss": 0.0107, "step": 113460 }, { "epoch": 72.31994901210962, "grad_norm": 0.34591278433799744, "learning_rate": 7.332631983718722e-05, "loss": 0.0224, "step": 113470 }, { "epoch": 72.32632249840663, "grad_norm": 0.2823264002799988, "learning_rate": 7.33214446751979e-05, "loss": 0.023, "step": 113480 }, { "epoch": 72.33269598470363, "grad_norm": 0.40549349784851074, "learning_rate": 7.3316569229831e-05, "loss": 0.0139, "step": 113490 }, { "epoch": 72.33906947100064, "grad_norm": 0.5716128349304199, "learning_rate": 7.331169350114577e-05, "loss": 0.016, "step": 113500 }, { "epoch": 72.34544295729764, "grad_norm": 0.22423169016838074, "learning_rate": 7.330681748920146e-05, "loss": 0.0146, "step": 113510 }, { "epoch": 72.35181644359464, "grad_norm": 0.2838788628578186, "learning_rate": 7.330194119405731e-05, "loss": 0.0265, "step": 113520 }, { "epoch": 72.35818992989165, "grad_norm": 0.25387415289878845, "learning_rate": 7.329706461577257e-05, "loss": 0.0195, "step": 113530 }, { "epoch": 72.36456341618866, "grad_norm": 0.5474233627319336, "learning_rate": 7.32921877544065e-05, "loss": 0.0144, "step": 113540 }, { "epoch": 72.37093690248565, "grad_norm": 0.22595922648906708, "learning_rate": 7.328731061001837e-05, "loss": 0.0119, "step": 113550 }, { "epoch": 72.37731038878266, "grad_norm": 0.3871650695800781, "learning_rate": 7.328243318266741e-05, "loss": 0.0221, "step": 113560 }, { "epoch": 72.38368387507967, "grad_norm": 0.2002009004354477, "learning_rate": 7.327755547241291e-05, "loss": 0.0097, "step": 113570 }, { "epoch": 72.39005736137668, "grad_norm": 0.5796409845352173, "learning_rate": 7.327267747931414e-05, "loss": 0.0169, "step": 113580 }, { "epoch": 72.39643084767367, "grad_norm": 0.3281729519367218, "learning_rate": 7.326779920343035e-05, "loss": 0.0096, "step": 113590 }, { "epoch": 72.40280433397068, "grad_norm": 0.2723643481731415, "learning_rate": 7.326292064482085e-05, "loss": 0.0252, "step": 113600 }, { "epoch": 72.40917782026769, "grad_norm": 0.33537307381629944, "learning_rate": 7.32580418035449e-05, "loss": 0.0184, "step": 113610 }, { "epoch": 72.4155513065647, "grad_norm": 0.43124207854270935, "learning_rate": 7.325316267966177e-05, "loss": 0.0096, "step": 113620 }, { "epoch": 72.4219247928617, "grad_norm": 0.3130299150943756, "learning_rate": 7.324828327323076e-05, "loss": 0.0198, "step": 113630 }, { "epoch": 72.4282982791587, "grad_norm": 0.4134792983531952, "learning_rate": 7.324340358431117e-05, "loss": 0.017, "step": 113640 }, { "epoch": 72.4346717654557, "grad_norm": 0.5460556149482727, "learning_rate": 7.323852361296227e-05, "loss": 0.0208, "step": 113650 }, { "epoch": 72.44104525175271, "grad_norm": 0.5248019695281982, "learning_rate": 7.323364335924337e-05, "loss": 0.0241, "step": 113660 }, { "epoch": 72.44741873804972, "grad_norm": 0.33820849657058716, "learning_rate": 7.322876282321379e-05, "loss": 0.0213, "step": 113670 }, { "epoch": 72.45379222434671, "grad_norm": 0.319362074136734, "learning_rate": 7.322388200493277e-05, "loss": 0.0163, "step": 113680 }, { "epoch": 72.46016571064372, "grad_norm": 0.3042007088661194, "learning_rate": 7.321900090445967e-05, "loss": 0.0276, "step": 113690 }, { "epoch": 72.46653919694073, "grad_norm": 0.5222786068916321, "learning_rate": 7.321411952185379e-05, "loss": 0.0106, "step": 113700 }, { "epoch": 72.47291268323774, "grad_norm": 0.44042277336120605, "learning_rate": 7.320923785717445e-05, "loss": 0.0202, "step": 113710 }, { "epoch": 72.47928616953473, "grad_norm": 0.2913137972354889, "learning_rate": 7.320435591048094e-05, "loss": 0.018, "step": 113720 }, { "epoch": 72.48565965583174, "grad_norm": 0.31539174914360046, "learning_rate": 7.319947368183262e-05, "loss": 0.0168, "step": 113730 }, { "epoch": 72.49203314212875, "grad_norm": 0.23334136605262756, "learning_rate": 7.319459117128877e-05, "loss": 0.0202, "step": 113740 }, { "epoch": 72.49840662842576, "grad_norm": 0.39079874753952026, "learning_rate": 7.318970837890875e-05, "loss": 0.0145, "step": 113750 }, { "epoch": 72.50478011472275, "grad_norm": 0.46653157472610474, "learning_rate": 7.318482530475186e-05, "loss": 0.0172, "step": 113760 }, { "epoch": 72.51115360101976, "grad_norm": 0.2498743236064911, "learning_rate": 7.317994194887747e-05, "loss": 0.0149, "step": 113770 }, { "epoch": 72.51752708731676, "grad_norm": 0.5402017831802368, "learning_rate": 7.317505831134489e-05, "loss": 0.0386, "step": 113780 }, { "epoch": 72.52390057361377, "grad_norm": 0.37883448600769043, "learning_rate": 7.317017439221347e-05, "loss": 0.0115, "step": 113790 }, { "epoch": 72.53027405991077, "grad_norm": 0.453969806432724, "learning_rate": 7.316529019154257e-05, "loss": 0.0206, "step": 113800 }, { "epoch": 72.53664754620777, "grad_norm": 0.34084638953208923, "learning_rate": 7.316040570939151e-05, "loss": 0.0115, "step": 113810 }, { "epoch": 72.54302103250478, "grad_norm": 0.4473154544830322, "learning_rate": 7.315552094581966e-05, "loss": 0.0113, "step": 113820 }, { "epoch": 72.54939451880179, "grad_norm": 0.2327394187450409, "learning_rate": 7.315063590088637e-05, "loss": 0.02, "step": 113830 }, { "epoch": 72.55576800509878, "grad_norm": 0.256333589553833, "learning_rate": 7.314575057465098e-05, "loss": 0.0147, "step": 113840 }, { "epoch": 72.56214149139579, "grad_norm": 0.4698444604873657, "learning_rate": 7.314086496717289e-05, "loss": 0.0259, "step": 113850 }, { "epoch": 72.5685149776928, "grad_norm": 0.32647955417633057, "learning_rate": 7.313597907851142e-05, "loss": 0.0122, "step": 113860 }, { "epoch": 72.57488846398981, "grad_norm": 0.48328351974487305, "learning_rate": 7.313109290872598e-05, "loss": 0.0309, "step": 113870 }, { "epoch": 72.5812619502868, "grad_norm": 0.2738121449947357, "learning_rate": 7.31262064578759e-05, "loss": 0.0177, "step": 113880 }, { "epoch": 72.58763543658381, "grad_norm": 0.3332400619983673, "learning_rate": 7.312131972602058e-05, "loss": 0.0139, "step": 113890 }, { "epoch": 72.59400892288082, "grad_norm": 0.5476122498512268, "learning_rate": 7.311643271321942e-05, "loss": 0.0143, "step": 113900 }, { "epoch": 72.60038240917783, "grad_norm": 0.4550575911998749, "learning_rate": 7.311154541953176e-05, "loss": 0.0199, "step": 113910 }, { "epoch": 72.60675589547482, "grad_norm": 0.25108546018600464, "learning_rate": 7.3106657845017e-05, "loss": 0.015, "step": 113920 }, { "epoch": 72.61312938177183, "grad_norm": 0.7585155963897705, "learning_rate": 7.310176998973453e-05, "loss": 0.0281, "step": 113930 }, { "epoch": 72.61950286806884, "grad_norm": 0.30048468708992004, "learning_rate": 7.309688185374376e-05, "loss": 0.0115, "step": 113940 }, { "epoch": 72.62587635436584, "grad_norm": 0.6492212414741516, "learning_rate": 7.309199343710405e-05, "loss": 0.0243, "step": 113950 }, { "epoch": 72.63224984066284, "grad_norm": 0.5122234225273132, "learning_rate": 7.308710473987483e-05, "loss": 0.0188, "step": 113960 }, { "epoch": 72.63862332695984, "grad_norm": 0.609464168548584, "learning_rate": 7.308221576211548e-05, "loss": 0.0226, "step": 113970 }, { "epoch": 72.64499681325685, "grad_norm": 0.30804866552352905, "learning_rate": 7.307732650388542e-05, "loss": 0.0148, "step": 113980 }, { "epoch": 72.65137029955386, "grad_norm": 0.4142485558986664, "learning_rate": 7.307243696524405e-05, "loss": 0.0182, "step": 113990 }, { "epoch": 72.65774378585085, "grad_norm": 0.30982667207717896, "learning_rate": 7.306754714625078e-05, "loss": 0.0175, "step": 114000 }, { "epoch": 72.66411727214786, "grad_norm": 0.3742220401763916, "learning_rate": 7.306265704696504e-05, "loss": 0.0194, "step": 114010 }, { "epoch": 72.67049075844487, "grad_norm": 0.6224552392959595, "learning_rate": 7.305776666744625e-05, "loss": 0.0201, "step": 114020 }, { "epoch": 72.67686424474188, "grad_norm": 0.35640913248062134, "learning_rate": 7.305287600775382e-05, "loss": 0.0174, "step": 114030 }, { "epoch": 72.68323773103887, "grad_norm": 0.2707843482494354, "learning_rate": 7.304798506794718e-05, "loss": 0.0103, "step": 114040 }, { "epoch": 72.68961121733588, "grad_norm": 0.4198243021965027, "learning_rate": 7.304309384808577e-05, "loss": 0.0125, "step": 114050 }, { "epoch": 72.69598470363289, "grad_norm": 0.3541308045387268, "learning_rate": 7.303820234822901e-05, "loss": 0.0195, "step": 114060 }, { "epoch": 72.7023581899299, "grad_norm": 0.38320985436439514, "learning_rate": 7.303331056843631e-05, "loss": 0.0182, "step": 114070 }, { "epoch": 72.70873167622689, "grad_norm": 0.5441397428512573, "learning_rate": 7.302841850876718e-05, "loss": 0.0191, "step": 114080 }, { "epoch": 72.7151051625239, "grad_norm": 0.40555664896965027, "learning_rate": 7.302352616928099e-05, "loss": 0.0204, "step": 114090 }, { "epoch": 72.7214786488209, "grad_norm": 0.270123690366745, "learning_rate": 7.301863355003725e-05, "loss": 0.017, "step": 114100 }, { "epoch": 72.72785213511791, "grad_norm": 0.401199609041214, "learning_rate": 7.301374065109535e-05, "loss": 0.0198, "step": 114110 }, { "epoch": 72.73422562141491, "grad_norm": 0.30923980474472046, "learning_rate": 7.30088474725148e-05, "loss": 0.0189, "step": 114120 }, { "epoch": 72.74059910771192, "grad_norm": 0.317912220954895, "learning_rate": 7.300395401435501e-05, "loss": 0.0146, "step": 114130 }, { "epoch": 72.74697259400892, "grad_norm": 0.3823492228984833, "learning_rate": 7.299906027667547e-05, "loss": 0.0138, "step": 114140 }, { "epoch": 72.75334608030593, "grad_norm": 0.6895157694816589, "learning_rate": 7.299416625953564e-05, "loss": 0.0189, "step": 114150 }, { "epoch": 72.75971956660293, "grad_norm": 0.6077188849449158, "learning_rate": 7.298927196299496e-05, "loss": 0.0215, "step": 114160 }, { "epoch": 72.76609305289993, "grad_norm": 0.447308212518692, "learning_rate": 7.298437738711292e-05, "loss": 0.0267, "step": 114170 }, { "epoch": 72.77246653919694, "grad_norm": 0.3975064754486084, "learning_rate": 7.297948253194899e-05, "loss": 0.0146, "step": 114180 }, { "epoch": 72.77884002549395, "grad_norm": 0.24420438706874847, "learning_rate": 7.297458739756267e-05, "loss": 0.0268, "step": 114190 }, { "epoch": 72.78521351179094, "grad_norm": 0.3586350083351135, "learning_rate": 7.296969198401342e-05, "loss": 0.0208, "step": 114200 }, { "epoch": 72.79158699808795, "grad_norm": 0.805204451084137, "learning_rate": 7.296479629136073e-05, "loss": 0.0238, "step": 114210 }, { "epoch": 72.79796048438496, "grad_norm": 0.48016560077667236, "learning_rate": 7.295990031966407e-05, "loss": 0.0191, "step": 114220 }, { "epoch": 72.80433397068197, "grad_norm": 0.4322308897972107, "learning_rate": 7.295500406898295e-05, "loss": 0.021, "step": 114230 }, { "epoch": 72.81070745697897, "grad_norm": 0.28259584307670593, "learning_rate": 7.295010753937686e-05, "loss": 0.0124, "step": 114240 }, { "epoch": 72.81708094327597, "grad_norm": 0.2607811391353607, "learning_rate": 7.294521073090527e-05, "loss": 0.0156, "step": 114250 }, { "epoch": 72.82345442957298, "grad_norm": 0.43748167157173157, "learning_rate": 7.294031364362773e-05, "loss": 0.0181, "step": 114260 }, { "epoch": 72.82982791586998, "grad_norm": 0.21345020830631256, "learning_rate": 7.29354162776037e-05, "loss": 0.017, "step": 114270 }, { "epoch": 72.83620140216699, "grad_norm": 0.5328529477119446, "learning_rate": 7.293051863289274e-05, "loss": 0.0157, "step": 114280 }, { "epoch": 72.84257488846399, "grad_norm": 0.9353928565979004, "learning_rate": 7.29256207095543e-05, "loss": 0.0181, "step": 114290 }, { "epoch": 72.848948374761, "grad_norm": 0.3485811650753021, "learning_rate": 7.292072250764795e-05, "loss": 0.0127, "step": 114300 }, { "epoch": 72.855321861058, "grad_norm": 0.39055851101875305, "learning_rate": 7.291582402723316e-05, "loss": 0.0103, "step": 114310 }, { "epoch": 72.86169534735501, "grad_norm": 0.4384983777999878, "learning_rate": 7.291092526836947e-05, "loss": 0.0194, "step": 114320 }, { "epoch": 72.868068833652, "grad_norm": 0.6139965057373047, "learning_rate": 7.290602623111642e-05, "loss": 0.0297, "step": 114330 }, { "epoch": 72.87444231994901, "grad_norm": 0.5330699682235718, "learning_rate": 7.290112691553351e-05, "loss": 0.0184, "step": 114340 }, { "epoch": 72.88081580624602, "grad_norm": 0.5116077065467834, "learning_rate": 7.28962273216803e-05, "loss": 0.0124, "step": 114350 }, { "epoch": 72.88718929254303, "grad_norm": 0.5688324570655823, "learning_rate": 7.28913274496163e-05, "loss": 0.0175, "step": 114360 }, { "epoch": 72.89356277884002, "grad_norm": 0.3196817636489868, "learning_rate": 7.288642729940106e-05, "loss": 0.0201, "step": 114370 }, { "epoch": 72.89993626513703, "grad_norm": 0.5364440679550171, "learning_rate": 7.288152687109413e-05, "loss": 0.0169, "step": 114380 }, { "epoch": 72.90630975143404, "grad_norm": 0.43039670586586, "learning_rate": 7.287662616475504e-05, "loss": 0.0181, "step": 114390 }, { "epoch": 72.91268323773104, "grad_norm": 0.220923513174057, "learning_rate": 7.287172518044334e-05, "loss": 0.0312, "step": 114400 }, { "epoch": 72.91905672402804, "grad_norm": 0.3401090204715729, "learning_rate": 7.286682391821858e-05, "loss": 0.0148, "step": 114410 }, { "epoch": 72.92543021032505, "grad_norm": 0.7129428386688232, "learning_rate": 7.286192237814033e-05, "loss": 0.0256, "step": 114420 }, { "epoch": 72.93180369662205, "grad_norm": 1.0186797380447388, "learning_rate": 7.285702056026813e-05, "loss": 0.0231, "step": 114430 }, { "epoch": 72.93817718291906, "grad_norm": 0.5374343991279602, "learning_rate": 7.285211846466154e-05, "loss": 0.0209, "step": 114440 }, { "epoch": 72.94455066921606, "grad_norm": 0.5231232047080994, "learning_rate": 7.284721609138014e-05, "loss": 0.02, "step": 114450 }, { "epoch": 72.95092415551306, "grad_norm": 0.3980078399181366, "learning_rate": 7.284231344048351e-05, "loss": 0.0133, "step": 114460 }, { "epoch": 72.95729764181007, "grad_norm": 0.6895831227302551, "learning_rate": 7.28374105120312e-05, "loss": 0.0209, "step": 114470 }, { "epoch": 72.96367112810708, "grad_norm": 0.47740039229393005, "learning_rate": 7.283250730608277e-05, "loss": 0.0272, "step": 114480 }, { "epoch": 72.97004461440407, "grad_norm": 0.41901305317878723, "learning_rate": 7.282760382269783e-05, "loss": 0.0249, "step": 114490 }, { "epoch": 72.97641810070108, "grad_norm": 0.3793869614601135, "learning_rate": 7.282270006193595e-05, "loss": 0.0242, "step": 114500 }, { "epoch": 72.98279158699809, "grad_norm": 0.27045634388923645, "learning_rate": 7.281779602385673e-05, "loss": 0.0139, "step": 114510 }, { "epoch": 72.9891650732951, "grad_norm": 0.29091525077819824, "learning_rate": 7.281289170851972e-05, "loss": 0.0179, "step": 114520 }, { "epoch": 72.99553855959209, "grad_norm": 0.27438822388648987, "learning_rate": 7.280798711598456e-05, "loss": 0.0174, "step": 114530 }, { "epoch": 73.0019120458891, "grad_norm": 0.4468340277671814, "learning_rate": 7.280308224631081e-05, "loss": 0.0235, "step": 114540 }, { "epoch": 73.00828553218611, "grad_norm": 0.4306821823120117, "learning_rate": 7.279817709955809e-05, "loss": 0.0182, "step": 114550 }, { "epoch": 73.01465901848312, "grad_norm": 0.3407873809337616, "learning_rate": 7.279327167578599e-05, "loss": 0.0157, "step": 114560 }, { "epoch": 73.02103250478011, "grad_norm": 0.4185697138309479, "learning_rate": 7.27883659750541e-05, "loss": 0.0151, "step": 114570 }, { "epoch": 73.02740599107712, "grad_norm": 0.37426844239234924, "learning_rate": 7.278345999742208e-05, "loss": 0.0252, "step": 114580 }, { "epoch": 73.03377947737413, "grad_norm": 0.8414224982261658, "learning_rate": 7.277855374294948e-05, "loss": 0.0253, "step": 114590 }, { "epoch": 73.04015296367113, "grad_norm": 0.4614699184894562, "learning_rate": 7.277364721169597e-05, "loss": 0.0172, "step": 114600 }, { "epoch": 73.04652644996813, "grad_norm": 0.697331428527832, "learning_rate": 7.276874040372113e-05, "loss": 0.0165, "step": 114610 }, { "epoch": 73.05289993626513, "grad_norm": 0.2998666763305664, "learning_rate": 7.27638333190846e-05, "loss": 0.0178, "step": 114620 }, { "epoch": 73.05927342256214, "grad_norm": 0.24222257733345032, "learning_rate": 7.275892595784601e-05, "loss": 0.0128, "step": 114630 }, { "epoch": 73.06564690885915, "grad_norm": 0.38814419507980347, "learning_rate": 7.275401832006496e-05, "loss": 0.0122, "step": 114640 }, { "epoch": 73.07202039515614, "grad_norm": 0.2249758094549179, "learning_rate": 7.274911040580114e-05, "loss": 0.0099, "step": 114650 }, { "epoch": 73.07839388145315, "grad_norm": 0.465024471282959, "learning_rate": 7.274420221511411e-05, "loss": 0.0221, "step": 114660 }, { "epoch": 73.08476736775016, "grad_norm": 0.5032816529273987, "learning_rate": 7.273929374806358e-05, "loss": 0.0132, "step": 114670 }, { "epoch": 73.09114085404717, "grad_norm": 0.3518093526363373, "learning_rate": 7.273438500470914e-05, "loss": 0.0231, "step": 114680 }, { "epoch": 73.09751434034416, "grad_norm": 0.18249323964118958, "learning_rate": 7.272947598511047e-05, "loss": 0.0197, "step": 114690 }, { "epoch": 73.10388782664117, "grad_norm": 0.2743316888809204, "learning_rate": 7.272456668932721e-05, "loss": 0.0178, "step": 114700 }, { "epoch": 73.11026131293818, "grad_norm": 0.48399490118026733, "learning_rate": 7.2719657117419e-05, "loss": 0.0178, "step": 114710 }, { "epoch": 73.11663479923519, "grad_norm": 0.3588631749153137, "learning_rate": 7.271474726944551e-05, "loss": 0.0216, "step": 114720 }, { "epoch": 73.12300828553218, "grad_norm": 0.3495731055736542, "learning_rate": 7.270983714546639e-05, "loss": 0.0144, "step": 114730 }, { "epoch": 73.12938177182919, "grad_norm": 0.36568087339401245, "learning_rate": 7.270492674554131e-05, "loss": 0.0241, "step": 114740 }, { "epoch": 73.1357552581262, "grad_norm": 0.26062512397766113, "learning_rate": 7.270001606972993e-05, "loss": 0.0106, "step": 114750 }, { "epoch": 73.1421287444232, "grad_norm": 0.5435271859169006, "learning_rate": 7.269510511809192e-05, "loss": 0.0251, "step": 114760 }, { "epoch": 73.1485022307202, "grad_norm": 0.3497425615787506, "learning_rate": 7.269019389068696e-05, "loss": 0.0185, "step": 114770 }, { "epoch": 73.1548757170172, "grad_norm": 0.6228663325309753, "learning_rate": 7.268528238757472e-05, "loss": 0.0105, "step": 114780 }, { "epoch": 73.16124920331421, "grad_norm": 0.6168334484100342, "learning_rate": 7.268037060881489e-05, "loss": 0.0161, "step": 114790 }, { "epoch": 73.16762268961122, "grad_norm": 0.37382084131240845, "learning_rate": 7.267545855446713e-05, "loss": 0.0148, "step": 114800 }, { "epoch": 73.17399617590821, "grad_norm": 0.4541049599647522, "learning_rate": 7.267054622459115e-05, "loss": 0.0159, "step": 114810 }, { "epoch": 73.18036966220522, "grad_norm": 0.2264547049999237, "learning_rate": 7.266563361924663e-05, "loss": 0.0181, "step": 114820 }, { "epoch": 73.18674314850223, "grad_norm": 0.45743653178215027, "learning_rate": 7.266072073849325e-05, "loss": 0.0158, "step": 114830 }, { "epoch": 73.19311663479924, "grad_norm": 0.5299502611160278, "learning_rate": 7.26558075823907e-05, "loss": 0.0127, "step": 114840 }, { "epoch": 73.19949012109623, "grad_norm": 0.24945341050624847, "learning_rate": 7.265089415099874e-05, "loss": 0.0214, "step": 114850 }, { "epoch": 73.20586360739324, "grad_norm": 0.6723679900169373, "learning_rate": 7.2645980444377e-05, "loss": 0.0231, "step": 114860 }, { "epoch": 73.21223709369025, "grad_norm": 0.3296510875225067, "learning_rate": 7.264106646258523e-05, "loss": 0.0125, "step": 114870 }, { "epoch": 73.21861057998726, "grad_norm": 0.47656840085983276, "learning_rate": 7.263615220568313e-05, "loss": 0.0164, "step": 114880 }, { "epoch": 73.22498406628426, "grad_norm": 0.421760231256485, "learning_rate": 7.263123767373039e-05, "loss": 0.0123, "step": 114890 }, { "epoch": 73.23135755258126, "grad_norm": 0.47046104073524475, "learning_rate": 7.262632286678675e-05, "loss": 0.028, "step": 114900 }, { "epoch": 73.23773103887827, "grad_norm": 0.2811891436576843, "learning_rate": 7.262140778491194e-05, "loss": 0.0108, "step": 114910 }, { "epoch": 73.24410452517527, "grad_norm": 0.33337539434432983, "learning_rate": 7.261649242816566e-05, "loss": 0.0176, "step": 114920 }, { "epoch": 73.25047801147228, "grad_norm": 0.575538158416748, "learning_rate": 7.261157679660763e-05, "loss": 0.0197, "step": 114930 }, { "epoch": 73.25685149776928, "grad_norm": 0.2701853811740875, "learning_rate": 7.260666089029761e-05, "loss": 0.0183, "step": 114940 }, { "epoch": 73.26322498406628, "grad_norm": 0.5189904570579529, "learning_rate": 7.260174470929531e-05, "loss": 0.015, "step": 114950 }, { "epoch": 73.26959847036329, "grad_norm": 0.5346341133117676, "learning_rate": 7.259682825366046e-05, "loss": 0.0124, "step": 114960 }, { "epoch": 73.2759719566603, "grad_norm": 0.6046198606491089, "learning_rate": 7.259191152345284e-05, "loss": 0.0193, "step": 114970 }, { "epoch": 73.2823454429573, "grad_norm": 0.23794376850128174, "learning_rate": 7.258699451873214e-05, "loss": 0.0185, "step": 114980 }, { "epoch": 73.2887189292543, "grad_norm": 0.5603199005126953, "learning_rate": 7.258207723955814e-05, "loss": 0.0142, "step": 114990 }, { "epoch": 73.29509241555131, "grad_norm": 0.23493841290473938, "learning_rate": 7.257715968599057e-05, "loss": 0.013, "step": 115000 }, { "epoch": 73.30146590184832, "grad_norm": 0.2693469822406769, "learning_rate": 7.25722418580892e-05, "loss": 0.0164, "step": 115010 }, { "epoch": 73.30783938814531, "grad_norm": 0.26224181056022644, "learning_rate": 7.25673237559138e-05, "loss": 0.0206, "step": 115020 }, { "epoch": 73.31421287444232, "grad_norm": 0.27086660265922546, "learning_rate": 7.256240537952408e-05, "loss": 0.0129, "step": 115030 }, { "epoch": 73.32058636073933, "grad_norm": 0.2586776912212372, "learning_rate": 7.255748672897985e-05, "loss": 0.0158, "step": 115040 }, { "epoch": 73.32695984703633, "grad_norm": 0.848334550857544, "learning_rate": 7.255256780434085e-05, "loss": 0.025, "step": 115050 }, { "epoch": 73.33333333333333, "grad_norm": 0.39143237471580505, "learning_rate": 7.254764860566686e-05, "loss": 0.0227, "step": 115060 }, { "epoch": 73.33970681963034, "grad_norm": 0.37229445576667786, "learning_rate": 7.254272913301765e-05, "loss": 0.0216, "step": 115070 }, { "epoch": 73.34608030592734, "grad_norm": 0.30652183294296265, "learning_rate": 7.2537809386453e-05, "loss": 0.0269, "step": 115080 }, { "epoch": 73.35245379222435, "grad_norm": 0.31502023339271545, "learning_rate": 7.253288936603269e-05, "loss": 0.0169, "step": 115090 }, { "epoch": 73.35882727852135, "grad_norm": 0.3458635210990906, "learning_rate": 7.252796907181649e-05, "loss": 0.013, "step": 115100 }, { "epoch": 73.36520076481835, "grad_norm": 0.3791997730731964, "learning_rate": 7.25230485038642e-05, "loss": 0.0174, "step": 115110 }, { "epoch": 73.37157425111536, "grad_norm": 0.5891581177711487, "learning_rate": 7.25181276622356e-05, "loss": 0.0249, "step": 115120 }, { "epoch": 73.37794773741237, "grad_norm": 0.35299086570739746, "learning_rate": 7.25132065469905e-05, "loss": 0.0144, "step": 115130 }, { "epoch": 73.38432122370936, "grad_norm": 0.31718409061431885, "learning_rate": 7.250828515818868e-05, "loss": 0.02, "step": 115140 }, { "epoch": 73.39069471000637, "grad_norm": 0.35293900966644287, "learning_rate": 7.250336349588994e-05, "loss": 0.0244, "step": 115150 }, { "epoch": 73.39706819630338, "grad_norm": 0.3104652166366577, "learning_rate": 7.249844156015408e-05, "loss": 0.0124, "step": 115160 }, { "epoch": 73.40344168260039, "grad_norm": 0.3197007477283478, "learning_rate": 7.249351935104094e-05, "loss": 0.017, "step": 115170 }, { "epoch": 73.40981516889738, "grad_norm": 0.36745020747184753, "learning_rate": 7.248859686861027e-05, "loss": 0.0181, "step": 115180 }, { "epoch": 73.41618865519439, "grad_norm": 0.6316176056861877, "learning_rate": 7.248367411292194e-05, "loss": 0.0164, "step": 115190 }, { "epoch": 73.4225621414914, "grad_norm": 0.4720275402069092, "learning_rate": 7.247875108403572e-05, "loss": 0.0168, "step": 115200 }, { "epoch": 73.4289356277884, "grad_norm": 0.3026297092437744, "learning_rate": 7.247382778201147e-05, "loss": 0.0144, "step": 115210 }, { "epoch": 73.4353091140854, "grad_norm": 0.5124156475067139, "learning_rate": 7.246890420690898e-05, "loss": 0.0206, "step": 115220 }, { "epoch": 73.4416826003824, "grad_norm": 0.8104950785636902, "learning_rate": 7.246398035878811e-05, "loss": 0.0173, "step": 115230 }, { "epoch": 73.44805608667941, "grad_norm": 0.45218324661254883, "learning_rate": 7.245905623770864e-05, "loss": 0.0229, "step": 115240 }, { "epoch": 73.45442957297642, "grad_norm": 0.2616206407546997, "learning_rate": 7.245413184373045e-05, "loss": 0.0151, "step": 115250 }, { "epoch": 73.46080305927342, "grad_norm": 0.31167906522750854, "learning_rate": 7.244920717691336e-05, "loss": 0.0139, "step": 115260 }, { "epoch": 73.46717654557042, "grad_norm": 0.6149519681930542, "learning_rate": 7.24442822373172e-05, "loss": 0.0174, "step": 115270 }, { "epoch": 73.47355003186743, "grad_norm": 0.7584963440895081, "learning_rate": 7.243935702500182e-05, "loss": 0.0192, "step": 115280 }, { "epoch": 73.47992351816444, "grad_norm": 0.35027632117271423, "learning_rate": 7.243443154002707e-05, "loss": 0.0107, "step": 115290 }, { "epoch": 73.48629700446143, "grad_norm": 0.25112730264663696, "learning_rate": 7.242950578245279e-05, "loss": 0.0163, "step": 115300 }, { "epoch": 73.49267049075844, "grad_norm": 0.3585738241672516, "learning_rate": 7.242457975233885e-05, "loss": 0.0135, "step": 115310 }, { "epoch": 73.49904397705545, "grad_norm": 0.5342162847518921, "learning_rate": 7.241965344974508e-05, "loss": 0.0134, "step": 115320 }, { "epoch": 73.50541746335246, "grad_norm": 0.3074877858161926, "learning_rate": 7.241472687473136e-05, "loss": 0.0123, "step": 115330 }, { "epoch": 73.51179094964945, "grad_norm": 0.3166631758213043, "learning_rate": 7.240980002735753e-05, "loss": 0.0124, "step": 115340 }, { "epoch": 73.51816443594646, "grad_norm": 0.3163415789604187, "learning_rate": 7.24048729076835e-05, "loss": 0.0104, "step": 115350 }, { "epoch": 73.52453792224347, "grad_norm": 0.3791068196296692, "learning_rate": 7.23999455157691e-05, "loss": 0.0164, "step": 115360 }, { "epoch": 73.53091140854048, "grad_norm": 0.4049588739871979, "learning_rate": 7.23950178516742e-05, "loss": 0.0181, "step": 115370 }, { "epoch": 73.53728489483747, "grad_norm": 0.27487578988075256, "learning_rate": 7.23900899154587e-05, "loss": 0.0145, "step": 115380 }, { "epoch": 73.54365838113448, "grad_norm": 0.5929043889045715, "learning_rate": 7.238516170718245e-05, "loss": 0.0164, "step": 115390 }, { "epoch": 73.55003186743149, "grad_norm": 0.5540155172348022, "learning_rate": 7.238023322690537e-05, "loss": 0.0168, "step": 115400 }, { "epoch": 73.5564053537285, "grad_norm": 0.4740724563598633, "learning_rate": 7.237530447468732e-05, "loss": 0.0153, "step": 115410 }, { "epoch": 73.56277884002549, "grad_norm": 0.6215957999229431, "learning_rate": 7.237037545058821e-05, "loss": 0.016, "step": 115420 }, { "epoch": 73.5691523263225, "grad_norm": 0.24161942303180695, "learning_rate": 7.236544615466789e-05, "loss": 0.0167, "step": 115430 }, { "epoch": 73.5755258126195, "grad_norm": 0.4381413459777832, "learning_rate": 7.236051658698629e-05, "loss": 0.0224, "step": 115440 }, { "epoch": 73.58189929891651, "grad_norm": 0.24443212151527405, "learning_rate": 7.235558674760331e-05, "loss": 0.0162, "step": 115450 }, { "epoch": 73.5882727852135, "grad_norm": 0.24450792372226715, "learning_rate": 7.235065663657883e-05, "loss": 0.0152, "step": 115460 }, { "epoch": 73.59464627151051, "grad_norm": 0.5791457295417786, "learning_rate": 7.234572625397279e-05, "loss": 0.0242, "step": 115470 }, { "epoch": 73.60101975780752, "grad_norm": 0.33753135800361633, "learning_rate": 7.234079559984505e-05, "loss": 0.0188, "step": 115480 }, { "epoch": 73.60739324410453, "grad_norm": 0.4831898510456085, "learning_rate": 7.233586467425558e-05, "loss": 0.018, "step": 115490 }, { "epoch": 73.61376673040154, "grad_norm": 0.536468505859375, "learning_rate": 7.233093347726426e-05, "loss": 0.0228, "step": 115500 }, { "epoch": 73.62014021669853, "grad_norm": 0.32197025418281555, "learning_rate": 7.232600200893101e-05, "loss": 0.0225, "step": 115510 }, { "epoch": 73.62651370299554, "grad_norm": 0.2870180308818817, "learning_rate": 7.232107026931576e-05, "loss": 0.018, "step": 115520 }, { "epoch": 73.63288718929255, "grad_norm": 0.4758233428001404, "learning_rate": 7.231613825847841e-05, "loss": 0.0176, "step": 115530 }, { "epoch": 73.63926067558955, "grad_norm": 0.5439596176147461, "learning_rate": 7.231120597647893e-05, "loss": 0.0135, "step": 115540 }, { "epoch": 73.64563416188655, "grad_norm": 0.22396385669708252, "learning_rate": 7.230627342337724e-05, "loss": 0.0119, "step": 115550 }, { "epoch": 73.65200764818356, "grad_norm": 0.30205848813056946, "learning_rate": 7.230134059923325e-05, "loss": 0.017, "step": 115560 }, { "epoch": 73.65838113448056, "grad_norm": 0.4344211220741272, "learning_rate": 7.229640750410692e-05, "loss": 0.0196, "step": 115570 }, { "epoch": 73.66475462077757, "grad_norm": 0.5814890265464783, "learning_rate": 7.229147413805819e-05, "loss": 0.016, "step": 115580 }, { "epoch": 73.67112810707457, "grad_norm": 0.5645961165428162, "learning_rate": 7.2286540501147e-05, "loss": 0.0196, "step": 115590 }, { "epoch": 73.67750159337157, "grad_norm": 0.5897126197814941, "learning_rate": 7.22816065934333e-05, "loss": 0.0223, "step": 115600 }, { "epoch": 73.68387507966858, "grad_norm": 0.2617107331752777, "learning_rate": 7.227667241497705e-05, "loss": 0.0166, "step": 115610 }, { "epoch": 73.69024856596559, "grad_norm": 0.3378779888153076, "learning_rate": 7.227173796583819e-05, "loss": 0.0201, "step": 115620 }, { "epoch": 73.69662205226258, "grad_norm": 0.36935514211654663, "learning_rate": 7.226680324607668e-05, "loss": 0.0169, "step": 115630 }, { "epoch": 73.70299553855959, "grad_norm": 0.33140432834625244, "learning_rate": 7.22618682557525e-05, "loss": 0.0155, "step": 115640 }, { "epoch": 73.7093690248566, "grad_norm": 0.3313341438770294, "learning_rate": 7.22569329949256e-05, "loss": 0.016, "step": 115650 }, { "epoch": 73.7157425111536, "grad_norm": 0.5798285603523254, "learning_rate": 7.225199746365594e-05, "loss": 0.0224, "step": 115660 }, { "epoch": 73.7221159974506, "grad_norm": 0.35749414563179016, "learning_rate": 7.224706166200352e-05, "loss": 0.0125, "step": 115670 }, { "epoch": 73.72848948374761, "grad_norm": 0.3144128918647766, "learning_rate": 7.224212559002829e-05, "loss": 0.0172, "step": 115680 }, { "epoch": 73.73486297004462, "grad_norm": 0.34504345059394836, "learning_rate": 7.223718924779023e-05, "loss": 0.0148, "step": 115690 }, { "epoch": 73.74123645634162, "grad_norm": 0.299193412065506, "learning_rate": 7.223225263534932e-05, "loss": 0.0262, "step": 115700 }, { "epoch": 73.74760994263862, "grad_norm": 0.605874240398407, "learning_rate": 7.222731575276555e-05, "loss": 0.0128, "step": 115710 }, { "epoch": 73.75398342893563, "grad_norm": 0.48805171251296997, "learning_rate": 7.222237860009892e-05, "loss": 0.017, "step": 115720 }, { "epoch": 73.76035691523263, "grad_norm": 0.25587064027786255, "learning_rate": 7.22174411774094e-05, "loss": 0.034, "step": 115730 }, { "epoch": 73.76673040152964, "grad_norm": 0.9557430744171143, "learning_rate": 7.221250348475698e-05, "loss": 0.024, "step": 115740 }, { "epoch": 73.77310388782664, "grad_norm": 0.41542255878448486, "learning_rate": 7.22075655222017e-05, "loss": 0.0174, "step": 115750 }, { "epoch": 73.77947737412364, "grad_norm": 0.6677271723747253, "learning_rate": 7.22026272898035e-05, "loss": 0.0169, "step": 115760 }, { "epoch": 73.78585086042065, "grad_norm": 0.5032698512077332, "learning_rate": 7.219768878762244e-05, "loss": 0.0106, "step": 115770 }, { "epoch": 73.79222434671766, "grad_norm": 0.7453247308731079, "learning_rate": 7.219275001571849e-05, "loss": 0.0133, "step": 115780 }, { "epoch": 73.79859783301465, "grad_norm": 0.4247680604457855, "learning_rate": 7.218781097415168e-05, "loss": 0.0138, "step": 115790 }, { "epoch": 73.80497131931166, "grad_norm": 0.5640400648117065, "learning_rate": 7.218287166298201e-05, "loss": 0.0201, "step": 115800 }, { "epoch": 73.81134480560867, "grad_norm": 0.33291658759117126, "learning_rate": 7.217793208226952e-05, "loss": 0.0182, "step": 115810 }, { "epoch": 73.81771829190568, "grad_norm": 0.4215115010738373, "learning_rate": 7.21729922320742e-05, "loss": 0.0164, "step": 115820 }, { "epoch": 73.82409177820267, "grad_norm": 0.3280388116836548, "learning_rate": 7.216805211245609e-05, "loss": 0.0225, "step": 115830 }, { "epoch": 73.83046526449968, "grad_norm": 0.3067978024482727, "learning_rate": 7.216311172347525e-05, "loss": 0.0121, "step": 115840 }, { "epoch": 73.83683875079669, "grad_norm": 0.7962836623191833, "learning_rate": 7.215817106519164e-05, "loss": 0.0155, "step": 115850 }, { "epoch": 73.8432122370937, "grad_norm": 0.23182956874370575, "learning_rate": 7.215323013766533e-05, "loss": 0.0232, "step": 115860 }, { "epoch": 73.84958572339069, "grad_norm": 0.48458367586135864, "learning_rate": 7.214828894095638e-05, "loss": 0.0208, "step": 115870 }, { "epoch": 73.8559592096877, "grad_norm": 0.19405071437358856, "learning_rate": 7.21433474751248e-05, "loss": 0.0152, "step": 115880 }, { "epoch": 73.8623326959847, "grad_norm": 0.5217065811157227, "learning_rate": 7.213840574023063e-05, "loss": 0.0177, "step": 115890 }, { "epoch": 73.86870618228171, "grad_norm": 0.26839619874954224, "learning_rate": 7.213346373633393e-05, "loss": 0.0197, "step": 115900 }, { "epoch": 73.8750796685787, "grad_norm": 0.2389526069164276, "learning_rate": 7.212852146349475e-05, "loss": 0.0102, "step": 115910 }, { "epoch": 73.88145315487571, "grad_norm": 0.26983219385147095, "learning_rate": 7.212357892177315e-05, "loss": 0.0202, "step": 115920 }, { "epoch": 73.88782664117272, "grad_norm": 0.2893909513950348, "learning_rate": 7.211863611122918e-05, "loss": 0.0191, "step": 115930 }, { "epoch": 73.89420012746973, "grad_norm": 0.2626626193523407, "learning_rate": 7.211369303192288e-05, "loss": 0.0214, "step": 115940 }, { "epoch": 73.90057361376672, "grad_norm": 0.4110710620880127, "learning_rate": 7.210874968391435e-05, "loss": 0.0191, "step": 115950 }, { "epoch": 73.90694710006373, "grad_norm": 0.3613704442977905, "learning_rate": 7.210380606726362e-05, "loss": 0.0196, "step": 115960 }, { "epoch": 73.91332058636074, "grad_norm": 0.5233836770057678, "learning_rate": 7.20988621820308e-05, "loss": 0.0261, "step": 115970 }, { "epoch": 73.91969407265775, "grad_norm": 0.28794923424720764, "learning_rate": 7.209391802827592e-05, "loss": 0.0221, "step": 115980 }, { "epoch": 73.92606755895474, "grad_norm": 0.6523963809013367, "learning_rate": 7.208897360605909e-05, "loss": 0.0244, "step": 115990 }, { "epoch": 73.93244104525175, "grad_norm": 0.2600138187408447, "learning_rate": 7.208402891544038e-05, "loss": 0.0128, "step": 116000 }, { "epoch": 73.93881453154876, "grad_norm": 0.6235347986221313, "learning_rate": 7.207908395647983e-05, "loss": 0.0169, "step": 116010 }, { "epoch": 73.94518801784577, "grad_norm": 0.23931175470352173, "learning_rate": 7.20741387292376e-05, "loss": 0.0139, "step": 116020 }, { "epoch": 73.95156150414276, "grad_norm": 0.5392659306526184, "learning_rate": 7.206919323377373e-05, "loss": 0.0161, "step": 116030 }, { "epoch": 73.95793499043977, "grad_norm": 0.5339849591255188, "learning_rate": 7.206424747014833e-05, "loss": 0.0282, "step": 116040 }, { "epoch": 73.96430847673678, "grad_norm": 0.4119715988636017, "learning_rate": 7.205930143842147e-05, "loss": 0.0154, "step": 116050 }, { "epoch": 73.97068196303378, "grad_norm": 0.3263351321220398, "learning_rate": 7.20543551386533e-05, "loss": 0.0177, "step": 116060 }, { "epoch": 73.97705544933078, "grad_norm": 0.621549129486084, "learning_rate": 7.204940857090389e-05, "loss": 0.0139, "step": 116070 }, { "epoch": 73.98342893562778, "grad_norm": 0.8502930998802185, "learning_rate": 7.204446173523333e-05, "loss": 0.0242, "step": 116080 }, { "epoch": 73.9898024219248, "grad_norm": 0.33467820286750793, "learning_rate": 7.203951463170176e-05, "loss": 0.0129, "step": 116090 }, { "epoch": 73.9961759082218, "grad_norm": 0.6854860186576843, "learning_rate": 7.203456726036927e-05, "loss": 0.0131, "step": 116100 }, { "epoch": 74.00254939451881, "grad_norm": 0.3075568377971649, "learning_rate": 7.2029619621296e-05, "loss": 0.0117, "step": 116110 }, { "epoch": 74.0089228808158, "grad_norm": 0.40048810839653015, "learning_rate": 7.202467171454202e-05, "loss": 0.0149, "step": 116120 }, { "epoch": 74.01529636711281, "grad_norm": 0.3538587987422943, "learning_rate": 7.201972354016752e-05, "loss": 0.0093, "step": 116130 }, { "epoch": 74.02166985340982, "grad_norm": 0.41513967514038086, "learning_rate": 7.201477509823255e-05, "loss": 0.0109, "step": 116140 }, { "epoch": 74.02804333970683, "grad_norm": 0.4846752882003784, "learning_rate": 7.200982638879733e-05, "loss": 0.0222, "step": 116150 }, { "epoch": 74.03441682600382, "grad_norm": 0.4269276261329651, "learning_rate": 7.20048774119219e-05, "loss": 0.0152, "step": 116160 }, { "epoch": 74.04079031230083, "grad_norm": 0.30178314447402954, "learning_rate": 7.199992816766644e-05, "loss": 0.0161, "step": 116170 }, { "epoch": 74.04716379859784, "grad_norm": 0.521265983581543, "learning_rate": 7.199497865609108e-05, "loss": 0.0207, "step": 116180 }, { "epoch": 74.05353728489484, "grad_norm": 0.4082285761833191, "learning_rate": 7.199002887725597e-05, "loss": 0.0222, "step": 116190 }, { "epoch": 74.05991077119184, "grad_norm": 0.3966520428657532, "learning_rate": 7.198507883122124e-05, "loss": 0.0155, "step": 116200 }, { "epoch": 74.06628425748885, "grad_norm": 0.31399857997894287, "learning_rate": 7.198012851804705e-05, "loss": 0.0192, "step": 116210 }, { "epoch": 74.07265774378585, "grad_norm": 0.5151280164718628, "learning_rate": 7.197517793779354e-05, "loss": 0.0126, "step": 116220 }, { "epoch": 74.07903123008286, "grad_norm": 0.8411592245101929, "learning_rate": 7.197022709052088e-05, "loss": 0.0207, "step": 116230 }, { "epoch": 74.08540471637986, "grad_norm": 0.29851576685905457, "learning_rate": 7.19652759762892e-05, "loss": 0.02, "step": 116240 }, { "epoch": 74.09177820267686, "grad_norm": 0.5149940848350525, "learning_rate": 7.19603245951587e-05, "loss": 0.0172, "step": 116250 }, { "epoch": 74.09815168897387, "grad_norm": 0.4416622221469879, "learning_rate": 7.19553729471895e-05, "loss": 0.0168, "step": 116260 }, { "epoch": 74.10452517527088, "grad_norm": 0.4956913888454437, "learning_rate": 7.195042103244181e-05, "loss": 0.0134, "step": 116270 }, { "epoch": 74.11089866156787, "grad_norm": 0.37001433968544006, "learning_rate": 7.194546885097576e-05, "loss": 0.0166, "step": 116280 }, { "epoch": 74.11727214786488, "grad_norm": 0.2170042246580124, "learning_rate": 7.194051640285157e-05, "loss": 0.0147, "step": 116290 }, { "epoch": 74.12364563416189, "grad_norm": 0.42293381690979004, "learning_rate": 7.193556368812936e-05, "loss": 0.0136, "step": 116300 }, { "epoch": 74.1300191204589, "grad_norm": 0.3197706937789917, "learning_rate": 7.193061070686936e-05, "loss": 0.0189, "step": 116310 }, { "epoch": 74.13639260675589, "grad_norm": 0.2805660367012024, "learning_rate": 7.192565745913173e-05, "loss": 0.0109, "step": 116320 }, { "epoch": 74.1427660930529, "grad_norm": 0.3782398998737335, "learning_rate": 7.192070394497667e-05, "loss": 0.0106, "step": 116330 }, { "epoch": 74.1491395793499, "grad_norm": 0.4187425374984741, "learning_rate": 7.191575016446435e-05, "loss": 0.0167, "step": 116340 }, { "epoch": 74.15551306564691, "grad_norm": 0.44776809215545654, "learning_rate": 7.191079611765496e-05, "loss": 0.0178, "step": 116350 }, { "epoch": 74.16188655194391, "grad_norm": 0.7005776762962341, "learning_rate": 7.190584180460874e-05, "loss": 0.0322, "step": 116360 }, { "epoch": 74.16826003824092, "grad_norm": 0.23176611959934235, "learning_rate": 7.190088722538582e-05, "loss": 0.022, "step": 116370 }, { "epoch": 74.17463352453792, "grad_norm": 0.4133422076702118, "learning_rate": 7.189593238004648e-05, "loss": 0.0209, "step": 116380 }, { "epoch": 74.18100701083493, "grad_norm": 0.49787554144859314, "learning_rate": 7.189097726865088e-05, "loss": 0.0175, "step": 116390 }, { "epoch": 74.18738049713193, "grad_norm": 0.40401026606559753, "learning_rate": 7.188602189125923e-05, "loss": 0.0137, "step": 116400 }, { "epoch": 74.19375398342893, "grad_norm": 0.579494297504425, "learning_rate": 7.188106624793175e-05, "loss": 0.0203, "step": 116410 }, { "epoch": 74.20012746972594, "grad_norm": 0.3028266131877899, "learning_rate": 7.187611033872865e-05, "loss": 0.0137, "step": 116420 }, { "epoch": 74.20650095602295, "grad_norm": 0.2473432868719101, "learning_rate": 7.187115416371018e-05, "loss": 0.0145, "step": 116430 }, { "epoch": 74.21287444231994, "grad_norm": 0.6884672045707703, "learning_rate": 7.18661977229365e-05, "loss": 0.0375, "step": 116440 }, { "epoch": 74.21924792861695, "grad_norm": 0.14972451329231262, "learning_rate": 7.18612410164679e-05, "loss": 0.0281, "step": 116450 }, { "epoch": 74.22562141491396, "grad_norm": 0.22430667281150818, "learning_rate": 7.185628404436458e-05, "loss": 0.0105, "step": 116460 }, { "epoch": 74.23199490121097, "grad_norm": 0.4549750089645386, "learning_rate": 7.185132680668676e-05, "loss": 0.0173, "step": 116470 }, { "epoch": 74.23836838750796, "grad_norm": 0.4119848906993866, "learning_rate": 7.184636930349472e-05, "loss": 0.0101, "step": 116480 }, { "epoch": 74.24474187380497, "grad_norm": 0.550010621547699, "learning_rate": 7.184141153484862e-05, "loss": 0.0242, "step": 116490 }, { "epoch": 74.25111536010198, "grad_norm": 0.5055806040763855, "learning_rate": 7.183645350080877e-05, "loss": 0.0238, "step": 116500 }, { "epoch": 74.25748884639899, "grad_norm": 0.2379290759563446, "learning_rate": 7.18314952014354e-05, "loss": 0.0128, "step": 116510 }, { "epoch": 74.26386233269598, "grad_norm": 0.5193620920181274, "learning_rate": 7.182653663678873e-05, "loss": 0.0191, "step": 116520 }, { "epoch": 74.27023581899299, "grad_norm": 0.3642948567867279, "learning_rate": 7.182157780692905e-05, "loss": 0.0157, "step": 116530 }, { "epoch": 74.27660930529, "grad_norm": 0.3807300627231598, "learning_rate": 7.18166187119166e-05, "loss": 0.0105, "step": 116540 }, { "epoch": 74.282982791587, "grad_norm": 0.4102737009525299, "learning_rate": 7.181165935181163e-05, "loss": 0.015, "step": 116550 }, { "epoch": 74.289356277884, "grad_norm": 0.35219132900238037, "learning_rate": 7.18066997266744e-05, "loss": 0.0165, "step": 116560 }, { "epoch": 74.295729764181, "grad_norm": 0.316741943359375, "learning_rate": 7.180173983656518e-05, "loss": 0.0245, "step": 116570 }, { "epoch": 74.30210325047801, "grad_norm": 0.6073954701423645, "learning_rate": 7.179677968154424e-05, "loss": 0.0146, "step": 116580 }, { "epoch": 74.30847673677502, "grad_norm": 0.45314398407936096, "learning_rate": 7.179181926167184e-05, "loss": 0.0142, "step": 116590 }, { "epoch": 74.31485022307201, "grad_norm": 0.4378422796726227, "learning_rate": 7.178685857700827e-05, "loss": 0.0128, "step": 116600 }, { "epoch": 74.32122370936902, "grad_norm": 0.8679520487785339, "learning_rate": 7.178189762761378e-05, "loss": 0.0211, "step": 116610 }, { "epoch": 74.32759719566603, "grad_norm": 0.2875358760356903, "learning_rate": 7.177693641354868e-05, "loss": 0.012, "step": 116620 }, { "epoch": 74.33397068196304, "grad_norm": 0.33438658714294434, "learning_rate": 7.177197493487326e-05, "loss": 0.0125, "step": 116630 }, { "epoch": 74.34034416826003, "grad_norm": 0.5345471501350403, "learning_rate": 7.176701319164777e-05, "loss": 0.0194, "step": 116640 }, { "epoch": 74.34671765455704, "grad_norm": 0.4707488715648651, "learning_rate": 7.176205118393251e-05, "loss": 0.0164, "step": 116650 }, { "epoch": 74.35309114085405, "grad_norm": 0.2933320999145508, "learning_rate": 7.17570889117878e-05, "loss": 0.0177, "step": 116660 }, { "epoch": 74.35946462715106, "grad_norm": 0.2661857306957245, "learning_rate": 7.17521263752739e-05, "loss": 0.0105, "step": 116670 }, { "epoch": 74.36583811344805, "grad_norm": 0.20921756327152252, "learning_rate": 7.174716357445115e-05, "loss": 0.0163, "step": 116680 }, { "epoch": 74.37221159974506, "grad_norm": 0.4425075948238373, "learning_rate": 7.17422005093798e-05, "loss": 0.0255, "step": 116690 }, { "epoch": 74.37858508604207, "grad_norm": 0.590911328792572, "learning_rate": 7.173723718012022e-05, "loss": 0.0164, "step": 116700 }, { "epoch": 74.38495857233907, "grad_norm": 0.37575894594192505, "learning_rate": 7.173227358673267e-05, "loss": 0.0265, "step": 116710 }, { "epoch": 74.39133205863607, "grad_norm": 0.35875996947288513, "learning_rate": 7.172730972927748e-05, "loss": 0.0162, "step": 116720 }, { "epoch": 74.39770554493307, "grad_norm": 0.32842519879341125, "learning_rate": 7.172234560781495e-05, "loss": 0.0209, "step": 116730 }, { "epoch": 74.40407903123008, "grad_norm": 0.3168003261089325, "learning_rate": 7.171738122240543e-05, "loss": 0.0157, "step": 116740 }, { "epoch": 74.41045251752709, "grad_norm": 0.29454731941223145, "learning_rate": 7.171241657310921e-05, "loss": 0.0168, "step": 116750 }, { "epoch": 74.4168260038241, "grad_norm": 0.4793085753917694, "learning_rate": 7.170745165998663e-05, "loss": 0.0105, "step": 116760 }, { "epoch": 74.42319949012109, "grad_norm": 0.5265362858772278, "learning_rate": 7.170248648309801e-05, "loss": 0.0166, "step": 116770 }, { "epoch": 74.4295729764181, "grad_norm": 0.15961089730262756, "learning_rate": 7.16975210425037e-05, "loss": 0.0177, "step": 116780 }, { "epoch": 74.43594646271511, "grad_norm": 0.3948343098163605, "learning_rate": 7.169255533826403e-05, "loss": 0.0178, "step": 116790 }, { "epoch": 74.44231994901212, "grad_norm": 0.3741171956062317, "learning_rate": 7.168758937043933e-05, "loss": 0.0213, "step": 116800 }, { "epoch": 74.44869343530911, "grad_norm": 0.2917334735393524, "learning_rate": 7.168262313908993e-05, "loss": 0.0234, "step": 116810 }, { "epoch": 74.45506692160612, "grad_norm": 0.3895876705646515, "learning_rate": 7.16776566442762e-05, "loss": 0.0199, "step": 116820 }, { "epoch": 74.46144040790313, "grad_norm": 0.22522903978824615, "learning_rate": 7.167268988605845e-05, "loss": 0.0154, "step": 116830 }, { "epoch": 74.46781389420013, "grad_norm": 0.4412127435207367, "learning_rate": 7.166772286449708e-05, "loss": 0.0232, "step": 116840 }, { "epoch": 74.47418738049713, "grad_norm": 0.4267738461494446, "learning_rate": 7.166275557965241e-05, "loss": 0.0192, "step": 116850 }, { "epoch": 74.48056086679414, "grad_norm": 0.41632208228111267, "learning_rate": 7.16577880315848e-05, "loss": 0.0135, "step": 116860 }, { "epoch": 74.48693435309114, "grad_norm": 0.4077979028224945, "learning_rate": 7.165282022035462e-05, "loss": 0.0245, "step": 116870 }, { "epoch": 74.49330783938815, "grad_norm": 0.3422371447086334, "learning_rate": 7.164785214602226e-05, "loss": 0.0314, "step": 116880 }, { "epoch": 74.49968132568515, "grad_norm": 0.45189401507377625, "learning_rate": 7.164288380864803e-05, "loss": 0.0215, "step": 116890 }, { "epoch": 74.50605481198215, "grad_norm": 0.8089586496353149, "learning_rate": 7.163791520829232e-05, "loss": 0.028, "step": 116900 }, { "epoch": 74.51242829827916, "grad_norm": 0.42305925488471985, "learning_rate": 7.163294634501553e-05, "loss": 0.0176, "step": 116910 }, { "epoch": 74.51880178457617, "grad_norm": 0.6178303360939026, "learning_rate": 7.162797721887801e-05, "loss": 0.0173, "step": 116920 }, { "epoch": 74.52517527087316, "grad_norm": 0.3254499137401581, "learning_rate": 7.162300782994015e-05, "loss": 0.0188, "step": 116930 }, { "epoch": 74.53154875717017, "grad_norm": 0.4442274868488312, "learning_rate": 7.161803817826231e-05, "loss": 0.0118, "step": 116940 }, { "epoch": 74.53792224346718, "grad_norm": 0.19798272848129272, "learning_rate": 7.161306826390491e-05, "loss": 0.014, "step": 116950 }, { "epoch": 74.54429572976419, "grad_norm": 0.49758380651474, "learning_rate": 7.160809808692835e-05, "loss": 0.0182, "step": 116960 }, { "epoch": 74.55066921606118, "grad_norm": 0.23390142619609833, "learning_rate": 7.160312764739295e-05, "loss": 0.0175, "step": 116970 }, { "epoch": 74.55704270235819, "grad_norm": 0.2280474305152893, "learning_rate": 7.159815694535919e-05, "loss": 0.0154, "step": 116980 }, { "epoch": 74.5634161886552, "grad_norm": 0.39231476187705994, "learning_rate": 7.15931859808874e-05, "loss": 0.0125, "step": 116990 }, { "epoch": 74.5697896749522, "grad_norm": 0.4148441255092621, "learning_rate": 7.158821475403803e-05, "loss": 0.013, "step": 117000 }, { "epoch": 74.5761631612492, "grad_norm": 0.37935805320739746, "learning_rate": 7.158324326487147e-05, "loss": 0.0162, "step": 117010 }, { "epoch": 74.5825366475462, "grad_norm": 0.2716043293476105, "learning_rate": 7.157827151344813e-05, "loss": 0.0116, "step": 117020 }, { "epoch": 74.58891013384321, "grad_norm": 0.2592540979385376, "learning_rate": 7.15732994998284e-05, "loss": 0.0139, "step": 117030 }, { "epoch": 74.59528362014022, "grad_norm": 0.552582323551178, "learning_rate": 7.156832722407271e-05, "loss": 0.0172, "step": 117040 }, { "epoch": 74.60165710643722, "grad_norm": 0.5816164016723633, "learning_rate": 7.156335468624151e-05, "loss": 0.0307, "step": 117050 }, { "epoch": 74.60803059273422, "grad_norm": 0.44948694109916687, "learning_rate": 7.155838188639517e-05, "loss": 0.0154, "step": 117060 }, { "epoch": 74.61440407903123, "grad_norm": 0.34117797017097473, "learning_rate": 7.155340882459414e-05, "loss": 0.019, "step": 117070 }, { "epoch": 74.62077756532824, "grad_norm": 0.32258760929107666, "learning_rate": 7.154843550089885e-05, "loss": 0.0211, "step": 117080 }, { "epoch": 74.62715105162523, "grad_norm": 0.2637089788913727, "learning_rate": 7.154346191536971e-05, "loss": 0.0176, "step": 117090 }, { "epoch": 74.63352453792224, "grad_norm": 0.4709986746311188, "learning_rate": 7.153848806806718e-05, "loss": 0.0176, "step": 117100 }, { "epoch": 74.63989802421925, "grad_norm": 0.41009974479675293, "learning_rate": 7.153351395905168e-05, "loss": 0.0179, "step": 117110 }, { "epoch": 74.64627151051626, "grad_norm": 0.6144423484802246, "learning_rate": 7.152853958838366e-05, "loss": 0.0136, "step": 117120 }, { "epoch": 74.65264499681325, "grad_norm": 0.3654935657978058, "learning_rate": 7.152356495612354e-05, "loss": 0.0193, "step": 117130 }, { "epoch": 74.65901848311026, "grad_norm": 0.2829592227935791, "learning_rate": 7.151859006233179e-05, "loss": 0.0112, "step": 117140 }, { "epoch": 74.66539196940727, "grad_norm": 0.6396158933639526, "learning_rate": 7.151361490706886e-05, "loss": 0.0197, "step": 117150 }, { "epoch": 74.67176545570427, "grad_norm": 0.2542078197002411, "learning_rate": 7.15086394903952e-05, "loss": 0.0139, "step": 117160 }, { "epoch": 74.67813894200127, "grad_norm": 0.29929178953170776, "learning_rate": 7.150366381237125e-05, "loss": 0.0255, "step": 117170 }, { "epoch": 74.68451242829828, "grad_norm": 0.545782208442688, "learning_rate": 7.149868787305749e-05, "loss": 0.0162, "step": 117180 }, { "epoch": 74.69088591459528, "grad_norm": 0.34052392840385437, "learning_rate": 7.149371167251438e-05, "loss": 0.0206, "step": 117190 }, { "epoch": 74.69725940089229, "grad_norm": 0.27173107862472534, "learning_rate": 7.148873521080237e-05, "loss": 0.0177, "step": 117200 }, { "epoch": 74.70363288718929, "grad_norm": 0.3928282856941223, "learning_rate": 7.148375848798194e-05, "loss": 0.0142, "step": 117210 }, { "epoch": 74.7100063734863, "grad_norm": 0.21264420449733734, "learning_rate": 7.147878150411356e-05, "loss": 0.0128, "step": 117220 }, { "epoch": 74.7163798597833, "grad_norm": 0.8242433667182922, "learning_rate": 7.147380425925771e-05, "loss": 0.019, "step": 117230 }, { "epoch": 74.72275334608031, "grad_norm": 0.9869304895401001, "learning_rate": 7.146882675347486e-05, "loss": 0.0136, "step": 117240 }, { "epoch": 74.7291268323773, "grad_norm": 0.42476144433021545, "learning_rate": 7.14638489868255e-05, "loss": 0.0282, "step": 117250 }, { "epoch": 74.73550031867431, "grad_norm": 0.22038203477859497, "learning_rate": 7.14588709593701e-05, "loss": 0.0161, "step": 117260 }, { "epoch": 74.74187380497132, "grad_norm": 0.38567858934402466, "learning_rate": 7.145389267116917e-05, "loss": 0.0101, "step": 117270 }, { "epoch": 74.74824729126833, "grad_norm": 0.23413538932800293, "learning_rate": 7.144891412228318e-05, "loss": 0.0193, "step": 117280 }, { "epoch": 74.75462077756532, "grad_norm": 0.24906465411186218, "learning_rate": 7.144393531277264e-05, "loss": 0.0149, "step": 117290 }, { "epoch": 74.76099426386233, "grad_norm": 0.4420420229434967, "learning_rate": 7.143895624269802e-05, "loss": 0.0144, "step": 117300 }, { "epoch": 74.76736775015934, "grad_norm": 0.5890982151031494, "learning_rate": 7.143397691211986e-05, "loss": 0.0173, "step": 117310 }, { "epoch": 74.77374123645635, "grad_norm": 0.5032193064689636, "learning_rate": 7.142899732109864e-05, "loss": 0.0272, "step": 117320 }, { "epoch": 74.78011472275334, "grad_norm": 0.31868594884872437, "learning_rate": 7.142401746969487e-05, "loss": 0.0194, "step": 117330 }, { "epoch": 74.78648820905035, "grad_norm": 0.4130663275718689, "learning_rate": 7.141903735796904e-05, "loss": 0.018, "step": 117340 }, { "epoch": 74.79286169534736, "grad_norm": 0.2160739302635193, "learning_rate": 7.141405698598172e-05, "loss": 0.018, "step": 117350 }, { "epoch": 74.79923518164436, "grad_norm": 0.2437848150730133, "learning_rate": 7.140907635379336e-05, "loss": 0.0194, "step": 117360 }, { "epoch": 74.80560866794136, "grad_norm": 0.3845278322696686, "learning_rate": 7.140409546146452e-05, "loss": 0.0192, "step": 117370 }, { "epoch": 74.81198215423836, "grad_norm": 0.254384845495224, "learning_rate": 7.139911430905571e-05, "loss": 0.0123, "step": 117380 }, { "epoch": 74.81835564053537, "grad_norm": 0.2830893099308014, "learning_rate": 7.139413289662745e-05, "loss": 0.0231, "step": 117390 }, { "epoch": 74.82472912683238, "grad_norm": 0.5557066798210144, "learning_rate": 7.138915122424029e-05, "loss": 0.0213, "step": 117400 }, { "epoch": 74.83110261312939, "grad_norm": 0.7901528477668762, "learning_rate": 7.138416929195474e-05, "loss": 0.0251, "step": 117410 }, { "epoch": 74.83747609942638, "grad_norm": 0.4900665283203125, "learning_rate": 7.137918709983134e-05, "loss": 0.0188, "step": 117420 }, { "epoch": 74.84384958572339, "grad_norm": 0.40884748101234436, "learning_rate": 7.137420464793063e-05, "loss": 0.0234, "step": 117430 }, { "epoch": 74.8502230720204, "grad_norm": 0.5936700105667114, "learning_rate": 7.136922193631317e-05, "loss": 0.0181, "step": 117440 }, { "epoch": 74.8565965583174, "grad_norm": 0.4353640675544739, "learning_rate": 7.136423896503948e-05, "loss": 0.0083, "step": 117450 }, { "epoch": 74.8629700446144, "grad_norm": 0.21858566999435425, "learning_rate": 7.13592557341701e-05, "loss": 0.0138, "step": 117460 }, { "epoch": 74.86934353091141, "grad_norm": 0.2760002613067627, "learning_rate": 7.135427224376561e-05, "loss": 0.0139, "step": 117470 }, { "epoch": 74.87571701720842, "grad_norm": 0.6029525399208069, "learning_rate": 7.134928849388655e-05, "loss": 0.0124, "step": 117480 }, { "epoch": 74.88209050350542, "grad_norm": 0.6593682169914246, "learning_rate": 7.134430448459348e-05, "loss": 0.0142, "step": 117490 }, { "epoch": 74.88846398980242, "grad_norm": 0.3489951193332672, "learning_rate": 7.133932021594694e-05, "loss": 0.0115, "step": 117500 }, { "epoch": 74.89483747609943, "grad_norm": 0.5161749720573425, "learning_rate": 7.133433568800752e-05, "loss": 0.0182, "step": 117510 }, { "epoch": 74.90121096239643, "grad_norm": 0.3992754817008972, "learning_rate": 7.13293509008358e-05, "loss": 0.015, "step": 117520 }, { "epoch": 74.90758444869344, "grad_norm": 0.5144767165184021, "learning_rate": 7.132436585449231e-05, "loss": 0.0113, "step": 117530 }, { "epoch": 74.91395793499044, "grad_norm": 0.34124207496643066, "learning_rate": 7.131938054903763e-05, "loss": 0.0172, "step": 117540 }, { "epoch": 74.92033142128744, "grad_norm": 0.46153032779693604, "learning_rate": 7.131439498453237e-05, "loss": 0.0285, "step": 117550 }, { "epoch": 74.92670490758445, "grad_norm": 0.5381916761398315, "learning_rate": 7.130940916103707e-05, "loss": 0.0126, "step": 117560 }, { "epoch": 74.93307839388146, "grad_norm": 0.6206822991371155, "learning_rate": 7.130442307861235e-05, "loss": 0.0243, "step": 117570 }, { "epoch": 74.93945188017845, "grad_norm": 0.282241553068161, "learning_rate": 7.129943673731876e-05, "loss": 0.0157, "step": 117580 }, { "epoch": 74.94582536647546, "grad_norm": 0.5282867550849915, "learning_rate": 7.12944501372169e-05, "loss": 0.0235, "step": 117590 }, { "epoch": 74.95219885277247, "grad_norm": 0.326990008354187, "learning_rate": 7.128946327836738e-05, "loss": 0.0219, "step": 117600 }, { "epoch": 74.95857233906948, "grad_norm": 0.27237987518310547, "learning_rate": 7.128447616083077e-05, "loss": 0.0229, "step": 117610 }, { "epoch": 74.96494582536647, "grad_norm": 0.30156657099723816, "learning_rate": 7.127948878466768e-05, "loss": 0.0232, "step": 117620 }, { "epoch": 74.97131931166348, "grad_norm": 0.563775360584259, "learning_rate": 7.127450114993871e-05, "loss": 0.0221, "step": 117630 }, { "epoch": 74.97769279796049, "grad_norm": 0.37508416175842285, "learning_rate": 7.126951325670447e-05, "loss": 0.0183, "step": 117640 }, { "epoch": 74.9840662842575, "grad_norm": 0.7179257869720459, "learning_rate": 7.126452510502554e-05, "loss": 0.0189, "step": 117650 }, { "epoch": 74.99043977055449, "grad_norm": 0.3970891237258911, "learning_rate": 7.125953669496259e-05, "loss": 0.0179, "step": 117660 }, { "epoch": 74.9968132568515, "grad_norm": 0.15423694252967834, "learning_rate": 7.125454802657617e-05, "loss": 0.0158, "step": 117670 }, { "epoch": 75.0031867431485, "grad_norm": 0.47962063550949097, "learning_rate": 7.124955909992693e-05, "loss": 0.0153, "step": 117680 }, { "epoch": 75.00956022944551, "grad_norm": 0.31093651056289673, "learning_rate": 7.124456991507548e-05, "loss": 0.0144, "step": 117690 }, { "epoch": 75.0159337157425, "grad_norm": 0.34098923206329346, "learning_rate": 7.123958047208246e-05, "loss": 0.0199, "step": 117700 }, { "epoch": 75.02230720203951, "grad_norm": 0.4469543397426605, "learning_rate": 7.123459077100848e-05, "loss": 0.0141, "step": 117710 }, { "epoch": 75.02868068833652, "grad_norm": 0.4541991651058197, "learning_rate": 7.122960081191417e-05, "loss": 0.0212, "step": 117720 }, { "epoch": 75.03505417463353, "grad_norm": 0.39003786444664, "learning_rate": 7.122461059486016e-05, "loss": 0.0158, "step": 117730 }, { "epoch": 75.04142766093052, "grad_norm": 0.5544167757034302, "learning_rate": 7.12196201199071e-05, "loss": 0.0208, "step": 117740 }, { "epoch": 75.04780114722753, "grad_norm": 0.27241212129592896, "learning_rate": 7.121462938711561e-05, "loss": 0.022, "step": 117750 }, { "epoch": 75.05417463352454, "grad_norm": 0.6401001811027527, "learning_rate": 7.120963839654635e-05, "loss": 0.0164, "step": 117760 }, { "epoch": 75.06054811982155, "grad_norm": 0.23535755276679993, "learning_rate": 7.120464714825998e-05, "loss": 0.014, "step": 117770 }, { "epoch": 75.06692160611854, "grad_norm": 0.30301934480667114, "learning_rate": 7.11996556423171e-05, "loss": 0.0134, "step": 117780 }, { "epoch": 75.07329509241555, "grad_norm": 0.38700070977211, "learning_rate": 7.119466387877839e-05, "loss": 0.0126, "step": 117790 }, { "epoch": 75.07966857871256, "grad_norm": 0.5298253297805786, "learning_rate": 7.11896718577045e-05, "loss": 0.0103, "step": 117800 }, { "epoch": 75.08604206500956, "grad_norm": 0.24552221596240997, "learning_rate": 7.11846795791561e-05, "loss": 0.0162, "step": 117810 }, { "epoch": 75.09241555130656, "grad_norm": 0.3840981721878052, "learning_rate": 7.117968704319384e-05, "loss": 0.0218, "step": 117820 }, { "epoch": 75.09878903760357, "grad_norm": 0.3721402883529663, "learning_rate": 7.117469424987839e-05, "loss": 0.016, "step": 117830 }, { "epoch": 75.10516252390057, "grad_norm": 0.42303621768951416, "learning_rate": 7.116970119927041e-05, "loss": 0.0199, "step": 117840 }, { "epoch": 75.11153601019758, "grad_norm": 0.5831379294395447, "learning_rate": 7.116470789143057e-05, "loss": 0.0251, "step": 117850 }, { "epoch": 75.11790949649458, "grad_norm": 0.40739545226097107, "learning_rate": 7.115971432641955e-05, "loss": 0.0149, "step": 117860 }, { "epoch": 75.12428298279158, "grad_norm": 0.8486936092376709, "learning_rate": 7.115472050429802e-05, "loss": 0.0301, "step": 117870 }, { "epoch": 75.13065646908859, "grad_norm": 0.5703004002571106, "learning_rate": 7.114972642512665e-05, "loss": 0.0212, "step": 117880 }, { "epoch": 75.1370299553856, "grad_norm": 0.350324422121048, "learning_rate": 7.114473208896616e-05, "loss": 0.0118, "step": 117890 }, { "epoch": 75.1434034416826, "grad_norm": 0.4853611886501312, "learning_rate": 7.113973749587719e-05, "loss": 0.0155, "step": 117900 }, { "epoch": 75.1497769279796, "grad_norm": 0.4184267222881317, "learning_rate": 7.113474264592045e-05, "loss": 0.0103, "step": 117910 }, { "epoch": 75.15615041427661, "grad_norm": 0.7300262451171875, "learning_rate": 7.112974753915665e-05, "loss": 0.0172, "step": 117920 }, { "epoch": 75.16252390057362, "grad_norm": 0.34682875871658325, "learning_rate": 7.112475217564647e-05, "loss": 0.0129, "step": 117930 }, { "epoch": 75.16889738687061, "grad_norm": 0.3113180696964264, "learning_rate": 7.111975655545059e-05, "loss": 0.0181, "step": 117940 }, { "epoch": 75.17527087316762, "grad_norm": 0.39826396107673645, "learning_rate": 7.111476067862973e-05, "loss": 0.0159, "step": 117950 }, { "epoch": 75.18164435946463, "grad_norm": 0.48751622438430786, "learning_rate": 7.110976454524459e-05, "loss": 0.019, "step": 117960 }, { "epoch": 75.18801784576164, "grad_norm": 0.3592997193336487, "learning_rate": 7.110476815535588e-05, "loss": 0.0152, "step": 117970 }, { "epoch": 75.19439133205863, "grad_norm": 0.1926124542951584, "learning_rate": 7.109977150902433e-05, "loss": 0.0108, "step": 117980 }, { "epoch": 75.20076481835564, "grad_norm": 0.5644398331642151, "learning_rate": 7.10947746063106e-05, "loss": 0.0158, "step": 117990 }, { "epoch": 75.20713830465264, "grad_norm": 0.45703577995300293, "learning_rate": 7.108977744727547e-05, "loss": 0.0171, "step": 118000 }, { "epoch": 75.21351179094965, "grad_norm": 0.4369147717952728, "learning_rate": 7.108478003197963e-05, "loss": 0.0096, "step": 118010 }, { "epoch": 75.21988527724666, "grad_norm": 0.45491623878479004, "learning_rate": 7.10797823604838e-05, "loss": 0.0104, "step": 118020 }, { "epoch": 75.22625876354365, "grad_norm": 0.453640341758728, "learning_rate": 7.107478443284869e-05, "loss": 0.0185, "step": 118030 }, { "epoch": 75.23263224984066, "grad_norm": 0.44394561648368835, "learning_rate": 7.106978624913509e-05, "loss": 0.0203, "step": 118040 }, { "epoch": 75.23900573613767, "grad_norm": 0.29633545875549316, "learning_rate": 7.106478780940366e-05, "loss": 0.0202, "step": 118050 }, { "epoch": 75.24537922243468, "grad_norm": 0.3647107183933258, "learning_rate": 7.10597891137152e-05, "loss": 0.016, "step": 118060 }, { "epoch": 75.25175270873167, "grad_norm": 0.24277363717556, "learning_rate": 7.105479016213039e-05, "loss": 0.0249, "step": 118070 }, { "epoch": 75.25812619502868, "grad_norm": 0.3760528564453125, "learning_rate": 7.104979095471003e-05, "loss": 0.0171, "step": 118080 }, { "epoch": 75.26449968132569, "grad_norm": 0.2430296689271927, "learning_rate": 7.10447914915148e-05, "loss": 0.0166, "step": 118090 }, { "epoch": 75.2708731676227, "grad_norm": 0.5031399726867676, "learning_rate": 7.103979177260551e-05, "loss": 0.0197, "step": 118100 }, { "epoch": 75.27724665391969, "grad_norm": 0.5476059317588806, "learning_rate": 7.103479179804287e-05, "loss": 0.0258, "step": 118110 }, { "epoch": 75.2836201402167, "grad_norm": 0.20177757740020752, "learning_rate": 7.102979156788767e-05, "loss": 0.0189, "step": 118120 }, { "epoch": 75.2899936265137, "grad_norm": 0.3523573875427246, "learning_rate": 7.102479108220063e-05, "loss": 0.0222, "step": 118130 }, { "epoch": 75.29636711281071, "grad_norm": 0.5069718360900879, "learning_rate": 7.101979034104254e-05, "loss": 0.0115, "step": 118140 }, { "epoch": 75.30274059910771, "grad_norm": 0.4522167444229126, "learning_rate": 7.101478934447414e-05, "loss": 0.0118, "step": 118150 }, { "epoch": 75.30911408540472, "grad_norm": 0.3507945239543915, "learning_rate": 7.100978809255622e-05, "loss": 0.0143, "step": 118160 }, { "epoch": 75.31548757170172, "grad_norm": 0.6283950805664062, "learning_rate": 7.100478658534952e-05, "loss": 0.0188, "step": 118170 }, { "epoch": 75.32186105799873, "grad_norm": 0.5681741833686829, "learning_rate": 7.099978482291484e-05, "loss": 0.0171, "step": 118180 }, { "epoch": 75.32823454429573, "grad_norm": 0.3189215064048767, "learning_rate": 7.099478280531295e-05, "loss": 0.0171, "step": 118190 }, { "epoch": 75.33460803059273, "grad_norm": 1.1620513200759888, "learning_rate": 7.098978053260464e-05, "loss": 0.0212, "step": 118200 }, { "epoch": 75.34098151688974, "grad_norm": 0.421426922082901, "learning_rate": 7.098477800485066e-05, "loss": 0.0173, "step": 118210 }, { "epoch": 75.34735500318675, "grad_norm": 0.2165140062570572, "learning_rate": 7.09797752221118e-05, "loss": 0.0119, "step": 118220 }, { "epoch": 75.35372848948374, "grad_norm": 0.7376498579978943, "learning_rate": 7.09747721844489e-05, "loss": 0.0173, "step": 118230 }, { "epoch": 75.36010197578075, "grad_norm": 0.22096237540245056, "learning_rate": 7.096976889192268e-05, "loss": 0.0256, "step": 118240 }, { "epoch": 75.36647546207776, "grad_norm": 0.3184158504009247, "learning_rate": 7.0964765344594e-05, "loss": 0.0179, "step": 118250 }, { "epoch": 75.37284894837477, "grad_norm": 0.28278058767318726, "learning_rate": 7.095976154252362e-05, "loss": 0.0167, "step": 118260 }, { "epoch": 75.37922243467176, "grad_norm": 0.3692629635334015, "learning_rate": 7.095475748577233e-05, "loss": 0.0213, "step": 118270 }, { "epoch": 75.38559592096877, "grad_norm": 0.4559711217880249, "learning_rate": 7.094975317440097e-05, "loss": 0.0175, "step": 118280 }, { "epoch": 75.39196940726578, "grad_norm": 0.3263126611709595, "learning_rate": 7.094474860847032e-05, "loss": 0.0154, "step": 118290 }, { "epoch": 75.39834289356278, "grad_norm": 0.2649509906768799, "learning_rate": 7.09397437880412e-05, "loss": 0.0132, "step": 118300 }, { "epoch": 75.40471637985978, "grad_norm": 0.36323386430740356, "learning_rate": 7.093473871317442e-05, "loss": 0.0103, "step": 118310 }, { "epoch": 75.41108986615679, "grad_norm": 0.20992234349250793, "learning_rate": 7.092973338393081e-05, "loss": 0.0182, "step": 118320 }, { "epoch": 75.4174633524538, "grad_norm": 0.47000283002853394, "learning_rate": 7.092472780037117e-05, "loss": 0.0189, "step": 118330 }, { "epoch": 75.4238368387508, "grad_norm": 0.3510236442089081, "learning_rate": 7.091972196255632e-05, "loss": 0.0159, "step": 118340 }, { "epoch": 75.4302103250478, "grad_norm": 0.325776606798172, "learning_rate": 7.091471587054712e-05, "loss": 0.0195, "step": 118350 }, { "epoch": 75.4365838113448, "grad_norm": 0.7340406775474548, "learning_rate": 7.090970952440435e-05, "loss": 0.0264, "step": 118360 }, { "epoch": 75.44295729764181, "grad_norm": 0.476479709148407, "learning_rate": 7.090470292418889e-05, "loss": 0.0155, "step": 118370 }, { "epoch": 75.44933078393882, "grad_norm": 0.3958926796913147, "learning_rate": 7.089969606996154e-05, "loss": 0.0141, "step": 118380 }, { "epoch": 75.45570427023581, "grad_norm": 0.3670749366283417, "learning_rate": 7.089468896178316e-05, "loss": 0.014, "step": 118390 }, { "epoch": 75.46207775653282, "grad_norm": 0.4032856523990631, "learning_rate": 7.088968159971457e-05, "loss": 0.0251, "step": 118400 }, { "epoch": 75.46845124282983, "grad_norm": 0.2790076732635498, "learning_rate": 7.088467398381663e-05, "loss": 0.0163, "step": 118410 }, { "epoch": 75.47482472912684, "grad_norm": 0.4327918291091919, "learning_rate": 7.087966611415019e-05, "loss": 0.0213, "step": 118420 }, { "epoch": 75.48119821542383, "grad_norm": 0.5310767889022827, "learning_rate": 7.087465799077608e-05, "loss": 0.0134, "step": 118430 }, { "epoch": 75.48757170172084, "grad_norm": 0.45143043994903564, "learning_rate": 7.086964961375518e-05, "loss": 0.0198, "step": 118440 }, { "epoch": 75.49394518801785, "grad_norm": 0.3090333640575409, "learning_rate": 7.086464098314832e-05, "loss": 0.0145, "step": 118450 }, { "epoch": 75.50031867431485, "grad_norm": 0.2475583404302597, "learning_rate": 7.085963209901638e-05, "loss": 0.0201, "step": 118460 }, { "epoch": 75.50669216061185, "grad_norm": 0.41350194811820984, "learning_rate": 7.08546229614202e-05, "loss": 0.0127, "step": 118470 }, { "epoch": 75.51306564690886, "grad_norm": 0.2784697115421295, "learning_rate": 7.084961357042068e-05, "loss": 0.0139, "step": 118480 }, { "epoch": 75.51943913320586, "grad_norm": 0.3643914759159088, "learning_rate": 7.084460392607868e-05, "loss": 0.0159, "step": 118490 }, { "epoch": 75.52581261950287, "grad_norm": 0.6785404086112976, "learning_rate": 7.083959402845504e-05, "loss": 0.0156, "step": 118500 }, { "epoch": 75.53218610579987, "grad_norm": 0.4057476818561554, "learning_rate": 7.083458387761065e-05, "loss": 0.0139, "step": 118510 }, { "epoch": 75.53855959209687, "grad_norm": 0.5370882153511047, "learning_rate": 7.082957347360642e-05, "loss": 0.023, "step": 118520 }, { "epoch": 75.54493307839388, "grad_norm": 0.5640504956245422, "learning_rate": 7.08245628165032e-05, "loss": 0.0163, "step": 118530 }, { "epoch": 75.55130656469089, "grad_norm": 0.3632296919822693, "learning_rate": 7.081955190636185e-05, "loss": 0.0108, "step": 118540 }, { "epoch": 75.55768005098788, "grad_norm": 0.5030195713043213, "learning_rate": 7.081454074324332e-05, "loss": 0.0118, "step": 118550 }, { "epoch": 75.56405353728489, "grad_norm": 0.6293888092041016, "learning_rate": 7.080952932720845e-05, "loss": 0.0154, "step": 118560 }, { "epoch": 75.5704270235819, "grad_norm": 0.4529464840888977, "learning_rate": 7.080451765831817e-05, "loss": 0.0201, "step": 118570 }, { "epoch": 75.57680050987891, "grad_norm": 0.31795647740364075, "learning_rate": 7.079950573663336e-05, "loss": 0.0163, "step": 118580 }, { "epoch": 75.5831739961759, "grad_norm": 0.3560052216053009, "learning_rate": 7.079449356221489e-05, "loss": 0.0167, "step": 118590 }, { "epoch": 75.58954748247291, "grad_norm": 0.20952339470386505, "learning_rate": 7.078948113512372e-05, "loss": 0.0188, "step": 118600 }, { "epoch": 75.59592096876992, "grad_norm": 0.4707447588443756, "learning_rate": 7.07844684554207e-05, "loss": 0.0256, "step": 118610 }, { "epoch": 75.60229445506693, "grad_norm": 0.7181344628334045, "learning_rate": 7.077945552316678e-05, "loss": 0.024, "step": 118620 }, { "epoch": 75.60866794136393, "grad_norm": 0.5379713177680969, "learning_rate": 7.077444233842285e-05, "loss": 0.017, "step": 118630 }, { "epoch": 75.61504142766093, "grad_norm": 0.4485417306423187, "learning_rate": 7.076942890124983e-05, "loss": 0.0243, "step": 118640 }, { "epoch": 75.62141491395793, "grad_norm": 0.3172924220561981, "learning_rate": 7.076441521170865e-05, "loss": 0.0176, "step": 118650 }, { "epoch": 75.62778840025494, "grad_norm": 0.49425095319747925, "learning_rate": 7.07594012698602e-05, "loss": 0.0162, "step": 118660 }, { "epoch": 75.63416188655195, "grad_norm": 0.26472529768943787, "learning_rate": 7.075438707576544e-05, "loss": 0.0183, "step": 118670 }, { "epoch": 75.64053537284894, "grad_norm": 0.3454269468784332, "learning_rate": 7.074937262948527e-05, "loss": 0.0124, "step": 118680 }, { "epoch": 75.64690885914595, "grad_norm": 0.6954070925712585, "learning_rate": 7.074435793108065e-05, "loss": 0.021, "step": 118690 }, { "epoch": 75.65328234544296, "grad_norm": 0.2862941026687622, "learning_rate": 7.073934298061247e-05, "loss": 0.0186, "step": 118700 }, { "epoch": 75.65965583173997, "grad_norm": 0.4144957363605499, "learning_rate": 7.073432777814172e-05, "loss": 0.0115, "step": 118710 }, { "epoch": 75.66602931803696, "grad_norm": 0.1956450641155243, "learning_rate": 7.072931232372929e-05, "loss": 0.0213, "step": 118720 }, { "epoch": 75.67240280433397, "grad_norm": 0.42568740248680115, "learning_rate": 7.072429661743616e-05, "loss": 0.0172, "step": 118730 }, { "epoch": 75.67877629063098, "grad_norm": 0.2669760584831238, "learning_rate": 7.071928065932324e-05, "loss": 0.0211, "step": 118740 }, { "epoch": 75.68514977692799, "grad_norm": 0.33795788884162903, "learning_rate": 7.071426444945151e-05, "loss": 0.0183, "step": 118750 }, { "epoch": 75.69152326322498, "grad_norm": 0.46806928515434265, "learning_rate": 7.070924798788191e-05, "loss": 0.0206, "step": 118760 }, { "epoch": 75.69789674952199, "grad_norm": 0.48012083768844604, "learning_rate": 7.070423127467539e-05, "loss": 0.0123, "step": 118770 }, { "epoch": 75.704270235819, "grad_norm": 0.2523433268070221, "learning_rate": 7.069921430989292e-05, "loss": 0.0118, "step": 118780 }, { "epoch": 75.710643722116, "grad_norm": 0.3420585095882416, "learning_rate": 7.069419709359544e-05, "loss": 0.0164, "step": 118790 }, { "epoch": 75.717017208413, "grad_norm": 0.29377248883247375, "learning_rate": 7.068917962584394e-05, "loss": 0.0119, "step": 118800 }, { "epoch": 75.72339069471, "grad_norm": 0.5734994411468506, "learning_rate": 7.068416190669937e-05, "loss": 0.0152, "step": 118810 }, { "epoch": 75.72976418100701, "grad_norm": 0.3722682595252991, "learning_rate": 7.067914393622269e-05, "loss": 0.0165, "step": 118820 }, { "epoch": 75.73613766730402, "grad_norm": 0.42218488454818726, "learning_rate": 7.06741257144749e-05, "loss": 0.0111, "step": 118830 }, { "epoch": 75.74251115360101, "grad_norm": 0.5269044637680054, "learning_rate": 7.066910724151696e-05, "loss": 0.0204, "step": 118840 }, { "epoch": 75.74888463989802, "grad_norm": 0.3886585235595703, "learning_rate": 7.066408851740986e-05, "loss": 0.0235, "step": 118850 }, { "epoch": 75.75525812619503, "grad_norm": 0.25223931670188904, "learning_rate": 7.065906954221455e-05, "loss": 0.0163, "step": 118860 }, { "epoch": 75.76163161249204, "grad_norm": 1.0044302940368652, "learning_rate": 7.065405031599207e-05, "loss": 0.0149, "step": 118870 }, { "epoch": 75.76800509878903, "grad_norm": 0.3749907314777374, "learning_rate": 7.064903083880335e-05, "loss": 0.0181, "step": 118880 }, { "epoch": 75.77437858508604, "grad_norm": 0.26379361748695374, "learning_rate": 7.064401111070944e-05, "loss": 0.0131, "step": 118890 }, { "epoch": 75.78075207138305, "grad_norm": 0.7485328912734985, "learning_rate": 7.063899113177128e-05, "loss": 0.025, "step": 118900 }, { "epoch": 75.78712555768006, "grad_norm": 0.3414473235607147, "learning_rate": 7.06339709020499e-05, "loss": 0.0158, "step": 118910 }, { "epoch": 75.79349904397705, "grad_norm": 0.32937902212142944, "learning_rate": 7.06289504216063e-05, "loss": 0.0221, "step": 118920 }, { "epoch": 75.79987253027406, "grad_norm": 0.34725791215896606, "learning_rate": 7.062392969050145e-05, "loss": 0.0146, "step": 118930 }, { "epoch": 75.80624601657107, "grad_norm": 0.7936139702796936, "learning_rate": 7.061890870879642e-05, "loss": 0.0205, "step": 118940 }, { "epoch": 75.81261950286807, "grad_norm": 0.26154300570487976, "learning_rate": 7.061388747655214e-05, "loss": 0.0149, "step": 118950 }, { "epoch": 75.81899298916507, "grad_norm": 0.4940328598022461, "learning_rate": 7.060886599382969e-05, "loss": 0.0205, "step": 118960 }, { "epoch": 75.82536647546208, "grad_norm": 0.27195629477500916, "learning_rate": 7.060384426069006e-05, "loss": 0.0206, "step": 118970 }, { "epoch": 75.83173996175908, "grad_norm": 0.38097116351127625, "learning_rate": 7.059882227719427e-05, "loss": 0.0167, "step": 118980 }, { "epoch": 75.83811344805609, "grad_norm": 0.23440241813659668, "learning_rate": 7.059380004340333e-05, "loss": 0.0158, "step": 118990 }, { "epoch": 75.84448693435309, "grad_norm": 0.606343150138855, "learning_rate": 7.058877755937828e-05, "loss": 0.0243, "step": 119000 }, { "epoch": 75.8508604206501, "grad_norm": 0.3098571002483368, "learning_rate": 7.058375482518014e-05, "loss": 0.0232, "step": 119010 }, { "epoch": 75.8572339069471, "grad_norm": 0.34063032269477844, "learning_rate": 7.057873184086994e-05, "loss": 0.012, "step": 119020 }, { "epoch": 75.86360739324411, "grad_norm": 0.4766545295715332, "learning_rate": 7.057370860650873e-05, "loss": 0.0121, "step": 119030 }, { "epoch": 75.8699808795411, "grad_norm": 0.21780772507190704, "learning_rate": 7.056868512215754e-05, "loss": 0.0241, "step": 119040 }, { "epoch": 75.87635436583811, "grad_norm": 0.426311731338501, "learning_rate": 7.05636613878774e-05, "loss": 0.0182, "step": 119050 }, { "epoch": 75.88272785213512, "grad_norm": 0.3660537302494049, "learning_rate": 7.055863740372936e-05, "loss": 0.013, "step": 119060 }, { "epoch": 75.88910133843213, "grad_norm": 0.2848835587501526, "learning_rate": 7.055361316977446e-05, "loss": 0.012, "step": 119070 }, { "epoch": 75.89547482472912, "grad_norm": 0.27099817991256714, "learning_rate": 7.054858868607377e-05, "loss": 0.0162, "step": 119080 }, { "epoch": 75.90184831102613, "grad_norm": 0.31384721398353577, "learning_rate": 7.05435639526883e-05, "loss": 0.0175, "step": 119090 }, { "epoch": 75.90822179732314, "grad_norm": 0.2895788252353668, "learning_rate": 7.053853896967914e-05, "loss": 0.0185, "step": 119100 }, { "epoch": 75.91459528362014, "grad_norm": 0.38014233112335205, "learning_rate": 7.053351373710735e-05, "loss": 0.0115, "step": 119110 }, { "epoch": 75.92096876991714, "grad_norm": 0.36204490065574646, "learning_rate": 7.052848825503398e-05, "loss": 0.0129, "step": 119120 }, { "epoch": 75.92734225621415, "grad_norm": 0.2689323127269745, "learning_rate": 7.052346252352008e-05, "loss": 0.0125, "step": 119130 }, { "epoch": 75.93371574251115, "grad_norm": 0.39021265506744385, "learning_rate": 7.051843654262676e-05, "loss": 0.0173, "step": 119140 }, { "epoch": 75.94008922880816, "grad_norm": 0.3947398066520691, "learning_rate": 7.051341031241505e-05, "loss": 0.0185, "step": 119150 }, { "epoch": 75.94646271510516, "grad_norm": 0.3476082384586334, "learning_rate": 7.050838383294603e-05, "loss": 0.0281, "step": 119160 }, { "epoch": 75.95283620140216, "grad_norm": 0.2957853078842163, "learning_rate": 7.050335710428078e-05, "loss": 0.0234, "step": 119170 }, { "epoch": 75.95920968769917, "grad_norm": 0.28981685638427734, "learning_rate": 7.04983301264804e-05, "loss": 0.0138, "step": 119180 }, { "epoch": 75.96558317399618, "grad_norm": 0.37735751271247864, "learning_rate": 7.049330289960594e-05, "loss": 0.0166, "step": 119190 }, { "epoch": 75.97195666029317, "grad_norm": 0.38681334257125854, "learning_rate": 7.048827542371851e-05, "loss": 0.0405, "step": 119200 }, { "epoch": 75.97833014659018, "grad_norm": 0.32490020990371704, "learning_rate": 7.048324769887919e-05, "loss": 0.0129, "step": 119210 }, { "epoch": 75.98470363288719, "grad_norm": 0.5360503196716309, "learning_rate": 7.047821972514907e-05, "loss": 0.015, "step": 119220 }, { "epoch": 75.9910771191842, "grad_norm": 0.7611535787582397, "learning_rate": 7.047319150258925e-05, "loss": 0.0205, "step": 119230 }, { "epoch": 75.99745060548119, "grad_norm": 0.2923852205276489, "learning_rate": 7.04681630312608e-05, "loss": 0.0196, "step": 119240 }, { "epoch": 76.0038240917782, "grad_norm": 0.5442422032356262, "learning_rate": 7.046313431122487e-05, "loss": 0.0153, "step": 119250 }, { "epoch": 76.0101975780752, "grad_norm": 0.47718143463134766, "learning_rate": 7.045810534254252e-05, "loss": 0.0266, "step": 119260 }, { "epoch": 76.01657106437222, "grad_norm": 0.5469136238098145, "learning_rate": 7.045307612527489e-05, "loss": 0.0178, "step": 119270 }, { "epoch": 76.02294455066922, "grad_norm": 0.6368625164031982, "learning_rate": 7.044804665948305e-05, "loss": 0.0258, "step": 119280 }, { "epoch": 76.02931803696622, "grad_norm": 0.48599711060523987, "learning_rate": 7.044301694522817e-05, "loss": 0.0169, "step": 119290 }, { "epoch": 76.03569152326322, "grad_norm": 0.33294788002967834, "learning_rate": 7.043798698257132e-05, "loss": 0.0151, "step": 119300 }, { "epoch": 76.04206500956023, "grad_norm": 0.3763900399208069, "learning_rate": 7.043295677157362e-05, "loss": 0.0147, "step": 119310 }, { "epoch": 76.04843849585724, "grad_norm": 0.3118327856063843, "learning_rate": 7.04279263122962e-05, "loss": 0.0105, "step": 119320 }, { "epoch": 76.05481198215423, "grad_norm": 0.25498318672180176, "learning_rate": 7.042289560480023e-05, "loss": 0.0137, "step": 119330 }, { "epoch": 76.06118546845124, "grad_norm": 0.4376508593559265, "learning_rate": 7.041786464914674e-05, "loss": 0.0182, "step": 119340 }, { "epoch": 76.06755895474825, "grad_norm": 0.23315739631652832, "learning_rate": 7.041283344539696e-05, "loss": 0.0135, "step": 119350 }, { "epoch": 76.07393244104526, "grad_norm": 0.4076998829841614, "learning_rate": 7.040780199361195e-05, "loss": 0.0225, "step": 119360 }, { "epoch": 76.08030592734225, "grad_norm": 0.24937200546264648, "learning_rate": 7.04027702938529e-05, "loss": 0.0123, "step": 119370 }, { "epoch": 76.08667941363926, "grad_norm": 0.5979923009872437, "learning_rate": 7.039773834618092e-05, "loss": 0.0256, "step": 119380 }, { "epoch": 76.09305289993627, "grad_norm": 0.34047478437423706, "learning_rate": 7.039270615065715e-05, "loss": 0.0141, "step": 119390 }, { "epoch": 76.09942638623328, "grad_norm": 0.41216030716896057, "learning_rate": 7.038767370734276e-05, "loss": 0.0159, "step": 119400 }, { "epoch": 76.10579987253027, "grad_norm": 0.48933830857276917, "learning_rate": 7.038264101629888e-05, "loss": 0.014, "step": 119410 }, { "epoch": 76.11217335882728, "grad_norm": 0.44872793555259705, "learning_rate": 7.037760807758666e-05, "loss": 0.0229, "step": 119420 }, { "epoch": 76.11854684512429, "grad_norm": 0.7171058654785156, "learning_rate": 7.037257489126728e-05, "loss": 0.0225, "step": 119430 }, { "epoch": 76.1249203314213, "grad_norm": 0.46266770362854004, "learning_rate": 7.036754145740186e-05, "loss": 0.0198, "step": 119440 }, { "epoch": 76.13129381771829, "grad_norm": 0.42266812920570374, "learning_rate": 7.03625077760516e-05, "loss": 0.0183, "step": 119450 }, { "epoch": 76.1376673040153, "grad_norm": 0.31465527415275574, "learning_rate": 7.035747384727764e-05, "loss": 0.017, "step": 119460 }, { "epoch": 76.1440407903123, "grad_norm": 0.3554028570652008, "learning_rate": 7.035243967114114e-05, "loss": 0.0307, "step": 119470 }, { "epoch": 76.15041427660931, "grad_norm": 0.3877589702606201, "learning_rate": 7.03474052477033e-05, "loss": 0.0348, "step": 119480 }, { "epoch": 76.1567877629063, "grad_norm": 0.3062985837459564, "learning_rate": 7.034237057702525e-05, "loss": 0.025, "step": 119490 }, { "epoch": 76.16316124920331, "grad_norm": 0.2593017816543579, "learning_rate": 7.033733565916821e-05, "loss": 0.0178, "step": 119500 }, { "epoch": 76.16953473550032, "grad_norm": 0.5586726665496826, "learning_rate": 7.033230049419333e-05, "loss": 0.0281, "step": 119510 }, { "epoch": 76.17590822179733, "grad_norm": 0.29772013425827026, "learning_rate": 7.032726508216182e-05, "loss": 0.0137, "step": 119520 }, { "epoch": 76.18228170809432, "grad_norm": 0.25305941700935364, "learning_rate": 7.032222942313482e-05, "loss": 0.016, "step": 119530 }, { "epoch": 76.18865519439133, "grad_norm": 0.3606245517730713, "learning_rate": 7.031719351717357e-05, "loss": 0.0132, "step": 119540 }, { "epoch": 76.19502868068834, "grad_norm": 0.26501190662384033, "learning_rate": 7.031215736433922e-05, "loss": 0.0102, "step": 119550 }, { "epoch": 76.20140216698535, "grad_norm": 0.2294761836528778, "learning_rate": 7.030712096469299e-05, "loss": 0.0139, "step": 119560 }, { "epoch": 76.20777565328234, "grad_norm": 0.3854679465293884, "learning_rate": 7.030208431829607e-05, "loss": 0.0108, "step": 119570 }, { "epoch": 76.21414913957935, "grad_norm": 0.3587517738342285, "learning_rate": 7.029704742520966e-05, "loss": 0.0148, "step": 119580 }, { "epoch": 76.22052262587636, "grad_norm": 0.29181337356567383, "learning_rate": 7.029201028549493e-05, "loss": 0.0117, "step": 119590 }, { "epoch": 76.22689611217336, "grad_norm": 0.26849082112312317, "learning_rate": 7.028697289921316e-05, "loss": 0.0198, "step": 119600 }, { "epoch": 76.23326959847036, "grad_norm": 0.4411601424217224, "learning_rate": 7.028193526642548e-05, "loss": 0.0167, "step": 119610 }, { "epoch": 76.23964308476737, "grad_norm": 0.7212768793106079, "learning_rate": 7.027689738719315e-05, "loss": 0.0248, "step": 119620 }, { "epoch": 76.24601657106437, "grad_norm": 0.3983801305294037, "learning_rate": 7.027185926157738e-05, "loss": 0.0128, "step": 119630 }, { "epoch": 76.25239005736138, "grad_norm": 0.2677469551563263, "learning_rate": 7.026682088963938e-05, "loss": 0.0118, "step": 119640 }, { "epoch": 76.25876354365838, "grad_norm": 0.4805612564086914, "learning_rate": 7.026178227144036e-05, "loss": 0.0099, "step": 119650 }, { "epoch": 76.26513702995538, "grad_norm": 0.6268524527549744, "learning_rate": 7.025674340704156e-05, "loss": 0.0196, "step": 119660 }, { "epoch": 76.27151051625239, "grad_norm": 0.22509434819221497, "learning_rate": 7.025170429650419e-05, "loss": 0.0243, "step": 119670 }, { "epoch": 76.2778840025494, "grad_norm": 0.2901289165019989, "learning_rate": 7.02466649398895e-05, "loss": 0.0131, "step": 119680 }, { "epoch": 76.2842574888464, "grad_norm": 0.31110355257987976, "learning_rate": 7.024162533725873e-05, "loss": 0.0077, "step": 119690 }, { "epoch": 76.2906309751434, "grad_norm": 0.3637370467185974, "learning_rate": 7.023658548867308e-05, "loss": 0.0169, "step": 119700 }, { "epoch": 76.29700446144041, "grad_norm": 0.3088567852973938, "learning_rate": 7.023154539419384e-05, "loss": 0.0151, "step": 119710 }, { "epoch": 76.30337794773742, "grad_norm": 0.22625067830085754, "learning_rate": 7.02265050538822e-05, "loss": 0.0118, "step": 119720 }, { "epoch": 76.30975143403441, "grad_norm": 0.3720156252384186, "learning_rate": 7.022146446779942e-05, "loss": 0.0162, "step": 119730 }, { "epoch": 76.31612492033142, "grad_norm": 0.4158764183521271, "learning_rate": 7.021642363600676e-05, "loss": 0.0221, "step": 119740 }, { "epoch": 76.32249840662843, "grad_norm": 0.3193633556365967, "learning_rate": 7.021138255856547e-05, "loss": 0.0147, "step": 119750 }, { "epoch": 76.32887189292543, "grad_norm": 0.7858723998069763, "learning_rate": 7.02063412355368e-05, "loss": 0.0164, "step": 119760 }, { "epoch": 76.33524537922243, "grad_norm": 0.5369969606399536, "learning_rate": 7.020129966698198e-05, "loss": 0.0253, "step": 119770 }, { "epoch": 76.34161886551944, "grad_norm": 0.5215790867805481, "learning_rate": 7.019625785296233e-05, "loss": 0.0154, "step": 119780 }, { "epoch": 76.34799235181644, "grad_norm": 0.30377906560897827, "learning_rate": 7.019121579353906e-05, "loss": 0.0127, "step": 119790 }, { "epoch": 76.35436583811345, "grad_norm": 0.363247275352478, "learning_rate": 7.018617348877346e-05, "loss": 0.0168, "step": 119800 }, { "epoch": 76.36073932441045, "grad_norm": 0.4339054822921753, "learning_rate": 7.01811309387268e-05, "loss": 0.0186, "step": 119810 }, { "epoch": 76.36711281070745, "grad_norm": 0.33712658286094666, "learning_rate": 7.017608814346035e-05, "loss": 0.0147, "step": 119820 }, { "epoch": 76.37348629700446, "grad_norm": 0.4331188499927521, "learning_rate": 7.017104510303536e-05, "loss": 0.0155, "step": 119830 }, { "epoch": 76.37985978330147, "grad_norm": 0.46566882729530334, "learning_rate": 7.016600181751313e-05, "loss": 0.0218, "step": 119840 }, { "epoch": 76.38623326959846, "grad_norm": 0.5132261514663696, "learning_rate": 7.016095828695496e-05, "loss": 0.0352, "step": 119850 }, { "epoch": 76.39260675589547, "grad_norm": 0.32452791929244995, "learning_rate": 7.01559145114221e-05, "loss": 0.0147, "step": 119860 }, { "epoch": 76.39898024219248, "grad_norm": 0.387556791305542, "learning_rate": 7.015087049097584e-05, "loss": 0.0115, "step": 119870 }, { "epoch": 76.40535372848949, "grad_norm": 0.6907685995101929, "learning_rate": 7.014582622567747e-05, "loss": 0.0205, "step": 119880 }, { "epoch": 76.4117272147865, "grad_norm": 0.5552417039871216, "learning_rate": 7.014078171558831e-05, "loss": 0.0104, "step": 119890 }, { "epoch": 76.41810070108349, "grad_norm": 0.23790223896503448, "learning_rate": 7.013573696076964e-05, "loss": 0.0222, "step": 119900 }, { "epoch": 76.4244741873805, "grad_norm": 0.41896069049835205, "learning_rate": 7.013069196128274e-05, "loss": 0.0186, "step": 119910 }, { "epoch": 76.4308476736775, "grad_norm": 0.5840412378311157, "learning_rate": 7.012564671718893e-05, "loss": 0.0172, "step": 119920 }, { "epoch": 76.43722115997451, "grad_norm": 0.43002501130104065, "learning_rate": 7.01206012285495e-05, "loss": 0.0159, "step": 119930 }, { "epoch": 76.4435946462715, "grad_norm": 0.2577081024646759, "learning_rate": 7.011555549542578e-05, "loss": 0.0177, "step": 119940 }, { "epoch": 76.44996813256851, "grad_norm": 0.4418856203556061, "learning_rate": 7.011050951787908e-05, "loss": 0.02, "step": 119950 }, { "epoch": 76.45634161886552, "grad_norm": 0.5544539093971252, "learning_rate": 7.010546329597069e-05, "loss": 0.0164, "step": 119960 }, { "epoch": 76.46271510516253, "grad_norm": 0.4831506907939911, "learning_rate": 7.010041682976193e-05, "loss": 0.0182, "step": 119970 }, { "epoch": 76.46908859145952, "grad_norm": 0.25018179416656494, "learning_rate": 7.009537011931413e-05, "loss": 0.0151, "step": 119980 }, { "epoch": 76.47546207775653, "grad_norm": 0.5172871947288513, "learning_rate": 7.009032316468861e-05, "loss": 0.0204, "step": 119990 }, { "epoch": 76.48183556405354, "grad_norm": 0.5293229818344116, "learning_rate": 7.00852759659467e-05, "loss": 0.0251, "step": 120000 }, { "epoch": 76.48820905035055, "grad_norm": 0.4393138289451599, "learning_rate": 7.008022852314974e-05, "loss": 0.0107, "step": 120010 }, { "epoch": 76.49458253664754, "grad_norm": 0.34186798334121704, "learning_rate": 7.007518083635903e-05, "loss": 0.0201, "step": 120020 }, { "epoch": 76.50095602294455, "grad_norm": 0.3845565617084503, "learning_rate": 7.00701329056359e-05, "loss": 0.0134, "step": 120030 }, { "epoch": 76.50732950924156, "grad_norm": 0.2562335133552551, "learning_rate": 7.006508473104174e-05, "loss": 0.0264, "step": 120040 }, { "epoch": 76.51370299553857, "grad_norm": 0.4415591061115265, "learning_rate": 7.006003631263784e-05, "loss": 0.038, "step": 120050 }, { "epoch": 76.52007648183556, "grad_norm": 0.3340368866920471, "learning_rate": 7.005498765048554e-05, "loss": 0.0135, "step": 120060 }, { "epoch": 76.52644996813257, "grad_norm": 0.5183016061782837, "learning_rate": 7.004993874464623e-05, "loss": 0.0199, "step": 120070 }, { "epoch": 76.53282345442958, "grad_norm": 0.6553449034690857, "learning_rate": 7.004488959518123e-05, "loss": 0.0144, "step": 120080 }, { "epoch": 76.53919694072658, "grad_norm": 0.2988850176334381, "learning_rate": 7.003984020215188e-05, "loss": 0.0146, "step": 120090 }, { "epoch": 76.54557042702358, "grad_norm": 0.2388104796409607, "learning_rate": 7.003479056561957e-05, "loss": 0.0096, "step": 120100 }, { "epoch": 76.55194391332059, "grad_norm": 0.19145923852920532, "learning_rate": 7.002974068564563e-05, "loss": 0.0208, "step": 120110 }, { "epoch": 76.5583173996176, "grad_norm": 0.6082251667976379, "learning_rate": 7.002469056229144e-05, "loss": 0.0157, "step": 120120 }, { "epoch": 76.5646908859146, "grad_norm": 0.6472329497337341, "learning_rate": 7.001964019561836e-05, "loss": 0.0251, "step": 120130 }, { "epoch": 76.5710643722116, "grad_norm": 0.3972470760345459, "learning_rate": 7.001458958568772e-05, "loss": 0.0127, "step": 120140 }, { "epoch": 76.5774378585086, "grad_norm": 0.6075165867805481, "learning_rate": 7.000953873256094e-05, "loss": 0.0222, "step": 120150 }, { "epoch": 76.58381134480561, "grad_norm": 0.18439415097236633, "learning_rate": 7.000448763629936e-05, "loss": 0.0107, "step": 120160 }, { "epoch": 76.59018483110262, "grad_norm": 0.3768170177936554, "learning_rate": 6.999943629696437e-05, "loss": 0.0156, "step": 120170 }, { "epoch": 76.59655831739961, "grad_norm": 0.4187471568584442, "learning_rate": 6.999438471461735e-05, "loss": 0.0261, "step": 120180 }, { "epoch": 76.60293180369662, "grad_norm": 0.4538813829421997, "learning_rate": 6.998933288931968e-05, "loss": 0.0151, "step": 120190 }, { "epoch": 76.60930528999363, "grad_norm": 0.2580612003803253, "learning_rate": 6.998428082113273e-05, "loss": 0.0107, "step": 120200 }, { "epoch": 76.61567877629064, "grad_norm": 0.7332178354263306, "learning_rate": 6.997922851011792e-05, "loss": 0.018, "step": 120210 }, { "epoch": 76.62205226258763, "grad_norm": 0.25342512130737305, "learning_rate": 6.997417595633661e-05, "loss": 0.0139, "step": 120220 }, { "epoch": 76.62842574888464, "grad_norm": 0.19051136076450348, "learning_rate": 6.99691231598502e-05, "loss": 0.0125, "step": 120230 }, { "epoch": 76.63479923518165, "grad_norm": 0.374942809343338, "learning_rate": 6.99640701207201e-05, "loss": 0.0131, "step": 120240 }, { "epoch": 76.64117272147865, "grad_norm": 0.32361602783203125, "learning_rate": 6.99590168390077e-05, "loss": 0.0138, "step": 120250 }, { "epoch": 76.64754620777565, "grad_norm": 0.32478588819503784, "learning_rate": 6.995396331477437e-05, "loss": 0.0157, "step": 120260 }, { "epoch": 76.65391969407266, "grad_norm": 0.7987448573112488, "learning_rate": 6.994890954808157e-05, "loss": 0.0168, "step": 120270 }, { "epoch": 76.66029318036966, "grad_norm": 0.38661864399909973, "learning_rate": 6.994385553899069e-05, "loss": 0.0118, "step": 120280 }, { "epoch": 76.66666666666667, "grad_norm": 0.4990396797657013, "learning_rate": 6.993880128756314e-05, "loss": 0.0117, "step": 120290 }, { "epoch": 76.67304015296367, "grad_norm": 0.2263706475496292, "learning_rate": 6.993374679386031e-05, "loss": 0.0157, "step": 120300 }, { "epoch": 76.67941363926067, "grad_norm": 0.25576871633529663, "learning_rate": 6.992869205794365e-05, "loss": 0.0192, "step": 120310 }, { "epoch": 76.68578712555768, "grad_norm": 0.4239945411682129, "learning_rate": 6.992363707987455e-05, "loss": 0.0106, "step": 120320 }, { "epoch": 76.69216061185469, "grad_norm": 0.5215150713920593, "learning_rate": 6.991858185971447e-05, "loss": 0.0165, "step": 120330 }, { "epoch": 76.69853409815168, "grad_norm": 0.39891546964645386, "learning_rate": 6.991352639752479e-05, "loss": 0.0132, "step": 120340 }, { "epoch": 76.70490758444869, "grad_norm": 0.5601545572280884, "learning_rate": 6.990847069336698e-05, "loss": 0.013, "step": 120350 }, { "epoch": 76.7112810707457, "grad_norm": 0.4320802390575409, "learning_rate": 6.990341474730244e-05, "loss": 0.0173, "step": 120360 }, { "epoch": 76.7176545570427, "grad_norm": 0.32223349809646606, "learning_rate": 6.989835855939263e-05, "loss": 0.031, "step": 120370 }, { "epoch": 76.7240280433397, "grad_norm": 0.5285472273826599, "learning_rate": 6.989330212969898e-05, "loss": 0.0155, "step": 120380 }, { "epoch": 76.73040152963671, "grad_norm": 0.5540227890014648, "learning_rate": 6.988824545828292e-05, "loss": 0.0143, "step": 120390 }, { "epoch": 76.73677501593372, "grad_norm": 0.317134290933609, "learning_rate": 6.98831885452059e-05, "loss": 0.0153, "step": 120400 }, { "epoch": 76.74314850223072, "grad_norm": 0.5720508694648743, "learning_rate": 6.987813139052937e-05, "loss": 0.0155, "step": 120410 }, { "epoch": 76.74952198852772, "grad_norm": 0.3130979537963867, "learning_rate": 6.987307399431477e-05, "loss": 0.0219, "step": 120420 }, { "epoch": 76.75589547482473, "grad_norm": 0.3876934349536896, "learning_rate": 6.986801635662356e-05, "loss": 0.0224, "step": 120430 }, { "epoch": 76.76226896112173, "grad_norm": 0.37582266330718994, "learning_rate": 6.986295847751719e-05, "loss": 0.0191, "step": 120440 }, { "epoch": 76.76864244741874, "grad_norm": 0.3939870595932007, "learning_rate": 6.985790035705713e-05, "loss": 0.0131, "step": 120450 }, { "epoch": 76.77501593371574, "grad_norm": 0.3004836440086365, "learning_rate": 6.985284199530481e-05, "loss": 0.0102, "step": 120460 }, { "epoch": 76.78138942001274, "grad_norm": 0.39778104424476624, "learning_rate": 6.984778339232174e-05, "loss": 0.0171, "step": 120470 }, { "epoch": 76.78776290630975, "grad_norm": 0.3575454652309418, "learning_rate": 6.984272454816936e-05, "loss": 0.0205, "step": 120480 }, { "epoch": 76.79413639260676, "grad_norm": 0.35549396276474, "learning_rate": 6.983766546290913e-05, "loss": 0.0219, "step": 120490 }, { "epoch": 76.80050987890377, "grad_norm": 0.4208260178565979, "learning_rate": 6.983260613660254e-05, "loss": 0.0136, "step": 120500 }, { "epoch": 76.80688336520076, "grad_norm": 0.37947744131088257, "learning_rate": 6.982754656931106e-05, "loss": 0.0131, "step": 120510 }, { "epoch": 76.81325685149777, "grad_norm": 0.34810665249824524, "learning_rate": 6.982248676109619e-05, "loss": 0.0204, "step": 120520 }, { "epoch": 76.81963033779478, "grad_norm": 0.5307003259658813, "learning_rate": 6.981742671201936e-05, "loss": 0.0241, "step": 120530 }, { "epoch": 76.82600382409179, "grad_norm": 0.4547986388206482, "learning_rate": 6.98123664221421e-05, "loss": 0.0156, "step": 120540 }, { "epoch": 76.83237731038878, "grad_norm": 0.3267855942249298, "learning_rate": 6.980730589152586e-05, "loss": 0.0179, "step": 120550 }, { "epoch": 76.83875079668579, "grad_norm": 0.40358152985572815, "learning_rate": 6.980224512023218e-05, "loss": 0.0138, "step": 120560 }, { "epoch": 76.8451242829828, "grad_norm": 0.464021772146225, "learning_rate": 6.979718410832251e-05, "loss": 0.0206, "step": 120570 }, { "epoch": 76.8514977692798, "grad_norm": 0.31981149315834045, "learning_rate": 6.979212285585837e-05, "loss": 0.0289, "step": 120580 }, { "epoch": 76.8578712555768, "grad_norm": 0.6751660108566284, "learning_rate": 6.978706136290125e-05, "loss": 0.0203, "step": 120590 }, { "epoch": 76.8642447418738, "grad_norm": 0.6354386806488037, "learning_rate": 6.978199962951264e-05, "loss": 0.0233, "step": 120600 }, { "epoch": 76.87061822817081, "grad_norm": 0.5301569104194641, "learning_rate": 6.977693765575407e-05, "loss": 0.0316, "step": 120610 }, { "epoch": 76.87699171446782, "grad_norm": 0.7087173461914062, "learning_rate": 6.977187544168703e-05, "loss": 0.0229, "step": 120620 }, { "epoch": 76.88336520076481, "grad_norm": 0.4721750319004059, "learning_rate": 6.976681298737304e-05, "loss": 0.0256, "step": 120630 }, { "epoch": 76.88973868706182, "grad_norm": 0.5929004549980164, "learning_rate": 6.976175029287359e-05, "loss": 0.0251, "step": 120640 }, { "epoch": 76.89611217335883, "grad_norm": 0.40192368626594543, "learning_rate": 6.975668735825025e-05, "loss": 0.0206, "step": 120650 }, { "epoch": 76.90248565965584, "grad_norm": 0.778732419013977, "learning_rate": 6.975162418356448e-05, "loss": 0.0328, "step": 120660 }, { "epoch": 76.90885914595283, "grad_norm": 0.3958010971546173, "learning_rate": 6.974656076887783e-05, "loss": 0.0163, "step": 120670 }, { "epoch": 76.91523263224984, "grad_norm": 0.40840908885002136, "learning_rate": 6.974149711425184e-05, "loss": 0.017, "step": 120680 }, { "epoch": 76.92160611854685, "grad_norm": 0.4346381723880768, "learning_rate": 6.9736433219748e-05, "loss": 0.0248, "step": 120690 }, { "epoch": 76.92797960484386, "grad_norm": 0.25700050592422485, "learning_rate": 6.973136908542788e-05, "loss": 0.0242, "step": 120700 }, { "epoch": 76.93435309114085, "grad_norm": 0.5753760933876038, "learning_rate": 6.972630471135297e-05, "loss": 0.0177, "step": 120710 }, { "epoch": 76.94072657743786, "grad_norm": 0.3882943391799927, "learning_rate": 6.972124009758486e-05, "loss": 0.0249, "step": 120720 }, { "epoch": 76.94710006373487, "grad_norm": 0.5235735177993774, "learning_rate": 6.971617524418504e-05, "loss": 0.026, "step": 120730 }, { "epoch": 76.95347355003187, "grad_norm": 0.28797486424446106, "learning_rate": 6.971111015121509e-05, "loss": 0.0185, "step": 120740 }, { "epoch": 76.95984703632887, "grad_norm": 0.49461284279823303, "learning_rate": 6.970604481873654e-05, "loss": 0.0211, "step": 120750 }, { "epoch": 76.96622052262587, "grad_norm": 0.45036908984184265, "learning_rate": 6.970097924681095e-05, "loss": 0.0265, "step": 120760 }, { "epoch": 76.97259400892288, "grad_norm": 0.6251679062843323, "learning_rate": 6.969591343549985e-05, "loss": 0.0161, "step": 120770 }, { "epoch": 76.97896749521989, "grad_norm": 0.3549080789089203, "learning_rate": 6.96908473848648e-05, "loss": 0.0198, "step": 120780 }, { "epoch": 76.98534098151688, "grad_norm": 0.6185206174850464, "learning_rate": 6.968578109496737e-05, "loss": 0.0226, "step": 120790 }, { "epoch": 76.99171446781389, "grad_norm": 0.3077760338783264, "learning_rate": 6.96807145658691e-05, "loss": 0.0228, "step": 120800 }, { "epoch": 76.9980879541109, "grad_norm": 0.218795046210289, "learning_rate": 6.967564779763159e-05, "loss": 0.0135, "step": 120810 }, { "epoch": 77.00446144040791, "grad_norm": 0.296843945980072, "learning_rate": 6.967058079031637e-05, "loss": 0.0256, "step": 120820 }, { "epoch": 77.0108349267049, "grad_norm": 0.32512179017066956, "learning_rate": 6.966551354398502e-05, "loss": 0.0131, "step": 120830 }, { "epoch": 77.01720841300191, "grad_norm": 0.3759230375289917, "learning_rate": 6.966044605869913e-05, "loss": 0.0247, "step": 120840 }, { "epoch": 77.02358189929892, "grad_norm": 0.35798123478889465, "learning_rate": 6.965537833452024e-05, "loss": 0.009, "step": 120850 }, { "epoch": 77.02995538559593, "grad_norm": 0.32153579592704773, "learning_rate": 6.965031037150996e-05, "loss": 0.0254, "step": 120860 }, { "epoch": 77.03632887189292, "grad_norm": 0.32079213857650757, "learning_rate": 6.964524216972983e-05, "loss": 0.0178, "step": 120870 }, { "epoch": 77.04270235818993, "grad_norm": 0.5773524641990662, "learning_rate": 6.964017372924147e-05, "loss": 0.0148, "step": 120880 }, { "epoch": 77.04907584448694, "grad_norm": 0.24179412424564362, "learning_rate": 6.963510505010646e-05, "loss": 0.0124, "step": 120890 }, { "epoch": 77.05544933078394, "grad_norm": 0.6741291880607605, "learning_rate": 6.963003613238639e-05, "loss": 0.0298, "step": 120900 }, { "epoch": 77.06182281708094, "grad_norm": 0.21513473987579346, "learning_rate": 6.962496697614283e-05, "loss": 0.0251, "step": 120910 }, { "epoch": 77.06819630337795, "grad_norm": 0.6130028367042542, "learning_rate": 6.961989758143741e-05, "loss": 0.0165, "step": 120920 }, { "epoch": 77.07456978967495, "grad_norm": 0.1981063336133957, "learning_rate": 6.961482794833171e-05, "loss": 0.0228, "step": 120930 }, { "epoch": 77.08094327597196, "grad_norm": 0.40909233689308167, "learning_rate": 6.960975807688732e-05, "loss": 0.0179, "step": 120940 }, { "epoch": 77.08731676226896, "grad_norm": 0.20108391344547272, "learning_rate": 6.960468796716585e-05, "loss": 0.0117, "step": 120950 }, { "epoch": 77.09369024856596, "grad_norm": 0.4932329058647156, "learning_rate": 6.959961761922894e-05, "loss": 0.016, "step": 120960 }, { "epoch": 77.10006373486297, "grad_norm": 0.460509717464447, "learning_rate": 6.959454703313815e-05, "loss": 0.0191, "step": 120970 }, { "epoch": 77.10643722115998, "grad_norm": 0.3285568356513977, "learning_rate": 6.958947620895511e-05, "loss": 0.0142, "step": 120980 }, { "epoch": 77.11281070745697, "grad_norm": 0.3266829550266266, "learning_rate": 6.958440514674144e-05, "loss": 0.0143, "step": 120990 }, { "epoch": 77.11918419375398, "grad_norm": 0.2708703279495239, "learning_rate": 6.957933384655877e-05, "loss": 0.0104, "step": 121000 }, { "epoch": 77.12555768005099, "grad_norm": 0.3283633887767792, "learning_rate": 6.957426230846868e-05, "loss": 0.0113, "step": 121010 }, { "epoch": 77.131931166348, "grad_norm": 0.13865064084529877, "learning_rate": 6.956919053253286e-05, "loss": 0.0156, "step": 121020 }, { "epoch": 77.13830465264499, "grad_norm": 0.45916593074798584, "learning_rate": 6.956411851881287e-05, "loss": 0.0132, "step": 121030 }, { "epoch": 77.144678138942, "grad_norm": 0.257213294506073, "learning_rate": 6.955904626737038e-05, "loss": 0.013, "step": 121040 }, { "epoch": 77.151051625239, "grad_norm": 0.3984711468219757, "learning_rate": 6.9553973778267e-05, "loss": 0.0233, "step": 121050 }, { "epoch": 77.15742511153601, "grad_norm": 0.49317264556884766, "learning_rate": 6.95489010515644e-05, "loss": 0.0149, "step": 121060 }, { "epoch": 77.16379859783301, "grad_norm": 0.3807309567928314, "learning_rate": 6.954382808732418e-05, "loss": 0.009, "step": 121070 }, { "epoch": 77.17017208413002, "grad_norm": 0.3420100510120392, "learning_rate": 6.953875488560799e-05, "loss": 0.0183, "step": 121080 }, { "epoch": 77.17654557042702, "grad_norm": 0.5449411273002625, "learning_rate": 6.953368144647749e-05, "loss": 0.0222, "step": 121090 }, { "epoch": 77.18291905672403, "grad_norm": 0.25654295086860657, "learning_rate": 6.95286077699943e-05, "loss": 0.0099, "step": 121100 }, { "epoch": 77.18929254302103, "grad_norm": 0.2492886781692505, "learning_rate": 6.952353385622011e-05, "loss": 0.0135, "step": 121110 }, { "epoch": 77.19566602931803, "grad_norm": 0.5579444766044617, "learning_rate": 6.951845970521654e-05, "loss": 0.0172, "step": 121120 }, { "epoch": 77.20203951561504, "grad_norm": 0.3125736713409424, "learning_rate": 6.951338531704526e-05, "loss": 0.0199, "step": 121130 }, { "epoch": 77.20841300191205, "grad_norm": 0.23040452599525452, "learning_rate": 6.950831069176793e-05, "loss": 0.0165, "step": 121140 }, { "epoch": 77.21478648820906, "grad_norm": 0.6188251376152039, "learning_rate": 6.95032358294462e-05, "loss": 0.0207, "step": 121150 }, { "epoch": 77.22115997450605, "grad_norm": 0.2550593614578247, "learning_rate": 6.949816073014174e-05, "loss": 0.015, "step": 121160 }, { "epoch": 77.22753346080306, "grad_norm": 0.33948245644569397, "learning_rate": 6.949308539391622e-05, "loss": 0.0097, "step": 121170 }, { "epoch": 77.23390694710007, "grad_norm": 0.23025597631931305, "learning_rate": 6.94880098208313e-05, "loss": 0.0125, "step": 121180 }, { "epoch": 77.24028043339707, "grad_norm": 0.7384238839149475, "learning_rate": 6.948293401094866e-05, "loss": 0.0222, "step": 121190 }, { "epoch": 77.24665391969407, "grad_norm": 0.265072226524353, "learning_rate": 6.947785796432999e-05, "loss": 0.0141, "step": 121200 }, { "epoch": 77.25302740599108, "grad_norm": 0.3657500147819519, "learning_rate": 6.947278168103694e-05, "loss": 0.0122, "step": 121210 }, { "epoch": 77.25940089228808, "grad_norm": 0.4296131134033203, "learning_rate": 6.946770516113122e-05, "loss": 0.017, "step": 121220 }, { "epoch": 77.26577437858509, "grad_norm": 0.23262923955917358, "learning_rate": 6.94626284046745e-05, "loss": 0.0108, "step": 121230 }, { "epoch": 77.27214786488209, "grad_norm": 0.3266640305519104, "learning_rate": 6.945755141172848e-05, "loss": 0.0155, "step": 121240 }, { "epoch": 77.2785213511791, "grad_norm": 0.35070186853408813, "learning_rate": 6.945247418235482e-05, "loss": 0.0185, "step": 121250 }, { "epoch": 77.2848948374761, "grad_norm": 0.2177131474018097, "learning_rate": 6.944739671661523e-05, "loss": 0.0116, "step": 121260 }, { "epoch": 77.29126832377311, "grad_norm": 0.3152422606945038, "learning_rate": 6.944231901457144e-05, "loss": 0.0116, "step": 121270 }, { "epoch": 77.2976418100701, "grad_norm": 0.18078164756298065, "learning_rate": 6.943724107628508e-05, "loss": 0.0174, "step": 121280 }, { "epoch": 77.30401529636711, "grad_norm": 0.21065227687358856, "learning_rate": 6.943216290181791e-05, "loss": 0.0175, "step": 121290 }, { "epoch": 77.31038878266412, "grad_norm": 0.17832772433757782, "learning_rate": 6.94270844912316e-05, "loss": 0.0089, "step": 121300 }, { "epoch": 77.31676226896113, "grad_norm": 0.45380157232284546, "learning_rate": 6.942200584458788e-05, "loss": 0.0183, "step": 121310 }, { "epoch": 77.32313575525812, "grad_norm": 0.48307523131370544, "learning_rate": 6.941692696194846e-05, "loss": 0.0152, "step": 121320 }, { "epoch": 77.32950924155513, "grad_norm": 0.32400044798851013, "learning_rate": 6.941184784337502e-05, "loss": 0.0168, "step": 121330 }, { "epoch": 77.33588272785214, "grad_norm": 0.3296479880809784, "learning_rate": 6.940676848892933e-05, "loss": 0.0212, "step": 121340 }, { "epoch": 77.34225621414915, "grad_norm": 0.224664106965065, "learning_rate": 6.940168889867306e-05, "loss": 0.0219, "step": 121350 }, { "epoch": 77.34862970044614, "grad_norm": 0.5276059508323669, "learning_rate": 6.939660907266796e-05, "loss": 0.0117, "step": 121360 }, { "epoch": 77.35500318674315, "grad_norm": 0.20155705511569977, "learning_rate": 6.939152901097575e-05, "loss": 0.0187, "step": 121370 }, { "epoch": 77.36137667304016, "grad_norm": 0.24087291955947876, "learning_rate": 6.938644871365814e-05, "loss": 0.0127, "step": 121380 }, { "epoch": 77.36775015933716, "grad_norm": 0.48428475856781006, "learning_rate": 6.938136818077687e-05, "loss": 0.0151, "step": 121390 }, { "epoch": 77.37412364563416, "grad_norm": 0.3187786340713501, "learning_rate": 6.93762874123937e-05, "loss": 0.0161, "step": 121400 }, { "epoch": 77.38049713193116, "grad_norm": 0.511940062046051, "learning_rate": 6.937120640857034e-05, "loss": 0.0155, "step": 121410 }, { "epoch": 77.38687061822817, "grad_norm": 0.3615458309650421, "learning_rate": 6.936612516936852e-05, "loss": 0.0161, "step": 121420 }, { "epoch": 77.39324410452518, "grad_norm": 0.34313055872917175, "learning_rate": 6.936104369485e-05, "loss": 0.0086, "step": 121430 }, { "epoch": 77.39961759082217, "grad_norm": 0.4214589595794678, "learning_rate": 6.935596198507652e-05, "loss": 0.0114, "step": 121440 }, { "epoch": 77.40599107711918, "grad_norm": 0.754612386226654, "learning_rate": 6.935088004010983e-05, "loss": 0.02, "step": 121450 }, { "epoch": 77.41236456341619, "grad_norm": 0.413110613822937, "learning_rate": 6.934579786001167e-05, "loss": 0.0144, "step": 121460 }, { "epoch": 77.4187380497132, "grad_norm": 0.365928590297699, "learning_rate": 6.93407154448438e-05, "loss": 0.0214, "step": 121470 }, { "epoch": 77.42511153601019, "grad_norm": 0.4973861873149872, "learning_rate": 6.9335632794668e-05, "loss": 0.0188, "step": 121480 }, { "epoch": 77.4314850223072, "grad_norm": 0.5191419124603271, "learning_rate": 6.933054990954597e-05, "loss": 0.0238, "step": 121490 }, { "epoch": 77.43785850860421, "grad_norm": 0.4629170000553131, "learning_rate": 6.932546678953953e-05, "loss": 0.0164, "step": 121500 }, { "epoch": 77.44423199490122, "grad_norm": 0.38150128722190857, "learning_rate": 6.932038343471041e-05, "loss": 0.018, "step": 121510 }, { "epoch": 77.45060548119821, "grad_norm": 0.3358640968799591, "learning_rate": 6.931529984512041e-05, "loss": 0.0195, "step": 121520 }, { "epoch": 77.45697896749522, "grad_norm": 0.331556111574173, "learning_rate": 6.931021602083125e-05, "loss": 0.017, "step": 121530 }, { "epoch": 77.46335245379223, "grad_norm": 0.4522685110569, "learning_rate": 6.930513196190477e-05, "loss": 0.0315, "step": 121540 }, { "epoch": 77.46972594008923, "grad_norm": 0.6638670563697815, "learning_rate": 6.930004766840268e-05, "loss": 0.0275, "step": 121550 }, { "epoch": 77.47609942638623, "grad_norm": 0.3923209309577942, "learning_rate": 6.92949631403868e-05, "loss": 0.0131, "step": 121560 }, { "epoch": 77.48247291268324, "grad_norm": 0.4300227761268616, "learning_rate": 6.928987837791891e-05, "loss": 0.02, "step": 121570 }, { "epoch": 77.48884639898024, "grad_norm": 0.4284272789955139, "learning_rate": 6.928479338106077e-05, "loss": 0.0189, "step": 121580 }, { "epoch": 77.49521988527725, "grad_norm": 0.351423442363739, "learning_rate": 6.927970814987419e-05, "loss": 0.0164, "step": 121590 }, { "epoch": 77.50159337157424, "grad_norm": 0.43340548872947693, "learning_rate": 6.927462268442093e-05, "loss": 0.0183, "step": 121600 }, { "epoch": 77.50796685787125, "grad_norm": 0.3433035910129547, "learning_rate": 6.926953698476284e-05, "loss": 0.0174, "step": 121610 }, { "epoch": 77.51434034416826, "grad_norm": 0.46647587418556213, "learning_rate": 6.926445105096167e-05, "loss": 0.0243, "step": 121620 }, { "epoch": 77.52071383046527, "grad_norm": 0.28962430357933044, "learning_rate": 6.925936488307923e-05, "loss": 0.0143, "step": 121630 }, { "epoch": 77.52708731676226, "grad_norm": 0.2893587350845337, "learning_rate": 6.925427848117731e-05, "loss": 0.0114, "step": 121640 }, { "epoch": 77.53346080305927, "grad_norm": 0.559873640537262, "learning_rate": 6.924919184531774e-05, "loss": 0.0219, "step": 121650 }, { "epoch": 77.53983428935628, "grad_norm": 0.34016141295433044, "learning_rate": 6.92441049755623e-05, "loss": 0.0136, "step": 121660 }, { "epoch": 77.54620777565329, "grad_norm": 0.23608146607875824, "learning_rate": 6.923901787197284e-05, "loss": 0.0135, "step": 121670 }, { "epoch": 77.55258126195028, "grad_norm": 0.24244184792041779, "learning_rate": 6.923393053461114e-05, "loss": 0.0191, "step": 121680 }, { "epoch": 77.55895474824729, "grad_norm": 0.5104697942733765, "learning_rate": 6.922884296353901e-05, "loss": 0.0136, "step": 121690 }, { "epoch": 77.5653282345443, "grad_norm": 0.33493471145629883, "learning_rate": 6.922375515881829e-05, "loss": 0.0177, "step": 121700 }, { "epoch": 77.5717017208413, "grad_norm": 0.3044620156288147, "learning_rate": 6.921866712051081e-05, "loss": 0.0199, "step": 121710 }, { "epoch": 77.5780752071383, "grad_norm": 0.372275710105896, "learning_rate": 6.921357884867836e-05, "loss": 0.0178, "step": 121720 }, { "epoch": 77.5844486934353, "grad_norm": 0.4293034076690674, "learning_rate": 6.92084903433828e-05, "loss": 0.0167, "step": 121730 }, { "epoch": 77.59082217973231, "grad_norm": 0.3320325016975403, "learning_rate": 6.920340160468593e-05, "loss": 0.0145, "step": 121740 }, { "epoch": 77.59719566602932, "grad_norm": 0.45226767659187317, "learning_rate": 6.919831263264962e-05, "loss": 0.0179, "step": 121750 }, { "epoch": 77.60356915232632, "grad_norm": 0.37698057293891907, "learning_rate": 6.919322342733567e-05, "loss": 0.0135, "step": 121760 }, { "epoch": 77.60994263862332, "grad_norm": 0.4043039083480835, "learning_rate": 6.918813398880595e-05, "loss": 0.0156, "step": 121770 }, { "epoch": 77.61631612492033, "grad_norm": 0.5637972354888916, "learning_rate": 6.918304431712227e-05, "loss": 0.0195, "step": 121780 }, { "epoch": 77.62268961121734, "grad_norm": 0.30011045932769775, "learning_rate": 6.91779544123465e-05, "loss": 0.013, "step": 121790 }, { "epoch": 77.62906309751435, "grad_norm": 0.3574414849281311, "learning_rate": 6.917286427454047e-05, "loss": 0.0155, "step": 121800 }, { "epoch": 77.63543658381134, "grad_norm": 0.43574419617652893, "learning_rate": 6.916777390376603e-05, "loss": 0.0124, "step": 121810 }, { "epoch": 77.64181007010835, "grad_norm": 0.4471307098865509, "learning_rate": 6.916268330008507e-05, "loss": 0.0185, "step": 121820 }, { "epoch": 77.64818355640536, "grad_norm": 0.3900284469127655, "learning_rate": 6.915759246355937e-05, "loss": 0.0146, "step": 121830 }, { "epoch": 77.65455704270236, "grad_norm": 0.28427740931510925, "learning_rate": 6.915250139425087e-05, "loss": 0.0176, "step": 121840 }, { "epoch": 77.66093052899936, "grad_norm": 0.4913356602191925, "learning_rate": 6.914741009222138e-05, "loss": 0.021, "step": 121850 }, { "epoch": 77.66730401529637, "grad_norm": 0.5813433527946472, "learning_rate": 6.914231855753278e-05, "loss": 0.0187, "step": 121860 }, { "epoch": 77.67367750159337, "grad_norm": 0.2852868139743805, "learning_rate": 6.913722679024693e-05, "loss": 0.0212, "step": 121870 }, { "epoch": 77.68005098789038, "grad_norm": 0.3290872871875763, "learning_rate": 6.913213479042572e-05, "loss": 0.0204, "step": 121880 }, { "epoch": 77.68642447418738, "grad_norm": 0.24395370483398438, "learning_rate": 6.9127042558131e-05, "loss": 0.0116, "step": 121890 }, { "epoch": 77.69279796048438, "grad_norm": 0.312213659286499, "learning_rate": 6.912195009342466e-05, "loss": 0.0213, "step": 121900 }, { "epoch": 77.69917144678139, "grad_norm": 0.35331323742866516, "learning_rate": 6.911685739636858e-05, "loss": 0.0103, "step": 121910 }, { "epoch": 77.7055449330784, "grad_norm": 0.3034515380859375, "learning_rate": 6.91117644670246e-05, "loss": 0.0151, "step": 121920 }, { "epoch": 77.7119184193754, "grad_norm": 0.35450872778892517, "learning_rate": 6.910667130545468e-05, "loss": 0.0138, "step": 121930 }, { "epoch": 77.7182919056724, "grad_norm": 0.26445645093917847, "learning_rate": 6.910157791172062e-05, "loss": 0.0182, "step": 121940 }, { "epoch": 77.72466539196941, "grad_norm": 0.40601012110710144, "learning_rate": 6.909648428588439e-05, "loss": 0.0201, "step": 121950 }, { "epoch": 77.73103887826642, "grad_norm": 0.37410447001457214, "learning_rate": 6.909139042800783e-05, "loss": 0.0244, "step": 121960 }, { "epoch": 77.73741236456341, "grad_norm": 0.2934815287590027, "learning_rate": 6.908629633815285e-05, "loss": 0.0075, "step": 121970 }, { "epoch": 77.74378585086042, "grad_norm": 0.45327824354171753, "learning_rate": 6.908120201638136e-05, "loss": 0.0238, "step": 121980 }, { "epoch": 77.75015933715743, "grad_norm": 0.4717637896537781, "learning_rate": 6.907610746275523e-05, "loss": 0.0233, "step": 121990 }, { "epoch": 77.75653282345444, "grad_norm": 0.3918783366680145, "learning_rate": 6.907101267733641e-05, "loss": 0.0134, "step": 122000 }, { "epoch": 77.76290630975143, "grad_norm": 0.3900119662284851, "learning_rate": 6.906591766018675e-05, "loss": 0.0148, "step": 122010 }, { "epoch": 77.76927979604844, "grad_norm": 0.5096040964126587, "learning_rate": 6.906082241136822e-05, "loss": 0.0176, "step": 122020 }, { "epoch": 77.77565328234544, "grad_norm": 0.6446693539619446, "learning_rate": 6.905572693094268e-05, "loss": 0.031, "step": 122030 }, { "epoch": 77.78202676864245, "grad_norm": 0.3987652659416199, "learning_rate": 6.905063121897207e-05, "loss": 0.0147, "step": 122040 }, { "epoch": 77.78840025493945, "grad_norm": 0.5443147420883179, "learning_rate": 6.904553527551831e-05, "loss": 0.0135, "step": 122050 }, { "epoch": 77.79477374123645, "grad_norm": 0.5124820470809937, "learning_rate": 6.90404391006433e-05, "loss": 0.0159, "step": 122060 }, { "epoch": 77.80114722753346, "grad_norm": 0.3089767098426819, "learning_rate": 6.903534269440902e-05, "loss": 0.0154, "step": 122070 }, { "epoch": 77.80752071383047, "grad_norm": 0.543022096157074, "learning_rate": 6.90302460568773e-05, "loss": 0.0165, "step": 122080 }, { "epoch": 77.81389420012746, "grad_norm": 0.37379905581474304, "learning_rate": 6.902514918811015e-05, "loss": 0.0221, "step": 122090 }, { "epoch": 77.82026768642447, "grad_norm": 0.3803763687610626, "learning_rate": 6.902005208816946e-05, "loss": 0.0123, "step": 122100 }, { "epoch": 77.82664117272148, "grad_norm": 0.21133831143379211, "learning_rate": 6.90149547571172e-05, "loss": 0.0185, "step": 122110 }, { "epoch": 77.83301465901849, "grad_norm": 0.5882111191749573, "learning_rate": 6.900985719501528e-05, "loss": 0.029, "step": 122120 }, { "epoch": 77.83938814531548, "grad_norm": 0.4107116460800171, "learning_rate": 6.900475940192564e-05, "loss": 0.0197, "step": 122130 }, { "epoch": 77.84576163161249, "grad_norm": 0.28202855587005615, "learning_rate": 6.899966137791022e-05, "loss": 0.0173, "step": 122140 }, { "epoch": 77.8521351179095, "grad_norm": 0.2856064736843109, "learning_rate": 6.8994563123031e-05, "loss": 0.0235, "step": 122150 }, { "epoch": 77.8585086042065, "grad_norm": 0.31416404247283936, "learning_rate": 6.89894646373499e-05, "loss": 0.0149, "step": 122160 }, { "epoch": 77.8648820905035, "grad_norm": 0.4551234245300293, "learning_rate": 6.898436592092885e-05, "loss": 0.0192, "step": 122170 }, { "epoch": 77.87125557680051, "grad_norm": 0.42161476612091064, "learning_rate": 6.897926697382986e-05, "loss": 0.0182, "step": 122180 }, { "epoch": 77.87762906309752, "grad_norm": 0.30027756094932556, "learning_rate": 6.897416779611485e-05, "loss": 0.0137, "step": 122190 }, { "epoch": 77.88400254939452, "grad_norm": 0.23313355445861816, "learning_rate": 6.896906838784577e-05, "loss": 0.0136, "step": 122200 }, { "epoch": 77.89037603569152, "grad_norm": 0.33784210681915283, "learning_rate": 6.896396874908462e-05, "loss": 0.018, "step": 122210 }, { "epoch": 77.89674952198853, "grad_norm": 0.48688432574272156, "learning_rate": 6.895886887989334e-05, "loss": 0.0208, "step": 122220 }, { "epoch": 77.90312300828553, "grad_norm": 0.4813896119594574, "learning_rate": 6.89537687803339e-05, "loss": 0.0227, "step": 122230 }, { "epoch": 77.90949649458254, "grad_norm": 0.41594409942626953, "learning_rate": 6.894866845046825e-05, "loss": 0.0151, "step": 122240 }, { "epoch": 77.91586998087953, "grad_norm": 0.5649278163909912, "learning_rate": 6.894356789035843e-05, "loss": 0.015, "step": 122250 }, { "epoch": 77.92224346717654, "grad_norm": 0.47628235816955566, "learning_rate": 6.893846710006635e-05, "loss": 0.0199, "step": 122260 }, { "epoch": 77.92861695347355, "grad_norm": 0.6539028286933899, "learning_rate": 6.893336607965402e-05, "loss": 0.0173, "step": 122270 }, { "epoch": 77.93499043977056, "grad_norm": 0.41415587067604065, "learning_rate": 6.892826482918342e-05, "loss": 0.0102, "step": 122280 }, { "epoch": 77.94136392606755, "grad_norm": 0.34293362498283386, "learning_rate": 6.892316334871652e-05, "loss": 0.0167, "step": 122290 }, { "epoch": 77.94773741236456, "grad_norm": 0.4101725220680237, "learning_rate": 6.891806163831532e-05, "loss": 0.0256, "step": 122300 }, { "epoch": 77.95411089866157, "grad_norm": 0.5718145370483398, "learning_rate": 6.891295969804181e-05, "loss": 0.0225, "step": 122310 }, { "epoch": 77.96048438495858, "grad_norm": 0.6408422589302063, "learning_rate": 6.8907857527958e-05, "loss": 0.0221, "step": 122320 }, { "epoch": 77.96685787125557, "grad_norm": 0.32891184091567993, "learning_rate": 6.890275512812584e-05, "loss": 0.0267, "step": 122330 }, { "epoch": 77.97323135755258, "grad_norm": 0.5236144065856934, "learning_rate": 6.889765249860738e-05, "loss": 0.0148, "step": 122340 }, { "epoch": 77.97960484384959, "grad_norm": 0.11620648205280304, "learning_rate": 6.889254963946459e-05, "loss": 0.016, "step": 122350 }, { "epoch": 77.9859783301466, "grad_norm": 0.6122729182243347, "learning_rate": 6.88874465507595e-05, "loss": 0.0153, "step": 122360 }, { "epoch": 77.99235181644359, "grad_norm": 0.14491863548755646, "learning_rate": 6.888234323255408e-05, "loss": 0.026, "step": 122370 }, { "epoch": 77.9987253027406, "grad_norm": 1.0197968482971191, "learning_rate": 6.887723968491037e-05, "loss": 0.0158, "step": 122380 }, { "epoch": 78.0050987890376, "grad_norm": 0.36731505393981934, "learning_rate": 6.887213590789038e-05, "loss": 0.011, "step": 122390 }, { "epoch": 78.01147227533461, "grad_norm": 0.8004700541496277, "learning_rate": 6.886703190155612e-05, "loss": 0.0169, "step": 122400 }, { "epoch": 78.01784576163162, "grad_norm": 0.5506035089492798, "learning_rate": 6.88619276659696e-05, "loss": 0.0264, "step": 122410 }, { "epoch": 78.02421924792861, "grad_norm": 0.3063824772834778, "learning_rate": 6.885682320119286e-05, "loss": 0.0196, "step": 122420 }, { "epoch": 78.03059273422562, "grad_norm": 0.28915634751319885, "learning_rate": 6.88517185072879e-05, "loss": 0.027, "step": 122430 }, { "epoch": 78.03696622052263, "grad_norm": 0.28449276089668274, "learning_rate": 6.884661358431678e-05, "loss": 0.0205, "step": 122440 }, { "epoch": 78.04333970681964, "grad_norm": 0.2267657369375229, "learning_rate": 6.884150843234149e-05, "loss": 0.0218, "step": 122450 }, { "epoch": 78.04971319311663, "grad_norm": 0.30294060707092285, "learning_rate": 6.883640305142411e-05, "loss": 0.0163, "step": 122460 }, { "epoch": 78.05608667941364, "grad_norm": 0.20734252035617828, "learning_rate": 6.883129744162662e-05, "loss": 0.0113, "step": 122470 }, { "epoch": 78.06246016571065, "grad_norm": 0.7200838923454285, "learning_rate": 6.882619160301112e-05, "loss": 0.0189, "step": 122480 }, { "epoch": 78.06883365200765, "grad_norm": 0.34621474146842957, "learning_rate": 6.882108553563959e-05, "loss": 0.01, "step": 122490 }, { "epoch": 78.07520713830465, "grad_norm": 0.3410166800022125, "learning_rate": 6.88159792395741e-05, "loss": 0.0087, "step": 122500 }, { "epoch": 78.08158062460166, "grad_norm": 0.4359540641307831, "learning_rate": 6.881087271487671e-05, "loss": 0.0119, "step": 122510 }, { "epoch": 78.08795411089866, "grad_norm": 0.19677318632602692, "learning_rate": 6.880576596160945e-05, "loss": 0.0132, "step": 122520 }, { "epoch": 78.09432759719567, "grad_norm": 0.3519301414489746, "learning_rate": 6.880065897983438e-05, "loss": 0.016, "step": 122530 }, { "epoch": 78.10070108349267, "grad_norm": 0.33513250946998596, "learning_rate": 6.879555176961355e-05, "loss": 0.0103, "step": 122540 }, { "epoch": 78.10707456978967, "grad_norm": 0.4781673848628998, "learning_rate": 6.879044433100904e-05, "loss": 0.0156, "step": 122550 }, { "epoch": 78.11344805608668, "grad_norm": 0.7219738364219666, "learning_rate": 6.878533666408286e-05, "loss": 0.0103, "step": 122560 }, { "epoch": 78.11982154238369, "grad_norm": 0.24531111121177673, "learning_rate": 6.878022876889711e-05, "loss": 0.0092, "step": 122570 }, { "epoch": 78.12619502868068, "grad_norm": 0.6606714725494385, "learning_rate": 6.877512064551386e-05, "loss": 0.0185, "step": 122580 }, { "epoch": 78.13256851497769, "grad_norm": 0.36494237184524536, "learning_rate": 6.877001229399516e-05, "loss": 0.0098, "step": 122590 }, { "epoch": 78.1389420012747, "grad_norm": 0.28938886523246765, "learning_rate": 6.876490371440308e-05, "loss": 0.0122, "step": 122600 }, { "epoch": 78.14531548757171, "grad_norm": 0.14077922701835632, "learning_rate": 6.875979490679972e-05, "loss": 0.0133, "step": 122610 }, { "epoch": 78.1516889738687, "grad_norm": 0.5218281149864197, "learning_rate": 6.875468587124714e-05, "loss": 0.0149, "step": 122620 }, { "epoch": 78.15806246016571, "grad_norm": 0.4053407311439514, "learning_rate": 6.874957660780741e-05, "loss": 0.0132, "step": 122630 }, { "epoch": 78.16443594646272, "grad_norm": 0.37988582253456116, "learning_rate": 6.874446711654263e-05, "loss": 0.0146, "step": 122640 }, { "epoch": 78.17080943275973, "grad_norm": 0.32150617241859436, "learning_rate": 6.873935739751486e-05, "loss": 0.0156, "step": 122650 }, { "epoch": 78.17718291905672, "grad_norm": 0.3478430509567261, "learning_rate": 6.873424745078623e-05, "loss": 0.0314, "step": 122660 }, { "epoch": 78.18355640535373, "grad_norm": 0.3678928017616272, "learning_rate": 6.872913727641877e-05, "loss": 0.01, "step": 122670 }, { "epoch": 78.18992989165073, "grad_norm": 0.3248390257358551, "learning_rate": 6.872402687447464e-05, "loss": 0.0143, "step": 122680 }, { "epoch": 78.19630337794774, "grad_norm": 0.27577346563339233, "learning_rate": 6.87189162450159e-05, "loss": 0.0118, "step": 122690 }, { "epoch": 78.20267686424474, "grad_norm": 0.42649486660957336, "learning_rate": 6.871380538810464e-05, "loss": 0.0144, "step": 122700 }, { "epoch": 78.20905035054174, "grad_norm": 0.6105812788009644, "learning_rate": 6.8708694303803e-05, "loss": 0.0178, "step": 122710 }, { "epoch": 78.21542383683875, "grad_norm": 0.4149724245071411, "learning_rate": 6.870358299217303e-05, "loss": 0.0116, "step": 122720 }, { "epoch": 78.22179732313576, "grad_norm": 0.3241637051105499, "learning_rate": 6.869847145327689e-05, "loss": 0.0103, "step": 122730 }, { "epoch": 78.22817080943275, "grad_norm": 0.353291392326355, "learning_rate": 6.869335968717666e-05, "loss": 0.021, "step": 122740 }, { "epoch": 78.23454429572976, "grad_norm": 0.3571188449859619, "learning_rate": 6.868824769393446e-05, "loss": 0.0193, "step": 122750 }, { "epoch": 78.24091778202677, "grad_norm": 0.4770832359790802, "learning_rate": 6.868313547361241e-05, "loss": 0.01, "step": 122760 }, { "epoch": 78.24729126832378, "grad_norm": 0.2476295828819275, "learning_rate": 6.867802302627262e-05, "loss": 0.0118, "step": 122770 }, { "epoch": 78.25366475462077, "grad_norm": 0.191560760140419, "learning_rate": 6.867291035197723e-05, "loss": 0.0215, "step": 122780 }, { "epoch": 78.26003824091778, "grad_norm": 0.36439791321754456, "learning_rate": 6.866779745078832e-05, "loss": 0.0168, "step": 122790 }, { "epoch": 78.26641172721479, "grad_norm": 0.38987231254577637, "learning_rate": 6.866268432276806e-05, "loss": 0.0233, "step": 122800 }, { "epoch": 78.2727852135118, "grad_norm": 0.42037877440452576, "learning_rate": 6.865757096797856e-05, "loss": 0.0165, "step": 122810 }, { "epoch": 78.27915869980879, "grad_norm": 0.4059255123138428, "learning_rate": 6.865245738648197e-05, "loss": 0.0117, "step": 122820 }, { "epoch": 78.2855321861058, "grad_norm": 0.4492069482803345, "learning_rate": 6.86473435783404e-05, "loss": 0.0124, "step": 122830 }, { "epoch": 78.2919056724028, "grad_norm": 0.4361444413661957, "learning_rate": 6.8642229543616e-05, "loss": 0.024, "step": 122840 }, { "epoch": 78.29827915869981, "grad_norm": 0.35346418619155884, "learning_rate": 6.863711528237093e-05, "loss": 0.0167, "step": 122850 }, { "epoch": 78.3046526449968, "grad_norm": 0.3377143442630768, "learning_rate": 6.86320007946673e-05, "loss": 0.0265, "step": 122860 }, { "epoch": 78.31102613129381, "grad_norm": 0.32617509365081787, "learning_rate": 6.862688608056727e-05, "loss": 0.0119, "step": 122870 }, { "epoch": 78.31739961759082, "grad_norm": 0.45718562602996826, "learning_rate": 6.862177114013297e-05, "loss": 0.0324, "step": 122880 }, { "epoch": 78.32377310388783, "grad_norm": 0.2335909754037857, "learning_rate": 6.861665597342659e-05, "loss": 0.0142, "step": 122890 }, { "epoch": 78.33014659018482, "grad_norm": 0.3062015175819397, "learning_rate": 6.861154058051024e-05, "loss": 0.0144, "step": 122900 }, { "epoch": 78.33652007648183, "grad_norm": 0.3479377329349518, "learning_rate": 6.860642496144612e-05, "loss": 0.0166, "step": 122910 }, { "epoch": 78.34289356277884, "grad_norm": 0.6701332926750183, "learning_rate": 6.860130911629636e-05, "loss": 0.0266, "step": 122920 }, { "epoch": 78.34926704907585, "grad_norm": 0.3406379818916321, "learning_rate": 6.859619304512315e-05, "loss": 0.0262, "step": 122930 }, { "epoch": 78.35564053537284, "grad_norm": 0.3330257534980774, "learning_rate": 6.859107674798863e-05, "loss": 0.0162, "step": 122940 }, { "epoch": 78.36201402166985, "grad_norm": 0.44840726256370544, "learning_rate": 6.858596022495496e-05, "loss": 0.0165, "step": 122950 }, { "epoch": 78.36838750796686, "grad_norm": 0.27725595235824585, "learning_rate": 6.858084347608433e-05, "loss": 0.0128, "step": 122960 }, { "epoch": 78.37476099426387, "grad_norm": 0.286251962184906, "learning_rate": 6.857572650143891e-05, "loss": 0.0278, "step": 122970 }, { "epoch": 78.38113448056086, "grad_norm": 0.39910373091697693, "learning_rate": 6.857060930108088e-05, "loss": 0.0181, "step": 122980 }, { "epoch": 78.38750796685787, "grad_norm": 0.1990775465965271, "learning_rate": 6.85654918750724e-05, "loss": 0.0198, "step": 122990 }, { "epoch": 78.39388145315488, "grad_norm": 0.19122517108917236, "learning_rate": 6.856037422347567e-05, "loss": 0.0159, "step": 123000 }, { "epoch": 78.40025493945188, "grad_norm": 0.2408302277326584, "learning_rate": 6.855525634635289e-05, "loss": 0.0242, "step": 123010 }, { "epoch": 78.40662842574889, "grad_norm": 0.26401007175445557, "learning_rate": 6.85501382437662e-05, "loss": 0.0121, "step": 123020 }, { "epoch": 78.41300191204589, "grad_norm": 0.24770453572273254, "learning_rate": 6.854501991577784e-05, "loss": 0.0183, "step": 123030 }, { "epoch": 78.4193753983429, "grad_norm": 0.30025139451026917, "learning_rate": 6.853990136244996e-05, "loss": 0.014, "step": 123040 }, { "epoch": 78.4257488846399, "grad_norm": 0.5051948428153992, "learning_rate": 6.853478258384476e-05, "loss": 0.0142, "step": 123050 }, { "epoch": 78.43212237093691, "grad_norm": 0.4362858831882477, "learning_rate": 6.852966358002449e-05, "loss": 0.0196, "step": 123060 }, { "epoch": 78.4384958572339, "grad_norm": 0.31164684891700745, "learning_rate": 6.852454435105129e-05, "loss": 0.0107, "step": 123070 }, { "epoch": 78.44486934353091, "grad_norm": 0.42629143595695496, "learning_rate": 6.851942489698737e-05, "loss": 0.0175, "step": 123080 }, { "epoch": 78.45124282982792, "grad_norm": 0.25977206230163574, "learning_rate": 6.851430521789499e-05, "loss": 0.0137, "step": 123090 }, { "epoch": 78.45761631612493, "grad_norm": 0.5357429385185242, "learning_rate": 6.850918531383629e-05, "loss": 0.03, "step": 123100 }, { "epoch": 78.46398980242192, "grad_norm": 0.49596142768859863, "learning_rate": 6.850406518487354e-05, "loss": 0.0306, "step": 123110 }, { "epoch": 78.47036328871893, "grad_norm": 0.21735702455043793, "learning_rate": 6.84989448310689e-05, "loss": 0.0205, "step": 123120 }, { "epoch": 78.47673677501594, "grad_norm": 0.4362592399120331, "learning_rate": 6.849382425248464e-05, "loss": 0.0122, "step": 123130 }, { "epoch": 78.48311026131294, "grad_norm": 0.45920586585998535, "learning_rate": 6.848870344918292e-05, "loss": 0.0172, "step": 123140 }, { "epoch": 78.48948374760994, "grad_norm": 0.19502292573451996, "learning_rate": 6.848358242122604e-05, "loss": 0.02, "step": 123150 }, { "epoch": 78.49585723390695, "grad_norm": 0.415803462266922, "learning_rate": 6.847846116867614e-05, "loss": 0.0189, "step": 123160 }, { "epoch": 78.50223072020395, "grad_norm": 0.4810436964035034, "learning_rate": 6.847333969159551e-05, "loss": 0.0158, "step": 123170 }, { "epoch": 78.50860420650096, "grad_norm": 0.39929306507110596, "learning_rate": 6.846821799004637e-05, "loss": 0.0367, "step": 123180 }, { "epoch": 78.51497769279796, "grad_norm": 0.2582939863204956, "learning_rate": 6.846309606409094e-05, "loss": 0.0182, "step": 123190 }, { "epoch": 78.52135117909496, "grad_norm": 0.31870555877685547, "learning_rate": 6.845797391379144e-05, "loss": 0.0172, "step": 123200 }, { "epoch": 78.52772466539197, "grad_norm": 0.2942270338535309, "learning_rate": 6.845285153921014e-05, "loss": 0.0125, "step": 123210 }, { "epoch": 78.53409815168898, "grad_norm": 0.336407333612442, "learning_rate": 6.844772894040929e-05, "loss": 0.0188, "step": 123220 }, { "epoch": 78.54047163798597, "grad_norm": 0.22782959043979645, "learning_rate": 6.84426061174511e-05, "loss": 0.012, "step": 123230 }, { "epoch": 78.54684512428298, "grad_norm": 0.2748051881790161, "learning_rate": 6.843748307039783e-05, "loss": 0.0231, "step": 123240 }, { "epoch": 78.55321861057999, "grad_norm": 0.2406461238861084, "learning_rate": 6.843235979931175e-05, "loss": 0.0135, "step": 123250 }, { "epoch": 78.559592096877, "grad_norm": 0.3856358230113983, "learning_rate": 6.84272363042551e-05, "loss": 0.0177, "step": 123260 }, { "epoch": 78.56596558317399, "grad_norm": 0.26624780893325806, "learning_rate": 6.842211258529009e-05, "loss": 0.0101, "step": 123270 }, { "epoch": 78.572339069471, "grad_norm": 0.2951872944831848, "learning_rate": 6.841698864247906e-05, "loss": 0.0165, "step": 123280 }, { "epoch": 78.578712555768, "grad_norm": 0.3321335017681122, "learning_rate": 6.841186447588421e-05, "loss": 0.0154, "step": 123290 }, { "epoch": 78.58508604206502, "grad_norm": 0.38842177391052246, "learning_rate": 6.840674008556782e-05, "loss": 0.0105, "step": 123300 }, { "epoch": 78.59145952836201, "grad_norm": 0.6308919191360474, "learning_rate": 6.840161547159216e-05, "loss": 0.0208, "step": 123310 }, { "epoch": 78.59783301465902, "grad_norm": 0.3894574046134949, "learning_rate": 6.839649063401951e-05, "loss": 0.0312, "step": 123320 }, { "epoch": 78.60420650095602, "grad_norm": 0.21011745929718018, "learning_rate": 6.839136557291214e-05, "loss": 0.0212, "step": 123330 }, { "epoch": 78.61057998725303, "grad_norm": 0.36269330978393555, "learning_rate": 6.838624028833229e-05, "loss": 0.0151, "step": 123340 }, { "epoch": 78.61695347355003, "grad_norm": 0.24412178993225098, "learning_rate": 6.838111478034228e-05, "loss": 0.0178, "step": 123350 }, { "epoch": 78.62332695984703, "grad_norm": 0.3549131453037262, "learning_rate": 6.837598904900434e-05, "loss": 0.032, "step": 123360 }, { "epoch": 78.62970044614404, "grad_norm": 0.41879183053970337, "learning_rate": 6.837086309438081e-05, "loss": 0.0185, "step": 123370 }, { "epoch": 78.63607393244105, "grad_norm": 0.3784198462963104, "learning_rate": 6.836573691653394e-05, "loss": 0.0131, "step": 123380 }, { "epoch": 78.64244741873804, "grad_norm": 0.5755842328071594, "learning_rate": 6.836061051552602e-05, "loss": 0.0184, "step": 123390 }, { "epoch": 78.64882090503505, "grad_norm": 0.4400772452354431, "learning_rate": 6.835548389141936e-05, "loss": 0.0223, "step": 123400 }, { "epoch": 78.65519439133206, "grad_norm": 0.3973684012889862, "learning_rate": 6.835035704427623e-05, "loss": 0.0185, "step": 123410 }, { "epoch": 78.66156787762907, "grad_norm": 0.4347219467163086, "learning_rate": 6.834522997415895e-05, "loss": 0.019, "step": 123420 }, { "epoch": 78.66794136392606, "grad_norm": 0.49584612250328064, "learning_rate": 6.834010268112979e-05, "loss": 0.0153, "step": 123430 }, { "epoch": 78.67431485022307, "grad_norm": 0.5026621222496033, "learning_rate": 6.833497516525108e-05, "loss": 0.0196, "step": 123440 }, { "epoch": 78.68068833652008, "grad_norm": 0.44700005650520325, "learning_rate": 6.83298474265851e-05, "loss": 0.0155, "step": 123450 }, { "epoch": 78.68706182281709, "grad_norm": 0.5249901413917542, "learning_rate": 6.832471946519416e-05, "loss": 0.014, "step": 123460 }, { "epoch": 78.69343530911408, "grad_norm": 0.5060945153236389, "learning_rate": 6.83195912811406e-05, "loss": 0.0165, "step": 123470 }, { "epoch": 78.69980879541109, "grad_norm": 0.3246866464614868, "learning_rate": 6.83144628744867e-05, "loss": 0.0145, "step": 123480 }, { "epoch": 78.7061822817081, "grad_norm": 0.3250579237937927, "learning_rate": 6.830933424529477e-05, "loss": 0.023, "step": 123490 }, { "epoch": 78.7125557680051, "grad_norm": 0.47065550088882446, "learning_rate": 6.830420539362716e-05, "loss": 0.0287, "step": 123500 }, { "epoch": 78.7189292543021, "grad_norm": 0.5402032732963562, "learning_rate": 6.829907631954617e-05, "loss": 0.0148, "step": 123510 }, { "epoch": 78.7253027405991, "grad_norm": 0.6417707204818726, "learning_rate": 6.829394702311411e-05, "loss": 0.0163, "step": 123520 }, { "epoch": 78.73167622689611, "grad_norm": 0.5371558666229248, "learning_rate": 6.828881750439334e-05, "loss": 0.0214, "step": 123530 }, { "epoch": 78.73804971319312, "grad_norm": 0.5166661739349365, "learning_rate": 6.828368776344615e-05, "loss": 0.0144, "step": 123540 }, { "epoch": 78.74442319949011, "grad_norm": 0.8493273854255676, "learning_rate": 6.827855780033492e-05, "loss": 0.0167, "step": 123550 }, { "epoch": 78.75079668578712, "grad_norm": 0.7652248740196228, "learning_rate": 6.827342761512193e-05, "loss": 0.0189, "step": 123560 }, { "epoch": 78.75717017208413, "grad_norm": 0.23265600204467773, "learning_rate": 6.826829720786955e-05, "loss": 0.0105, "step": 123570 }, { "epoch": 78.76354365838114, "grad_norm": 0.38748666644096375, "learning_rate": 6.826316657864012e-05, "loss": 0.0135, "step": 123580 }, { "epoch": 78.76991714467813, "grad_norm": 0.4733855128288269, "learning_rate": 6.825803572749596e-05, "loss": 0.0147, "step": 123590 }, { "epoch": 78.77629063097514, "grad_norm": 0.2693250775337219, "learning_rate": 6.825290465449943e-05, "loss": 0.0154, "step": 123600 }, { "epoch": 78.78266411727215, "grad_norm": 0.3528660535812378, "learning_rate": 6.824777335971287e-05, "loss": 0.01, "step": 123610 }, { "epoch": 78.78903760356916, "grad_norm": 0.2684353291988373, "learning_rate": 6.824264184319865e-05, "loss": 0.0133, "step": 123620 }, { "epoch": 78.79541108986615, "grad_norm": 0.5610007047653198, "learning_rate": 6.823751010501909e-05, "loss": 0.0151, "step": 123630 }, { "epoch": 78.80178457616316, "grad_norm": 0.29709136486053467, "learning_rate": 6.823237814523658e-05, "loss": 0.0148, "step": 123640 }, { "epoch": 78.80815806246017, "grad_norm": 0.24438971281051636, "learning_rate": 6.822724596391344e-05, "loss": 0.0114, "step": 123650 }, { "epoch": 78.81453154875717, "grad_norm": 0.29563039541244507, "learning_rate": 6.822211356111208e-05, "loss": 0.0207, "step": 123660 }, { "epoch": 78.82090503505418, "grad_norm": 0.9763038158416748, "learning_rate": 6.821698093689481e-05, "loss": 0.0135, "step": 123670 }, { "epoch": 78.82727852135118, "grad_norm": 0.4119942784309387, "learning_rate": 6.821184809132404e-05, "loss": 0.0144, "step": 123680 }, { "epoch": 78.83365200764818, "grad_norm": 0.3480439782142639, "learning_rate": 6.820671502446211e-05, "loss": 0.0183, "step": 123690 }, { "epoch": 78.84002549394519, "grad_norm": 0.45418667793273926, "learning_rate": 6.820158173637142e-05, "loss": 0.0146, "step": 123700 }, { "epoch": 78.8463989802422, "grad_norm": 0.6552848815917969, "learning_rate": 6.81964482271143e-05, "loss": 0.0159, "step": 123710 }, { "epoch": 78.8527724665392, "grad_norm": 0.42525896430015564, "learning_rate": 6.819131449675317e-05, "loss": 0.0161, "step": 123720 }, { "epoch": 78.8591459528362, "grad_norm": 0.30015668272972107, "learning_rate": 6.818618054535039e-05, "loss": 0.011, "step": 123730 }, { "epoch": 78.86551943913321, "grad_norm": 0.6558993458747864, "learning_rate": 6.818104637296835e-05, "loss": 0.0134, "step": 123740 }, { "epoch": 78.87189292543022, "grad_norm": 0.27336692810058594, "learning_rate": 6.817591197966941e-05, "loss": 0.0156, "step": 123750 }, { "epoch": 78.87826641172721, "grad_norm": 0.46464288234710693, "learning_rate": 6.8170777365516e-05, "loss": 0.0107, "step": 123760 }, { "epoch": 78.88463989802422, "grad_norm": 0.5010867118835449, "learning_rate": 6.816564253057048e-05, "loss": 0.0189, "step": 123770 }, { "epoch": 78.89101338432123, "grad_norm": 0.4154602885246277, "learning_rate": 6.816050747489526e-05, "loss": 0.0147, "step": 123780 }, { "epoch": 78.89738687061823, "grad_norm": 0.4458504617214203, "learning_rate": 6.815537219855273e-05, "loss": 0.013, "step": 123790 }, { "epoch": 78.90376035691523, "grad_norm": 0.2146216779947281, "learning_rate": 6.815023670160529e-05, "loss": 0.0156, "step": 123800 }, { "epoch": 78.91013384321224, "grad_norm": 0.19892099499702454, "learning_rate": 6.814510098411532e-05, "loss": 0.0317, "step": 123810 }, { "epoch": 78.91650732950924, "grad_norm": 0.3107859492301941, "learning_rate": 6.813996504614524e-05, "loss": 0.0134, "step": 123820 }, { "epoch": 78.92288081580625, "grad_norm": 0.6733356714248657, "learning_rate": 6.813482888775746e-05, "loss": 0.0233, "step": 123830 }, { "epoch": 78.92925430210325, "grad_norm": 0.31189224123954773, "learning_rate": 6.81296925090144e-05, "loss": 0.0169, "step": 123840 }, { "epoch": 78.93562778840025, "grad_norm": 0.18726645410060883, "learning_rate": 6.812455590997845e-05, "loss": 0.0188, "step": 123850 }, { "epoch": 78.94200127469726, "grad_norm": 0.4826422929763794, "learning_rate": 6.811941909071205e-05, "loss": 0.0112, "step": 123860 }, { "epoch": 78.94837476099427, "grad_norm": 0.5559147000312805, "learning_rate": 6.811428205127756e-05, "loss": 0.0214, "step": 123870 }, { "epoch": 78.95474824729126, "grad_norm": 0.32084429264068604, "learning_rate": 6.810914479173749e-05, "loss": 0.0105, "step": 123880 }, { "epoch": 78.96112173358827, "grad_norm": 0.6392157077789307, "learning_rate": 6.810400731215419e-05, "loss": 0.0274, "step": 123890 }, { "epoch": 78.96749521988528, "grad_norm": 0.32043755054473877, "learning_rate": 6.80988696125901e-05, "loss": 0.0175, "step": 123900 }, { "epoch": 78.97386870618229, "grad_norm": 0.45797303318977356, "learning_rate": 6.809373169310766e-05, "loss": 0.0159, "step": 123910 }, { "epoch": 78.98024219247928, "grad_norm": 0.24299174547195435, "learning_rate": 6.80885935537693e-05, "loss": 0.0205, "step": 123920 }, { "epoch": 78.98661567877629, "grad_norm": 0.28977566957473755, "learning_rate": 6.808345519463744e-05, "loss": 0.0241, "step": 123930 }, { "epoch": 78.9929891650733, "grad_norm": 0.7594200372695923, "learning_rate": 6.807831661577455e-05, "loss": 0.0165, "step": 123940 }, { "epoch": 78.9993626513703, "grad_norm": 0.3583877384662628, "learning_rate": 6.807317781724302e-05, "loss": 0.0126, "step": 123950 }, { "epoch": 79.0057361376673, "grad_norm": 0.2698872983455658, "learning_rate": 6.806803879910533e-05, "loss": 0.0195, "step": 123960 }, { "epoch": 79.0121096239643, "grad_norm": 0.2844659686088562, "learning_rate": 6.80628995614239e-05, "loss": 0.0142, "step": 123970 }, { "epoch": 79.01848311026131, "grad_norm": 0.39553436636924744, "learning_rate": 6.805776010426119e-05, "loss": 0.0224, "step": 123980 }, { "epoch": 79.02485659655832, "grad_norm": 0.34019938111305237, "learning_rate": 6.805262042767966e-05, "loss": 0.0124, "step": 123990 }, { "epoch": 79.03123008285532, "grad_norm": 0.6314989924430847, "learning_rate": 6.804748053174172e-05, "loss": 0.023, "step": 124000 }, { "epoch": 79.03760356915232, "grad_norm": 0.31770408153533936, "learning_rate": 6.804234041650988e-05, "loss": 0.0142, "step": 124010 }, { "epoch": 79.04397705544933, "grad_norm": 0.581812858581543, "learning_rate": 6.803720008204655e-05, "loss": 0.0282, "step": 124020 }, { "epoch": 79.05035054174634, "grad_norm": 0.3969649076461792, "learning_rate": 6.803205952841421e-05, "loss": 0.0169, "step": 124030 }, { "epoch": 79.05672402804333, "grad_norm": 0.5441523790359497, "learning_rate": 6.802691875567533e-05, "loss": 0.0156, "step": 124040 }, { "epoch": 79.06309751434034, "grad_norm": 0.3006282448768616, "learning_rate": 6.802177776389236e-05, "loss": 0.0132, "step": 124050 }, { "epoch": 79.06947100063735, "grad_norm": 0.5456255674362183, "learning_rate": 6.801663655312778e-05, "loss": 0.0197, "step": 124060 }, { "epoch": 79.07584448693436, "grad_norm": 0.3816102147102356, "learning_rate": 6.801149512344404e-05, "loss": 0.0115, "step": 124070 }, { "epoch": 79.08221797323135, "grad_norm": 0.2637821137905121, "learning_rate": 6.800635347490366e-05, "loss": 0.0139, "step": 124080 }, { "epoch": 79.08859145952836, "grad_norm": 0.3666629493236542, "learning_rate": 6.800121160756906e-05, "loss": 0.0164, "step": 124090 }, { "epoch": 79.09496494582537, "grad_norm": 0.5454000234603882, "learning_rate": 6.799606952150275e-05, "loss": 0.023, "step": 124100 }, { "epoch": 79.10133843212238, "grad_norm": 0.27976348996162415, "learning_rate": 6.79909272167672e-05, "loss": 0.0203, "step": 124110 }, { "epoch": 79.10771191841937, "grad_norm": 0.9234357476234436, "learning_rate": 6.798578469342491e-05, "loss": 0.0222, "step": 124120 }, { "epoch": 79.11408540471638, "grad_norm": 0.3904895484447479, "learning_rate": 6.798064195153834e-05, "loss": 0.0183, "step": 124130 }, { "epoch": 79.12045889101339, "grad_norm": 0.3978463113307953, "learning_rate": 6.797549899117e-05, "loss": 0.0159, "step": 124140 }, { "epoch": 79.1268323773104, "grad_norm": 0.48421919345855713, "learning_rate": 6.797035581238238e-05, "loss": 0.0219, "step": 124150 }, { "epoch": 79.13320586360739, "grad_norm": 0.4921092092990875, "learning_rate": 6.796521241523795e-05, "loss": 0.0177, "step": 124160 }, { "epoch": 79.1395793499044, "grad_norm": 0.5840530395507812, "learning_rate": 6.796006879979927e-05, "loss": 0.0162, "step": 124170 }, { "epoch": 79.1459528362014, "grad_norm": 0.6610860228538513, "learning_rate": 6.795492496612874e-05, "loss": 0.0264, "step": 124180 }, { "epoch": 79.15232632249841, "grad_norm": 0.3641248643398285, "learning_rate": 6.794978091428896e-05, "loss": 0.0161, "step": 124190 }, { "epoch": 79.1586998087954, "grad_norm": 0.3426484763622284, "learning_rate": 6.794463664434237e-05, "loss": 0.0142, "step": 124200 }, { "epoch": 79.16507329509241, "grad_norm": 0.5096228122711182, "learning_rate": 6.793949215635152e-05, "loss": 0.0119, "step": 124210 }, { "epoch": 79.17144678138942, "grad_norm": 0.47674989700317383, "learning_rate": 6.793434745037891e-05, "loss": 0.0197, "step": 124220 }, { "epoch": 79.17782026768643, "grad_norm": 0.3248569071292877, "learning_rate": 6.792920252648703e-05, "loss": 0.0151, "step": 124230 }, { "epoch": 79.18419375398342, "grad_norm": 0.430462509393692, "learning_rate": 6.792405738473841e-05, "loss": 0.0149, "step": 124240 }, { "epoch": 79.19056724028043, "grad_norm": 0.30890515446662903, "learning_rate": 6.791891202519558e-05, "loss": 0.0126, "step": 124250 }, { "epoch": 79.19694072657744, "grad_norm": 0.40231853723526, "learning_rate": 6.791376644792105e-05, "loss": 0.0128, "step": 124260 }, { "epoch": 79.20331421287445, "grad_norm": 0.4438477158546448, "learning_rate": 6.790862065297732e-05, "loss": 0.0351, "step": 124270 }, { "epoch": 79.20968769917145, "grad_norm": 0.5814865231513977, "learning_rate": 6.790347464042698e-05, "loss": 0.0178, "step": 124280 }, { "epoch": 79.21606118546845, "grad_norm": 0.3947550356388092, "learning_rate": 6.789832841033249e-05, "loss": 0.0132, "step": 124290 }, { "epoch": 79.22243467176546, "grad_norm": 0.5559412837028503, "learning_rate": 6.789318196275642e-05, "loss": 0.0201, "step": 124300 }, { "epoch": 79.22880815806246, "grad_norm": 0.37741851806640625, "learning_rate": 6.788803529776131e-05, "loss": 0.0196, "step": 124310 }, { "epoch": 79.23518164435947, "grad_norm": 0.31301259994506836, "learning_rate": 6.788288841540966e-05, "loss": 0.0138, "step": 124320 }, { "epoch": 79.24155513065647, "grad_norm": 0.6295177936553955, "learning_rate": 6.787774131576404e-05, "loss": 0.021, "step": 124330 }, { "epoch": 79.24792861695347, "grad_norm": 0.27624720335006714, "learning_rate": 6.787259399888697e-05, "loss": 0.0127, "step": 124340 }, { "epoch": 79.25430210325048, "grad_norm": 0.5697582364082336, "learning_rate": 6.786744646484103e-05, "loss": 0.0186, "step": 124350 }, { "epoch": 79.26067558954749, "grad_norm": 0.2906741499900818, "learning_rate": 6.786229871368873e-05, "loss": 0.0225, "step": 124360 }, { "epoch": 79.26704907584448, "grad_norm": 0.4944457709789276, "learning_rate": 6.785715074549265e-05, "loss": 0.0228, "step": 124370 }, { "epoch": 79.27342256214149, "grad_norm": 0.26455989480018616, "learning_rate": 6.785200256031534e-05, "loss": 0.0219, "step": 124380 }, { "epoch": 79.2797960484385, "grad_norm": 0.30772116780281067, "learning_rate": 6.784685415821931e-05, "loss": 0.0132, "step": 124390 }, { "epoch": 79.2861695347355, "grad_norm": 0.4650837182998657, "learning_rate": 6.784170553926716e-05, "loss": 0.0268, "step": 124400 }, { "epoch": 79.2925430210325, "grad_norm": 0.23812608420848846, "learning_rate": 6.783655670352145e-05, "loss": 0.0127, "step": 124410 }, { "epoch": 79.29891650732951, "grad_norm": 0.6308455467224121, "learning_rate": 6.783140765104474e-05, "loss": 0.0139, "step": 124420 }, { "epoch": 79.30528999362652, "grad_norm": 0.2753642797470093, "learning_rate": 6.782625838189957e-05, "loss": 0.0137, "step": 124430 }, { "epoch": 79.31166347992352, "grad_norm": 0.3419796824455261, "learning_rate": 6.782110889614857e-05, "loss": 0.0243, "step": 124440 }, { "epoch": 79.31803696622052, "grad_norm": 0.51396644115448, "learning_rate": 6.781595919385424e-05, "loss": 0.0224, "step": 124450 }, { "epoch": 79.32441045251753, "grad_norm": 0.40464600920677185, "learning_rate": 6.781080927507919e-05, "loss": 0.0206, "step": 124460 }, { "epoch": 79.33078393881453, "grad_norm": 0.5071026086807251, "learning_rate": 6.780565913988599e-05, "loss": 0.0153, "step": 124470 }, { "epoch": 79.33715742511154, "grad_norm": 0.5155244469642639, "learning_rate": 6.780050878833722e-05, "loss": 0.0202, "step": 124480 }, { "epoch": 79.34353091140854, "grad_norm": 0.5394009351730347, "learning_rate": 6.779535822049547e-05, "loss": 0.0151, "step": 124490 }, { "epoch": 79.34990439770554, "grad_norm": 0.6136733889579773, "learning_rate": 6.77902074364233e-05, "loss": 0.0191, "step": 124500 }, { "epoch": 79.35627788400255, "grad_norm": 0.3775118589401245, "learning_rate": 6.778505643618334e-05, "loss": 0.0099, "step": 124510 }, { "epoch": 79.36265137029956, "grad_norm": 0.45429179072380066, "learning_rate": 6.777990521983812e-05, "loss": 0.0142, "step": 124520 }, { "epoch": 79.36902485659655, "grad_norm": 0.4027152955532074, "learning_rate": 6.77747537874503e-05, "loss": 0.0178, "step": 124530 }, { "epoch": 79.37539834289356, "grad_norm": 0.27589869499206543, "learning_rate": 6.77696021390824e-05, "loss": 0.0203, "step": 124540 }, { "epoch": 79.38177182919057, "grad_norm": 0.5553091764450073, "learning_rate": 6.776445027479707e-05, "loss": 0.0153, "step": 124550 }, { "epoch": 79.38814531548758, "grad_norm": 0.2851516008377075, "learning_rate": 6.775929819465691e-05, "loss": 0.0139, "step": 124560 }, { "epoch": 79.39451880178457, "grad_norm": 0.33986395597457886, "learning_rate": 6.775414589872451e-05, "loss": 0.0239, "step": 124570 }, { "epoch": 79.40089228808158, "grad_norm": 0.43518608808517456, "learning_rate": 6.774899338706247e-05, "loss": 0.0185, "step": 124580 }, { "epoch": 79.40726577437859, "grad_norm": 0.4383088946342468, "learning_rate": 6.77438406597334e-05, "loss": 0.016, "step": 124590 }, { "epoch": 79.4136392606756, "grad_norm": 0.35903435945510864, "learning_rate": 6.773868771679991e-05, "loss": 0.0174, "step": 124600 }, { "epoch": 79.42001274697259, "grad_norm": 0.5579381585121155, "learning_rate": 6.77335345583246e-05, "loss": 0.0281, "step": 124610 }, { "epoch": 79.4263862332696, "grad_norm": 0.4031221568584442, "learning_rate": 6.772838118437013e-05, "loss": 0.0256, "step": 124620 }, { "epoch": 79.4327597195666, "grad_norm": 0.4691297113895416, "learning_rate": 6.77232275949991e-05, "loss": 0.0131, "step": 124630 }, { "epoch": 79.43913320586361, "grad_norm": 0.5654774308204651, "learning_rate": 6.771807379027408e-05, "loss": 0.0241, "step": 124640 }, { "epoch": 79.4455066921606, "grad_norm": 0.5557680726051331, "learning_rate": 6.771291977025778e-05, "loss": 0.016, "step": 124650 }, { "epoch": 79.45188017845761, "grad_norm": 0.22522112727165222, "learning_rate": 6.770776553501275e-05, "loss": 0.0174, "step": 124660 }, { "epoch": 79.45825366475462, "grad_norm": 0.23699238896369934, "learning_rate": 6.770261108460166e-05, "loss": 0.0118, "step": 124670 }, { "epoch": 79.46462715105163, "grad_norm": 0.25070518255233765, "learning_rate": 6.769745641908712e-05, "loss": 0.013, "step": 124680 }, { "epoch": 79.47100063734862, "grad_norm": 0.3604833781719208, "learning_rate": 6.769230153853179e-05, "loss": 0.0094, "step": 124690 }, { "epoch": 79.47737412364563, "grad_norm": 0.47783318161964417, "learning_rate": 6.768714644299829e-05, "loss": 0.0186, "step": 124700 }, { "epoch": 79.48374760994264, "grad_norm": 0.30699992179870605, "learning_rate": 6.768199113254925e-05, "loss": 0.0152, "step": 124710 }, { "epoch": 79.49012109623965, "grad_norm": 0.32178446650505066, "learning_rate": 6.767683560724733e-05, "loss": 0.011, "step": 124720 }, { "epoch": 79.49649458253664, "grad_norm": 0.4799908399581909, "learning_rate": 6.767167986715516e-05, "loss": 0.0167, "step": 124730 }, { "epoch": 79.50286806883365, "grad_norm": 0.16650621592998505, "learning_rate": 6.766652391233542e-05, "loss": 0.014, "step": 124740 }, { "epoch": 79.50924155513066, "grad_norm": 0.4994553029537201, "learning_rate": 6.766136774285071e-05, "loss": 0.0168, "step": 124750 }, { "epoch": 79.51561504142767, "grad_norm": 0.4704348146915436, "learning_rate": 6.765621135876372e-05, "loss": 0.0075, "step": 124760 }, { "epoch": 79.52198852772466, "grad_norm": 0.4244696795940399, "learning_rate": 6.765105476013709e-05, "loss": 0.0192, "step": 124770 }, { "epoch": 79.52836201402167, "grad_norm": 0.5259035229682922, "learning_rate": 6.764589794703348e-05, "loss": 0.0134, "step": 124780 }, { "epoch": 79.53473550031867, "grad_norm": 0.25295814871788025, "learning_rate": 6.764074091951554e-05, "loss": 0.0142, "step": 124790 }, { "epoch": 79.54110898661568, "grad_norm": 0.5579010844230652, "learning_rate": 6.763558367764595e-05, "loss": 0.0226, "step": 124800 }, { "epoch": 79.54748247291268, "grad_norm": 0.4375065267086029, "learning_rate": 6.763042622148736e-05, "loss": 0.0121, "step": 124810 }, { "epoch": 79.55385595920968, "grad_norm": 0.2199287712574005, "learning_rate": 6.762526855110245e-05, "loss": 0.0158, "step": 124820 }, { "epoch": 79.56022944550669, "grad_norm": 0.3574201464653015, "learning_rate": 6.762011066655389e-05, "loss": 0.0119, "step": 124830 }, { "epoch": 79.5666029318037, "grad_norm": 0.32563516497612, "learning_rate": 6.761495256790435e-05, "loss": 0.0131, "step": 124840 }, { "epoch": 79.5729764181007, "grad_norm": 0.3432236313819885, "learning_rate": 6.76097942552165e-05, "loss": 0.0118, "step": 124850 }, { "epoch": 79.5793499043977, "grad_norm": 0.17910881340503693, "learning_rate": 6.760463572855302e-05, "loss": 0.0115, "step": 124860 }, { "epoch": 79.58572339069471, "grad_norm": 0.47910794615745544, "learning_rate": 6.759947698797661e-05, "loss": 0.0154, "step": 124870 }, { "epoch": 79.59209687699172, "grad_norm": 0.6364296078681946, "learning_rate": 6.759431803354992e-05, "loss": 0.0167, "step": 124880 }, { "epoch": 79.59847036328873, "grad_norm": 0.3372046649456024, "learning_rate": 6.758915886533566e-05, "loss": 0.018, "step": 124890 }, { "epoch": 79.60484384958572, "grad_norm": 0.5765656232833862, "learning_rate": 6.758399948339653e-05, "loss": 0.0203, "step": 124900 }, { "epoch": 79.61121733588273, "grad_norm": 0.5448253154754639, "learning_rate": 6.757883988779517e-05, "loss": 0.0173, "step": 124910 }, { "epoch": 79.61759082217974, "grad_norm": 0.470099538564682, "learning_rate": 6.757368007859434e-05, "loss": 0.0139, "step": 124920 }, { "epoch": 79.62396430847674, "grad_norm": 0.5501277446746826, "learning_rate": 6.756852005585668e-05, "loss": 0.0129, "step": 124930 }, { "epoch": 79.63033779477374, "grad_norm": 0.1706847995519638, "learning_rate": 6.756335981964492e-05, "loss": 0.0101, "step": 124940 }, { "epoch": 79.63671128107075, "grad_norm": 0.2411523461341858, "learning_rate": 6.755819937002177e-05, "loss": 0.0117, "step": 124950 }, { "epoch": 79.64308476736775, "grad_norm": 0.4624946415424347, "learning_rate": 6.75530387070499e-05, "loss": 0.0284, "step": 124960 }, { "epoch": 79.64945825366476, "grad_norm": 0.20828039944171906, "learning_rate": 6.754787783079206e-05, "loss": 0.0135, "step": 124970 }, { "epoch": 79.65583173996176, "grad_norm": 0.45076823234558105, "learning_rate": 6.754271674131091e-05, "loss": 0.016, "step": 124980 }, { "epoch": 79.66220522625876, "grad_norm": 0.19420182704925537, "learning_rate": 6.753755543866922e-05, "loss": 0.0203, "step": 124990 }, { "epoch": 79.66857871255577, "grad_norm": 0.323278546333313, "learning_rate": 6.753239392292965e-05, "loss": 0.0142, "step": 125000 }, { "epoch": 79.67495219885278, "grad_norm": 0.19359712302684784, "learning_rate": 6.752723219415494e-05, "loss": 0.0169, "step": 125010 }, { "epoch": 79.68132568514977, "grad_norm": 0.3531534671783447, "learning_rate": 6.752207025240782e-05, "loss": 0.016, "step": 125020 }, { "epoch": 79.68769917144678, "grad_norm": 0.44073188304901123, "learning_rate": 6.7516908097751e-05, "loss": 0.0158, "step": 125030 }, { "epoch": 79.69407265774379, "grad_norm": 0.42768749594688416, "learning_rate": 6.75117457302472e-05, "loss": 0.0145, "step": 125040 }, { "epoch": 79.7004461440408, "grad_norm": 0.3125505745410919, "learning_rate": 6.750658314995916e-05, "loss": 0.0104, "step": 125050 }, { "epoch": 79.70681963033779, "grad_norm": 0.27193912863731384, "learning_rate": 6.750142035694962e-05, "loss": 0.0149, "step": 125060 }, { "epoch": 79.7131931166348, "grad_norm": 0.16986165940761566, "learning_rate": 6.74962573512813e-05, "loss": 0.0141, "step": 125070 }, { "epoch": 79.7195666029318, "grad_norm": 0.49531176686286926, "learning_rate": 6.749109413301691e-05, "loss": 0.0151, "step": 125080 }, { "epoch": 79.72594008922881, "grad_norm": 0.4428061544895172, "learning_rate": 6.748593070221923e-05, "loss": 0.0185, "step": 125090 }, { "epoch": 79.73231357552581, "grad_norm": 0.38770952820777893, "learning_rate": 6.748076705895097e-05, "loss": 0.0186, "step": 125100 }, { "epoch": 79.73868706182282, "grad_norm": 0.5201800465583801, "learning_rate": 6.74756032032749e-05, "loss": 0.0155, "step": 125110 }, { "epoch": 79.74506054811982, "grad_norm": 0.6536782383918762, "learning_rate": 6.747043913525373e-05, "loss": 0.013, "step": 125120 }, { "epoch": 79.75143403441683, "grad_norm": 0.32139503955841064, "learning_rate": 6.746527485495025e-05, "loss": 0.0166, "step": 125130 }, { "epoch": 79.75780752071383, "grad_norm": 0.2603851556777954, "learning_rate": 6.746011036242717e-05, "loss": 0.0155, "step": 125140 }, { "epoch": 79.76418100701083, "grad_norm": 0.3858010470867157, "learning_rate": 6.745494565774728e-05, "loss": 0.0188, "step": 125150 }, { "epoch": 79.77055449330784, "grad_norm": 0.33930355310440063, "learning_rate": 6.744978074097332e-05, "loss": 0.0153, "step": 125160 }, { "epoch": 79.77692797960485, "grad_norm": 0.5987381935119629, "learning_rate": 6.744461561216805e-05, "loss": 0.0178, "step": 125170 }, { "epoch": 79.78330146590184, "grad_norm": 0.3014277517795563, "learning_rate": 6.743945027139423e-05, "loss": 0.0196, "step": 125180 }, { "epoch": 79.78967495219885, "grad_norm": 0.2712571322917938, "learning_rate": 6.74342847187146e-05, "loss": 0.0219, "step": 125190 }, { "epoch": 79.79604843849586, "grad_norm": 0.5042886734008789, "learning_rate": 6.742911895419198e-05, "loss": 0.0242, "step": 125200 }, { "epoch": 79.80242192479287, "grad_norm": 0.3758966624736786, "learning_rate": 6.74239529778891e-05, "loss": 0.0096, "step": 125210 }, { "epoch": 79.80879541108986, "grad_norm": 0.4786214530467987, "learning_rate": 6.741878678986873e-05, "loss": 0.0123, "step": 125220 }, { "epoch": 79.81516889738687, "grad_norm": 0.35215839743614197, "learning_rate": 6.741362039019367e-05, "loss": 0.0164, "step": 125230 }, { "epoch": 79.82154238368388, "grad_norm": 0.3008069694042206, "learning_rate": 6.740845377892667e-05, "loss": 0.0175, "step": 125240 }, { "epoch": 79.82791586998088, "grad_norm": 0.2933357357978821, "learning_rate": 6.740328695613053e-05, "loss": 0.0114, "step": 125250 }, { "epoch": 79.83428935627788, "grad_norm": 0.3213497996330261, "learning_rate": 6.7398119921868e-05, "loss": 0.0153, "step": 125260 }, { "epoch": 79.84066284257489, "grad_norm": 0.3545714318752289, "learning_rate": 6.739295267620192e-05, "loss": 0.0066, "step": 125270 }, { "epoch": 79.8470363288719, "grad_norm": 0.2940559983253479, "learning_rate": 6.738778521919501e-05, "loss": 0.0224, "step": 125280 }, { "epoch": 79.8534098151689, "grad_norm": 0.6588410139083862, "learning_rate": 6.738261755091012e-05, "loss": 0.0135, "step": 125290 }, { "epoch": 79.8597833014659, "grad_norm": 0.38771483302116394, "learning_rate": 6.737744967141e-05, "loss": 0.0264, "step": 125300 }, { "epoch": 79.8661567877629, "grad_norm": 0.39279651641845703, "learning_rate": 6.737228158075747e-05, "loss": 0.0251, "step": 125310 }, { "epoch": 79.87253027405991, "grad_norm": 0.5684505105018616, "learning_rate": 6.73671132790153e-05, "loss": 0.0175, "step": 125320 }, { "epoch": 79.87890376035692, "grad_norm": 0.2768491804599762, "learning_rate": 6.736194476624632e-05, "loss": 0.012, "step": 125330 }, { "epoch": 79.88527724665391, "grad_norm": 0.44127583503723145, "learning_rate": 6.735677604251332e-05, "loss": 0.0116, "step": 125340 }, { "epoch": 79.89165073295092, "grad_norm": 0.581598162651062, "learning_rate": 6.735160710787909e-05, "loss": 0.0183, "step": 125350 }, { "epoch": 79.89802421924793, "grad_norm": 0.25878554582595825, "learning_rate": 6.734643796240647e-05, "loss": 0.0142, "step": 125360 }, { "epoch": 79.90439770554494, "grad_norm": 0.5962002873420715, "learning_rate": 6.734126860615824e-05, "loss": 0.0161, "step": 125370 }, { "epoch": 79.91077119184193, "grad_norm": 0.23590587079524994, "learning_rate": 6.733609903919722e-05, "loss": 0.0146, "step": 125380 }, { "epoch": 79.91714467813894, "grad_norm": 0.3890500068664551, "learning_rate": 6.733092926158623e-05, "loss": 0.0146, "step": 125390 }, { "epoch": 79.92351816443595, "grad_norm": 0.3760499358177185, "learning_rate": 6.732575927338808e-05, "loss": 0.0113, "step": 125400 }, { "epoch": 79.92989165073296, "grad_norm": 0.32836225628852844, "learning_rate": 6.73205890746656e-05, "loss": 0.0134, "step": 125410 }, { "epoch": 79.93626513702995, "grad_norm": 0.37851718068122864, "learning_rate": 6.731541866548164e-05, "loss": 0.0137, "step": 125420 }, { "epoch": 79.94263862332696, "grad_norm": 0.4003143906593323, "learning_rate": 6.731024804589896e-05, "loss": 0.0166, "step": 125430 }, { "epoch": 79.94901210962396, "grad_norm": 0.3304022252559662, "learning_rate": 6.730507721598043e-05, "loss": 0.0153, "step": 125440 }, { "epoch": 79.95538559592097, "grad_norm": 0.3601759374141693, "learning_rate": 6.729990617578889e-05, "loss": 0.0212, "step": 125450 }, { "epoch": 79.96175908221797, "grad_norm": 0.5092607140541077, "learning_rate": 6.729473492538713e-05, "loss": 0.0302, "step": 125460 }, { "epoch": 79.96813256851497, "grad_norm": 0.346640944480896, "learning_rate": 6.728956346483802e-05, "loss": 0.0115, "step": 125470 }, { "epoch": 79.97450605481198, "grad_norm": 0.5884768962860107, "learning_rate": 6.72843917942044e-05, "loss": 0.0203, "step": 125480 }, { "epoch": 79.98087954110899, "grad_norm": 0.39155152440071106, "learning_rate": 6.727921991354909e-05, "loss": 0.0097, "step": 125490 }, { "epoch": 79.98725302740598, "grad_norm": 0.48818764090538025, "learning_rate": 6.727404782293494e-05, "loss": 0.0171, "step": 125500 }, { "epoch": 79.99362651370299, "grad_norm": 0.4754868447780609, "learning_rate": 6.726887552242482e-05, "loss": 0.0133, "step": 125510 }, { "epoch": 80.0, "grad_norm": 0.4132716953754425, "learning_rate": 6.726370301208154e-05, "loss": 0.0129, "step": 125520 }, { "epoch": 80.00637348629701, "grad_norm": 0.26044467091560364, "learning_rate": 6.725853029196797e-05, "loss": 0.0196, "step": 125530 }, { "epoch": 80.01274697259402, "grad_norm": 0.34793469309806824, "learning_rate": 6.725335736214697e-05, "loss": 0.0114, "step": 125540 }, { "epoch": 80.01912045889101, "grad_norm": 0.26328396797180176, "learning_rate": 6.724818422268137e-05, "loss": 0.0157, "step": 125550 }, { "epoch": 80.02549394518802, "grad_norm": 0.5081735253334045, "learning_rate": 6.724301087363407e-05, "loss": 0.0197, "step": 125560 }, { "epoch": 80.03186743148503, "grad_norm": 0.39210349321365356, "learning_rate": 6.72378373150679e-05, "loss": 0.0155, "step": 125570 }, { "epoch": 80.03824091778203, "grad_norm": 0.3234591484069824, "learning_rate": 6.723266354704574e-05, "loss": 0.0152, "step": 125580 }, { "epoch": 80.04461440407903, "grad_norm": 0.46330076456069946, "learning_rate": 6.722748956963043e-05, "loss": 0.0154, "step": 125590 }, { "epoch": 80.05098789037604, "grad_norm": 0.5016193389892578, "learning_rate": 6.722231538288486e-05, "loss": 0.0155, "step": 125600 }, { "epoch": 80.05736137667304, "grad_norm": 0.3275073766708374, "learning_rate": 6.721714098687191e-05, "loss": 0.0133, "step": 125610 }, { "epoch": 80.06373486297005, "grad_norm": 0.7065107226371765, "learning_rate": 6.721196638165443e-05, "loss": 0.0225, "step": 125620 }, { "epoch": 80.07010834926704, "grad_norm": 0.3828933835029602, "learning_rate": 6.720679156729532e-05, "loss": 0.0146, "step": 125630 }, { "epoch": 80.07648183556405, "grad_norm": 0.4238526225090027, "learning_rate": 6.720161654385743e-05, "loss": 0.0268, "step": 125640 }, { "epoch": 80.08285532186106, "grad_norm": 0.2987675964832306, "learning_rate": 6.719644131140367e-05, "loss": 0.0224, "step": 125650 }, { "epoch": 80.08922880815807, "grad_norm": 0.42434296011924744, "learning_rate": 6.71912658699969e-05, "loss": 0.02, "step": 125660 }, { "epoch": 80.09560229445506, "grad_norm": 0.24993960559368134, "learning_rate": 6.718609021970003e-05, "loss": 0.0143, "step": 125670 }, { "epoch": 80.10197578075207, "grad_norm": 0.4516327977180481, "learning_rate": 6.718091436057593e-05, "loss": 0.0098, "step": 125680 }, { "epoch": 80.10834926704908, "grad_norm": 0.3569120764732361, "learning_rate": 6.717573829268749e-05, "loss": 0.0135, "step": 125690 }, { "epoch": 80.11472275334609, "grad_norm": 0.4617933928966522, "learning_rate": 6.717056201609764e-05, "loss": 0.0165, "step": 125700 }, { "epoch": 80.12109623964308, "grad_norm": 0.6558597683906555, "learning_rate": 6.716538553086923e-05, "loss": 0.0143, "step": 125710 }, { "epoch": 80.12746972594009, "grad_norm": 0.4239666759967804, "learning_rate": 6.716020883706518e-05, "loss": 0.0209, "step": 125720 }, { "epoch": 80.1338432122371, "grad_norm": 0.2596076428890228, "learning_rate": 6.715503193474839e-05, "loss": 0.0194, "step": 125730 }, { "epoch": 80.1402166985341, "grad_norm": 0.26778683066368103, "learning_rate": 6.714985482398179e-05, "loss": 0.0135, "step": 125740 }, { "epoch": 80.1465901848311, "grad_norm": 0.2040417641401291, "learning_rate": 6.714467750482823e-05, "loss": 0.0097, "step": 125750 }, { "epoch": 80.1529636711281, "grad_norm": 0.45506352186203003, "learning_rate": 6.713949997735067e-05, "loss": 0.0215, "step": 125760 }, { "epoch": 80.15933715742511, "grad_norm": 0.38872867822647095, "learning_rate": 6.713432224161199e-05, "loss": 0.0177, "step": 125770 }, { "epoch": 80.16571064372212, "grad_norm": 0.18489481508731842, "learning_rate": 6.712914429767511e-05, "loss": 0.0125, "step": 125780 }, { "epoch": 80.17208413001912, "grad_norm": 0.3248577415943146, "learning_rate": 6.712396614560297e-05, "loss": 0.0131, "step": 125790 }, { "epoch": 80.17845761631612, "grad_norm": 0.43897250294685364, "learning_rate": 6.711878778545848e-05, "loss": 0.0237, "step": 125800 }, { "epoch": 80.18483110261313, "grad_norm": 0.453967422246933, "learning_rate": 6.711360921730454e-05, "loss": 0.0141, "step": 125810 }, { "epoch": 80.19120458891014, "grad_norm": 0.2577480375766754, "learning_rate": 6.710843044120409e-05, "loss": 0.0145, "step": 125820 }, { "epoch": 80.19757807520713, "grad_norm": 0.32862624526023865, "learning_rate": 6.710325145722007e-05, "loss": 0.014, "step": 125830 }, { "epoch": 80.20395156150414, "grad_norm": 0.508571982383728, "learning_rate": 6.70980722654154e-05, "loss": 0.0178, "step": 125840 }, { "epoch": 80.21032504780115, "grad_norm": 0.37741729617118835, "learning_rate": 6.709289286585299e-05, "loss": 0.0113, "step": 125850 }, { "epoch": 80.21669853409816, "grad_norm": 0.688094973564148, "learning_rate": 6.708771325859581e-05, "loss": 0.0172, "step": 125860 }, { "epoch": 80.22307202039515, "grad_norm": 0.48461732268333435, "learning_rate": 6.708253344370676e-05, "loss": 0.0126, "step": 125870 }, { "epoch": 80.22944550669216, "grad_norm": 0.316806823015213, "learning_rate": 6.707735342124882e-05, "loss": 0.0101, "step": 125880 }, { "epoch": 80.23581899298917, "grad_norm": 0.4391910433769226, "learning_rate": 6.707217319128489e-05, "loss": 0.0203, "step": 125890 }, { "epoch": 80.24219247928617, "grad_norm": 0.3578508496284485, "learning_rate": 6.706699275387796e-05, "loss": 0.0309, "step": 125900 }, { "epoch": 80.24856596558317, "grad_norm": 0.35560983419418335, "learning_rate": 6.706181210909095e-05, "loss": 0.0139, "step": 125910 }, { "epoch": 80.25493945188018, "grad_norm": 0.3086462914943695, "learning_rate": 6.705663125698682e-05, "loss": 0.0159, "step": 125920 }, { "epoch": 80.26131293817718, "grad_norm": 0.5664320588111877, "learning_rate": 6.705145019762849e-05, "loss": 0.019, "step": 125930 }, { "epoch": 80.26768642447419, "grad_norm": 0.3132663667201996, "learning_rate": 6.704626893107897e-05, "loss": 0.0095, "step": 125940 }, { "epoch": 80.27405991077119, "grad_norm": 0.48230257630348206, "learning_rate": 6.704108745740117e-05, "loss": 0.0218, "step": 125950 }, { "epoch": 80.2804333970682, "grad_norm": 0.6967734694480896, "learning_rate": 6.703590577665808e-05, "loss": 0.02, "step": 125960 }, { "epoch": 80.2868068833652, "grad_norm": 0.391135573387146, "learning_rate": 6.703072388891263e-05, "loss": 0.0169, "step": 125970 }, { "epoch": 80.29318036966221, "grad_norm": 0.41381028294563293, "learning_rate": 6.702554179422783e-05, "loss": 0.0249, "step": 125980 }, { "epoch": 80.2995538559592, "grad_norm": 1.3740564584732056, "learning_rate": 6.702035949266658e-05, "loss": 0.027, "step": 125990 }, { "epoch": 80.30592734225621, "grad_norm": 0.3906605541706085, "learning_rate": 6.701517698429194e-05, "loss": 0.0208, "step": 126000 }, { "epoch": 80.31230082855322, "grad_norm": 0.3945595622062683, "learning_rate": 6.70099942691668e-05, "loss": 0.0189, "step": 126010 }, { "epoch": 80.31867431485023, "grad_norm": 0.6461365222930908, "learning_rate": 6.700481134735421e-05, "loss": 0.0234, "step": 126020 }, { "epoch": 80.32504780114722, "grad_norm": 0.6638172268867493, "learning_rate": 6.699962821891706e-05, "loss": 0.0148, "step": 126030 }, { "epoch": 80.33142128744423, "grad_norm": 0.5848672389984131, "learning_rate": 6.699444488391842e-05, "loss": 0.0255, "step": 126040 }, { "epoch": 80.33779477374124, "grad_norm": 0.3781450092792511, "learning_rate": 6.698926134242119e-05, "loss": 0.0155, "step": 126050 }, { "epoch": 80.34416826003824, "grad_norm": 0.6222744584083557, "learning_rate": 6.698407759448843e-05, "loss": 0.0187, "step": 126060 }, { "epoch": 80.35054174633524, "grad_norm": 0.767084538936615, "learning_rate": 6.697889364018306e-05, "loss": 0.019, "step": 126070 }, { "epoch": 80.35691523263225, "grad_norm": 0.43843960762023926, "learning_rate": 6.697370947956812e-05, "loss": 0.0177, "step": 126080 }, { "epoch": 80.36328871892925, "grad_norm": 0.4191601872444153, "learning_rate": 6.696852511270659e-05, "loss": 0.0128, "step": 126090 }, { "epoch": 80.36966220522626, "grad_norm": 0.3066982328891754, "learning_rate": 6.696334053966144e-05, "loss": 0.01, "step": 126100 }, { "epoch": 80.37603569152326, "grad_norm": 0.2755712568759918, "learning_rate": 6.69581557604957e-05, "loss": 0.0239, "step": 126110 }, { "epoch": 80.38240917782026, "grad_norm": 0.34613803029060364, "learning_rate": 6.695297077527236e-05, "loss": 0.0128, "step": 126120 }, { "epoch": 80.38878266411727, "grad_norm": 0.5091246366500854, "learning_rate": 6.694778558405441e-05, "loss": 0.0203, "step": 126130 }, { "epoch": 80.39515615041428, "grad_norm": 0.2727690637111664, "learning_rate": 6.694260018690485e-05, "loss": 0.0175, "step": 126140 }, { "epoch": 80.40152963671127, "grad_norm": 0.5497962832450867, "learning_rate": 6.693741458388674e-05, "loss": 0.0232, "step": 126150 }, { "epoch": 80.40790312300828, "grad_norm": 0.35026729106903076, "learning_rate": 6.693222877506302e-05, "loss": 0.0179, "step": 126160 }, { "epoch": 80.41427660930529, "grad_norm": 0.23851560056209564, "learning_rate": 6.692704276049674e-05, "loss": 0.0227, "step": 126170 }, { "epoch": 80.4206500956023, "grad_norm": 0.5691242218017578, "learning_rate": 6.692185654025092e-05, "loss": 0.0217, "step": 126180 }, { "epoch": 80.4270235818993, "grad_norm": 0.501129686832428, "learning_rate": 6.691667011438855e-05, "loss": 0.0092, "step": 126190 }, { "epoch": 80.4333970681963, "grad_norm": 0.38817015290260315, "learning_rate": 6.691148348297268e-05, "loss": 0.0152, "step": 126200 }, { "epoch": 80.43977055449331, "grad_norm": 0.6953297853469849, "learning_rate": 6.69062966460663e-05, "loss": 0.0101, "step": 126210 }, { "epoch": 80.44614404079032, "grad_norm": 0.4346459209918976, "learning_rate": 6.690110960373246e-05, "loss": 0.0115, "step": 126220 }, { "epoch": 80.45251752708732, "grad_norm": 0.4636843204498291, "learning_rate": 6.689592235603419e-05, "loss": 0.0182, "step": 126230 }, { "epoch": 80.45889101338432, "grad_norm": 0.3680875301361084, "learning_rate": 6.68907349030345e-05, "loss": 0.0221, "step": 126240 }, { "epoch": 80.46526449968133, "grad_norm": 0.4416764974594116, "learning_rate": 6.688554724479646e-05, "loss": 0.0242, "step": 126250 }, { "epoch": 80.47163798597833, "grad_norm": 0.37517237663269043, "learning_rate": 6.688035938138303e-05, "loss": 0.0116, "step": 126260 }, { "epoch": 80.47801147227534, "grad_norm": 0.35759076476097107, "learning_rate": 6.687517131285735e-05, "loss": 0.017, "step": 126270 }, { "epoch": 80.48438495857233, "grad_norm": 0.3682042062282562, "learning_rate": 6.686998303928237e-05, "loss": 0.0204, "step": 126280 }, { "epoch": 80.49075844486934, "grad_norm": 0.6756048798561096, "learning_rate": 6.686479456072118e-05, "loss": 0.0194, "step": 126290 }, { "epoch": 80.49713193116635, "grad_norm": 0.6227577328681946, "learning_rate": 6.685960587723681e-05, "loss": 0.0241, "step": 126300 }, { "epoch": 80.50350541746336, "grad_norm": 0.4742136299610138, "learning_rate": 6.685441698889232e-05, "loss": 0.0211, "step": 126310 }, { "epoch": 80.50987890376035, "grad_norm": 0.2923162579536438, "learning_rate": 6.684922789575074e-05, "loss": 0.0222, "step": 126320 }, { "epoch": 80.51625239005736, "grad_norm": 0.5508079528808594, "learning_rate": 6.684403859787514e-05, "loss": 0.0159, "step": 126330 }, { "epoch": 80.52262587635437, "grad_norm": 0.3177070617675781, "learning_rate": 6.683884909532857e-05, "loss": 0.0129, "step": 126340 }, { "epoch": 80.52899936265138, "grad_norm": 0.5068110227584839, "learning_rate": 6.683365938817406e-05, "loss": 0.0128, "step": 126350 }, { "epoch": 80.53537284894837, "grad_norm": 0.33872923254966736, "learning_rate": 6.682846947647472e-05, "loss": 0.0167, "step": 126360 }, { "epoch": 80.54174633524538, "grad_norm": 0.23440749943256378, "learning_rate": 6.682327936029357e-05, "loss": 0.0156, "step": 126370 }, { "epoch": 80.54811982154239, "grad_norm": 0.16772273182868958, "learning_rate": 6.68180890396937e-05, "loss": 0.0114, "step": 126380 }, { "epoch": 80.5544933078394, "grad_norm": 0.7510534524917603, "learning_rate": 6.681289851473818e-05, "loss": 0.0197, "step": 126390 }, { "epoch": 80.56086679413639, "grad_norm": 0.3557160496711731, "learning_rate": 6.680770778549006e-05, "loss": 0.0185, "step": 126400 }, { "epoch": 80.5672402804334, "grad_norm": 0.3356439173221588, "learning_rate": 6.680251685201242e-05, "loss": 0.0247, "step": 126410 }, { "epoch": 80.5736137667304, "grad_norm": 0.6048323512077332, "learning_rate": 6.679732571436834e-05, "loss": 0.0279, "step": 126420 }, { "epoch": 80.57998725302741, "grad_norm": 0.3814190924167633, "learning_rate": 6.679213437262088e-05, "loss": 0.0211, "step": 126430 }, { "epoch": 80.5863607393244, "grad_norm": 0.35746559500694275, "learning_rate": 6.678694282683315e-05, "loss": 0.0139, "step": 126440 }, { "epoch": 80.59273422562141, "grad_norm": 0.38779416680336, "learning_rate": 6.678175107706821e-05, "loss": 0.0123, "step": 126450 }, { "epoch": 80.59910771191842, "grad_norm": 0.4195821285247803, "learning_rate": 6.677655912338913e-05, "loss": 0.0175, "step": 126460 }, { "epoch": 80.60548119821543, "grad_norm": 0.2711597979068756, "learning_rate": 6.677136696585906e-05, "loss": 0.0228, "step": 126470 }, { "epoch": 80.61185468451242, "grad_norm": 0.5140336155891418, "learning_rate": 6.676617460454102e-05, "loss": 0.0135, "step": 126480 }, { "epoch": 80.61822817080943, "grad_norm": 0.3927972912788391, "learning_rate": 6.676098203949812e-05, "loss": 0.0099, "step": 126490 }, { "epoch": 80.62460165710644, "grad_norm": 0.12783552706241608, "learning_rate": 6.675578927079348e-05, "loss": 0.0162, "step": 126500 }, { "epoch": 80.63097514340345, "grad_norm": 0.2168223112821579, "learning_rate": 6.675059629849017e-05, "loss": 0.015, "step": 126510 }, { "epoch": 80.63734862970044, "grad_norm": 0.31361573934555054, "learning_rate": 6.674540312265131e-05, "loss": 0.0141, "step": 126520 }, { "epoch": 80.64372211599745, "grad_norm": 0.6191925406455994, "learning_rate": 6.674020974333998e-05, "loss": 0.0162, "step": 126530 }, { "epoch": 80.65009560229446, "grad_norm": 0.3114016652107239, "learning_rate": 6.67350161606193e-05, "loss": 0.0139, "step": 126540 }, { "epoch": 80.65646908859146, "grad_norm": 0.34173473715782166, "learning_rate": 6.672982237455237e-05, "loss": 0.0117, "step": 126550 }, { "epoch": 80.66284257488846, "grad_norm": 0.488564133644104, "learning_rate": 6.672462838520232e-05, "loss": 0.0208, "step": 126560 }, { "epoch": 80.66921606118547, "grad_norm": 0.2971140444278717, "learning_rate": 6.671943419263223e-05, "loss": 0.0212, "step": 126570 }, { "epoch": 80.67558954748247, "grad_norm": 0.40928199887275696, "learning_rate": 6.671423979690523e-05, "loss": 0.0099, "step": 126580 }, { "epoch": 80.68196303377948, "grad_norm": 0.40064939856529236, "learning_rate": 6.670904519808443e-05, "loss": 0.0216, "step": 126590 }, { "epoch": 80.68833652007648, "grad_norm": 0.5325694680213928, "learning_rate": 6.670385039623295e-05, "loss": 0.0105, "step": 126600 }, { "epoch": 80.69471000637348, "grad_norm": 0.15836618840694427, "learning_rate": 6.669865539141394e-05, "loss": 0.0198, "step": 126610 }, { "epoch": 80.70108349267049, "grad_norm": 0.5138508677482605, "learning_rate": 6.669346018369049e-05, "loss": 0.0148, "step": 126620 }, { "epoch": 80.7074569789675, "grad_norm": 0.3755810856819153, "learning_rate": 6.668826477312574e-05, "loss": 0.0178, "step": 126630 }, { "epoch": 80.7138304652645, "grad_norm": 0.22888292372226715, "learning_rate": 6.668306915978281e-05, "loss": 0.0164, "step": 126640 }, { "epoch": 80.7202039515615, "grad_norm": 0.2579191327095032, "learning_rate": 6.667787334372484e-05, "loss": 0.0155, "step": 126650 }, { "epoch": 80.72657743785851, "grad_norm": 0.265672892332077, "learning_rate": 6.667267732501496e-05, "loss": 0.0087, "step": 126660 }, { "epoch": 80.73295092415552, "grad_norm": 0.8445600867271423, "learning_rate": 6.666748110371632e-05, "loss": 0.0274, "step": 126670 }, { "epoch": 80.73932441045251, "grad_norm": 0.6260095238685608, "learning_rate": 6.666228467989203e-05, "loss": 0.0168, "step": 126680 }, { "epoch": 80.74569789674952, "grad_norm": 0.25650957226753235, "learning_rate": 6.665708805360527e-05, "loss": 0.0265, "step": 126690 }, { "epoch": 80.75207138304653, "grad_norm": 0.44495564699172974, "learning_rate": 6.665189122491915e-05, "loss": 0.0146, "step": 126700 }, { "epoch": 80.75844486934353, "grad_norm": 0.3959968388080597, "learning_rate": 6.664669419389684e-05, "loss": 0.0196, "step": 126710 }, { "epoch": 80.76481835564053, "grad_norm": 0.7253341674804688, "learning_rate": 6.664149696060146e-05, "loss": 0.0148, "step": 126720 }, { "epoch": 80.77119184193754, "grad_norm": 0.3937951326370239, "learning_rate": 6.663629952509619e-05, "loss": 0.0136, "step": 126730 }, { "epoch": 80.77756532823454, "grad_norm": 0.3259277939796448, "learning_rate": 6.663110188744417e-05, "loss": 0.0086, "step": 126740 }, { "epoch": 80.78393881453155, "grad_norm": 0.3008105754852295, "learning_rate": 6.662590404770857e-05, "loss": 0.0144, "step": 126750 }, { "epoch": 80.79031230082855, "grad_norm": 0.29666775465011597, "learning_rate": 6.662070600595252e-05, "loss": 0.0217, "step": 126760 }, { "epoch": 80.79668578712555, "grad_norm": 0.6915923953056335, "learning_rate": 6.661550776223922e-05, "loss": 0.0254, "step": 126770 }, { "epoch": 80.80305927342256, "grad_norm": 0.3394468426704407, "learning_rate": 6.66103093166318e-05, "loss": 0.0125, "step": 126780 }, { "epoch": 80.80943275971957, "grad_norm": 0.3648572862148285, "learning_rate": 6.660511066919344e-05, "loss": 0.0179, "step": 126790 }, { "epoch": 80.81580624601658, "grad_norm": 0.31661874055862427, "learning_rate": 6.659991181998733e-05, "loss": 0.0172, "step": 126800 }, { "epoch": 80.82217973231357, "grad_norm": 0.18575285375118256, "learning_rate": 6.659471276907658e-05, "loss": 0.0169, "step": 126810 }, { "epoch": 80.82855321861058, "grad_norm": 0.19795109331607819, "learning_rate": 6.658951351652444e-05, "loss": 0.0208, "step": 126820 }, { "epoch": 80.83492670490759, "grad_norm": 0.41461706161499023, "learning_rate": 6.658431406239402e-05, "loss": 0.0121, "step": 126830 }, { "epoch": 80.8413001912046, "grad_norm": 0.1891605705022812, "learning_rate": 6.657911440674855e-05, "loss": 0.0129, "step": 126840 }, { "epoch": 80.84767367750159, "grad_norm": 0.2017437368631363, "learning_rate": 6.657391454965117e-05, "loss": 0.0155, "step": 126850 }, { "epoch": 80.8540471637986, "grad_norm": 0.2239898294210434, "learning_rate": 6.656871449116508e-05, "loss": 0.0291, "step": 126860 }, { "epoch": 80.8604206500956, "grad_norm": 0.48643994331359863, "learning_rate": 6.656351423135346e-05, "loss": 0.0204, "step": 126870 }, { "epoch": 80.86679413639261, "grad_norm": 0.5520294308662415, "learning_rate": 6.655831377027952e-05, "loss": 0.0216, "step": 126880 }, { "epoch": 80.8731676226896, "grad_norm": 0.6381688117980957, "learning_rate": 6.655311310800643e-05, "loss": 0.0215, "step": 126890 }, { "epoch": 80.87954110898661, "grad_norm": 0.29683762788772583, "learning_rate": 6.654791224459737e-05, "loss": 0.0204, "step": 126900 }, { "epoch": 80.88591459528362, "grad_norm": 0.5710203051567078, "learning_rate": 6.654271118011557e-05, "loss": 0.016, "step": 126910 }, { "epoch": 80.89228808158063, "grad_norm": 0.4914308786392212, "learning_rate": 6.65375099146242e-05, "loss": 0.0198, "step": 126920 }, { "epoch": 80.89866156787762, "grad_norm": 0.32553476095199585, "learning_rate": 6.653230844818647e-05, "loss": 0.0124, "step": 126930 }, { "epoch": 80.90503505417463, "grad_norm": 0.39785251021385193, "learning_rate": 6.652710678086558e-05, "loss": 0.0128, "step": 126940 }, { "epoch": 80.91140854047164, "grad_norm": 0.40677180886268616, "learning_rate": 6.652190491272476e-05, "loss": 0.0212, "step": 126950 }, { "epoch": 80.91778202676865, "grad_norm": 0.4438180923461914, "learning_rate": 6.651670284382718e-05, "loss": 0.0229, "step": 126960 }, { "epoch": 80.92415551306564, "grad_norm": 0.911753237247467, "learning_rate": 6.651150057423607e-05, "loss": 0.0095, "step": 126970 }, { "epoch": 80.93052899936265, "grad_norm": 0.6160798072814941, "learning_rate": 6.650629810401465e-05, "loss": 0.0226, "step": 126980 }, { "epoch": 80.93690248565966, "grad_norm": 0.4749008119106293, "learning_rate": 6.65010954332261e-05, "loss": 0.0172, "step": 126990 }, { "epoch": 80.94327597195667, "grad_norm": 0.29766616225242615, "learning_rate": 6.649589256193367e-05, "loss": 0.0189, "step": 127000 }, { "epoch": 80.94964945825366, "grad_norm": 0.47174158692359924, "learning_rate": 6.649068949020058e-05, "loss": 0.0127, "step": 127010 }, { "epoch": 80.95602294455067, "grad_norm": 0.32872432470321655, "learning_rate": 6.648548621809003e-05, "loss": 0.0171, "step": 127020 }, { "epoch": 80.96239643084768, "grad_norm": 0.28067752718925476, "learning_rate": 6.648028274566526e-05, "loss": 0.0123, "step": 127030 }, { "epoch": 80.96876991714468, "grad_norm": 0.23956497013568878, "learning_rate": 6.647507907298949e-05, "loss": 0.0196, "step": 127040 }, { "epoch": 80.97514340344168, "grad_norm": 0.29510581493377686, "learning_rate": 6.646987520012597e-05, "loss": 0.0127, "step": 127050 }, { "epoch": 80.98151688973869, "grad_norm": 0.23039419949054718, "learning_rate": 6.64646711271379e-05, "loss": 0.0203, "step": 127060 }, { "epoch": 80.9878903760357, "grad_norm": 0.253165602684021, "learning_rate": 6.645946685408856e-05, "loss": 0.0147, "step": 127070 }, { "epoch": 80.9942638623327, "grad_norm": 0.4904017746448517, "learning_rate": 6.645426238104113e-05, "loss": 0.0208, "step": 127080 }, { "epoch": 81.0006373486297, "grad_norm": 0.7988378405570984, "learning_rate": 6.644905770805889e-05, "loss": 0.0122, "step": 127090 }, { "epoch": 81.0070108349267, "grad_norm": 0.43837496638298035, "learning_rate": 6.644385283520504e-05, "loss": 0.0202, "step": 127100 }, { "epoch": 81.01338432122371, "grad_norm": 0.2410060167312622, "learning_rate": 6.64386477625429e-05, "loss": 0.0118, "step": 127110 }, { "epoch": 81.01975780752072, "grad_norm": 0.38906365633010864, "learning_rate": 6.643344249013562e-05, "loss": 0.0139, "step": 127120 }, { "epoch": 81.02613129381771, "grad_norm": 0.3276345729827881, "learning_rate": 6.642823701804653e-05, "loss": 0.0181, "step": 127130 }, { "epoch": 81.03250478011472, "grad_norm": 0.6254652142524719, "learning_rate": 6.642303134633885e-05, "loss": 0.0253, "step": 127140 }, { "epoch": 81.03887826641173, "grad_norm": 0.8940832614898682, "learning_rate": 6.641782547507581e-05, "loss": 0.0192, "step": 127150 }, { "epoch": 81.04525175270874, "grad_norm": 0.6184812784194946, "learning_rate": 6.641261940432071e-05, "loss": 0.016, "step": 127160 }, { "epoch": 81.05162523900573, "grad_norm": 0.47905582189559937, "learning_rate": 6.640741313413678e-05, "loss": 0.0138, "step": 127170 }, { "epoch": 81.05799872530274, "grad_norm": 0.4322684705257416, "learning_rate": 6.640220666458729e-05, "loss": 0.0142, "step": 127180 }, { "epoch": 81.06437221159975, "grad_norm": 0.3111843168735504, "learning_rate": 6.63969999957355e-05, "loss": 0.013, "step": 127190 }, { "epoch": 81.07074569789675, "grad_norm": 0.3068602383136749, "learning_rate": 6.639179312764469e-05, "loss": 0.0176, "step": 127200 }, { "epoch": 81.07711918419375, "grad_norm": 0.3148795962333679, "learning_rate": 6.638658606037811e-05, "loss": 0.0217, "step": 127210 }, { "epoch": 81.08349267049076, "grad_norm": 0.17243891954421997, "learning_rate": 6.638137879399903e-05, "loss": 0.0108, "step": 127220 }, { "epoch": 81.08986615678776, "grad_norm": 0.2453475147485733, "learning_rate": 6.637617132857075e-05, "loss": 0.021, "step": 127230 }, { "epoch": 81.09623964308477, "grad_norm": 0.15608802437782288, "learning_rate": 6.63709636641565e-05, "loss": 0.0111, "step": 127240 }, { "epoch": 81.10261312938177, "grad_norm": 0.42830178141593933, "learning_rate": 6.636575580081961e-05, "loss": 0.0148, "step": 127250 }, { "epoch": 81.10898661567877, "grad_norm": 0.21630507707595825, "learning_rate": 6.636054773862332e-05, "loss": 0.0137, "step": 127260 }, { "epoch": 81.11536010197578, "grad_norm": 0.21450023353099823, "learning_rate": 6.635533947763093e-05, "loss": 0.0106, "step": 127270 }, { "epoch": 81.12173358827279, "grad_norm": 0.2385348379611969, "learning_rate": 6.635013101790573e-05, "loss": 0.0143, "step": 127280 }, { "epoch": 81.12810707456978, "grad_norm": 0.3169389069080353, "learning_rate": 6.6344922359511e-05, "loss": 0.0208, "step": 127290 }, { "epoch": 81.13448056086679, "grad_norm": 0.48389434814453125, "learning_rate": 6.633971350251002e-05, "loss": 0.0204, "step": 127300 }, { "epoch": 81.1408540471638, "grad_norm": 0.34694239497184753, "learning_rate": 6.63345044469661e-05, "loss": 0.0188, "step": 127310 }, { "epoch": 81.1472275334608, "grad_norm": 0.36945629119873047, "learning_rate": 6.632929519294253e-05, "loss": 0.0264, "step": 127320 }, { "epoch": 81.1536010197578, "grad_norm": 0.4333002269268036, "learning_rate": 6.632408574050261e-05, "loss": 0.012, "step": 127330 }, { "epoch": 81.15997450605481, "grad_norm": 0.4793090522289276, "learning_rate": 6.631887608970963e-05, "loss": 0.0173, "step": 127340 }, { "epoch": 81.16634799235182, "grad_norm": 0.5290791988372803, "learning_rate": 6.631366624062689e-05, "loss": 0.0147, "step": 127350 }, { "epoch": 81.17272147864882, "grad_norm": 0.3516162633895874, "learning_rate": 6.630845619331773e-05, "loss": 0.0196, "step": 127360 }, { "epoch": 81.17909496494582, "grad_norm": 0.21909646689891815, "learning_rate": 6.63032459478454e-05, "loss": 0.0153, "step": 127370 }, { "epoch": 81.18546845124283, "grad_norm": 0.5707698464393616, "learning_rate": 6.629803550427326e-05, "loss": 0.0167, "step": 127380 }, { "epoch": 81.19184193753983, "grad_norm": 0.29888975620269775, "learning_rate": 6.629282486266458e-05, "loss": 0.018, "step": 127390 }, { "epoch": 81.19821542383684, "grad_norm": 0.3173043131828308, "learning_rate": 6.628761402308273e-05, "loss": 0.0123, "step": 127400 }, { "epoch": 81.20458891013385, "grad_norm": 0.29929792881011963, "learning_rate": 6.628240298559096e-05, "loss": 0.026, "step": 127410 }, { "epoch": 81.21096239643084, "grad_norm": 0.2227499634027481, "learning_rate": 6.627719175025261e-05, "loss": 0.0081, "step": 127420 }, { "epoch": 81.21733588272785, "grad_norm": 0.23302891850471497, "learning_rate": 6.627198031713105e-05, "loss": 0.0174, "step": 127430 }, { "epoch": 81.22370936902486, "grad_norm": 0.3801746666431427, "learning_rate": 6.626676868628954e-05, "loss": 0.0091, "step": 127440 }, { "epoch": 81.23008285532187, "grad_norm": 0.3989483118057251, "learning_rate": 6.626155685779143e-05, "loss": 0.0186, "step": 127450 }, { "epoch": 81.23645634161886, "grad_norm": 0.5053424835205078, "learning_rate": 6.625634483170007e-05, "loss": 0.0231, "step": 127460 }, { "epoch": 81.24282982791587, "grad_norm": 0.7118179798126221, "learning_rate": 6.625113260807875e-05, "loss": 0.0198, "step": 127470 }, { "epoch": 81.24920331421288, "grad_norm": 0.38172170519828796, "learning_rate": 6.624592018699084e-05, "loss": 0.01, "step": 127480 }, { "epoch": 81.25557680050989, "grad_norm": 0.16911453008651733, "learning_rate": 6.624070756849964e-05, "loss": 0.0238, "step": 127490 }, { "epoch": 81.26195028680688, "grad_norm": 0.3202678859233856, "learning_rate": 6.623549475266855e-05, "loss": 0.017, "step": 127500 }, { "epoch": 81.26832377310389, "grad_norm": 0.42701807618141174, "learning_rate": 6.623028173956083e-05, "loss": 0.0155, "step": 127510 }, { "epoch": 81.2746972594009, "grad_norm": 0.26585322618484497, "learning_rate": 6.622506852923988e-05, "loss": 0.0106, "step": 127520 }, { "epoch": 81.2810707456979, "grad_norm": 0.23967081308364868, "learning_rate": 6.621985512176905e-05, "loss": 0.0093, "step": 127530 }, { "epoch": 81.2874442319949, "grad_norm": 0.2693195641040802, "learning_rate": 6.621464151721164e-05, "loss": 0.0148, "step": 127540 }, { "epoch": 81.2938177182919, "grad_norm": 0.2564235031604767, "learning_rate": 6.620942771563104e-05, "loss": 0.0125, "step": 127550 }, { "epoch": 81.30019120458891, "grad_norm": 0.34550246596336365, "learning_rate": 6.620421371709055e-05, "loss": 0.0197, "step": 127560 }, { "epoch": 81.30656469088592, "grad_norm": 0.26747581362724304, "learning_rate": 6.61989995216536e-05, "loss": 0.0128, "step": 127570 }, { "epoch": 81.31293817718291, "grad_norm": 0.1312360018491745, "learning_rate": 6.619378512938351e-05, "loss": 0.0139, "step": 127580 }, { "epoch": 81.31931166347992, "grad_norm": 0.4966040849685669, "learning_rate": 6.618857054034365e-05, "loss": 0.0158, "step": 127590 }, { "epoch": 81.32568514977693, "grad_norm": 0.4939136803150177, "learning_rate": 6.618335575459734e-05, "loss": 0.0266, "step": 127600 }, { "epoch": 81.33205863607394, "grad_norm": 0.2748204171657562, "learning_rate": 6.617814077220801e-05, "loss": 0.0297, "step": 127610 }, { "epoch": 81.33843212237093, "grad_norm": 0.3028007745742798, "learning_rate": 6.617292559323899e-05, "loss": 0.0151, "step": 127620 }, { "epoch": 81.34480560866794, "grad_norm": 0.4295285642147064, "learning_rate": 6.616771021775363e-05, "loss": 0.0195, "step": 127630 }, { "epoch": 81.35117909496495, "grad_norm": 0.3242585062980652, "learning_rate": 6.616249464581535e-05, "loss": 0.013, "step": 127640 }, { "epoch": 81.35755258126196, "grad_norm": 0.28624168038368225, "learning_rate": 6.615727887748749e-05, "loss": 0.0116, "step": 127650 }, { "epoch": 81.36392606755895, "grad_norm": 0.18329140543937683, "learning_rate": 6.615206291283344e-05, "loss": 0.0141, "step": 127660 }, { "epoch": 81.37029955385596, "grad_norm": 0.3322015106678009, "learning_rate": 6.614684675191658e-05, "loss": 0.0201, "step": 127670 }, { "epoch": 81.37667304015297, "grad_norm": 0.38822224736213684, "learning_rate": 6.614163039480027e-05, "loss": 0.0117, "step": 127680 }, { "epoch": 81.38304652644997, "grad_norm": 0.5598871111869812, "learning_rate": 6.613641384154794e-05, "loss": 0.0185, "step": 127690 }, { "epoch": 81.38942001274697, "grad_norm": 0.30626851320266724, "learning_rate": 6.613119709222292e-05, "loss": 0.0141, "step": 127700 }, { "epoch": 81.39579349904398, "grad_norm": 0.24263986945152283, "learning_rate": 6.612598014688864e-05, "loss": 0.0157, "step": 127710 }, { "epoch": 81.40216698534098, "grad_norm": 0.19710896909236908, "learning_rate": 6.612076300560847e-05, "loss": 0.0126, "step": 127720 }, { "epoch": 81.40854047163799, "grad_norm": 0.43520018458366394, "learning_rate": 6.611554566844581e-05, "loss": 0.014, "step": 127730 }, { "epoch": 81.41491395793498, "grad_norm": 0.23662208020687103, "learning_rate": 6.611032813546405e-05, "loss": 0.0148, "step": 127740 }, { "epoch": 81.421287444232, "grad_norm": 0.26984718441963196, "learning_rate": 6.61051104067266e-05, "loss": 0.0157, "step": 127750 }, { "epoch": 81.427660930529, "grad_norm": 0.4992947280406952, "learning_rate": 6.609989248229684e-05, "loss": 0.0195, "step": 127760 }, { "epoch": 81.43403441682601, "grad_norm": 0.40786072611808777, "learning_rate": 6.609467436223819e-05, "loss": 0.0171, "step": 127770 }, { "epoch": 81.440407903123, "grad_norm": 0.3281767964363098, "learning_rate": 6.608945604661407e-05, "loss": 0.0133, "step": 127780 }, { "epoch": 81.44678138942001, "grad_norm": 0.3373706042766571, "learning_rate": 6.608423753548785e-05, "loss": 0.0191, "step": 127790 }, { "epoch": 81.45315487571702, "grad_norm": 0.36204326152801514, "learning_rate": 6.607901882892296e-05, "loss": 0.0144, "step": 127800 }, { "epoch": 81.45952836201403, "grad_norm": 0.2809852957725525, "learning_rate": 6.607379992698281e-05, "loss": 0.0138, "step": 127810 }, { "epoch": 81.46590184831102, "grad_norm": 0.24130919575691223, "learning_rate": 6.606858082973082e-05, "loss": 0.0138, "step": 127820 }, { "epoch": 81.47227533460803, "grad_norm": 0.40450048446655273, "learning_rate": 6.606336153723038e-05, "loss": 0.0104, "step": 127830 }, { "epoch": 81.47864882090504, "grad_norm": 0.2732711732387543, "learning_rate": 6.605814204954496e-05, "loss": 0.0206, "step": 127840 }, { "epoch": 81.48502230720204, "grad_norm": 0.5569290518760681, "learning_rate": 6.605292236673794e-05, "loss": 0.0175, "step": 127850 }, { "epoch": 81.49139579349904, "grad_norm": 0.26931798458099365, "learning_rate": 6.604770248887275e-05, "loss": 0.0182, "step": 127860 }, { "epoch": 81.49776927979605, "grad_norm": 0.4693372845649719, "learning_rate": 6.604248241601285e-05, "loss": 0.0254, "step": 127870 }, { "epoch": 81.50414276609305, "grad_norm": 0.42395004630088806, "learning_rate": 6.60372621482216e-05, "loss": 0.0139, "step": 127880 }, { "epoch": 81.51051625239006, "grad_norm": 0.4101807475090027, "learning_rate": 6.603204168556251e-05, "loss": 0.0143, "step": 127890 }, { "epoch": 81.51688973868706, "grad_norm": 0.20198851823806763, "learning_rate": 6.602682102809896e-05, "loss": 0.0115, "step": 127900 }, { "epoch": 81.52326322498406, "grad_norm": 0.2849816679954529, "learning_rate": 6.60216001758944e-05, "loss": 0.012, "step": 127910 }, { "epoch": 81.52963671128107, "grad_norm": 0.43186095356941223, "learning_rate": 6.601637912901227e-05, "loss": 0.0168, "step": 127920 }, { "epoch": 81.53601019757808, "grad_norm": 0.37529632449150085, "learning_rate": 6.601115788751602e-05, "loss": 0.0225, "step": 127930 }, { "epoch": 81.54238368387507, "grad_norm": 0.16300275921821594, "learning_rate": 6.600593645146909e-05, "loss": 0.0245, "step": 127940 }, { "epoch": 81.54875717017208, "grad_norm": 0.4236455261707306, "learning_rate": 6.600071482093491e-05, "loss": 0.0125, "step": 127950 }, { "epoch": 81.55513065646909, "grad_norm": 0.6016177535057068, "learning_rate": 6.599549299597695e-05, "loss": 0.0162, "step": 127960 }, { "epoch": 81.5615041427661, "grad_norm": 0.42569202184677124, "learning_rate": 6.599027097665863e-05, "loss": 0.0184, "step": 127970 }, { "epoch": 81.56787762906309, "grad_norm": 0.3088478147983551, "learning_rate": 6.598504876304343e-05, "loss": 0.0117, "step": 127980 }, { "epoch": 81.5742511153601, "grad_norm": 0.34524720907211304, "learning_rate": 6.59798263551948e-05, "loss": 0.0128, "step": 127990 }, { "epoch": 81.5806246016571, "grad_norm": 0.3614957630634308, "learning_rate": 6.597460375317619e-05, "loss": 0.0134, "step": 128000 }, { "epoch": 81.58699808795411, "grad_norm": 0.18238256871700287, "learning_rate": 6.596938095705105e-05, "loss": 0.0092, "step": 128010 }, { "epoch": 81.59337157425111, "grad_norm": 0.6764192581176758, "learning_rate": 6.596415796688287e-05, "loss": 0.0183, "step": 128020 }, { "epoch": 81.59974506054812, "grad_norm": 0.7099656462669373, "learning_rate": 6.59589347827351e-05, "loss": 0.029, "step": 128030 }, { "epoch": 81.60611854684512, "grad_norm": 0.663221001625061, "learning_rate": 6.595371140467117e-05, "loss": 0.016, "step": 128040 }, { "epoch": 81.61249203314213, "grad_norm": 0.4533836245536804, "learning_rate": 6.594848783275462e-05, "loss": 0.0155, "step": 128050 }, { "epoch": 81.61886551943914, "grad_norm": 0.3718253970146179, "learning_rate": 6.594326406704887e-05, "loss": 0.0131, "step": 128060 }, { "epoch": 81.62523900573613, "grad_norm": 0.32075798511505127, "learning_rate": 6.593804010761741e-05, "loss": 0.0162, "step": 128070 }, { "epoch": 81.63161249203314, "grad_norm": 0.41828203201293945, "learning_rate": 6.593281595452372e-05, "loss": 0.0141, "step": 128080 }, { "epoch": 81.63798597833015, "grad_norm": 0.13964274525642395, "learning_rate": 6.592759160783127e-05, "loss": 0.0092, "step": 128090 }, { "epoch": 81.64435946462716, "grad_norm": 0.29005011916160583, "learning_rate": 6.592236706760354e-05, "loss": 0.0123, "step": 128100 }, { "epoch": 81.65073295092415, "grad_norm": 0.5831001996994019, "learning_rate": 6.591714233390402e-05, "loss": 0.0235, "step": 128110 }, { "epoch": 81.65710643722116, "grad_norm": 0.20778940618038177, "learning_rate": 6.591191740679619e-05, "loss": 0.0214, "step": 128120 }, { "epoch": 81.66347992351817, "grad_norm": 0.7422183156013489, "learning_rate": 6.590669228634354e-05, "loss": 0.0223, "step": 128130 }, { "epoch": 81.66985340981518, "grad_norm": 0.483907014131546, "learning_rate": 6.590146697260957e-05, "loss": 0.0145, "step": 128140 }, { "epoch": 81.67622689611217, "grad_norm": 0.3647027313709259, "learning_rate": 6.589624146565775e-05, "loss": 0.0125, "step": 128150 }, { "epoch": 81.68260038240918, "grad_norm": 0.2211553305387497, "learning_rate": 6.58910157655516e-05, "loss": 0.0157, "step": 128160 }, { "epoch": 81.68897386870619, "grad_norm": 0.8959919810295105, "learning_rate": 6.588578987235459e-05, "loss": 0.0163, "step": 128170 }, { "epoch": 81.6953473550032, "grad_norm": 0.289741575717926, "learning_rate": 6.588056378613023e-05, "loss": 0.0125, "step": 128180 }, { "epoch": 81.70172084130019, "grad_norm": 0.1956545114517212, "learning_rate": 6.587533750694206e-05, "loss": 0.0245, "step": 128190 }, { "epoch": 81.7080943275972, "grad_norm": 0.751069188117981, "learning_rate": 6.587011103485352e-05, "loss": 0.0165, "step": 128200 }, { "epoch": 81.7144678138942, "grad_norm": 0.44637590646743774, "learning_rate": 6.586488436992816e-05, "loss": 0.0131, "step": 128210 }, { "epoch": 81.72084130019121, "grad_norm": 0.35311785340309143, "learning_rate": 6.585965751222947e-05, "loss": 0.0121, "step": 128220 }, { "epoch": 81.7272147864882, "grad_norm": 0.2387092262506485, "learning_rate": 6.585443046182097e-05, "loss": 0.0125, "step": 128230 }, { "epoch": 81.73358827278521, "grad_norm": 0.3070967495441437, "learning_rate": 6.584920321876617e-05, "loss": 0.0202, "step": 128240 }, { "epoch": 81.73996175908222, "grad_norm": 0.23998437821865082, "learning_rate": 6.584397578312859e-05, "loss": 0.0129, "step": 128250 }, { "epoch": 81.74633524537923, "grad_norm": 0.1991690993309021, "learning_rate": 6.583874815497175e-05, "loss": 0.0162, "step": 128260 }, { "epoch": 81.75270873167622, "grad_norm": 0.2373322695493698, "learning_rate": 6.583352033435914e-05, "loss": 0.0146, "step": 128270 }, { "epoch": 81.75908221797323, "grad_norm": 0.48872095346450806, "learning_rate": 6.582829232135433e-05, "loss": 0.0134, "step": 128280 }, { "epoch": 81.76545570427024, "grad_norm": 0.20950189232826233, "learning_rate": 6.582306411602081e-05, "loss": 0.0113, "step": 128290 }, { "epoch": 81.77182919056725, "grad_norm": 0.2452247440814972, "learning_rate": 6.581783571842216e-05, "loss": 0.0116, "step": 128300 }, { "epoch": 81.77820267686424, "grad_norm": 0.48979443311691284, "learning_rate": 6.581260712862183e-05, "loss": 0.0127, "step": 128310 }, { "epoch": 81.78457616316125, "grad_norm": 0.2909833788871765, "learning_rate": 6.580737834668341e-05, "loss": 0.0258, "step": 128320 }, { "epoch": 81.79094964945826, "grad_norm": 0.5084080100059509, "learning_rate": 6.58021493726704e-05, "loss": 0.019, "step": 128330 }, { "epoch": 81.79732313575526, "grad_norm": 0.5048345923423767, "learning_rate": 6.579692020664639e-05, "loss": 0.0138, "step": 128340 }, { "epoch": 81.80369662205226, "grad_norm": 0.23233503103256226, "learning_rate": 6.579169084867486e-05, "loss": 0.0147, "step": 128350 }, { "epoch": 81.81007010834927, "grad_norm": 0.44075509905815125, "learning_rate": 6.578646129881938e-05, "loss": 0.0103, "step": 128360 }, { "epoch": 81.81644359464627, "grad_norm": 0.3790730834007263, "learning_rate": 6.57812315571435e-05, "loss": 0.0271, "step": 128370 }, { "epoch": 81.82281708094328, "grad_norm": 0.2620769441127777, "learning_rate": 6.577600162371074e-05, "loss": 0.0231, "step": 128380 }, { "epoch": 81.82919056724027, "grad_norm": 0.37984275817871094, "learning_rate": 6.577077149858468e-05, "loss": 0.0177, "step": 128390 }, { "epoch": 81.83556405353728, "grad_norm": 0.2711968421936035, "learning_rate": 6.576554118182885e-05, "loss": 0.0141, "step": 128400 }, { "epoch": 81.84193753983429, "grad_norm": 0.312850683927536, "learning_rate": 6.576031067350681e-05, "loss": 0.0155, "step": 128410 }, { "epoch": 81.8483110261313, "grad_norm": 0.6217361092567444, "learning_rate": 6.575507997368212e-05, "loss": 0.0166, "step": 128420 }, { "epoch": 81.85468451242829, "grad_norm": 0.2619372606277466, "learning_rate": 6.574984908241833e-05, "loss": 0.0177, "step": 128430 }, { "epoch": 81.8610579987253, "grad_norm": 0.1730402708053589, "learning_rate": 6.5744617999779e-05, "loss": 0.0204, "step": 128440 }, { "epoch": 81.86743148502231, "grad_norm": 0.5761787295341492, "learning_rate": 6.573938672582769e-05, "loss": 0.0183, "step": 128450 }, { "epoch": 81.87380497131932, "grad_norm": 0.22725649178028107, "learning_rate": 6.573415526062798e-05, "loss": 0.0187, "step": 128460 }, { "epoch": 81.88017845761631, "grad_norm": 0.29454728960990906, "learning_rate": 6.572892360424342e-05, "loss": 0.0176, "step": 128470 }, { "epoch": 81.88655194391332, "grad_norm": 0.22280116379261017, "learning_rate": 6.57236917567376e-05, "loss": 0.0117, "step": 128480 }, { "epoch": 81.89292543021033, "grad_norm": 0.4208506643772125, "learning_rate": 6.571845971817407e-05, "loss": 0.0288, "step": 128490 }, { "epoch": 81.89929891650733, "grad_norm": 0.532431960105896, "learning_rate": 6.571322748861643e-05, "loss": 0.0351, "step": 128500 }, { "epoch": 81.90567240280433, "grad_norm": 0.3497880697250366, "learning_rate": 6.570799506812824e-05, "loss": 0.0239, "step": 128510 }, { "epoch": 81.91204588910134, "grad_norm": 0.6106723546981812, "learning_rate": 6.570276245677305e-05, "loss": 0.0134, "step": 128520 }, { "epoch": 81.91841937539834, "grad_norm": 0.42493873834609985, "learning_rate": 6.569752965461448e-05, "loss": 0.0131, "step": 128530 }, { "epoch": 81.92479286169535, "grad_norm": 0.3729230761528015, "learning_rate": 6.569229666171613e-05, "loss": 0.0159, "step": 128540 }, { "epoch": 81.93116634799235, "grad_norm": 0.3722967803478241, "learning_rate": 6.568706347814154e-05, "loss": 0.0124, "step": 128550 }, { "epoch": 81.93753983428935, "grad_norm": 0.28316137194633484, "learning_rate": 6.56818301039543e-05, "loss": 0.0234, "step": 128560 }, { "epoch": 81.94391332058636, "grad_norm": 0.6038443446159363, "learning_rate": 6.567659653921805e-05, "loss": 0.0158, "step": 128570 }, { "epoch": 81.95028680688337, "grad_norm": 0.5959979295730591, "learning_rate": 6.567136278399634e-05, "loss": 0.0154, "step": 128580 }, { "epoch": 81.95666029318036, "grad_norm": 0.31616199016571045, "learning_rate": 6.566612883835276e-05, "loss": 0.0157, "step": 128590 }, { "epoch": 81.96303377947737, "grad_norm": 0.26260602474212646, "learning_rate": 6.566089470235095e-05, "loss": 0.0113, "step": 128600 }, { "epoch": 81.96940726577438, "grad_norm": 0.28833502531051636, "learning_rate": 6.565566037605447e-05, "loss": 0.0234, "step": 128610 }, { "epoch": 81.97578075207139, "grad_norm": 0.24887816607952118, "learning_rate": 6.565042585952693e-05, "loss": 0.011, "step": 128620 }, { "epoch": 81.98215423836838, "grad_norm": 0.21246270835399628, "learning_rate": 6.564519115283194e-05, "loss": 0.0142, "step": 128630 }, { "epoch": 81.98852772466539, "grad_norm": 0.8406859040260315, "learning_rate": 6.563995625603312e-05, "loss": 0.0311, "step": 128640 }, { "epoch": 81.9949012109624, "grad_norm": 0.36603865027427673, "learning_rate": 6.563472116919405e-05, "loss": 0.0143, "step": 128650 }, { "epoch": 82.0012746972594, "grad_norm": 0.2759566307067871, "learning_rate": 6.562948589237837e-05, "loss": 0.0217, "step": 128660 }, { "epoch": 82.00764818355641, "grad_norm": 0.33693820238113403, "learning_rate": 6.562425042564967e-05, "loss": 0.0131, "step": 128670 }, { "epoch": 82.0140216698534, "grad_norm": 0.6064636707305908, "learning_rate": 6.561901476907157e-05, "loss": 0.0315, "step": 128680 }, { "epoch": 82.02039515615041, "grad_norm": 0.3349491059780121, "learning_rate": 6.56137789227077e-05, "loss": 0.012, "step": 128690 }, { "epoch": 82.02676864244742, "grad_norm": 0.432344913482666, "learning_rate": 6.560854288662167e-05, "loss": 0.0134, "step": 128700 }, { "epoch": 82.03314212874443, "grad_norm": 0.563667893409729, "learning_rate": 6.560330666087711e-05, "loss": 0.0239, "step": 128710 }, { "epoch": 82.03951561504142, "grad_norm": 0.7653446197509766, "learning_rate": 6.559807024553765e-05, "loss": 0.0201, "step": 128720 }, { "epoch": 82.04588910133843, "grad_norm": 0.40037381649017334, "learning_rate": 6.55928336406669e-05, "loss": 0.0197, "step": 128730 }, { "epoch": 82.05226258763544, "grad_norm": 0.4187275469303131, "learning_rate": 6.558759684632851e-05, "loss": 0.0151, "step": 128740 }, { "epoch": 82.05863607393245, "grad_norm": 0.5199059247970581, "learning_rate": 6.558235986258608e-05, "loss": 0.0233, "step": 128750 }, { "epoch": 82.06500956022944, "grad_norm": 0.29151448607444763, "learning_rate": 6.557712268950329e-05, "loss": 0.0141, "step": 128760 }, { "epoch": 82.07138304652645, "grad_norm": 0.2693638503551483, "learning_rate": 6.557188532714374e-05, "loss": 0.01, "step": 128770 }, { "epoch": 82.07775653282346, "grad_norm": 0.3176136612892151, "learning_rate": 6.556664777557109e-05, "loss": 0.0122, "step": 128780 }, { "epoch": 82.08413001912047, "grad_norm": 0.43023332953453064, "learning_rate": 6.556141003484895e-05, "loss": 0.0161, "step": 128790 }, { "epoch": 82.09050350541746, "grad_norm": 0.5234009027481079, "learning_rate": 6.5556172105041e-05, "loss": 0.0136, "step": 128800 }, { "epoch": 82.09687699171447, "grad_norm": 0.37645745277404785, "learning_rate": 6.555093398621088e-05, "loss": 0.0117, "step": 128810 }, { "epoch": 82.10325047801147, "grad_norm": 0.27266931533813477, "learning_rate": 6.554569567842221e-05, "loss": 0.0149, "step": 128820 }, { "epoch": 82.10962396430848, "grad_norm": 0.34110593795776367, "learning_rate": 6.554045718173867e-05, "loss": 0.0118, "step": 128830 }, { "epoch": 82.11599745060548, "grad_norm": 0.46295127272605896, "learning_rate": 6.553521849622391e-05, "loss": 0.02, "step": 128840 }, { "epoch": 82.12237093690248, "grad_norm": 0.3209141194820404, "learning_rate": 6.552997962194157e-05, "loss": 0.0118, "step": 128850 }, { "epoch": 82.12874442319949, "grad_norm": 0.2872767746448517, "learning_rate": 6.552474055895531e-05, "loss": 0.0117, "step": 128860 }, { "epoch": 82.1351179094965, "grad_norm": 0.25526511669158936, "learning_rate": 6.55195013073288e-05, "loss": 0.0141, "step": 128870 }, { "epoch": 82.1414913957935, "grad_norm": 0.4619637727737427, "learning_rate": 6.55142618671257e-05, "loss": 0.0185, "step": 128880 }, { "epoch": 82.1478648820905, "grad_norm": 0.9284997582435608, "learning_rate": 6.550902223840967e-05, "loss": 0.0151, "step": 128890 }, { "epoch": 82.15423836838751, "grad_norm": 0.43440350890159607, "learning_rate": 6.550378242124437e-05, "loss": 0.0124, "step": 128900 }, { "epoch": 82.16061185468452, "grad_norm": 0.24251821637153625, "learning_rate": 6.549854241569347e-05, "loss": 0.0159, "step": 128910 }, { "epoch": 82.16698534098151, "grad_norm": 0.35185766220092773, "learning_rate": 6.549330222182067e-05, "loss": 0.0132, "step": 128920 }, { "epoch": 82.17335882727852, "grad_norm": 0.35577017068862915, "learning_rate": 6.54880618396896e-05, "loss": 0.0315, "step": 128930 }, { "epoch": 82.17973231357553, "grad_norm": 0.2870731055736542, "learning_rate": 6.548282126936397e-05, "loss": 0.0128, "step": 128940 }, { "epoch": 82.18610579987254, "grad_norm": 0.3392900824546814, "learning_rate": 6.547758051090742e-05, "loss": 0.0166, "step": 128950 }, { "epoch": 82.19247928616953, "grad_norm": 0.4017775058746338, "learning_rate": 6.547233956438368e-05, "loss": 0.0185, "step": 128960 }, { "epoch": 82.19885277246654, "grad_norm": 0.15848763287067413, "learning_rate": 6.546709842985639e-05, "loss": 0.0116, "step": 128970 }, { "epoch": 82.20522625876355, "grad_norm": 0.8483330011367798, "learning_rate": 6.546185710738926e-05, "loss": 0.0211, "step": 128980 }, { "epoch": 82.21159974506055, "grad_norm": 0.41901329159736633, "learning_rate": 6.545661559704596e-05, "loss": 0.0168, "step": 128990 }, { "epoch": 82.21797323135755, "grad_norm": 0.4196595847606659, "learning_rate": 6.545137389889018e-05, "loss": 0.0248, "step": 129000 }, { "epoch": 82.22434671765456, "grad_norm": 0.26421019434928894, "learning_rate": 6.544613201298564e-05, "loss": 0.0272, "step": 129010 }, { "epoch": 82.23072020395156, "grad_norm": 0.35071468353271484, "learning_rate": 6.544088993939599e-05, "loss": 0.0206, "step": 129020 }, { "epoch": 82.23709369024857, "grad_norm": 0.20422649383544922, "learning_rate": 6.543564767818497e-05, "loss": 0.01, "step": 129030 }, { "epoch": 82.24346717654556, "grad_norm": 0.3799484670162201, "learning_rate": 6.543040522941624e-05, "loss": 0.0117, "step": 129040 }, { "epoch": 82.24984066284257, "grad_norm": 0.34045475721359253, "learning_rate": 6.542516259315352e-05, "loss": 0.0096, "step": 129050 }, { "epoch": 82.25621414913958, "grad_norm": 0.49087706208229065, "learning_rate": 6.541991976946052e-05, "loss": 0.0189, "step": 129060 }, { "epoch": 82.26258763543659, "grad_norm": 0.5129827857017517, "learning_rate": 6.541467675840093e-05, "loss": 0.0174, "step": 129070 }, { "epoch": 82.26896112173358, "grad_norm": 0.3261740505695343, "learning_rate": 6.540943356003847e-05, "loss": 0.015, "step": 129080 }, { "epoch": 82.27533460803059, "grad_norm": 0.2976730763912201, "learning_rate": 6.540419017443683e-05, "loss": 0.0153, "step": 129090 }, { "epoch": 82.2817080943276, "grad_norm": 0.4898168444633484, "learning_rate": 6.539894660165975e-05, "loss": 0.0189, "step": 129100 }, { "epoch": 82.2880815806246, "grad_norm": 0.4398637115955353, "learning_rate": 6.539370284177092e-05, "loss": 0.0152, "step": 129110 }, { "epoch": 82.2944550669216, "grad_norm": 0.24904365837574005, "learning_rate": 6.538845889483407e-05, "loss": 0.0113, "step": 129120 }, { "epoch": 82.30082855321861, "grad_norm": 0.13801807165145874, "learning_rate": 6.538321476091292e-05, "loss": 0.0184, "step": 129130 }, { "epoch": 82.30720203951562, "grad_norm": 0.4048437178134918, "learning_rate": 6.537797044007118e-05, "loss": 0.0183, "step": 129140 }, { "epoch": 82.31357552581262, "grad_norm": 0.32958391308784485, "learning_rate": 6.537272593237257e-05, "loss": 0.0179, "step": 129150 }, { "epoch": 82.31994901210962, "grad_norm": 0.44573330879211426, "learning_rate": 6.536748123788084e-05, "loss": 0.0157, "step": 129160 }, { "epoch": 82.32632249840663, "grad_norm": 0.44613009691238403, "learning_rate": 6.536223635665971e-05, "loss": 0.0185, "step": 129170 }, { "epoch": 82.33269598470363, "grad_norm": 0.2173072099685669, "learning_rate": 6.535699128877288e-05, "loss": 0.0164, "step": 129180 }, { "epoch": 82.33906947100064, "grad_norm": 0.4045519232749939, "learning_rate": 6.535174603428413e-05, "loss": 0.0191, "step": 129190 }, { "epoch": 82.34544295729764, "grad_norm": 0.4143335223197937, "learning_rate": 6.534650059325715e-05, "loss": 0.011, "step": 129200 }, { "epoch": 82.35181644359464, "grad_norm": 0.24158607423305511, "learning_rate": 6.534125496575573e-05, "loss": 0.0102, "step": 129210 }, { "epoch": 82.35818992989165, "grad_norm": 0.5204176306724548, "learning_rate": 6.533600915184356e-05, "loss": 0.0148, "step": 129220 }, { "epoch": 82.36456341618866, "grad_norm": 0.23025505244731903, "learning_rate": 6.533076315158438e-05, "loss": 0.0153, "step": 129230 }, { "epoch": 82.37093690248565, "grad_norm": 0.16277673840522766, "learning_rate": 6.532551696504198e-05, "loss": 0.0153, "step": 129240 }, { "epoch": 82.37731038878266, "grad_norm": 0.3283427655696869, "learning_rate": 6.532027059228006e-05, "loss": 0.0166, "step": 129250 }, { "epoch": 82.38368387507967, "grad_norm": 0.2775934338569641, "learning_rate": 6.53150240333624e-05, "loss": 0.0129, "step": 129260 }, { "epoch": 82.39005736137668, "grad_norm": 0.3321723937988281, "learning_rate": 6.530977728835274e-05, "loss": 0.0105, "step": 129270 }, { "epoch": 82.39643084767367, "grad_norm": 0.2265108823776245, "learning_rate": 6.530453035731482e-05, "loss": 0.0127, "step": 129280 }, { "epoch": 82.40280433397068, "grad_norm": 0.5545596480369568, "learning_rate": 6.529928324031242e-05, "loss": 0.0168, "step": 129290 }, { "epoch": 82.40917782026769, "grad_norm": 0.36006003618240356, "learning_rate": 6.529403593740929e-05, "loss": 0.0106, "step": 129300 }, { "epoch": 82.4155513065647, "grad_norm": 0.26289042830467224, "learning_rate": 6.528878844866916e-05, "loss": 0.0122, "step": 129310 }, { "epoch": 82.4219247928617, "grad_norm": 0.188180074095726, "learning_rate": 6.528354077415582e-05, "loss": 0.0134, "step": 129320 }, { "epoch": 82.4282982791587, "grad_norm": 0.38672778010368347, "learning_rate": 6.527829291393304e-05, "loss": 0.0141, "step": 129330 }, { "epoch": 82.4346717654557, "grad_norm": 0.5343759059906006, "learning_rate": 6.527304486806456e-05, "loss": 0.0179, "step": 129340 }, { "epoch": 82.44104525175271, "grad_norm": 0.2163008153438568, "learning_rate": 6.526779663661419e-05, "loss": 0.0101, "step": 129350 }, { "epoch": 82.44741873804972, "grad_norm": 0.41879984736442566, "learning_rate": 6.526254821964565e-05, "loss": 0.0147, "step": 129360 }, { "epoch": 82.45379222434671, "grad_norm": 0.8188528418540955, "learning_rate": 6.525729961722276e-05, "loss": 0.0201, "step": 129370 }, { "epoch": 82.46016571064372, "grad_norm": 0.864723265171051, "learning_rate": 6.525205082940925e-05, "loss": 0.0158, "step": 129380 }, { "epoch": 82.46653919694073, "grad_norm": 0.47532087564468384, "learning_rate": 6.524680185626894e-05, "loss": 0.0176, "step": 129390 }, { "epoch": 82.47291268323774, "grad_norm": 0.44338297843933105, "learning_rate": 6.52415526978656e-05, "loss": 0.0165, "step": 129400 }, { "epoch": 82.47928616953473, "grad_norm": 0.42772194743156433, "learning_rate": 6.523630335426298e-05, "loss": 0.0162, "step": 129410 }, { "epoch": 82.48565965583174, "grad_norm": 0.5249453186988831, "learning_rate": 6.523105382552491e-05, "loss": 0.013, "step": 129420 }, { "epoch": 82.49203314212875, "grad_norm": 0.6009605526924133, "learning_rate": 6.522580411171515e-05, "loss": 0.018, "step": 129430 }, { "epoch": 82.49840662842576, "grad_norm": 0.46704569458961487, "learning_rate": 6.522055421289749e-05, "loss": 0.0135, "step": 129440 }, { "epoch": 82.50478011472275, "grad_norm": 0.3131047487258911, "learning_rate": 6.521530412913573e-05, "loss": 0.0136, "step": 129450 }, { "epoch": 82.51115360101976, "grad_norm": 0.6475018262863159, "learning_rate": 6.521005386049366e-05, "loss": 0.0243, "step": 129460 }, { "epoch": 82.51752708731676, "grad_norm": 0.21416524052619934, "learning_rate": 6.520480340703508e-05, "loss": 0.0173, "step": 129470 }, { "epoch": 82.52390057361377, "grad_norm": 0.5421136021614075, "learning_rate": 6.519955276882375e-05, "loss": 0.0211, "step": 129480 }, { "epoch": 82.53027405991077, "grad_norm": 0.40036237239837646, "learning_rate": 6.519430194592355e-05, "loss": 0.0148, "step": 129490 }, { "epoch": 82.53664754620777, "grad_norm": 0.4670766592025757, "learning_rate": 6.51890509383982e-05, "loss": 0.0117, "step": 129500 }, { "epoch": 82.54302103250478, "grad_norm": 0.3858421742916107, "learning_rate": 6.518379974631156e-05, "loss": 0.0172, "step": 129510 }, { "epoch": 82.54939451880179, "grad_norm": 0.3115980625152588, "learning_rate": 6.51785483697274e-05, "loss": 0.0198, "step": 129520 }, { "epoch": 82.55576800509878, "grad_norm": 0.15540215373039246, "learning_rate": 6.517329680870956e-05, "loss": 0.0117, "step": 129530 }, { "epoch": 82.56214149139579, "grad_norm": 0.3400847613811493, "learning_rate": 6.516804506332183e-05, "loss": 0.0129, "step": 129540 }, { "epoch": 82.5685149776928, "grad_norm": 0.6210516095161438, "learning_rate": 6.516279313362804e-05, "loss": 0.0239, "step": 129550 }, { "epoch": 82.57488846398981, "grad_norm": 0.6158271431922913, "learning_rate": 6.515754101969199e-05, "loss": 0.0227, "step": 129560 }, { "epoch": 82.5812619502868, "grad_norm": 0.3773817718029022, "learning_rate": 6.515228872157752e-05, "loss": 0.0202, "step": 129570 }, { "epoch": 82.58763543658381, "grad_norm": 0.275083065032959, "learning_rate": 6.514703623934841e-05, "loss": 0.0186, "step": 129580 }, { "epoch": 82.59400892288082, "grad_norm": 0.5150445699691772, "learning_rate": 6.514178357306852e-05, "loss": 0.0225, "step": 129590 }, { "epoch": 82.60038240917783, "grad_norm": 0.2508932948112488, "learning_rate": 6.513653072280166e-05, "loss": 0.022, "step": 129600 }, { "epoch": 82.60675589547482, "grad_norm": 0.25177550315856934, "learning_rate": 6.513127768861169e-05, "loss": 0.014, "step": 129610 }, { "epoch": 82.61312938177183, "grad_norm": 0.3836192488670349, "learning_rate": 6.512602447056238e-05, "loss": 0.0193, "step": 129620 }, { "epoch": 82.61950286806884, "grad_norm": 0.26287126541137695, "learning_rate": 6.51207710687176e-05, "loss": 0.0142, "step": 129630 }, { "epoch": 82.62587635436584, "grad_norm": 0.2941252291202545, "learning_rate": 6.511551748314117e-05, "loss": 0.0148, "step": 129640 }, { "epoch": 82.63224984066284, "grad_norm": 0.4325369894504547, "learning_rate": 6.511026371389693e-05, "loss": 0.0126, "step": 129650 }, { "epoch": 82.63862332695984, "grad_norm": 0.35974374413490295, "learning_rate": 6.510500976104872e-05, "loss": 0.0147, "step": 129660 }, { "epoch": 82.64499681325685, "grad_norm": 0.8904023766517639, "learning_rate": 6.50997556246604e-05, "loss": 0.0244, "step": 129670 }, { "epoch": 82.65137029955386, "grad_norm": 0.4447114169597626, "learning_rate": 6.509450130479577e-05, "loss": 0.0179, "step": 129680 }, { "epoch": 82.65774378585085, "grad_norm": 0.378740519285202, "learning_rate": 6.50892468015187e-05, "loss": 0.0162, "step": 129690 }, { "epoch": 82.66411727214786, "grad_norm": 0.6596896052360535, "learning_rate": 6.508399211489303e-05, "loss": 0.0219, "step": 129700 }, { "epoch": 82.67049075844487, "grad_norm": 0.4516632854938507, "learning_rate": 6.507873724498264e-05, "loss": 0.0134, "step": 129710 }, { "epoch": 82.67686424474188, "grad_norm": 0.45621487498283386, "learning_rate": 6.507348219185134e-05, "loss": 0.0122, "step": 129720 }, { "epoch": 82.68323773103887, "grad_norm": 0.3335203528404236, "learning_rate": 6.506822695556298e-05, "loss": 0.0242, "step": 129730 }, { "epoch": 82.68961121733588, "grad_norm": 0.46405524015426636, "learning_rate": 6.506297153618147e-05, "loss": 0.0154, "step": 129740 }, { "epoch": 82.69598470363289, "grad_norm": 0.30529388785362244, "learning_rate": 6.505771593377061e-05, "loss": 0.0166, "step": 129750 }, { "epoch": 82.7023581899299, "grad_norm": 0.6802051663398743, "learning_rate": 6.50524601483943e-05, "loss": 0.0158, "step": 129760 }, { "epoch": 82.70873167622689, "grad_norm": 0.4951118528842926, "learning_rate": 6.504720418011637e-05, "loss": 0.0152, "step": 129770 }, { "epoch": 82.7151051625239, "grad_norm": 0.4721625745296478, "learning_rate": 6.504194802900072e-05, "loss": 0.0129, "step": 129780 }, { "epoch": 82.7214786488209, "grad_norm": 0.4822039008140564, "learning_rate": 6.50366916951112e-05, "loss": 0.0277, "step": 129790 }, { "epoch": 82.72785213511791, "grad_norm": 0.3109472990036011, "learning_rate": 6.503143517851167e-05, "loss": 0.0108, "step": 129800 }, { "epoch": 82.73422562141491, "grad_norm": 0.48140203952789307, "learning_rate": 6.5026178479266e-05, "loss": 0.0155, "step": 129810 }, { "epoch": 82.74059910771192, "grad_norm": 0.39591076970100403, "learning_rate": 6.502092159743809e-05, "loss": 0.0209, "step": 129820 }, { "epoch": 82.74697259400892, "grad_norm": 0.23049214482307434, "learning_rate": 6.50156645330918e-05, "loss": 0.0128, "step": 129830 }, { "epoch": 82.75334608030593, "grad_norm": 0.2730417251586914, "learning_rate": 6.5010407286291e-05, "loss": 0.0293, "step": 129840 }, { "epoch": 82.75971956660293, "grad_norm": 0.24825452268123627, "learning_rate": 6.50051498570996e-05, "loss": 0.0175, "step": 129850 }, { "epoch": 82.76609305289993, "grad_norm": 0.5648978352546692, "learning_rate": 6.499989224558143e-05, "loss": 0.0121, "step": 129860 }, { "epoch": 82.77246653919694, "grad_norm": 0.3948519229888916, "learning_rate": 6.499463445180043e-05, "loss": 0.0106, "step": 129870 }, { "epoch": 82.77884002549395, "grad_norm": 0.6222568154335022, "learning_rate": 6.498937647582046e-05, "loss": 0.0169, "step": 129880 }, { "epoch": 82.78521351179094, "grad_norm": 0.43854808807373047, "learning_rate": 6.49841183177054e-05, "loss": 0.0239, "step": 129890 }, { "epoch": 82.79158699808795, "grad_norm": 0.37719249725341797, "learning_rate": 6.497885997751918e-05, "loss": 0.018, "step": 129900 }, { "epoch": 82.79796048438496, "grad_norm": 0.4126072824001312, "learning_rate": 6.497360145532565e-05, "loss": 0.0136, "step": 129910 }, { "epoch": 82.80433397068197, "grad_norm": 0.442438542842865, "learning_rate": 6.496834275118872e-05, "loss": 0.0142, "step": 129920 }, { "epoch": 82.81070745697897, "grad_norm": 0.44087305665016174, "learning_rate": 6.496308386517232e-05, "loss": 0.0101, "step": 129930 }, { "epoch": 82.81708094327597, "grad_norm": 0.46413296461105347, "learning_rate": 6.49578247973403e-05, "loss": 0.0097, "step": 129940 }, { "epoch": 82.82345442957298, "grad_norm": 0.2547758221626282, "learning_rate": 6.49525655477566e-05, "loss": 0.0105, "step": 129950 }, { "epoch": 82.82982791586998, "grad_norm": 0.5181885957717896, "learning_rate": 6.494730611648509e-05, "loss": 0.0163, "step": 129960 }, { "epoch": 82.83620140216699, "grad_norm": 0.2818664610385895, "learning_rate": 6.494204650358973e-05, "loss": 0.0128, "step": 129970 }, { "epoch": 82.84257488846399, "grad_norm": 0.6530259847640991, "learning_rate": 6.493678670913436e-05, "loss": 0.0127, "step": 129980 }, { "epoch": 82.848948374761, "grad_norm": 0.30274635553359985, "learning_rate": 6.493152673318296e-05, "loss": 0.029, "step": 129990 }, { "epoch": 82.855321861058, "grad_norm": 0.5398173928260803, "learning_rate": 6.49262665757994e-05, "loss": 0.0272, "step": 130000 }, { "epoch": 82.86169534735501, "grad_norm": 0.31484073400497437, "learning_rate": 6.49210062370476e-05, "loss": 0.0146, "step": 130010 }, { "epoch": 82.868068833652, "grad_norm": 0.338051438331604, "learning_rate": 6.491574571699149e-05, "loss": 0.0147, "step": 130020 }, { "epoch": 82.87444231994901, "grad_norm": 0.4722456634044647, "learning_rate": 6.4910485015695e-05, "loss": 0.0237, "step": 130030 }, { "epoch": 82.88081580624602, "grad_norm": 0.20754572749137878, "learning_rate": 6.490522413322205e-05, "loss": 0.0111, "step": 130040 }, { "epoch": 82.88718929254303, "grad_norm": 0.25654107332229614, "learning_rate": 6.489996306963652e-05, "loss": 0.0155, "step": 130050 }, { "epoch": 82.89356277884002, "grad_norm": 0.43719482421875, "learning_rate": 6.489470182500239e-05, "loss": 0.0136, "step": 130060 }, { "epoch": 82.89993626513703, "grad_norm": 0.4859764277935028, "learning_rate": 6.488944039938355e-05, "loss": 0.0172, "step": 130070 }, { "epoch": 82.90630975143404, "grad_norm": 0.35515403747558594, "learning_rate": 6.488417879284399e-05, "loss": 0.0269, "step": 130080 }, { "epoch": 82.91268323773104, "grad_norm": 0.4701721966266632, "learning_rate": 6.487891700544757e-05, "loss": 0.0114, "step": 130090 }, { "epoch": 82.91905672402804, "grad_norm": 0.2615286707878113, "learning_rate": 6.487365503725827e-05, "loss": 0.014, "step": 130100 }, { "epoch": 82.92543021032505, "grad_norm": 0.27387645840644836, "learning_rate": 6.486839288834003e-05, "loss": 0.0105, "step": 130110 }, { "epoch": 82.93180369662205, "grad_norm": 0.5262463092803955, "learning_rate": 6.486313055875676e-05, "loss": 0.0179, "step": 130120 }, { "epoch": 82.93817718291906, "grad_norm": 0.5102649331092834, "learning_rate": 6.485786804857244e-05, "loss": 0.0211, "step": 130130 }, { "epoch": 82.94455066921606, "grad_norm": 0.2652188539505005, "learning_rate": 6.485260535785099e-05, "loss": 0.0179, "step": 130140 }, { "epoch": 82.95092415551306, "grad_norm": 0.22846364974975586, "learning_rate": 6.484734248665634e-05, "loss": 0.0208, "step": 130150 }, { "epoch": 82.95729764181007, "grad_norm": 0.35175174474716187, "learning_rate": 6.484207943505249e-05, "loss": 0.0163, "step": 130160 }, { "epoch": 82.96367112810708, "grad_norm": 0.3670831620693207, "learning_rate": 6.483681620310334e-05, "loss": 0.0195, "step": 130170 }, { "epoch": 82.97004461440407, "grad_norm": 0.21414296329021454, "learning_rate": 6.483155279087287e-05, "loss": 0.0161, "step": 130180 }, { "epoch": 82.97641810070108, "grad_norm": 0.38655275106430054, "learning_rate": 6.482628919842503e-05, "loss": 0.0269, "step": 130190 }, { "epoch": 82.98279158699809, "grad_norm": 0.5764787793159485, "learning_rate": 6.48210254258238e-05, "loss": 0.0166, "step": 130200 }, { "epoch": 82.9891650732951, "grad_norm": 0.23272280395030975, "learning_rate": 6.481576147313308e-05, "loss": 0.0165, "step": 130210 }, { "epoch": 82.99553855959209, "grad_norm": 0.3530321419239044, "learning_rate": 6.48104973404169e-05, "loss": 0.0178, "step": 130220 }, { "epoch": 83.0019120458891, "grad_norm": 0.5218722820281982, "learning_rate": 6.480523302773918e-05, "loss": 0.0168, "step": 130230 }, { "epoch": 83.00828553218611, "grad_norm": 0.31366652250289917, "learning_rate": 6.47999685351639e-05, "loss": 0.0139, "step": 130240 }, { "epoch": 83.01465901848312, "grad_norm": 0.3952755928039551, "learning_rate": 6.479470386275503e-05, "loss": 0.0198, "step": 130250 }, { "epoch": 83.02103250478011, "grad_norm": 0.319271445274353, "learning_rate": 6.478943901057654e-05, "loss": 0.0164, "step": 130260 }, { "epoch": 83.02740599107712, "grad_norm": 0.2707465887069702, "learning_rate": 6.478417397869241e-05, "loss": 0.0093, "step": 130270 }, { "epoch": 83.03377947737413, "grad_norm": 0.43230336904525757, "learning_rate": 6.47789087671666e-05, "loss": 0.0164, "step": 130280 }, { "epoch": 83.04015296367113, "grad_norm": 0.2873488664627075, "learning_rate": 6.477364337606309e-05, "loss": 0.0109, "step": 130290 }, { "epoch": 83.04652644996813, "grad_norm": 0.4224412739276886, "learning_rate": 6.476837780544588e-05, "loss": 0.0305, "step": 130300 }, { "epoch": 83.05289993626513, "grad_norm": 0.24483788013458252, "learning_rate": 6.476311205537892e-05, "loss": 0.0136, "step": 130310 }, { "epoch": 83.05927342256214, "grad_norm": 0.32233986258506775, "learning_rate": 6.475784612592622e-05, "loss": 0.0242, "step": 130320 }, { "epoch": 83.06564690885915, "grad_norm": 0.5004836916923523, "learning_rate": 6.475258001715175e-05, "loss": 0.0269, "step": 130330 }, { "epoch": 83.07202039515614, "grad_norm": 0.46104440093040466, "learning_rate": 6.47473137291195e-05, "loss": 0.0272, "step": 130340 }, { "epoch": 83.07839388145315, "grad_norm": 0.47214803099632263, "learning_rate": 6.474204726189348e-05, "loss": 0.0177, "step": 130350 }, { "epoch": 83.08476736775016, "grad_norm": 0.7090554237365723, "learning_rate": 6.473678061553765e-05, "loss": 0.0194, "step": 130360 }, { "epoch": 83.09114085404717, "grad_norm": 0.5701329708099365, "learning_rate": 6.473151379011603e-05, "loss": 0.02, "step": 130370 }, { "epoch": 83.09751434034416, "grad_norm": 0.4428839087486267, "learning_rate": 6.47262467856926e-05, "loss": 0.0136, "step": 130380 }, { "epoch": 83.10388782664117, "grad_norm": 0.23075000941753387, "learning_rate": 6.472097960233139e-05, "loss": 0.0144, "step": 130390 }, { "epoch": 83.11026131293818, "grad_norm": 0.3052363693714142, "learning_rate": 6.471571224009636e-05, "loss": 0.0191, "step": 130400 }, { "epoch": 83.11663479923519, "grad_norm": 0.32743799686431885, "learning_rate": 6.471044469905153e-05, "loss": 0.0323, "step": 130410 }, { "epoch": 83.12300828553218, "grad_norm": 0.6052147746086121, "learning_rate": 6.470517697926093e-05, "loss": 0.0194, "step": 130420 }, { "epoch": 83.12938177182919, "grad_norm": 0.49153396487236023, "learning_rate": 6.469990908078854e-05, "loss": 0.0109, "step": 130430 }, { "epoch": 83.1357552581262, "grad_norm": 0.32386717200279236, "learning_rate": 6.469464100369835e-05, "loss": 0.0137, "step": 130440 }, { "epoch": 83.1421287444232, "grad_norm": 0.3301803767681122, "learning_rate": 6.468937274805442e-05, "loss": 0.0159, "step": 130450 }, { "epoch": 83.1485022307202, "grad_norm": 0.2878493666648865, "learning_rate": 6.468410431392072e-05, "loss": 0.012, "step": 130460 }, { "epoch": 83.1548757170172, "grad_norm": 0.23186154663562775, "learning_rate": 6.467883570136132e-05, "loss": 0.0112, "step": 130470 }, { "epoch": 83.16124920331421, "grad_norm": 0.35240933299064636, "learning_rate": 6.467356691044019e-05, "loss": 0.0126, "step": 130480 }, { "epoch": 83.16762268961122, "grad_norm": 0.7586543560028076, "learning_rate": 6.466829794122136e-05, "loss": 0.0225, "step": 130490 }, { "epoch": 83.17399617590821, "grad_norm": 0.25580722093582153, "learning_rate": 6.466302879376887e-05, "loss": 0.0153, "step": 130500 }, { "epoch": 83.18036966220522, "grad_norm": 0.38005533814430237, "learning_rate": 6.465775946814674e-05, "loss": 0.0144, "step": 130510 }, { "epoch": 83.18674314850223, "grad_norm": 0.33872169256210327, "learning_rate": 6.465248996441899e-05, "loss": 0.0138, "step": 130520 }, { "epoch": 83.19311663479924, "grad_norm": 0.2315920889377594, "learning_rate": 6.464722028264966e-05, "loss": 0.0117, "step": 130530 }, { "epoch": 83.19949012109623, "grad_norm": 0.26399996876716614, "learning_rate": 6.464195042290277e-05, "loss": 0.0091, "step": 130540 }, { "epoch": 83.20586360739324, "grad_norm": 0.3699013292789459, "learning_rate": 6.463668038524234e-05, "loss": 0.021, "step": 130550 }, { "epoch": 83.21223709369025, "grad_norm": 0.21113309264183044, "learning_rate": 6.463141016973245e-05, "loss": 0.0099, "step": 130560 }, { "epoch": 83.21861057998726, "grad_norm": 0.6450787782669067, "learning_rate": 6.46261397764371e-05, "loss": 0.0136, "step": 130570 }, { "epoch": 83.22498406628426, "grad_norm": 0.2625298798084259, "learning_rate": 6.462086920542034e-05, "loss": 0.0132, "step": 130580 }, { "epoch": 83.23135755258126, "grad_norm": 0.23705129325389862, "learning_rate": 6.461559845674621e-05, "loss": 0.0177, "step": 130590 }, { "epoch": 83.23773103887827, "grad_norm": 0.5440493226051331, "learning_rate": 6.461032753047878e-05, "loss": 0.0222, "step": 130600 }, { "epoch": 83.24410452517527, "grad_norm": 0.23699304461479187, "learning_rate": 6.460505642668206e-05, "loss": 0.0171, "step": 130610 }, { "epoch": 83.25047801147228, "grad_norm": 0.42001861333847046, "learning_rate": 6.459978514542011e-05, "loss": 0.0138, "step": 130620 }, { "epoch": 83.25685149776928, "grad_norm": 0.3001210689544678, "learning_rate": 6.4594513686757e-05, "loss": 0.024, "step": 130630 }, { "epoch": 83.26322498406628, "grad_norm": 0.2653592526912689, "learning_rate": 6.458924205075676e-05, "loss": 0.0296, "step": 130640 }, { "epoch": 83.26959847036329, "grad_norm": 0.4558902680873871, "learning_rate": 6.458397023748346e-05, "loss": 0.0128, "step": 130650 }, { "epoch": 83.2759719566603, "grad_norm": 0.4752671420574188, "learning_rate": 6.457869824700116e-05, "loss": 0.0117, "step": 130660 }, { "epoch": 83.2823454429573, "grad_norm": 0.20817027986049652, "learning_rate": 6.45734260793739e-05, "loss": 0.0099, "step": 130670 }, { "epoch": 83.2887189292543, "grad_norm": 0.3515315353870392, "learning_rate": 6.456815373466575e-05, "loss": 0.024, "step": 130680 }, { "epoch": 83.29509241555131, "grad_norm": 0.5317285060882568, "learning_rate": 6.456288121294079e-05, "loss": 0.0113, "step": 130690 }, { "epoch": 83.30146590184832, "grad_norm": 0.26010337471961975, "learning_rate": 6.455760851426307e-05, "loss": 0.0098, "step": 130700 }, { "epoch": 83.30783938814531, "grad_norm": 0.40324848890304565, "learning_rate": 6.455233563869665e-05, "loss": 0.0267, "step": 130710 }, { "epoch": 83.31421287444232, "grad_norm": 0.15192781388759613, "learning_rate": 6.454706258630564e-05, "loss": 0.019, "step": 130720 }, { "epoch": 83.32058636073933, "grad_norm": 0.37542569637298584, "learning_rate": 6.454178935715405e-05, "loss": 0.0175, "step": 130730 }, { "epoch": 83.32695984703633, "grad_norm": 0.1971392035484314, "learning_rate": 6.4536515951306e-05, "loss": 0.0183, "step": 130740 }, { "epoch": 83.33333333333333, "grad_norm": 0.3407244384288788, "learning_rate": 6.453124236882557e-05, "loss": 0.0112, "step": 130750 }, { "epoch": 83.33970681963034, "grad_norm": 0.2081877589225769, "learning_rate": 6.452596860977681e-05, "loss": 0.0119, "step": 130760 }, { "epoch": 83.34608030592734, "grad_norm": 0.2930329442024231, "learning_rate": 6.452069467422384e-05, "loss": 0.0093, "step": 130770 }, { "epoch": 83.35245379222435, "grad_norm": 0.41556593775749207, "learning_rate": 6.451542056223069e-05, "loss": 0.0113, "step": 130780 }, { "epoch": 83.35882727852135, "grad_norm": 0.31988170742988586, "learning_rate": 6.451014627386149e-05, "loss": 0.0161, "step": 130790 }, { "epoch": 83.36520076481835, "grad_norm": 0.17635323107242584, "learning_rate": 6.45048718091803e-05, "loss": 0.0124, "step": 130800 }, { "epoch": 83.37157425111536, "grad_norm": 0.2683553993701935, "learning_rate": 6.449959716825124e-05, "loss": 0.0154, "step": 130810 }, { "epoch": 83.37794773741237, "grad_norm": 0.6304535865783691, "learning_rate": 6.449432235113839e-05, "loss": 0.0132, "step": 130820 }, { "epoch": 83.38432122370936, "grad_norm": 0.42291203141212463, "learning_rate": 6.448904735790582e-05, "loss": 0.0131, "step": 130830 }, { "epoch": 83.39069471000637, "grad_norm": 0.3073234558105469, "learning_rate": 6.448377218861765e-05, "loss": 0.0258, "step": 130840 }, { "epoch": 83.39706819630338, "grad_norm": 0.4783661365509033, "learning_rate": 6.447849684333797e-05, "loss": 0.0119, "step": 130850 }, { "epoch": 83.40344168260039, "grad_norm": 0.24375177919864655, "learning_rate": 6.44732213221309e-05, "loss": 0.0188, "step": 130860 }, { "epoch": 83.40981516889738, "grad_norm": 0.2408261001110077, "learning_rate": 6.44679456250605e-05, "loss": 0.0226, "step": 130870 }, { "epoch": 83.41618865519439, "grad_norm": 0.29230937361717224, "learning_rate": 6.446266975219093e-05, "loss": 0.0101, "step": 130880 }, { "epoch": 83.4225621414914, "grad_norm": 0.2254548966884613, "learning_rate": 6.445739370358625e-05, "loss": 0.0139, "step": 130890 }, { "epoch": 83.4289356277884, "grad_norm": 0.6339810490608215, "learning_rate": 6.44521174793106e-05, "loss": 0.0145, "step": 130900 }, { "epoch": 83.4353091140854, "grad_norm": 0.4953218996524811, "learning_rate": 6.444684107942805e-05, "loss": 0.0219, "step": 130910 }, { "epoch": 83.4416826003824, "grad_norm": 0.34992629289627075, "learning_rate": 6.444156450400276e-05, "loss": 0.0142, "step": 130920 }, { "epoch": 83.44805608667941, "grad_norm": 0.6577014923095703, "learning_rate": 6.443628775309883e-05, "loss": 0.0168, "step": 130930 }, { "epoch": 83.45442957297642, "grad_norm": 0.32354632019996643, "learning_rate": 6.443101082678036e-05, "loss": 0.0115, "step": 130940 }, { "epoch": 83.46080305927342, "grad_norm": 0.29779189825057983, "learning_rate": 6.44257337251115e-05, "loss": 0.0114, "step": 130950 }, { "epoch": 83.46717654557042, "grad_norm": 0.6189366579055786, "learning_rate": 6.442045644815634e-05, "loss": 0.0181, "step": 130960 }, { "epoch": 83.47355003186743, "grad_norm": 0.36857569217681885, "learning_rate": 6.441517899597904e-05, "loss": 0.0144, "step": 130970 }, { "epoch": 83.47992351816444, "grad_norm": 0.38785719871520996, "learning_rate": 6.440990136864368e-05, "loss": 0.0207, "step": 130980 }, { "epoch": 83.48629700446143, "grad_norm": 0.21416784822940826, "learning_rate": 6.440462356621441e-05, "loss": 0.0171, "step": 130990 }, { "epoch": 83.49267049075844, "grad_norm": 0.7020421028137207, "learning_rate": 6.439934558875537e-05, "loss": 0.0147, "step": 131000 }, { "epoch": 83.49904397705545, "grad_norm": 0.2223960906267166, "learning_rate": 6.439406743633071e-05, "loss": 0.0131, "step": 131010 }, { "epoch": 83.50541746335246, "grad_norm": 0.8205466866493225, "learning_rate": 6.438878910900452e-05, "loss": 0.0187, "step": 131020 }, { "epoch": 83.51179094964945, "grad_norm": 0.3020743727684021, "learning_rate": 6.438351060684096e-05, "loss": 0.0186, "step": 131030 }, { "epoch": 83.51816443594646, "grad_norm": 0.43736720085144043, "learning_rate": 6.437823192990415e-05, "loss": 0.0142, "step": 131040 }, { "epoch": 83.52453792224347, "grad_norm": 0.5254865884780884, "learning_rate": 6.437295307825828e-05, "loss": 0.0093, "step": 131050 }, { "epoch": 83.53091140854048, "grad_norm": 0.2729909420013428, "learning_rate": 6.436767405196743e-05, "loss": 0.0167, "step": 131060 }, { "epoch": 83.53728489483747, "grad_norm": 0.3024376928806305, "learning_rate": 6.43623948510958e-05, "loss": 0.0115, "step": 131070 }, { "epoch": 83.54365838113448, "grad_norm": 0.3646664321422577, "learning_rate": 6.435711547570748e-05, "loss": 0.0109, "step": 131080 }, { "epoch": 83.55003186743149, "grad_norm": 0.6028757691383362, "learning_rate": 6.435183592586669e-05, "loss": 0.0237, "step": 131090 }, { "epoch": 83.5564053537285, "grad_norm": 0.16825492680072784, "learning_rate": 6.43465562016375e-05, "loss": 0.0145, "step": 131100 }, { "epoch": 83.56277884002549, "grad_norm": 0.6626102328300476, "learning_rate": 6.434127630308416e-05, "loss": 0.0326, "step": 131110 }, { "epoch": 83.5691523263225, "grad_norm": 0.3023071885108948, "learning_rate": 6.433599623027072e-05, "loss": 0.0124, "step": 131120 }, { "epoch": 83.5755258126195, "grad_norm": 0.3567672371864319, "learning_rate": 6.433071598326142e-05, "loss": 0.0128, "step": 131130 }, { "epoch": 83.58189929891651, "grad_norm": 0.42792874574661255, "learning_rate": 6.432543556212038e-05, "loss": 0.0192, "step": 131140 }, { "epoch": 83.5882727852135, "grad_norm": 0.1426030844449997, "learning_rate": 6.432015496691179e-05, "loss": 0.0232, "step": 131150 }, { "epoch": 83.59464627151051, "grad_norm": 0.16682283580303192, "learning_rate": 6.431487419769978e-05, "loss": 0.0131, "step": 131160 }, { "epoch": 83.60101975780752, "grad_norm": 0.30909544229507446, "learning_rate": 6.430959325454852e-05, "loss": 0.0187, "step": 131170 }, { "epoch": 83.60739324410453, "grad_norm": 0.27484557032585144, "learning_rate": 6.430431213752222e-05, "loss": 0.0138, "step": 131180 }, { "epoch": 83.61376673040154, "grad_norm": 0.5512232780456543, "learning_rate": 6.4299030846685e-05, "loss": 0.0177, "step": 131190 }, { "epoch": 83.62014021669853, "grad_norm": 0.3164660334587097, "learning_rate": 6.429374938210108e-05, "loss": 0.013, "step": 131200 }, { "epoch": 83.62651370299554, "grad_norm": 0.5080949664115906, "learning_rate": 6.42884677438346e-05, "loss": 0.0129, "step": 131210 }, { "epoch": 83.63288718929255, "grad_norm": 0.650016725063324, "learning_rate": 6.428318593194974e-05, "loss": 0.0202, "step": 131220 }, { "epoch": 83.63926067558955, "grad_norm": 0.8462024331092834, "learning_rate": 6.427790394651067e-05, "loss": 0.016, "step": 131230 }, { "epoch": 83.64563416188655, "grad_norm": 0.5229950547218323, "learning_rate": 6.42726217875816e-05, "loss": 0.0195, "step": 131240 }, { "epoch": 83.65200764818356, "grad_norm": 0.3396455645561218, "learning_rate": 6.426733945522672e-05, "loss": 0.0155, "step": 131250 }, { "epoch": 83.65838113448056, "grad_norm": 0.3319931924343109, "learning_rate": 6.426205694951016e-05, "loss": 0.0154, "step": 131260 }, { "epoch": 83.66475462077757, "grad_norm": 0.5990366339683533, "learning_rate": 6.425677427049616e-05, "loss": 0.0163, "step": 131270 }, { "epoch": 83.67112810707457, "grad_norm": 0.4167865812778473, "learning_rate": 6.425149141824889e-05, "loss": 0.0107, "step": 131280 }, { "epoch": 83.67750159337157, "grad_norm": 0.27583926916122437, "learning_rate": 6.424620839283255e-05, "loss": 0.0121, "step": 131290 }, { "epoch": 83.68387507966858, "grad_norm": 0.252920925617218, "learning_rate": 6.424092519431132e-05, "loss": 0.0138, "step": 131300 }, { "epoch": 83.69024856596559, "grad_norm": 0.6350367665290833, "learning_rate": 6.423564182274942e-05, "loss": 0.0156, "step": 131310 }, { "epoch": 83.69662205226258, "grad_norm": 0.7880165576934814, "learning_rate": 6.423035827821101e-05, "loss": 0.016, "step": 131320 }, { "epoch": 83.70299553855959, "grad_norm": 0.4576292634010315, "learning_rate": 6.42250745607603e-05, "loss": 0.0129, "step": 131330 }, { "epoch": 83.7093690248566, "grad_norm": 0.30360135436058044, "learning_rate": 6.421979067046153e-05, "loss": 0.0139, "step": 131340 }, { "epoch": 83.7157425111536, "grad_norm": 0.32643282413482666, "learning_rate": 6.421450660737887e-05, "loss": 0.0169, "step": 131350 }, { "epoch": 83.7221159974506, "grad_norm": 0.42639976739883423, "learning_rate": 6.420922237157654e-05, "loss": 0.0206, "step": 131360 }, { "epoch": 83.72848948374761, "grad_norm": 0.4696943759918213, "learning_rate": 6.420393796311874e-05, "loss": 0.0209, "step": 131370 }, { "epoch": 83.73486297004462, "grad_norm": 0.36730653047561646, "learning_rate": 6.419865338206968e-05, "loss": 0.0118, "step": 131380 }, { "epoch": 83.74123645634162, "grad_norm": 0.38335904479026794, "learning_rate": 6.419336862849356e-05, "loss": 0.0132, "step": 131390 }, { "epoch": 83.74760994263862, "grad_norm": 0.32962265610694885, "learning_rate": 6.418808370245462e-05, "loss": 0.0228, "step": 131400 }, { "epoch": 83.75398342893563, "grad_norm": 0.5191961526870728, "learning_rate": 6.418279860401708e-05, "loss": 0.0211, "step": 131410 }, { "epoch": 83.76035691523263, "grad_norm": 0.44134435057640076, "learning_rate": 6.417751333324513e-05, "loss": 0.0246, "step": 131420 }, { "epoch": 83.76673040152964, "grad_norm": 0.4587678015232086, "learning_rate": 6.417222789020303e-05, "loss": 0.0123, "step": 131430 }, { "epoch": 83.77310388782664, "grad_norm": 0.3685089349746704, "learning_rate": 6.416694227495496e-05, "loss": 0.014, "step": 131440 }, { "epoch": 83.77947737412364, "grad_norm": 0.23504896461963654, "learning_rate": 6.416165648756517e-05, "loss": 0.0142, "step": 131450 }, { "epoch": 83.78585086042065, "grad_norm": 0.4158063232898712, "learning_rate": 6.415637052809788e-05, "loss": 0.0143, "step": 131460 }, { "epoch": 83.79222434671766, "grad_norm": 0.2853351831436157, "learning_rate": 6.415108439661733e-05, "loss": 0.014, "step": 131470 }, { "epoch": 83.79859783301465, "grad_norm": 0.3286244869232178, "learning_rate": 6.414579809318775e-05, "loss": 0.0155, "step": 131480 }, { "epoch": 83.80497131931166, "grad_norm": 0.4894074499607086, "learning_rate": 6.414051161787334e-05, "loss": 0.0183, "step": 131490 }, { "epoch": 83.81134480560867, "grad_norm": 0.2464938759803772, "learning_rate": 6.413522497073839e-05, "loss": 0.0116, "step": 131500 }, { "epoch": 83.81771829190568, "grad_norm": 0.15380670130252838, "learning_rate": 6.41299381518471e-05, "loss": 0.0121, "step": 131510 }, { "epoch": 83.82409177820267, "grad_norm": 0.5270677208900452, "learning_rate": 6.412465116126371e-05, "loss": 0.0123, "step": 131520 }, { "epoch": 83.83046526449968, "grad_norm": 0.35303524136543274, "learning_rate": 6.41193639990525e-05, "loss": 0.0105, "step": 131530 }, { "epoch": 83.83683875079669, "grad_norm": 0.36112990975379944, "learning_rate": 6.411407666527767e-05, "loss": 0.0202, "step": 131540 }, { "epoch": 83.8432122370937, "grad_norm": 0.3976000249385834, "learning_rate": 6.410878916000348e-05, "loss": 0.0129, "step": 131550 }, { "epoch": 83.84958572339069, "grad_norm": 0.3795400857925415, "learning_rate": 6.410350148329419e-05, "loss": 0.0116, "step": 131560 }, { "epoch": 83.8559592096877, "grad_norm": 0.31290340423583984, "learning_rate": 6.409821363521403e-05, "loss": 0.012, "step": 131570 }, { "epoch": 83.8623326959847, "grad_norm": 0.4476834535598755, "learning_rate": 6.409292561582727e-05, "loss": 0.015, "step": 131580 }, { "epoch": 83.86870618228171, "grad_norm": 0.4359557330608368, "learning_rate": 6.408763742519815e-05, "loss": 0.0228, "step": 131590 }, { "epoch": 83.8750796685787, "grad_norm": 0.3220438063144684, "learning_rate": 6.408234906339094e-05, "loss": 0.0149, "step": 131600 }, { "epoch": 83.88145315487571, "grad_norm": 0.3882409632205963, "learning_rate": 6.40770605304699e-05, "loss": 0.0147, "step": 131610 }, { "epoch": 83.88782664117272, "grad_norm": 0.34638234972953796, "learning_rate": 6.407177182649927e-05, "loss": 0.0289, "step": 131620 }, { "epoch": 83.89420012746973, "grad_norm": 0.3443375527858734, "learning_rate": 6.406648295154333e-05, "loss": 0.0115, "step": 131630 }, { "epoch": 83.90057361376672, "grad_norm": 0.23756569623947144, "learning_rate": 6.406119390566637e-05, "loss": 0.0102, "step": 131640 }, { "epoch": 83.90694710006373, "grad_norm": 0.2672894299030304, "learning_rate": 6.405590468893258e-05, "loss": 0.0184, "step": 131650 }, { "epoch": 83.91332058636074, "grad_norm": 0.3958364427089691, "learning_rate": 6.40506153014063e-05, "loss": 0.0136, "step": 131660 }, { "epoch": 83.91969407265775, "grad_norm": 0.2411256730556488, "learning_rate": 6.404532574315179e-05, "loss": 0.0205, "step": 131670 }, { "epoch": 83.92606755895474, "grad_norm": 0.3888923227787018, "learning_rate": 6.404003601423329e-05, "loss": 0.01, "step": 131680 }, { "epoch": 83.93244104525175, "grad_norm": 0.27967751026153564, "learning_rate": 6.40347461147151e-05, "loss": 0.0156, "step": 131690 }, { "epoch": 83.93881453154876, "grad_norm": 0.34002095460891724, "learning_rate": 6.40294560446615e-05, "loss": 0.011, "step": 131700 }, { "epoch": 83.94518801784577, "grad_norm": 0.38990098237991333, "learning_rate": 6.402416580413676e-05, "loss": 0.014, "step": 131710 }, { "epoch": 83.95156150414276, "grad_norm": 0.2602407932281494, "learning_rate": 6.401887539320517e-05, "loss": 0.0115, "step": 131720 }, { "epoch": 83.95793499043977, "grad_norm": 0.5918154716491699, "learning_rate": 6.4013584811931e-05, "loss": 0.0139, "step": 131730 }, { "epoch": 83.96430847673678, "grad_norm": 0.260928213596344, "learning_rate": 6.400829406037856e-05, "loss": 0.0112, "step": 131740 }, { "epoch": 83.97068196303378, "grad_norm": 0.5408462285995483, "learning_rate": 6.40030031386121e-05, "loss": 0.0135, "step": 131750 }, { "epoch": 83.97705544933078, "grad_norm": 0.43934616446495056, "learning_rate": 6.399771204669594e-05, "loss": 0.0135, "step": 131760 }, { "epoch": 83.98342893562778, "grad_norm": 0.27909010648727417, "learning_rate": 6.399242078469437e-05, "loss": 0.0173, "step": 131770 }, { "epoch": 83.9898024219248, "grad_norm": 0.2956847548484802, "learning_rate": 6.398712935267166e-05, "loss": 0.0152, "step": 131780 }, { "epoch": 83.9961759082218, "grad_norm": 0.555381178855896, "learning_rate": 6.398183775069214e-05, "loss": 0.0122, "step": 131790 }, { "epoch": 84.00254939451881, "grad_norm": 0.1625937521457672, "learning_rate": 6.397654597882008e-05, "loss": 0.0151, "step": 131800 }, { "epoch": 84.0089228808158, "grad_norm": 0.17557282745838165, "learning_rate": 6.397125403711979e-05, "loss": 0.0173, "step": 131810 }, { "epoch": 84.01529636711281, "grad_norm": 0.5437086820602417, "learning_rate": 6.396596192565557e-05, "loss": 0.026, "step": 131820 }, { "epoch": 84.02166985340982, "grad_norm": 0.308824360370636, "learning_rate": 6.396066964449173e-05, "loss": 0.0181, "step": 131830 }, { "epoch": 84.02804333970683, "grad_norm": 0.41311559081077576, "learning_rate": 6.395537719369257e-05, "loss": 0.0142, "step": 131840 }, { "epoch": 84.03441682600382, "grad_norm": 0.42375484108924866, "learning_rate": 6.395008457332241e-05, "loss": 0.0179, "step": 131850 }, { "epoch": 84.04079031230083, "grad_norm": 0.5022709965705872, "learning_rate": 6.394479178344554e-05, "loss": 0.0189, "step": 131860 }, { "epoch": 84.04716379859784, "grad_norm": 0.18057653307914734, "learning_rate": 6.393949882412629e-05, "loss": 0.01, "step": 131870 }, { "epoch": 84.05353728489484, "grad_norm": 0.3134704530239105, "learning_rate": 6.393420569542897e-05, "loss": 0.0204, "step": 131880 }, { "epoch": 84.05991077119184, "grad_norm": 0.31656214594841003, "learning_rate": 6.39289123974179e-05, "loss": 0.019, "step": 131890 }, { "epoch": 84.06628425748885, "grad_norm": 0.20310260355472565, "learning_rate": 6.392361893015737e-05, "loss": 0.0133, "step": 131900 }, { "epoch": 84.07265774378585, "grad_norm": 0.6723842620849609, "learning_rate": 6.391832529371176e-05, "loss": 0.013, "step": 131910 }, { "epoch": 84.07903123008286, "grad_norm": 0.3332955539226532, "learning_rate": 6.391303148814534e-05, "loss": 0.0154, "step": 131920 }, { "epoch": 84.08540471637986, "grad_norm": 0.4290412664413452, "learning_rate": 6.390773751352243e-05, "loss": 0.0175, "step": 131930 }, { "epoch": 84.09177820267686, "grad_norm": 0.435187429189682, "learning_rate": 6.390244336990741e-05, "loss": 0.0204, "step": 131940 }, { "epoch": 84.09815168897387, "grad_norm": 0.5755150318145752, "learning_rate": 6.389714905736456e-05, "loss": 0.0243, "step": 131950 }, { "epoch": 84.10452517527088, "grad_norm": 0.36383485794067383, "learning_rate": 6.389185457595823e-05, "loss": 0.0139, "step": 131960 }, { "epoch": 84.11089866156787, "grad_norm": 0.2933783531188965, "learning_rate": 6.388655992575275e-05, "loss": 0.01, "step": 131970 }, { "epoch": 84.11727214786488, "grad_norm": 0.2072402834892273, "learning_rate": 6.388126510681247e-05, "loss": 0.0124, "step": 131980 }, { "epoch": 84.12364563416189, "grad_norm": 0.43248656392097473, "learning_rate": 6.38759701192017e-05, "loss": 0.0126, "step": 131990 }, { "epoch": 84.1300191204589, "grad_norm": 0.37408560514450073, "learning_rate": 6.387067496298479e-05, "loss": 0.0119, "step": 132000 }, { "epoch": 84.13639260675589, "grad_norm": 0.4798544645309448, "learning_rate": 6.386537963822607e-05, "loss": 0.0111, "step": 132010 }, { "epoch": 84.1427660930529, "grad_norm": 0.37076908349990845, "learning_rate": 6.386008414498991e-05, "loss": 0.0179, "step": 132020 }, { "epoch": 84.1491395793499, "grad_norm": 0.49032679200172424, "learning_rate": 6.385478848334066e-05, "loss": 0.0145, "step": 132030 }, { "epoch": 84.15551306564691, "grad_norm": 0.384633332490921, "learning_rate": 6.384949265334263e-05, "loss": 0.0121, "step": 132040 }, { "epoch": 84.16188655194391, "grad_norm": 0.26955392956733704, "learning_rate": 6.384419665506019e-05, "loss": 0.0223, "step": 132050 }, { "epoch": 84.16826003824092, "grad_norm": 0.41410180926322937, "learning_rate": 6.38389004885577e-05, "loss": 0.0107, "step": 132060 }, { "epoch": 84.17463352453792, "grad_norm": 0.3894641101360321, "learning_rate": 6.383360415389949e-05, "loss": 0.0169, "step": 132070 }, { "epoch": 84.18100701083493, "grad_norm": 0.33781740069389343, "learning_rate": 6.382830765114992e-05, "loss": 0.0116, "step": 132080 }, { "epoch": 84.18738049713193, "grad_norm": 0.2935221791267395, "learning_rate": 6.382301098037338e-05, "loss": 0.0116, "step": 132090 }, { "epoch": 84.19375398342893, "grad_norm": 0.4740540087223053, "learning_rate": 6.381771414163418e-05, "loss": 0.0152, "step": 132100 }, { "epoch": 84.20012746972594, "grad_norm": 0.3439091145992279, "learning_rate": 6.381241713499673e-05, "loss": 0.0127, "step": 132110 }, { "epoch": 84.20650095602295, "grad_norm": 0.435615599155426, "learning_rate": 6.380711996052535e-05, "loss": 0.0153, "step": 132120 }, { "epoch": 84.21287444231994, "grad_norm": 0.4336388409137726, "learning_rate": 6.380182261828445e-05, "loss": 0.019, "step": 132130 }, { "epoch": 84.21924792861695, "grad_norm": 0.43241459131240845, "learning_rate": 6.379652510833836e-05, "loss": 0.01, "step": 132140 }, { "epoch": 84.22562141491396, "grad_norm": 0.40375828742980957, "learning_rate": 6.379122743075145e-05, "loss": 0.0175, "step": 132150 }, { "epoch": 84.23199490121097, "grad_norm": 0.33931031823158264, "learning_rate": 6.378592958558814e-05, "loss": 0.0216, "step": 132160 }, { "epoch": 84.23836838750796, "grad_norm": 0.6289101839065552, "learning_rate": 6.378063157291275e-05, "loss": 0.0151, "step": 132170 }, { "epoch": 84.24474187380497, "grad_norm": 0.293766051530838, "learning_rate": 6.377533339278967e-05, "loss": 0.0097, "step": 132180 }, { "epoch": 84.25111536010198, "grad_norm": 0.4110451638698578, "learning_rate": 6.377003504528328e-05, "loss": 0.013, "step": 132190 }, { "epoch": 84.25748884639899, "grad_norm": 0.23686547577381134, "learning_rate": 6.376473653045798e-05, "loss": 0.0151, "step": 132200 }, { "epoch": 84.26386233269598, "grad_norm": 0.3428138494491577, "learning_rate": 6.375943784837813e-05, "loss": 0.0144, "step": 132210 }, { "epoch": 84.27023581899299, "grad_norm": 0.44701525568962097, "learning_rate": 6.375413899910812e-05, "loss": 0.0251, "step": 132220 }, { "epoch": 84.27660930529, "grad_norm": 0.3759009540081024, "learning_rate": 6.374883998271233e-05, "loss": 0.0158, "step": 132230 }, { "epoch": 84.282982791587, "grad_norm": 0.34278950095176697, "learning_rate": 6.374354079925515e-05, "loss": 0.0137, "step": 132240 }, { "epoch": 84.289356277884, "grad_norm": 0.6174064874649048, "learning_rate": 6.373824144880098e-05, "loss": 0.0125, "step": 132250 }, { "epoch": 84.295729764181, "grad_norm": 0.2763179838657379, "learning_rate": 6.373294193141419e-05, "loss": 0.0087, "step": 132260 }, { "epoch": 84.30210325047801, "grad_norm": 0.3914702832698822, "learning_rate": 6.372764224715922e-05, "loss": 0.018, "step": 132270 }, { "epoch": 84.30847673677502, "grad_norm": 0.7302253246307373, "learning_rate": 6.37223423961004e-05, "loss": 0.0143, "step": 132280 }, { "epoch": 84.31485022307201, "grad_norm": 0.27895912528038025, "learning_rate": 6.371704237830219e-05, "loss": 0.0091, "step": 132290 }, { "epoch": 84.32122370936902, "grad_norm": 0.6531727910041809, "learning_rate": 6.371174219382897e-05, "loss": 0.0179, "step": 132300 }, { "epoch": 84.32759719566603, "grad_norm": 0.586678147315979, "learning_rate": 6.37064418427451e-05, "loss": 0.0116, "step": 132310 }, { "epoch": 84.33397068196304, "grad_norm": 0.41012513637542725, "learning_rate": 6.370114132511505e-05, "loss": 0.0198, "step": 132320 }, { "epoch": 84.34034416826003, "grad_norm": 0.44647157192230225, "learning_rate": 6.369584064100318e-05, "loss": 0.0201, "step": 132330 }, { "epoch": 84.34671765455704, "grad_norm": 0.5076871514320374, "learning_rate": 6.369053979047395e-05, "loss": 0.0112, "step": 132340 }, { "epoch": 84.35309114085405, "grad_norm": 0.5035755038261414, "learning_rate": 6.368523877359169e-05, "loss": 0.0106, "step": 132350 }, { "epoch": 84.35946462715106, "grad_norm": 0.3012668490409851, "learning_rate": 6.36799375904209e-05, "loss": 0.0113, "step": 132360 }, { "epoch": 84.36583811344805, "grad_norm": 0.3172634541988373, "learning_rate": 6.367463624102592e-05, "loss": 0.0125, "step": 132370 }, { "epoch": 84.37221159974506, "grad_norm": 0.2987251877784729, "learning_rate": 6.366933472547121e-05, "loss": 0.0241, "step": 132380 }, { "epoch": 84.37858508604207, "grad_norm": 0.22357140481472015, "learning_rate": 6.366403304382118e-05, "loss": 0.0097, "step": 132390 }, { "epoch": 84.38495857233907, "grad_norm": 0.18303710222244263, "learning_rate": 6.365873119614024e-05, "loss": 0.0154, "step": 132400 }, { "epoch": 84.39133205863607, "grad_norm": 0.5183787941932678, "learning_rate": 6.365342918249283e-05, "loss": 0.0189, "step": 132410 }, { "epoch": 84.39770554493307, "grad_norm": 0.4597075879573822, "learning_rate": 6.364812700294334e-05, "loss": 0.016, "step": 132420 }, { "epoch": 84.40407903123008, "grad_norm": 0.317028671503067, "learning_rate": 6.364282465755626e-05, "loss": 0.0084, "step": 132430 }, { "epoch": 84.41045251752709, "grad_norm": 0.37524738907814026, "learning_rate": 6.363752214639595e-05, "loss": 0.015, "step": 132440 }, { "epoch": 84.4168260038241, "grad_norm": 0.4002605974674225, "learning_rate": 6.363221946952688e-05, "loss": 0.0156, "step": 132450 }, { "epoch": 84.42319949012109, "grad_norm": 0.4532420337200165, "learning_rate": 6.362691662701347e-05, "loss": 0.0172, "step": 132460 }, { "epoch": 84.4295729764181, "grad_norm": 0.41567230224609375, "learning_rate": 6.362161361892017e-05, "loss": 0.0164, "step": 132470 }, { "epoch": 84.43594646271511, "grad_norm": 0.6296383738517761, "learning_rate": 6.361631044531138e-05, "loss": 0.0205, "step": 132480 }, { "epoch": 84.44231994901212, "grad_norm": 0.6126928925514221, "learning_rate": 6.361100710625158e-05, "loss": 0.0112, "step": 132490 }, { "epoch": 84.44869343530911, "grad_norm": 0.3111700713634491, "learning_rate": 6.360570360180518e-05, "loss": 0.0164, "step": 132500 }, { "epoch": 84.45506692160612, "grad_norm": 0.1879359930753708, "learning_rate": 6.360039993203664e-05, "loss": 0.0088, "step": 132510 }, { "epoch": 84.46144040790313, "grad_norm": 0.1623217612504959, "learning_rate": 6.35950960970104e-05, "loss": 0.0127, "step": 132520 }, { "epoch": 84.46781389420013, "grad_norm": 0.38969889283180237, "learning_rate": 6.358979209679091e-05, "loss": 0.0165, "step": 132530 }, { "epoch": 84.47418738049713, "grad_norm": 0.44266587495803833, "learning_rate": 6.35844879314426e-05, "loss": 0.0168, "step": 132540 }, { "epoch": 84.48056086679414, "grad_norm": 0.8355053067207336, "learning_rate": 6.357918360102996e-05, "loss": 0.0153, "step": 132550 }, { "epoch": 84.48693435309114, "grad_norm": 0.2436361163854599, "learning_rate": 6.357387910561741e-05, "loss": 0.0141, "step": 132560 }, { "epoch": 84.49330783938815, "grad_norm": 0.3407157361507416, "learning_rate": 6.356857444526941e-05, "loss": 0.0165, "step": 132570 }, { "epoch": 84.49968132568515, "grad_norm": 0.21425671875476837, "learning_rate": 6.356326962005041e-05, "loss": 0.0146, "step": 132580 }, { "epoch": 84.50605481198215, "grad_norm": 0.4523334205150604, "learning_rate": 6.355796463002491e-05, "loss": 0.0157, "step": 132590 }, { "epoch": 84.51242829827916, "grad_norm": 0.24077092111110687, "learning_rate": 6.35526594752573e-05, "loss": 0.0219, "step": 132600 }, { "epoch": 84.51880178457617, "grad_norm": 0.3217729926109314, "learning_rate": 6.35473541558121e-05, "loss": 0.0245, "step": 132610 }, { "epoch": 84.52517527087316, "grad_norm": 0.2782188951969147, "learning_rate": 6.354204867175376e-05, "loss": 0.0179, "step": 132620 }, { "epoch": 84.53154875717017, "grad_norm": 0.5291478633880615, "learning_rate": 6.353674302314674e-05, "loss": 0.0127, "step": 132630 }, { "epoch": 84.53792224346718, "grad_norm": 0.35054436326026917, "learning_rate": 6.353143721005551e-05, "loss": 0.0115, "step": 132640 }, { "epoch": 84.54429572976419, "grad_norm": 0.30296698212623596, "learning_rate": 6.352613123254454e-05, "loss": 0.0147, "step": 132650 }, { "epoch": 84.55066921606118, "grad_norm": 0.21419648826122284, "learning_rate": 6.35208250906783e-05, "loss": 0.0097, "step": 132660 }, { "epoch": 84.55704270235819, "grad_norm": 0.5438587069511414, "learning_rate": 6.351551878452127e-05, "loss": 0.0161, "step": 132670 }, { "epoch": 84.5634161886552, "grad_norm": 0.3780128061771393, "learning_rate": 6.351021231413794e-05, "loss": 0.0219, "step": 132680 }, { "epoch": 84.5697896749522, "grad_norm": 0.3828471302986145, "learning_rate": 6.350490567959277e-05, "loss": 0.016, "step": 132690 }, { "epoch": 84.5761631612492, "grad_norm": 0.4249163568019867, "learning_rate": 6.349959888095024e-05, "loss": 0.0211, "step": 132700 }, { "epoch": 84.5825366475462, "grad_norm": 0.42051926255226135, "learning_rate": 6.349429191827486e-05, "loss": 0.0232, "step": 132710 }, { "epoch": 84.58891013384321, "grad_norm": 0.36169934272766113, "learning_rate": 6.348898479163105e-05, "loss": 0.0188, "step": 132720 }, { "epoch": 84.59528362014022, "grad_norm": 0.33523333072662354, "learning_rate": 6.348367750108337e-05, "loss": 0.0145, "step": 132730 }, { "epoch": 84.60165710643722, "grad_norm": 0.23613344132900238, "learning_rate": 6.347837004669625e-05, "loss": 0.0119, "step": 132740 }, { "epoch": 84.60803059273422, "grad_norm": 0.42854073643684387, "learning_rate": 6.347306242853425e-05, "loss": 0.0206, "step": 132750 }, { "epoch": 84.61440407903123, "grad_norm": 0.6108058094978333, "learning_rate": 6.346775464666179e-05, "loss": 0.0127, "step": 132760 }, { "epoch": 84.62077756532824, "grad_norm": 0.37020063400268555, "learning_rate": 6.346244670114341e-05, "loss": 0.011, "step": 132770 }, { "epoch": 84.62715105162523, "grad_norm": 0.3274647891521454, "learning_rate": 6.345713859204359e-05, "loss": 0.0172, "step": 132780 }, { "epoch": 84.63352453792224, "grad_norm": 0.30806925892829895, "learning_rate": 6.345183031942681e-05, "loss": 0.0232, "step": 132790 }, { "epoch": 84.63989802421925, "grad_norm": 0.3634806275367737, "learning_rate": 6.344652188335762e-05, "loss": 0.0381, "step": 132800 }, { "epoch": 84.64627151051626, "grad_norm": 0.6183315515518188, "learning_rate": 6.344121328390047e-05, "loss": 0.0162, "step": 132810 }, { "epoch": 84.65264499681325, "grad_norm": 0.3550736904144287, "learning_rate": 6.34359045211199e-05, "loss": 0.02, "step": 132820 }, { "epoch": 84.65901848311026, "grad_norm": 0.2917729616165161, "learning_rate": 6.34305955950804e-05, "loss": 0.0136, "step": 132830 }, { "epoch": 84.66539196940727, "grad_norm": 0.6440348625183105, "learning_rate": 6.34252865058465e-05, "loss": 0.0107, "step": 132840 }, { "epoch": 84.67176545570427, "grad_norm": 0.22162780165672302, "learning_rate": 6.341997725348268e-05, "loss": 0.0195, "step": 132850 }, { "epoch": 84.67813894200127, "grad_norm": 0.38527876138687134, "learning_rate": 6.341466783805345e-05, "loss": 0.0197, "step": 132860 }, { "epoch": 84.68451242829828, "grad_norm": 0.2322668582201004, "learning_rate": 6.340935825962336e-05, "loss": 0.0131, "step": 132870 }, { "epoch": 84.69088591459528, "grad_norm": 0.2872855067253113, "learning_rate": 6.34040485182569e-05, "loss": 0.015, "step": 132880 }, { "epoch": 84.69725940089229, "grad_norm": 0.30338212847709656, "learning_rate": 6.33987386140186e-05, "loss": 0.0199, "step": 132890 }, { "epoch": 84.70363288718929, "grad_norm": 0.2765369117259979, "learning_rate": 6.339342854697297e-05, "loss": 0.0147, "step": 132900 }, { "epoch": 84.7100063734863, "grad_norm": 0.23655110597610474, "learning_rate": 6.338811831718454e-05, "loss": 0.0179, "step": 132910 }, { "epoch": 84.7163798597833, "grad_norm": 0.34605106711387634, "learning_rate": 6.338280792471782e-05, "loss": 0.0109, "step": 132920 }, { "epoch": 84.72275334608031, "grad_norm": 0.24316366016864777, "learning_rate": 6.337749736963738e-05, "loss": 0.0197, "step": 132930 }, { "epoch": 84.7291268323773, "grad_norm": 0.2694697678089142, "learning_rate": 6.33721866520077e-05, "loss": 0.016, "step": 132940 }, { "epoch": 84.73550031867431, "grad_norm": 0.28915509581565857, "learning_rate": 6.336687577189331e-05, "loss": 0.0152, "step": 132950 }, { "epoch": 84.74187380497132, "grad_norm": 0.3840300440788269, "learning_rate": 6.336156472935877e-05, "loss": 0.0152, "step": 132960 }, { "epoch": 84.74824729126833, "grad_norm": 0.2756035029888153, "learning_rate": 6.33562535244686e-05, "loss": 0.0135, "step": 132970 }, { "epoch": 84.75462077756532, "grad_norm": 0.6210448145866394, "learning_rate": 6.335094215728735e-05, "loss": 0.0189, "step": 132980 }, { "epoch": 84.76099426386233, "grad_norm": 0.2618992328643799, "learning_rate": 6.334563062787953e-05, "loss": 0.0102, "step": 132990 }, { "epoch": 84.76736775015934, "grad_norm": 0.5321381092071533, "learning_rate": 6.334031893630972e-05, "loss": 0.0209, "step": 133000 }, { "epoch": 84.77374123645635, "grad_norm": 0.20945508778095245, "learning_rate": 6.333500708264243e-05, "loss": 0.0183, "step": 133010 }, { "epoch": 84.78011472275334, "grad_norm": 0.45514699816703796, "learning_rate": 6.332969506694222e-05, "loss": 0.0129, "step": 133020 }, { "epoch": 84.78648820905035, "grad_norm": 0.2396371066570282, "learning_rate": 6.332438288927362e-05, "loss": 0.0072, "step": 133030 }, { "epoch": 84.79286169534736, "grad_norm": 0.37780600786209106, "learning_rate": 6.331907054970118e-05, "loss": 0.0277, "step": 133040 }, { "epoch": 84.79923518164436, "grad_norm": 0.44274020195007324, "learning_rate": 6.331375804828947e-05, "loss": 0.023, "step": 133050 }, { "epoch": 84.80560866794136, "grad_norm": 0.32657960057258606, "learning_rate": 6.330844538510304e-05, "loss": 0.0092, "step": 133060 }, { "epoch": 84.81198215423836, "grad_norm": 0.3026450276374817, "learning_rate": 6.330313256020641e-05, "loss": 0.0134, "step": 133070 }, { "epoch": 84.81835564053537, "grad_norm": 0.30539625883102417, "learning_rate": 6.329781957366416e-05, "loss": 0.0115, "step": 133080 }, { "epoch": 84.82472912683238, "grad_norm": 0.5972595810890198, "learning_rate": 6.329250642554087e-05, "loss": 0.0095, "step": 133090 }, { "epoch": 84.83110261312939, "grad_norm": 0.2875621020793915, "learning_rate": 6.328719311590107e-05, "loss": 0.0153, "step": 133100 }, { "epoch": 84.83747609942638, "grad_norm": 0.5255768895149231, "learning_rate": 6.328187964480931e-05, "loss": 0.0307, "step": 133110 }, { "epoch": 84.84384958572339, "grad_norm": 0.5431597828865051, "learning_rate": 6.327656601233018e-05, "loss": 0.0173, "step": 133120 }, { "epoch": 84.8502230720204, "grad_norm": 0.31676965951919556, "learning_rate": 6.327125221852825e-05, "loss": 0.0235, "step": 133130 }, { "epoch": 84.8565965583174, "grad_norm": 0.5273497700691223, "learning_rate": 6.326593826346806e-05, "loss": 0.0149, "step": 133140 }, { "epoch": 84.8629700446144, "grad_norm": 0.4538824260234833, "learning_rate": 6.326062414721419e-05, "loss": 0.0147, "step": 133150 }, { "epoch": 84.86934353091141, "grad_norm": 0.43102169036865234, "learning_rate": 6.325530986983124e-05, "loss": 0.0132, "step": 133160 }, { "epoch": 84.87571701720842, "grad_norm": 0.4645758867263794, "learning_rate": 6.324999543138374e-05, "loss": 0.0129, "step": 133170 }, { "epoch": 84.88209050350542, "grad_norm": 0.5049845576286316, "learning_rate": 6.324468083193631e-05, "loss": 0.0206, "step": 133180 }, { "epoch": 84.88846398980242, "grad_norm": 0.20524127781391144, "learning_rate": 6.323936607155348e-05, "loss": 0.0093, "step": 133190 }, { "epoch": 84.89483747609943, "grad_norm": 0.3368780314922333, "learning_rate": 6.323405115029985e-05, "loss": 0.0271, "step": 133200 }, { "epoch": 84.90121096239643, "grad_norm": 0.21306592226028442, "learning_rate": 6.322873606824001e-05, "loss": 0.0108, "step": 133210 }, { "epoch": 84.90758444869344, "grad_norm": 0.21386587619781494, "learning_rate": 6.322342082543856e-05, "loss": 0.0116, "step": 133220 }, { "epoch": 84.91395793499044, "grad_norm": 0.4438561201095581, "learning_rate": 6.321810542196002e-05, "loss": 0.0246, "step": 133230 }, { "epoch": 84.92033142128744, "grad_norm": 0.3850349187850952, "learning_rate": 6.321278985786904e-05, "loss": 0.0161, "step": 133240 }, { "epoch": 84.92670490758445, "grad_norm": 0.37547019124031067, "learning_rate": 6.320747413323019e-05, "loss": 0.0167, "step": 133250 }, { "epoch": 84.93307839388146, "grad_norm": 0.5453292727470398, "learning_rate": 6.320215824810807e-05, "loss": 0.0161, "step": 133260 }, { "epoch": 84.93945188017845, "grad_norm": 0.5185712575912476, "learning_rate": 6.319684220256723e-05, "loss": 0.0111, "step": 133270 }, { "epoch": 84.94582536647546, "grad_norm": 0.354581356048584, "learning_rate": 6.319152599667232e-05, "loss": 0.0132, "step": 133280 }, { "epoch": 84.95219885277247, "grad_norm": 0.47317975759506226, "learning_rate": 6.318620963048791e-05, "loss": 0.0228, "step": 133290 }, { "epoch": 84.95857233906948, "grad_norm": 0.4889851212501526, "learning_rate": 6.318089310407862e-05, "loss": 0.0141, "step": 133300 }, { "epoch": 84.96494582536647, "grad_norm": 0.5003365278244019, "learning_rate": 6.317557641750901e-05, "loss": 0.0203, "step": 133310 }, { "epoch": 84.97131931166348, "grad_norm": 0.49199649691581726, "learning_rate": 6.317025957084372e-05, "loss": 0.0218, "step": 133320 }, { "epoch": 84.97769279796049, "grad_norm": 0.31489062309265137, "learning_rate": 6.316494256414733e-05, "loss": 0.0178, "step": 133330 }, { "epoch": 84.9840662842575, "grad_norm": 0.46738216280937195, "learning_rate": 6.315962539748447e-05, "loss": 0.0224, "step": 133340 }, { "epoch": 84.99043977055449, "grad_norm": 0.34236806631088257, "learning_rate": 6.315430807091973e-05, "loss": 0.0122, "step": 133350 }, { "epoch": 84.9968132568515, "grad_norm": 0.4099246859550476, "learning_rate": 6.314899058451773e-05, "loss": 0.0187, "step": 133360 }, { "epoch": 85.0031867431485, "grad_norm": 0.621935248374939, "learning_rate": 6.314367293834309e-05, "loss": 0.0224, "step": 133370 }, { "epoch": 85.00956022944551, "grad_norm": 0.3139297068119049, "learning_rate": 6.313835513246038e-05, "loss": 0.012, "step": 133380 }, { "epoch": 85.0159337157425, "grad_norm": 0.5313336849212646, "learning_rate": 6.313303716693428e-05, "loss": 0.0139, "step": 133390 }, { "epoch": 85.02230720203951, "grad_norm": 0.34900930523872375, "learning_rate": 6.312771904182937e-05, "loss": 0.0175, "step": 133400 }, { "epoch": 85.02868068833652, "grad_norm": 0.315639853477478, "learning_rate": 6.312240075721029e-05, "loss": 0.01, "step": 133410 }, { "epoch": 85.03505417463353, "grad_norm": 0.5061761140823364, "learning_rate": 6.311708231314164e-05, "loss": 0.012, "step": 133420 }, { "epoch": 85.04142766093052, "grad_norm": 0.21184015274047852, "learning_rate": 6.311176370968806e-05, "loss": 0.0153, "step": 133430 }, { "epoch": 85.04780114722753, "grad_norm": 0.49752309918403625, "learning_rate": 6.310644494691417e-05, "loss": 0.0128, "step": 133440 }, { "epoch": 85.05417463352454, "grad_norm": 0.5022903084754944, "learning_rate": 6.31011260248846e-05, "loss": 0.0146, "step": 133450 }, { "epoch": 85.06054811982155, "grad_norm": 0.46964195370674133, "learning_rate": 6.309580694366397e-05, "loss": 0.024, "step": 133460 }, { "epoch": 85.06692160611854, "grad_norm": 0.23028768599033356, "learning_rate": 6.309048770331692e-05, "loss": 0.01, "step": 133470 }, { "epoch": 85.07329509241555, "grad_norm": 0.3037722706794739, "learning_rate": 6.30851683039081e-05, "loss": 0.0106, "step": 133480 }, { "epoch": 85.07966857871256, "grad_norm": 0.16038651764392853, "learning_rate": 6.307984874550213e-05, "loss": 0.017, "step": 133490 }, { "epoch": 85.08604206500956, "grad_norm": 0.23609168827533722, "learning_rate": 6.307452902816364e-05, "loss": 0.0184, "step": 133500 }, { "epoch": 85.09241555130656, "grad_norm": 0.4985032081604004, "learning_rate": 6.306920915195728e-05, "loss": 0.0322, "step": 133510 }, { "epoch": 85.09878903760357, "grad_norm": 0.8257176280021667, "learning_rate": 6.30638891169477e-05, "loss": 0.0239, "step": 133520 }, { "epoch": 85.10516252390057, "grad_norm": 0.2936190366744995, "learning_rate": 6.305856892319952e-05, "loss": 0.012, "step": 133530 }, { "epoch": 85.11153601019758, "grad_norm": 0.34523454308509827, "learning_rate": 6.305324857077739e-05, "loss": 0.0166, "step": 133540 }, { "epoch": 85.11790949649458, "grad_norm": 0.28248846530914307, "learning_rate": 6.304792805974599e-05, "loss": 0.0238, "step": 133550 }, { "epoch": 85.12428298279158, "grad_norm": 0.4663640558719635, "learning_rate": 6.304260739016992e-05, "loss": 0.026, "step": 133560 }, { "epoch": 85.13065646908859, "grad_norm": 0.25383907556533813, "learning_rate": 6.303728656211387e-05, "loss": 0.0103, "step": 133570 }, { "epoch": 85.1370299553856, "grad_norm": 0.30031919479370117, "learning_rate": 6.303196557564249e-05, "loss": 0.0122, "step": 133580 }, { "epoch": 85.1434034416826, "grad_norm": 0.657975971698761, "learning_rate": 6.302664443082041e-05, "loss": 0.0196, "step": 133590 }, { "epoch": 85.1497769279796, "grad_norm": 0.39710748195648193, "learning_rate": 6.30213231277123e-05, "loss": 0.0183, "step": 133600 }, { "epoch": 85.15615041427661, "grad_norm": 0.3141266405582428, "learning_rate": 6.30160016663828e-05, "loss": 0.0105, "step": 133610 }, { "epoch": 85.16252390057362, "grad_norm": 0.306724488735199, "learning_rate": 6.301068004689662e-05, "loss": 0.0153, "step": 133620 }, { "epoch": 85.16889738687061, "grad_norm": 0.7298356294631958, "learning_rate": 6.300535826931838e-05, "loss": 0.0176, "step": 133630 }, { "epoch": 85.17527087316762, "grad_norm": 0.35468268394470215, "learning_rate": 6.300003633371277e-05, "loss": 0.014, "step": 133640 }, { "epoch": 85.18164435946463, "grad_norm": 0.30317091941833496, "learning_rate": 6.299471424014443e-05, "loss": 0.0138, "step": 133650 }, { "epoch": 85.18801784576164, "grad_norm": 0.47335687279701233, "learning_rate": 6.298939198867806e-05, "loss": 0.0144, "step": 133660 }, { "epoch": 85.19439133205863, "grad_norm": 0.616987407207489, "learning_rate": 6.298406957937832e-05, "loss": 0.0137, "step": 133670 }, { "epoch": 85.20076481835564, "grad_norm": 0.29102134704589844, "learning_rate": 6.297874701230984e-05, "loss": 0.0097, "step": 133680 }, { "epoch": 85.20713830465264, "grad_norm": 0.5683382153511047, "learning_rate": 6.297342428753736e-05, "loss": 0.0138, "step": 133690 }, { "epoch": 85.21351179094965, "grad_norm": 0.32119202613830566, "learning_rate": 6.29681014051255e-05, "loss": 0.0123, "step": 133700 }, { "epoch": 85.21988527724666, "grad_norm": 0.5160893201828003, "learning_rate": 6.2962778365139e-05, "loss": 0.0236, "step": 133710 }, { "epoch": 85.22625876354365, "grad_norm": 0.35252028703689575, "learning_rate": 6.295745516764247e-05, "loss": 0.0077, "step": 133720 }, { "epoch": 85.23263224984066, "grad_norm": 0.31651997566223145, "learning_rate": 6.295213181270064e-05, "loss": 0.0187, "step": 133730 }, { "epoch": 85.23900573613767, "grad_norm": 0.7795909643173218, "learning_rate": 6.294680830037819e-05, "loss": 0.0248, "step": 133740 }, { "epoch": 85.24537922243468, "grad_norm": 0.35686424374580383, "learning_rate": 6.294148463073976e-05, "loss": 0.0123, "step": 133750 }, { "epoch": 85.25175270873167, "grad_norm": 0.42998915910720825, "learning_rate": 6.293616080385012e-05, "loss": 0.0088, "step": 133760 }, { "epoch": 85.25812619502868, "grad_norm": 0.27756747603416443, "learning_rate": 6.293083681977387e-05, "loss": 0.0139, "step": 133770 }, { "epoch": 85.26449968132569, "grad_norm": 0.5360267162322998, "learning_rate": 6.292551267857578e-05, "loss": 0.0123, "step": 133780 }, { "epoch": 85.2708731676227, "grad_norm": 0.37585267424583435, "learning_rate": 6.292018838032047e-05, "loss": 0.0125, "step": 133790 }, { "epoch": 85.27724665391969, "grad_norm": 0.431610643863678, "learning_rate": 6.29148639250727e-05, "loss": 0.0086, "step": 133800 }, { "epoch": 85.2836201402167, "grad_norm": 0.386801153421402, "learning_rate": 6.290953931289713e-05, "loss": 0.0103, "step": 133810 }, { "epoch": 85.2899936265137, "grad_norm": 0.3985896408557892, "learning_rate": 6.290421454385849e-05, "loss": 0.0111, "step": 133820 }, { "epoch": 85.29636711281071, "grad_norm": 0.630875289440155, "learning_rate": 6.289888961802143e-05, "loss": 0.0164, "step": 133830 }, { "epoch": 85.30274059910771, "grad_norm": 0.28274887800216675, "learning_rate": 6.28935645354507e-05, "loss": 0.0116, "step": 133840 }, { "epoch": 85.30911408540472, "grad_norm": 0.308476984500885, "learning_rate": 6.288823929621098e-05, "loss": 0.0171, "step": 133850 }, { "epoch": 85.31548757170172, "grad_norm": 0.4567665755748749, "learning_rate": 6.288291390036698e-05, "loss": 0.0158, "step": 133860 }, { "epoch": 85.32186105799873, "grad_norm": 0.3329651653766632, "learning_rate": 6.287758834798341e-05, "loss": 0.0187, "step": 133870 }, { "epoch": 85.32823454429573, "grad_norm": 0.40971994400024414, "learning_rate": 6.2872262639125e-05, "loss": 0.0223, "step": 133880 }, { "epoch": 85.33460803059273, "grad_norm": 0.25484341382980347, "learning_rate": 6.286693677385643e-05, "loss": 0.0118, "step": 133890 }, { "epoch": 85.34098151688974, "grad_norm": 0.48147693276405334, "learning_rate": 6.286161075224243e-05, "loss": 0.0148, "step": 133900 }, { "epoch": 85.34735500318675, "grad_norm": 0.336171418428421, "learning_rate": 6.285628457434772e-05, "loss": 0.0124, "step": 133910 }, { "epoch": 85.35372848948374, "grad_norm": 0.5017493963241577, "learning_rate": 6.285095824023703e-05, "loss": 0.0124, "step": 133920 }, { "epoch": 85.36010197578075, "grad_norm": 0.2354864925146103, "learning_rate": 6.284563174997505e-05, "loss": 0.0164, "step": 133930 }, { "epoch": 85.36647546207776, "grad_norm": 0.3976515829563141, "learning_rate": 6.28403051036265e-05, "loss": 0.0108, "step": 133940 }, { "epoch": 85.37284894837477, "grad_norm": 0.1592789739370346, "learning_rate": 6.283497830125614e-05, "loss": 0.0155, "step": 133950 }, { "epoch": 85.37922243467176, "grad_norm": 0.3049215078353882, "learning_rate": 6.282965134292868e-05, "loss": 0.0162, "step": 133960 }, { "epoch": 85.38559592096877, "grad_norm": 0.3121669292449951, "learning_rate": 6.282432422870883e-05, "loss": 0.0162, "step": 133970 }, { "epoch": 85.39196940726578, "grad_norm": 0.43142643570899963, "learning_rate": 6.281899695866133e-05, "loss": 0.0169, "step": 133980 }, { "epoch": 85.39834289356278, "grad_norm": 0.371711403131485, "learning_rate": 6.281366953285094e-05, "loss": 0.0182, "step": 133990 }, { "epoch": 85.40471637985978, "grad_norm": 0.4096178710460663, "learning_rate": 6.280834195134234e-05, "loss": 0.0145, "step": 134000 }, { "epoch": 85.41108986615679, "grad_norm": 0.8119091987609863, "learning_rate": 6.28030142142003e-05, "loss": 0.0275, "step": 134010 }, { "epoch": 85.4174633524538, "grad_norm": 0.3857542872428894, "learning_rate": 6.279768632148955e-05, "loss": 0.0211, "step": 134020 }, { "epoch": 85.4238368387508, "grad_norm": 0.354531854391098, "learning_rate": 6.279235827327484e-05, "loss": 0.0131, "step": 134030 }, { "epoch": 85.4302103250478, "grad_norm": 0.2181270718574524, "learning_rate": 6.278703006962087e-05, "loss": 0.0179, "step": 134040 }, { "epoch": 85.4365838113448, "grad_norm": 0.423397958278656, "learning_rate": 6.278170171059243e-05, "loss": 0.024, "step": 134050 }, { "epoch": 85.44295729764181, "grad_norm": 0.46015915274620056, "learning_rate": 6.277637319625425e-05, "loss": 0.0126, "step": 134060 }, { "epoch": 85.44933078393882, "grad_norm": 0.39262688159942627, "learning_rate": 6.277104452667106e-05, "loss": 0.0217, "step": 134070 }, { "epoch": 85.45570427023581, "grad_norm": 0.1965094357728958, "learning_rate": 6.276571570190764e-05, "loss": 0.0105, "step": 134080 }, { "epoch": 85.46207775653282, "grad_norm": 0.35573405027389526, "learning_rate": 6.27603867220287e-05, "loss": 0.0139, "step": 134090 }, { "epoch": 85.46845124282983, "grad_norm": 0.2517143785953522, "learning_rate": 6.275505758709903e-05, "loss": 0.0117, "step": 134100 }, { "epoch": 85.47482472912684, "grad_norm": 0.3976518213748932, "learning_rate": 6.274972829718335e-05, "loss": 0.0134, "step": 134110 }, { "epoch": 85.48119821542383, "grad_norm": 0.3639957010746002, "learning_rate": 6.274439885234645e-05, "loss": 0.0192, "step": 134120 }, { "epoch": 85.48757170172084, "grad_norm": 0.5718289613723755, "learning_rate": 6.273906925265306e-05, "loss": 0.0177, "step": 134130 }, { "epoch": 85.49394518801785, "grad_norm": 0.2690408229827881, "learning_rate": 6.273373949816797e-05, "loss": 0.013, "step": 134140 }, { "epoch": 85.50031867431485, "grad_norm": 0.5210650563240051, "learning_rate": 6.27284095889559e-05, "loss": 0.0142, "step": 134150 }, { "epoch": 85.50669216061185, "grad_norm": 0.28243574500083923, "learning_rate": 6.272307952508164e-05, "loss": 0.0126, "step": 134160 }, { "epoch": 85.51306564690886, "grad_norm": 0.2538597881793976, "learning_rate": 6.271774930660997e-05, "loss": 0.0097, "step": 134170 }, { "epoch": 85.51943913320586, "grad_norm": 0.4388836622238159, "learning_rate": 6.271241893360561e-05, "loss": 0.0128, "step": 134180 }, { "epoch": 85.52581261950287, "grad_norm": 0.37384334206581116, "learning_rate": 6.270708840613337e-05, "loss": 0.0337, "step": 134190 }, { "epoch": 85.53218610579987, "grad_norm": 0.7318226099014282, "learning_rate": 6.270175772425802e-05, "loss": 0.0147, "step": 134200 }, { "epoch": 85.53855959209687, "grad_norm": 0.29239773750305176, "learning_rate": 6.26964268880443e-05, "loss": 0.0076, "step": 134210 }, { "epoch": 85.54493307839388, "grad_norm": 0.6223142147064209, "learning_rate": 6.269109589755701e-05, "loss": 0.0283, "step": 134220 }, { "epoch": 85.55130656469089, "grad_norm": 0.19736704230308533, "learning_rate": 6.268576475286095e-05, "loss": 0.0102, "step": 134230 }, { "epoch": 85.55768005098788, "grad_norm": 0.9032602310180664, "learning_rate": 6.268043345402085e-05, "loss": 0.0228, "step": 134240 }, { "epoch": 85.56405353728489, "grad_norm": 0.2433578222990036, "learning_rate": 6.26751020011015e-05, "loss": 0.0119, "step": 134250 }, { "epoch": 85.5704270235819, "grad_norm": 0.5338777899742126, "learning_rate": 6.266977039416771e-05, "loss": 0.0116, "step": 134260 }, { "epoch": 85.57680050987891, "grad_norm": 0.28763332962989807, "learning_rate": 6.266443863328424e-05, "loss": 0.0165, "step": 134270 }, { "epoch": 85.5831739961759, "grad_norm": 0.32613518834114075, "learning_rate": 6.26591067185159e-05, "loss": 0.0113, "step": 134280 }, { "epoch": 85.58954748247291, "grad_norm": 0.1906759887933731, "learning_rate": 6.265377464992743e-05, "loss": 0.0139, "step": 134290 }, { "epoch": 85.59592096876992, "grad_norm": 0.5102590322494507, "learning_rate": 6.264844242758368e-05, "loss": 0.0181, "step": 134300 }, { "epoch": 85.60229445506693, "grad_norm": 0.2676417827606201, "learning_rate": 6.26431100515494e-05, "loss": 0.0111, "step": 134310 }, { "epoch": 85.60866794136393, "grad_norm": 0.2396395355463028, "learning_rate": 6.263777752188939e-05, "loss": 0.024, "step": 134320 }, { "epoch": 85.61504142766093, "grad_norm": 0.5930048227310181, "learning_rate": 6.263244483866846e-05, "loss": 0.0165, "step": 134330 }, { "epoch": 85.62141491395793, "grad_norm": 0.19959649443626404, "learning_rate": 6.26271120019514e-05, "loss": 0.0229, "step": 134340 }, { "epoch": 85.62778840025494, "grad_norm": 0.24350512027740479, "learning_rate": 6.2621779011803e-05, "loss": 0.0148, "step": 134350 }, { "epoch": 85.63416188655195, "grad_norm": 0.23164118826389313, "learning_rate": 6.261644586828806e-05, "loss": 0.0175, "step": 134360 }, { "epoch": 85.64053537284894, "grad_norm": 0.7929355502128601, "learning_rate": 6.26111125714714e-05, "loss": 0.0161, "step": 134370 }, { "epoch": 85.64690885914595, "grad_norm": 0.26290762424468994, "learning_rate": 6.260577912141783e-05, "loss": 0.0182, "step": 134380 }, { "epoch": 85.65328234544296, "grad_norm": 0.45211490988731384, "learning_rate": 6.260044551819213e-05, "loss": 0.0161, "step": 134390 }, { "epoch": 85.65965583173997, "grad_norm": 0.602666974067688, "learning_rate": 6.259511176185913e-05, "loss": 0.0179, "step": 134400 }, { "epoch": 85.66602931803696, "grad_norm": 0.5500831604003906, "learning_rate": 6.258977785248362e-05, "loss": 0.0368, "step": 134410 }, { "epoch": 85.67240280433397, "grad_norm": 0.543546199798584, "learning_rate": 6.258444379013041e-05, "loss": 0.0131, "step": 134420 }, { "epoch": 85.67877629063098, "grad_norm": 0.45119503140449524, "learning_rate": 6.257910957486433e-05, "loss": 0.0156, "step": 134430 }, { "epoch": 85.68514977692799, "grad_norm": 0.3457984924316406, "learning_rate": 6.25737752067502e-05, "loss": 0.0124, "step": 134440 }, { "epoch": 85.69152326322498, "grad_norm": 0.4464523196220398, "learning_rate": 6.256844068585284e-05, "loss": 0.0127, "step": 134450 }, { "epoch": 85.69789674952199, "grad_norm": 0.4291567802429199, "learning_rate": 6.256310601223705e-05, "loss": 0.0188, "step": 134460 }, { "epoch": 85.704270235819, "grad_norm": 0.3365817368030548, "learning_rate": 6.255777118596766e-05, "loss": 0.0095, "step": 134470 }, { "epoch": 85.710643722116, "grad_norm": 0.2675381898880005, "learning_rate": 6.255243620710949e-05, "loss": 0.0084, "step": 134480 }, { "epoch": 85.717017208413, "grad_norm": 0.625413179397583, "learning_rate": 6.254710107572737e-05, "loss": 0.0202, "step": 134490 }, { "epoch": 85.72339069471, "grad_norm": 0.28728628158569336, "learning_rate": 6.254176579188612e-05, "loss": 0.0133, "step": 134500 }, { "epoch": 85.72976418100701, "grad_norm": 0.5359876155853271, "learning_rate": 6.253643035565058e-05, "loss": 0.0328, "step": 134510 }, { "epoch": 85.73613766730402, "grad_norm": 0.2798033654689789, "learning_rate": 6.253109476708558e-05, "loss": 0.0151, "step": 134520 }, { "epoch": 85.74251115360101, "grad_norm": 0.35901927947998047, "learning_rate": 6.252575902625595e-05, "loss": 0.0118, "step": 134530 }, { "epoch": 85.74888463989802, "grad_norm": 0.5089362859725952, "learning_rate": 6.25204231332265e-05, "loss": 0.0164, "step": 134540 }, { "epoch": 85.75525812619503, "grad_norm": 0.28204217553138733, "learning_rate": 6.251508708806211e-05, "loss": 0.015, "step": 134550 }, { "epoch": 85.76163161249204, "grad_norm": 0.32606256008148193, "learning_rate": 6.250975089082759e-05, "loss": 0.0172, "step": 134560 }, { "epoch": 85.76800509878903, "grad_norm": 0.40596604347229004, "learning_rate": 6.250441454158778e-05, "loss": 0.0148, "step": 134570 }, { "epoch": 85.77437858508604, "grad_norm": 0.22786960005760193, "learning_rate": 6.249907804040754e-05, "loss": 0.0094, "step": 134580 }, { "epoch": 85.78075207138305, "grad_norm": 0.2618916928768158, "learning_rate": 6.249374138735168e-05, "loss": 0.0276, "step": 134590 }, { "epoch": 85.78712555768006, "grad_norm": 0.4987781345844269, "learning_rate": 6.248840458248508e-05, "loss": 0.0134, "step": 134600 }, { "epoch": 85.79349904397705, "grad_norm": 0.4804847538471222, "learning_rate": 6.248306762587257e-05, "loss": 0.0213, "step": 134610 }, { "epoch": 85.79987253027406, "grad_norm": 0.21705777943134308, "learning_rate": 6.2477730517579e-05, "loss": 0.015, "step": 134620 }, { "epoch": 85.80624601657107, "grad_norm": 0.495553582906723, "learning_rate": 6.247239325766922e-05, "loss": 0.0298, "step": 134630 }, { "epoch": 85.81261950286807, "grad_norm": 0.23771072924137115, "learning_rate": 6.24670558462081e-05, "loss": 0.02, "step": 134640 }, { "epoch": 85.81899298916507, "grad_norm": 0.37462735176086426, "learning_rate": 6.246171828326048e-05, "loss": 0.0136, "step": 134650 }, { "epoch": 85.82536647546208, "grad_norm": 0.22058837115764618, "learning_rate": 6.24563805688912e-05, "loss": 0.0206, "step": 134660 }, { "epoch": 85.83173996175908, "grad_norm": 0.7076476216316223, "learning_rate": 6.245104270316516e-05, "loss": 0.0193, "step": 134670 }, { "epoch": 85.83811344805609, "grad_norm": 0.45809313654899597, "learning_rate": 6.244570468614717e-05, "loss": 0.0207, "step": 134680 }, { "epoch": 85.84448693435309, "grad_norm": 0.3392370939254761, "learning_rate": 6.244036651790212e-05, "loss": 0.0164, "step": 134690 }, { "epoch": 85.8508604206501, "grad_norm": 0.6087072491645813, "learning_rate": 6.243502819849488e-05, "loss": 0.0191, "step": 134700 }, { "epoch": 85.8572339069471, "grad_norm": 0.3491577208042145, "learning_rate": 6.242968972799032e-05, "loss": 0.0104, "step": 134710 }, { "epoch": 85.86360739324411, "grad_norm": 0.5296474695205688, "learning_rate": 6.242435110645329e-05, "loss": 0.0098, "step": 134720 }, { "epoch": 85.8699808795411, "grad_norm": 0.23673203587532043, "learning_rate": 6.241901233394864e-05, "loss": 0.0123, "step": 134730 }, { "epoch": 85.87635436583811, "grad_norm": 0.5240015983581543, "learning_rate": 6.241367341054128e-05, "loss": 0.0124, "step": 134740 }, { "epoch": 85.88272785213512, "grad_norm": 0.5271756649017334, "learning_rate": 6.240833433629606e-05, "loss": 0.0082, "step": 134750 }, { "epoch": 85.88910133843213, "grad_norm": 0.6531194448471069, "learning_rate": 6.240299511127787e-05, "loss": 0.0158, "step": 134760 }, { "epoch": 85.89547482472912, "grad_norm": 0.4254186749458313, "learning_rate": 6.239765573555157e-05, "loss": 0.012, "step": 134770 }, { "epoch": 85.90184831102613, "grad_norm": 0.20464026927947998, "learning_rate": 6.239231620918207e-05, "loss": 0.0114, "step": 134780 }, { "epoch": 85.90822179732314, "grad_norm": 0.3883301913738251, "learning_rate": 6.238697653223422e-05, "loss": 0.02, "step": 134790 }, { "epoch": 85.91459528362014, "grad_norm": 0.6050004363059998, "learning_rate": 6.23816367047729e-05, "loss": 0.0174, "step": 134800 }, { "epoch": 85.92096876991714, "grad_norm": 0.2954604923725128, "learning_rate": 6.2376296726863e-05, "loss": 0.0134, "step": 134810 }, { "epoch": 85.92734225621415, "grad_norm": 0.3161962926387787, "learning_rate": 6.237095659856941e-05, "loss": 0.0138, "step": 134820 }, { "epoch": 85.93371574251115, "grad_norm": 0.30038729310035706, "learning_rate": 6.236561631995703e-05, "loss": 0.0187, "step": 134830 }, { "epoch": 85.94008922880816, "grad_norm": 0.8933187127113342, "learning_rate": 6.236027589109073e-05, "loss": 0.0152, "step": 134840 }, { "epoch": 85.94646271510516, "grad_norm": 0.3523712158203125, "learning_rate": 6.235493531203542e-05, "loss": 0.0181, "step": 134850 }, { "epoch": 85.95283620140216, "grad_norm": 0.40961790084838867, "learning_rate": 6.234959458285595e-05, "loss": 0.0099, "step": 134860 }, { "epoch": 85.95920968769917, "grad_norm": 0.32076549530029297, "learning_rate": 6.234425370361728e-05, "loss": 0.0133, "step": 134870 }, { "epoch": 85.96558317399618, "grad_norm": 0.2806931138038635, "learning_rate": 6.233891267438425e-05, "loss": 0.018, "step": 134880 }, { "epoch": 85.97195666029317, "grad_norm": 0.4500206410884857, "learning_rate": 6.23335714952218e-05, "loss": 0.0125, "step": 134890 }, { "epoch": 85.97833014659018, "grad_norm": 0.4107638895511627, "learning_rate": 6.23282301661948e-05, "loss": 0.0151, "step": 134900 }, { "epoch": 85.98470363288719, "grad_norm": 1.0616971254348755, "learning_rate": 6.232288868736815e-05, "loss": 0.0248, "step": 134910 }, { "epoch": 85.9910771191842, "grad_norm": 0.7169125080108643, "learning_rate": 6.231754705880679e-05, "loss": 0.0164, "step": 134920 }, { "epoch": 85.99745060548119, "grad_norm": 0.2793859839439392, "learning_rate": 6.231220528057559e-05, "loss": 0.0089, "step": 134930 }, { "epoch": 86.0038240917782, "grad_norm": 0.6584393382072449, "learning_rate": 6.230686335273947e-05, "loss": 0.0136, "step": 134940 }, { "epoch": 86.0101975780752, "grad_norm": 0.4580683708190918, "learning_rate": 6.230152127536337e-05, "loss": 0.0088, "step": 134950 }, { "epoch": 86.01657106437222, "grad_norm": 0.6158672571182251, "learning_rate": 6.229617904851214e-05, "loss": 0.0159, "step": 134960 }, { "epoch": 86.02294455066922, "grad_norm": 0.5652865171432495, "learning_rate": 6.229083667225073e-05, "loss": 0.0231, "step": 134970 }, { "epoch": 86.02931803696622, "grad_norm": 0.5845011472702026, "learning_rate": 6.228549414664406e-05, "loss": 0.0119, "step": 134980 }, { "epoch": 86.03569152326322, "grad_norm": 0.3714328408241272, "learning_rate": 6.228015147175703e-05, "loss": 0.0199, "step": 134990 }, { "epoch": 86.04206500956023, "grad_norm": 0.5463960766792297, "learning_rate": 6.227480864765456e-05, "loss": 0.0169, "step": 135000 }, { "epoch": 86.04843849585724, "grad_norm": 0.21562433242797852, "learning_rate": 6.226946567440158e-05, "loss": 0.0133, "step": 135010 }, { "epoch": 86.05481198215423, "grad_norm": 0.3090166449546814, "learning_rate": 6.2264122552063e-05, "loss": 0.014, "step": 135020 }, { "epoch": 86.06118546845124, "grad_norm": 0.2305338829755783, "learning_rate": 6.225877928070377e-05, "loss": 0.0143, "step": 135030 }, { "epoch": 86.06755895474825, "grad_norm": 0.433553546667099, "learning_rate": 6.22534358603888e-05, "loss": 0.0114, "step": 135040 }, { "epoch": 86.07393244104526, "grad_norm": 0.3043432831764221, "learning_rate": 6.224809229118299e-05, "loss": 0.0145, "step": 135050 }, { "epoch": 86.08030592734225, "grad_norm": 0.2510030269622803, "learning_rate": 6.22427485731513e-05, "loss": 0.019, "step": 135060 }, { "epoch": 86.08667941363926, "grad_norm": 0.19729898869991302, "learning_rate": 6.223740470635867e-05, "loss": 0.0138, "step": 135070 }, { "epoch": 86.09305289993627, "grad_norm": 0.3960331678390503, "learning_rate": 6.223206069087002e-05, "loss": 0.0149, "step": 135080 }, { "epoch": 86.09942638623328, "grad_norm": 0.31779107451438904, "learning_rate": 6.222671652675027e-05, "loss": 0.0169, "step": 135090 }, { "epoch": 86.10579987253027, "grad_norm": 0.21725769340991974, "learning_rate": 6.222137221406439e-05, "loss": 0.0162, "step": 135100 }, { "epoch": 86.11217335882728, "grad_norm": 0.48256155848503113, "learning_rate": 6.221602775287728e-05, "loss": 0.0195, "step": 135110 }, { "epoch": 86.11854684512429, "grad_norm": 0.17646318674087524, "learning_rate": 6.22106831432539e-05, "loss": 0.0095, "step": 135120 }, { "epoch": 86.1249203314213, "grad_norm": 0.39146724343299866, "learning_rate": 6.22053383852592e-05, "loss": 0.0206, "step": 135130 }, { "epoch": 86.13129381771829, "grad_norm": 0.49523472785949707, "learning_rate": 6.219999347895811e-05, "loss": 0.0152, "step": 135140 }, { "epoch": 86.1376673040153, "grad_norm": 0.1939106434583664, "learning_rate": 6.219464842441558e-05, "loss": 0.0087, "step": 135150 }, { "epoch": 86.1440407903123, "grad_norm": 0.5579684376716614, "learning_rate": 6.218930322169655e-05, "loss": 0.013, "step": 135160 }, { "epoch": 86.15041427660931, "grad_norm": 0.42011529207229614, "learning_rate": 6.2183957870866e-05, "loss": 0.0133, "step": 135170 }, { "epoch": 86.1567877629063, "grad_norm": 0.33165866136550903, "learning_rate": 6.217861237198884e-05, "loss": 0.0105, "step": 135180 }, { "epoch": 86.16316124920331, "grad_norm": 0.16643035411834717, "learning_rate": 6.217326672513004e-05, "loss": 0.0125, "step": 135190 }, { "epoch": 86.16953473550032, "grad_norm": 0.18035632371902466, "learning_rate": 6.216792093035457e-05, "loss": 0.0115, "step": 135200 }, { "epoch": 86.17590822179733, "grad_norm": 0.6001988053321838, "learning_rate": 6.216257498772734e-05, "loss": 0.0313, "step": 135210 }, { "epoch": 86.18228170809432, "grad_norm": 0.22946247458457947, "learning_rate": 6.21572288973134e-05, "loss": 0.0222, "step": 135220 }, { "epoch": 86.18865519439133, "grad_norm": 0.5913975238800049, "learning_rate": 6.21518826591776e-05, "loss": 0.0292, "step": 135230 }, { "epoch": 86.19502868068834, "grad_norm": 0.3534841239452362, "learning_rate": 6.214653627338498e-05, "loss": 0.0102, "step": 135240 }, { "epoch": 86.20140216698535, "grad_norm": 0.20196394622325897, "learning_rate": 6.214118974000044e-05, "loss": 0.0146, "step": 135250 }, { "epoch": 86.20777565328234, "grad_norm": 0.26523587107658386, "learning_rate": 6.213584305908901e-05, "loss": 0.0125, "step": 135260 }, { "epoch": 86.21414913957935, "grad_norm": 0.4626877009868622, "learning_rate": 6.213049623071562e-05, "loss": 0.0129, "step": 135270 }, { "epoch": 86.22052262587636, "grad_norm": 0.5268779397010803, "learning_rate": 6.212514925494526e-05, "loss": 0.023, "step": 135280 }, { "epoch": 86.22689611217336, "grad_norm": 0.27981695532798767, "learning_rate": 6.211980213184287e-05, "loss": 0.0167, "step": 135290 }, { "epoch": 86.23326959847036, "grad_norm": 0.3353387415409088, "learning_rate": 6.211445486147346e-05, "loss": 0.0128, "step": 135300 }, { "epoch": 86.23964308476737, "grad_norm": 0.7728697061538696, "learning_rate": 6.210910744390197e-05, "loss": 0.0183, "step": 135310 }, { "epoch": 86.24601657106437, "grad_norm": 0.4499590992927551, "learning_rate": 6.210375987919341e-05, "loss": 0.0157, "step": 135320 }, { "epoch": 86.25239005736138, "grad_norm": 0.1418088972568512, "learning_rate": 6.209841216741271e-05, "loss": 0.0071, "step": 135330 }, { "epoch": 86.25876354365838, "grad_norm": 0.2138492912054062, "learning_rate": 6.20930643086249e-05, "loss": 0.0084, "step": 135340 }, { "epoch": 86.26513702995538, "grad_norm": 0.4573669731616974, "learning_rate": 6.208771630289493e-05, "loss": 0.0121, "step": 135350 }, { "epoch": 86.27151051625239, "grad_norm": 0.9010759592056274, "learning_rate": 6.208236815028781e-05, "loss": 0.0175, "step": 135360 }, { "epoch": 86.2778840025494, "grad_norm": 0.4801674485206604, "learning_rate": 6.207701985086849e-05, "loss": 0.0194, "step": 135370 }, { "epoch": 86.2842574888464, "grad_norm": 0.49767372012138367, "learning_rate": 6.2071671404702e-05, "loss": 0.0231, "step": 135380 }, { "epoch": 86.2906309751434, "grad_norm": 0.649364709854126, "learning_rate": 6.206632281185328e-05, "loss": 0.0193, "step": 135390 }, { "epoch": 86.29700446144041, "grad_norm": 0.34288161993026733, "learning_rate": 6.206097407238736e-05, "loss": 0.0168, "step": 135400 }, { "epoch": 86.30337794773742, "grad_norm": 0.25144845247268677, "learning_rate": 6.205562518636921e-05, "loss": 0.0138, "step": 135410 }, { "epoch": 86.30975143403441, "grad_norm": 0.43395283818244934, "learning_rate": 6.205027615386385e-05, "loss": 0.0172, "step": 135420 }, { "epoch": 86.31612492033142, "grad_norm": 0.30412763357162476, "learning_rate": 6.204492697493624e-05, "loss": 0.0126, "step": 135430 }, { "epoch": 86.32249840662843, "grad_norm": 0.2962929308414459, "learning_rate": 6.20395776496514e-05, "loss": 0.0123, "step": 135440 }, { "epoch": 86.32887189292543, "grad_norm": 0.2828255593776703, "learning_rate": 6.203422817807433e-05, "loss": 0.0148, "step": 135450 }, { "epoch": 86.33524537922243, "grad_norm": 0.665999174118042, "learning_rate": 6.202887856027001e-05, "loss": 0.0179, "step": 135460 }, { "epoch": 86.34161886551944, "grad_norm": 0.3620237410068512, "learning_rate": 6.202352879630347e-05, "loss": 0.0063, "step": 135470 }, { "epoch": 86.34799235181644, "grad_norm": 0.32914337515830994, "learning_rate": 6.20181788862397e-05, "loss": 0.014, "step": 135480 }, { "epoch": 86.35436583811345, "grad_norm": 0.3205731213092804, "learning_rate": 6.201282883014371e-05, "loss": 0.0165, "step": 135490 }, { "epoch": 86.36073932441045, "grad_norm": 0.341469943523407, "learning_rate": 6.20074786280805e-05, "loss": 0.0129, "step": 135500 }, { "epoch": 86.36711281070745, "grad_norm": 0.2432442158460617, "learning_rate": 6.200212828011511e-05, "loss": 0.0103, "step": 135510 }, { "epoch": 86.37348629700446, "grad_norm": 0.20887355506420135, "learning_rate": 6.199677778631251e-05, "loss": 0.0075, "step": 135520 }, { "epoch": 86.37985978330147, "grad_norm": 0.5486912727355957, "learning_rate": 6.199142714673774e-05, "loss": 0.0134, "step": 135530 }, { "epoch": 86.38623326959846, "grad_norm": 0.5357344150543213, "learning_rate": 6.19860763614558e-05, "loss": 0.0161, "step": 135540 }, { "epoch": 86.39260675589547, "grad_norm": 0.6181392073631287, "learning_rate": 6.198072543053173e-05, "loss": 0.0117, "step": 135550 }, { "epoch": 86.39898024219248, "grad_norm": 0.40764039754867554, "learning_rate": 6.197537435403052e-05, "loss": 0.0095, "step": 135560 }, { "epoch": 86.40535372848949, "grad_norm": 0.20449642837047577, "learning_rate": 6.197002313201721e-05, "loss": 0.0196, "step": 135570 }, { "epoch": 86.4117272147865, "grad_norm": 0.3925415575504303, "learning_rate": 6.196467176455683e-05, "loss": 0.0088, "step": 135580 }, { "epoch": 86.41810070108349, "grad_norm": 0.3782031536102295, "learning_rate": 6.195932025171436e-05, "loss": 0.0195, "step": 135590 }, { "epoch": 86.4244741873805, "grad_norm": 0.3787313401699066, "learning_rate": 6.195396859355488e-05, "loss": 0.011, "step": 135600 }, { "epoch": 86.4308476736775, "grad_norm": 0.2139919102191925, "learning_rate": 6.194861679014341e-05, "loss": 0.0141, "step": 135610 }, { "epoch": 86.43722115997451, "grad_norm": 0.26279762387275696, "learning_rate": 6.194326484154493e-05, "loss": 0.0182, "step": 135620 }, { "epoch": 86.4435946462715, "grad_norm": 0.451067179441452, "learning_rate": 6.193791274782453e-05, "loss": 0.0168, "step": 135630 }, { "epoch": 86.44996813256851, "grad_norm": 0.47460275888442993, "learning_rate": 6.19325605090472e-05, "loss": 0.0109, "step": 135640 }, { "epoch": 86.45634161886552, "grad_norm": 0.455019474029541, "learning_rate": 6.192720812527802e-05, "loss": 0.0108, "step": 135650 }, { "epoch": 86.46271510516253, "grad_norm": 0.4023546874523163, "learning_rate": 6.192185559658198e-05, "loss": 0.014, "step": 135660 }, { "epoch": 86.46908859145952, "grad_norm": 0.5042657256126404, "learning_rate": 6.191650292302414e-05, "loss": 0.027, "step": 135670 }, { "epoch": 86.47546207775653, "grad_norm": 0.2793891727924347, "learning_rate": 6.191115010466956e-05, "loss": 0.0098, "step": 135680 }, { "epoch": 86.48183556405354, "grad_norm": 0.24504855275154114, "learning_rate": 6.190579714158324e-05, "loss": 0.0158, "step": 135690 }, { "epoch": 86.48820905035055, "grad_norm": 0.5600149035453796, "learning_rate": 6.190044403383024e-05, "loss": 0.013, "step": 135700 }, { "epoch": 86.49458253664754, "grad_norm": 0.5385920405387878, "learning_rate": 6.189509078147561e-05, "loss": 0.0155, "step": 135710 }, { "epoch": 86.50095602294455, "grad_norm": 0.35997435450553894, "learning_rate": 6.188973738458441e-05, "loss": 0.016, "step": 135720 }, { "epoch": 86.50732950924156, "grad_norm": 0.6061364412307739, "learning_rate": 6.188438384322166e-05, "loss": 0.015, "step": 135730 }, { "epoch": 86.51370299553857, "grad_norm": 0.4451411962509155, "learning_rate": 6.187903015745243e-05, "loss": 0.0156, "step": 135740 }, { "epoch": 86.52007648183556, "grad_norm": 0.5176655650138855, "learning_rate": 6.187367632734177e-05, "loss": 0.0112, "step": 135750 }, { "epoch": 86.52644996813257, "grad_norm": 0.37311169505119324, "learning_rate": 6.186832235295474e-05, "loss": 0.0117, "step": 135760 }, { "epoch": 86.53282345442958, "grad_norm": 0.24009248614311218, "learning_rate": 6.186296823435639e-05, "loss": 0.0089, "step": 135770 }, { "epoch": 86.53919694072658, "grad_norm": 0.20149506628513336, "learning_rate": 6.185761397161175e-05, "loss": 0.0131, "step": 135780 }, { "epoch": 86.54557042702358, "grad_norm": 0.5700642466545105, "learning_rate": 6.185225956478593e-05, "loss": 0.0121, "step": 135790 }, { "epoch": 86.55194391332059, "grad_norm": 0.33434686064720154, "learning_rate": 6.184690501394394e-05, "loss": 0.0151, "step": 135800 }, { "epoch": 86.5583173996176, "grad_norm": 0.3418165445327759, "learning_rate": 6.184155031915089e-05, "loss": 0.0163, "step": 135810 }, { "epoch": 86.5646908859146, "grad_norm": 0.23490965366363525, "learning_rate": 6.183619548047183e-05, "loss": 0.0164, "step": 135820 }, { "epoch": 86.5710643722116, "grad_norm": 0.22476395964622498, "learning_rate": 6.183084049797181e-05, "loss": 0.0174, "step": 135830 }, { "epoch": 86.5774378585086, "grad_norm": 0.29840990900993347, "learning_rate": 6.182548537171591e-05, "loss": 0.0147, "step": 135840 }, { "epoch": 86.58381134480561, "grad_norm": 0.27412399649620056, "learning_rate": 6.182013010176918e-05, "loss": 0.0108, "step": 135850 }, { "epoch": 86.59018483110262, "grad_norm": 0.28438565135002136, "learning_rate": 6.181477468819673e-05, "loss": 0.016, "step": 135860 }, { "epoch": 86.59655831739961, "grad_norm": 0.379455029964447, "learning_rate": 6.18094191310636e-05, "loss": 0.0181, "step": 135870 }, { "epoch": 86.60293180369662, "grad_norm": 0.5012579560279846, "learning_rate": 6.180406343043489e-05, "loss": 0.022, "step": 135880 }, { "epoch": 86.60930528999363, "grad_norm": 0.4783099889755249, "learning_rate": 6.179870758637565e-05, "loss": 0.0279, "step": 135890 }, { "epoch": 86.61567877629064, "grad_norm": 0.2859387993812561, "learning_rate": 6.179335159895099e-05, "loss": 0.0107, "step": 135900 }, { "epoch": 86.62205226258763, "grad_norm": 0.5143576860427856, "learning_rate": 6.178799546822595e-05, "loss": 0.0185, "step": 135910 }, { "epoch": 86.62842574888464, "grad_norm": 0.43622055649757385, "learning_rate": 6.178263919426566e-05, "loss": 0.0205, "step": 135920 }, { "epoch": 86.63479923518165, "grad_norm": 0.5519411563873291, "learning_rate": 6.177728277713516e-05, "loss": 0.0139, "step": 135930 }, { "epoch": 86.64117272147865, "grad_norm": 0.6829575300216675, "learning_rate": 6.177192621689957e-05, "loss": 0.0186, "step": 135940 }, { "epoch": 86.64754620777565, "grad_norm": 0.38545626401901245, "learning_rate": 6.176656951362395e-05, "loss": 0.0257, "step": 135950 }, { "epoch": 86.65391969407266, "grad_norm": 0.27389034628868103, "learning_rate": 6.17612126673734e-05, "loss": 0.0326, "step": 135960 }, { "epoch": 86.66029318036966, "grad_norm": 0.21246659755706787, "learning_rate": 6.175585567821303e-05, "loss": 0.0106, "step": 135970 }, { "epoch": 86.66666666666667, "grad_norm": 0.2679625153541565, "learning_rate": 6.17504985462079e-05, "loss": 0.0129, "step": 135980 }, { "epoch": 86.67304015296367, "grad_norm": 0.26560646295547485, "learning_rate": 6.174514127142312e-05, "loss": 0.0203, "step": 135990 }, { "epoch": 86.67941363926067, "grad_norm": 0.432272732257843, "learning_rate": 6.173978385392377e-05, "loss": 0.0236, "step": 136000 }, { "epoch": 86.68578712555768, "grad_norm": 0.6627709269523621, "learning_rate": 6.173442629377498e-05, "loss": 0.0202, "step": 136010 }, { "epoch": 86.69216061185469, "grad_norm": 0.5887337923049927, "learning_rate": 6.172906859104183e-05, "loss": 0.0145, "step": 136020 }, { "epoch": 86.69853409815168, "grad_norm": 0.47394421696662903, "learning_rate": 6.172371074578942e-05, "loss": 0.0158, "step": 136030 }, { "epoch": 86.70490758444869, "grad_norm": 0.4661499857902527, "learning_rate": 6.171835275808286e-05, "loss": 0.0247, "step": 136040 }, { "epoch": 86.7112810707457, "grad_norm": 0.27699604630470276, "learning_rate": 6.171299462798724e-05, "loss": 0.0196, "step": 136050 }, { "epoch": 86.7176545570427, "grad_norm": 0.31430625915527344, "learning_rate": 6.170763635556767e-05, "loss": 0.0132, "step": 136060 }, { "epoch": 86.7240280433397, "grad_norm": 0.866858184337616, "learning_rate": 6.170227794088928e-05, "loss": 0.0147, "step": 136070 }, { "epoch": 86.73040152963671, "grad_norm": 0.3130281865596771, "learning_rate": 6.169691938401716e-05, "loss": 0.0136, "step": 136080 }, { "epoch": 86.73677501593372, "grad_norm": 0.566281259059906, "learning_rate": 6.169156068501643e-05, "loss": 0.0157, "step": 136090 }, { "epoch": 86.74314850223072, "grad_norm": 0.24743251502513885, "learning_rate": 6.16862018439522e-05, "loss": 0.0191, "step": 136100 }, { "epoch": 86.74952198852772, "grad_norm": 0.3813864588737488, "learning_rate": 6.168084286088958e-05, "loss": 0.0201, "step": 136110 }, { "epoch": 86.75589547482473, "grad_norm": 0.3566763699054718, "learning_rate": 6.167548373589368e-05, "loss": 0.0248, "step": 136120 }, { "epoch": 86.76226896112173, "grad_norm": 0.4163026213645935, "learning_rate": 6.167012446902964e-05, "loss": 0.0241, "step": 136130 }, { "epoch": 86.76864244741874, "grad_norm": 0.3649134039878845, "learning_rate": 6.166476506036257e-05, "loss": 0.0166, "step": 136140 }, { "epoch": 86.77501593371574, "grad_norm": 0.19617947936058044, "learning_rate": 6.165940550995759e-05, "loss": 0.0183, "step": 136150 }, { "epoch": 86.78138942001274, "grad_norm": 0.20727837085723877, "learning_rate": 6.165404581787983e-05, "loss": 0.0086, "step": 136160 }, { "epoch": 86.78776290630975, "grad_norm": 0.2628084719181061, "learning_rate": 6.164868598419439e-05, "loss": 0.0138, "step": 136170 }, { "epoch": 86.79413639260676, "grad_norm": 0.7333850860595703, "learning_rate": 6.164332600896645e-05, "loss": 0.0184, "step": 136180 }, { "epoch": 86.80050987890377, "grad_norm": 0.19258369505405426, "learning_rate": 6.163796589226108e-05, "loss": 0.0093, "step": 136190 }, { "epoch": 86.80688336520076, "grad_norm": 0.48198413848876953, "learning_rate": 6.163260563414344e-05, "loss": 0.0245, "step": 136200 }, { "epoch": 86.81325685149777, "grad_norm": 0.5114803910255432, "learning_rate": 6.162724523467867e-05, "loss": 0.0131, "step": 136210 }, { "epoch": 86.81963033779478, "grad_norm": 0.2419361174106598, "learning_rate": 6.162188469393188e-05, "loss": 0.0121, "step": 136220 }, { "epoch": 86.82600382409179, "grad_norm": 0.5240670442581177, "learning_rate": 6.161652401196822e-05, "loss": 0.0141, "step": 136230 }, { "epoch": 86.83237731038878, "grad_norm": 0.49525779485702515, "learning_rate": 6.161116318885283e-05, "loss": 0.0099, "step": 136240 }, { "epoch": 86.83875079668579, "grad_norm": 0.38083189725875854, "learning_rate": 6.160580222465085e-05, "loss": 0.0165, "step": 136250 }, { "epoch": 86.8451242829828, "grad_norm": 0.5293951630592346, "learning_rate": 6.16004411194274e-05, "loss": 0.0174, "step": 136260 }, { "epoch": 86.8514977692798, "grad_norm": 0.2664691209793091, "learning_rate": 6.159507987324765e-05, "loss": 0.0166, "step": 136270 }, { "epoch": 86.8578712555768, "grad_norm": 0.20196571946144104, "learning_rate": 6.158971848617673e-05, "loss": 0.0271, "step": 136280 }, { "epoch": 86.8642447418738, "grad_norm": 0.14810959994792938, "learning_rate": 6.158435695827979e-05, "loss": 0.018, "step": 136290 }, { "epoch": 86.87061822817081, "grad_norm": 0.2296142280101776, "learning_rate": 6.157899528962197e-05, "loss": 0.0138, "step": 136300 }, { "epoch": 86.87699171446782, "grad_norm": 0.15450845658779144, "learning_rate": 6.157363348026842e-05, "loss": 0.0151, "step": 136310 }, { "epoch": 86.88336520076481, "grad_norm": 0.48244673013687134, "learning_rate": 6.15682715302843e-05, "loss": 0.0218, "step": 136320 }, { "epoch": 86.88973868706182, "grad_norm": 0.3250318467617035, "learning_rate": 6.156290943973476e-05, "loss": 0.011, "step": 136330 }, { "epoch": 86.89611217335883, "grad_norm": 0.3073652386665344, "learning_rate": 6.155754720868496e-05, "loss": 0.013, "step": 136340 }, { "epoch": 86.90248565965584, "grad_norm": 0.3023199141025543, "learning_rate": 6.155218483720004e-05, "loss": 0.0113, "step": 136350 }, { "epoch": 86.90885914595283, "grad_norm": 0.46760857105255127, "learning_rate": 6.154682232534517e-05, "loss": 0.0118, "step": 136360 }, { "epoch": 86.91523263224984, "grad_norm": 0.3629573583602905, "learning_rate": 6.154145967318551e-05, "loss": 0.0163, "step": 136370 }, { "epoch": 86.92160611854685, "grad_norm": 0.4137875437736511, "learning_rate": 6.15360968807862e-05, "loss": 0.0185, "step": 136380 }, { "epoch": 86.92797960484386, "grad_norm": 0.2590354084968567, "learning_rate": 6.153073394821244e-05, "loss": 0.012, "step": 136390 }, { "epoch": 86.93435309114085, "grad_norm": 0.36579543352127075, "learning_rate": 6.152537087552938e-05, "loss": 0.0127, "step": 136400 }, { "epoch": 86.94072657743786, "grad_norm": 0.2452632337808609, "learning_rate": 6.152000766280218e-05, "loss": 0.0101, "step": 136410 }, { "epoch": 86.94710006373487, "grad_norm": 0.3453703224658966, "learning_rate": 6.151464431009599e-05, "loss": 0.0141, "step": 136420 }, { "epoch": 86.95347355003187, "grad_norm": 0.46439236402511597, "learning_rate": 6.150928081747603e-05, "loss": 0.0152, "step": 136430 }, { "epoch": 86.95984703632887, "grad_norm": 0.18831168115139008, "learning_rate": 6.150391718500741e-05, "loss": 0.0197, "step": 136440 }, { "epoch": 86.96622052262587, "grad_norm": 0.2990996539592743, "learning_rate": 6.149855341275536e-05, "loss": 0.0199, "step": 136450 }, { "epoch": 86.97259400892288, "grad_norm": 0.2614096701145172, "learning_rate": 6.149318950078501e-05, "loss": 0.0089, "step": 136460 }, { "epoch": 86.97896749521989, "grad_norm": 0.4590532183647156, "learning_rate": 6.148782544916157e-05, "loss": 0.0238, "step": 136470 }, { "epoch": 86.98534098151688, "grad_norm": 0.4219270348548889, "learning_rate": 6.14824612579502e-05, "loss": 0.0169, "step": 136480 }, { "epoch": 86.99171446781389, "grad_norm": 0.412659227848053, "learning_rate": 6.147709692721608e-05, "loss": 0.0299, "step": 136490 }, { "epoch": 86.9980879541109, "grad_norm": 0.28372716903686523, "learning_rate": 6.147173245702442e-05, "loss": 0.0168, "step": 136500 }, { "epoch": 87.00446144040791, "grad_norm": 0.36065414547920227, "learning_rate": 6.146636784744036e-05, "loss": 0.0107, "step": 136510 }, { "epoch": 87.0108349267049, "grad_norm": 0.21776160597801208, "learning_rate": 6.14610030985291e-05, "loss": 0.0103, "step": 136520 }, { "epoch": 87.01720841300191, "grad_norm": 0.377972811460495, "learning_rate": 6.145563821035582e-05, "loss": 0.0081, "step": 136530 }, { "epoch": 87.02358189929892, "grad_norm": 0.32573941349983215, "learning_rate": 6.145027318298576e-05, "loss": 0.0147, "step": 136540 }, { "epoch": 87.02995538559593, "grad_norm": 0.19913597404956818, "learning_rate": 6.144490801648404e-05, "loss": 0.0086, "step": 136550 }, { "epoch": 87.03632887189292, "grad_norm": 0.15542803704738617, "learning_rate": 6.143954271091589e-05, "loss": 0.0133, "step": 136560 }, { "epoch": 87.04270235818993, "grad_norm": 0.19216316938400269, "learning_rate": 6.14341772663465e-05, "loss": 0.0112, "step": 136570 }, { "epoch": 87.04907584448694, "grad_norm": 0.3057840168476105, "learning_rate": 6.142881168284104e-05, "loss": 0.0141, "step": 136580 }, { "epoch": 87.05544933078394, "grad_norm": 0.6105427145957947, "learning_rate": 6.142344596046475e-05, "loss": 0.0262, "step": 136590 }, { "epoch": 87.06182281708094, "grad_norm": 0.439746230840683, "learning_rate": 6.14180800992828e-05, "loss": 0.0213, "step": 136600 }, { "epoch": 87.06819630337795, "grad_norm": 0.24846015870571136, "learning_rate": 6.141271409936039e-05, "loss": 0.0165, "step": 136610 }, { "epoch": 87.07456978967495, "grad_norm": 0.2127065807580948, "learning_rate": 6.140734796076273e-05, "loss": 0.0252, "step": 136620 }, { "epoch": 87.08094327597196, "grad_norm": 0.22384266555309296, "learning_rate": 6.140198168355502e-05, "loss": 0.015, "step": 136630 }, { "epoch": 87.08731676226896, "grad_norm": 0.3309597671031952, "learning_rate": 6.139661526780247e-05, "loss": 0.0099, "step": 136640 }, { "epoch": 87.09369024856596, "grad_norm": 0.3901521563529968, "learning_rate": 6.13912487135703e-05, "loss": 0.0191, "step": 136650 }, { "epoch": 87.10006373486297, "grad_norm": 0.4719346761703491, "learning_rate": 6.138588202092367e-05, "loss": 0.0145, "step": 136660 }, { "epoch": 87.10643722115998, "grad_norm": 0.3398447334766388, "learning_rate": 6.138051518992784e-05, "loss": 0.0121, "step": 136670 }, { "epoch": 87.11281070745697, "grad_norm": 0.5052052736282349, "learning_rate": 6.137514822064802e-05, "loss": 0.0226, "step": 136680 }, { "epoch": 87.11918419375398, "grad_norm": 0.3320371210575104, "learning_rate": 6.13697811131494e-05, "loss": 0.0115, "step": 136690 }, { "epoch": 87.12555768005099, "grad_norm": 0.15817736089229584, "learning_rate": 6.136441386749719e-05, "loss": 0.0177, "step": 136700 }, { "epoch": 87.131931166348, "grad_norm": 0.25440025329589844, "learning_rate": 6.135904648375663e-05, "loss": 0.0132, "step": 136710 }, { "epoch": 87.13830465264499, "grad_norm": 0.9420782923698425, "learning_rate": 6.135367896199294e-05, "loss": 0.0233, "step": 136720 }, { "epoch": 87.144678138942, "grad_norm": 0.38487347960472107, "learning_rate": 6.134831130227132e-05, "loss": 0.0103, "step": 136730 }, { "epoch": 87.151051625239, "grad_norm": 0.25661471486091614, "learning_rate": 6.134294350465701e-05, "loss": 0.013, "step": 136740 }, { "epoch": 87.15742511153601, "grad_norm": 0.25249311327934265, "learning_rate": 6.133757556921523e-05, "loss": 0.0132, "step": 136750 }, { "epoch": 87.16379859783301, "grad_norm": 0.6065831780433655, "learning_rate": 6.133220749601119e-05, "loss": 0.0161, "step": 136760 }, { "epoch": 87.17017208413002, "grad_norm": 0.36369338631629944, "learning_rate": 6.132683928511014e-05, "loss": 0.0139, "step": 136770 }, { "epoch": 87.17654557042702, "grad_norm": 0.5983407497406006, "learning_rate": 6.13214709365773e-05, "loss": 0.0151, "step": 136780 }, { "epoch": 87.18291905672403, "grad_norm": 0.25521352887153625, "learning_rate": 6.131610245047789e-05, "loss": 0.0116, "step": 136790 }, { "epoch": 87.18929254302103, "grad_norm": 0.33713090419769287, "learning_rate": 6.131073382687715e-05, "loss": 0.0094, "step": 136800 }, { "epoch": 87.19566602931803, "grad_norm": 0.4561706781387329, "learning_rate": 6.130536506584032e-05, "loss": 0.0265, "step": 136810 }, { "epoch": 87.20203951561504, "grad_norm": 0.6540097594261169, "learning_rate": 6.129999616743264e-05, "loss": 0.0143, "step": 136820 }, { "epoch": 87.20841300191205, "grad_norm": 0.5582777857780457, "learning_rate": 6.129462713171932e-05, "loss": 0.0151, "step": 136830 }, { "epoch": 87.21478648820906, "grad_norm": 0.38014474511146545, "learning_rate": 6.128925795876562e-05, "loss": 0.022, "step": 136840 }, { "epoch": 87.22115997450605, "grad_norm": 0.6847732663154602, "learning_rate": 6.128388864863678e-05, "loss": 0.0227, "step": 136850 }, { "epoch": 87.22753346080306, "grad_norm": 0.41780000925064087, "learning_rate": 6.127851920139804e-05, "loss": 0.0117, "step": 136860 }, { "epoch": 87.23390694710007, "grad_norm": 0.35452067852020264, "learning_rate": 6.127314961711464e-05, "loss": 0.0215, "step": 136870 }, { "epoch": 87.24028043339707, "grad_norm": 0.17160610854625702, "learning_rate": 6.126777989585185e-05, "loss": 0.0214, "step": 136880 }, { "epoch": 87.24665391969407, "grad_norm": 0.47455185651779175, "learning_rate": 6.126241003767487e-05, "loss": 0.021, "step": 136890 }, { "epoch": 87.25302740599108, "grad_norm": 0.27651721239089966, "learning_rate": 6.125704004264898e-05, "loss": 0.0122, "step": 136900 }, { "epoch": 87.25940089228808, "grad_norm": 0.3424924910068512, "learning_rate": 6.125166991083942e-05, "loss": 0.0141, "step": 136910 }, { "epoch": 87.26577437858509, "grad_norm": 0.37724751234054565, "learning_rate": 6.124629964231145e-05, "loss": 0.0099, "step": 136920 }, { "epoch": 87.27214786488209, "grad_norm": 0.3622196316719055, "learning_rate": 6.124092923713033e-05, "loss": 0.0199, "step": 136930 }, { "epoch": 87.2785213511791, "grad_norm": 0.15154772996902466, "learning_rate": 6.12355586953613e-05, "loss": 0.0094, "step": 136940 }, { "epoch": 87.2848948374761, "grad_norm": 0.2815091609954834, "learning_rate": 6.123018801706963e-05, "loss": 0.0127, "step": 136950 }, { "epoch": 87.29126832377311, "grad_norm": 0.20359069108963013, "learning_rate": 6.122481720232058e-05, "loss": 0.0125, "step": 136960 }, { "epoch": 87.2976418100701, "grad_norm": 0.643278181552887, "learning_rate": 6.121944625117939e-05, "loss": 0.013, "step": 136970 }, { "epoch": 87.30401529636711, "grad_norm": 0.24554336071014404, "learning_rate": 6.121407516371134e-05, "loss": 0.0114, "step": 136980 }, { "epoch": 87.31038878266412, "grad_norm": 0.2733408212661743, "learning_rate": 6.120870393998168e-05, "loss": 0.0088, "step": 136990 }, { "epoch": 87.31676226896113, "grad_norm": 0.3001162111759186, "learning_rate": 6.120333258005569e-05, "loss": 0.0199, "step": 137000 }, { "epoch": 87.32313575525812, "grad_norm": 0.25674840807914734, "learning_rate": 6.119796108399862e-05, "loss": 0.017, "step": 137010 }, { "epoch": 87.32950924155513, "grad_norm": 0.3565239906311035, "learning_rate": 6.119258945187577e-05, "loss": 0.0196, "step": 137020 }, { "epoch": 87.33588272785214, "grad_norm": 0.26449480652809143, "learning_rate": 6.118721768375238e-05, "loss": 0.0089, "step": 137030 }, { "epoch": 87.34225621414915, "grad_norm": 0.2986873388290405, "learning_rate": 6.118184577969374e-05, "loss": 0.0098, "step": 137040 }, { "epoch": 87.34862970044614, "grad_norm": 0.5817490816116333, "learning_rate": 6.117647373976512e-05, "loss": 0.016, "step": 137050 }, { "epoch": 87.35500318674315, "grad_norm": 0.1572684645652771, "learning_rate": 6.117110156403178e-05, "loss": 0.0116, "step": 137060 }, { "epoch": 87.36137667304016, "grad_norm": 0.2864267826080322, "learning_rate": 6.116572925255902e-05, "loss": 0.0167, "step": 137070 }, { "epoch": 87.36775015933716, "grad_norm": 0.3274248540401459, "learning_rate": 6.116035680541208e-05, "loss": 0.0106, "step": 137080 }, { "epoch": 87.37412364563416, "grad_norm": 0.40401580929756165, "learning_rate": 6.11549842226563e-05, "loss": 0.0158, "step": 137090 }, { "epoch": 87.38049713193116, "grad_norm": 0.40803083777427673, "learning_rate": 6.114961150435691e-05, "loss": 0.0176, "step": 137100 }, { "epoch": 87.38687061822817, "grad_norm": 0.662192165851593, "learning_rate": 6.114423865057922e-05, "loss": 0.0199, "step": 137110 }, { "epoch": 87.39324410452518, "grad_norm": 0.4642355144023895, "learning_rate": 6.11388656613885e-05, "loss": 0.016, "step": 137120 }, { "epoch": 87.39961759082217, "grad_norm": 0.6393150091171265, "learning_rate": 6.113349253685006e-05, "loss": 0.0181, "step": 137130 }, { "epoch": 87.40599107711918, "grad_norm": 0.29093173146247864, "learning_rate": 6.112811927702917e-05, "loss": 0.0142, "step": 137140 }, { "epoch": 87.41236456341619, "grad_norm": 0.5215398073196411, "learning_rate": 6.112274588199111e-05, "loss": 0.0143, "step": 137150 }, { "epoch": 87.4187380497132, "grad_norm": 0.49087029695510864, "learning_rate": 6.111737235180121e-05, "loss": 0.0146, "step": 137160 }, { "epoch": 87.42511153601019, "grad_norm": 0.20683076977729797, "learning_rate": 6.111199868652472e-05, "loss": 0.0153, "step": 137170 }, { "epoch": 87.4314850223072, "grad_norm": 0.40812626481056213, "learning_rate": 6.110662488622696e-05, "loss": 0.0177, "step": 137180 }, { "epoch": 87.43785850860421, "grad_norm": 0.3843812048435211, "learning_rate": 6.110125095097321e-05, "loss": 0.0115, "step": 137190 }, { "epoch": 87.44423199490122, "grad_norm": 0.3612349033355713, "learning_rate": 6.109587688082881e-05, "loss": 0.0085, "step": 137200 }, { "epoch": 87.45060548119821, "grad_norm": 0.3207167387008667, "learning_rate": 6.109050267585901e-05, "loss": 0.011, "step": 137210 }, { "epoch": 87.45697896749522, "grad_norm": 0.41377007961273193, "learning_rate": 6.108512833612913e-05, "loss": 0.0125, "step": 137220 }, { "epoch": 87.46335245379223, "grad_norm": 0.43529173731803894, "learning_rate": 6.107975386170448e-05, "loss": 0.0201, "step": 137230 }, { "epoch": 87.46972594008923, "grad_norm": 0.5296605825424194, "learning_rate": 6.107437925265035e-05, "loss": 0.0115, "step": 137240 }, { "epoch": 87.47609942638623, "grad_norm": 0.4453665614128113, "learning_rate": 6.106900450903209e-05, "loss": 0.0155, "step": 137250 }, { "epoch": 87.48247291268324, "grad_norm": 0.6092602014541626, "learning_rate": 6.106362963091493e-05, "loss": 0.0236, "step": 137260 }, { "epoch": 87.48884639898024, "grad_norm": 0.38884127140045166, "learning_rate": 6.105825461836425e-05, "loss": 0.0151, "step": 137270 }, { "epoch": 87.49521988527725, "grad_norm": 0.2980034351348877, "learning_rate": 6.10528794714453e-05, "loss": 0.0107, "step": 137280 }, { "epoch": 87.50159337157424, "grad_norm": 0.49606195092201233, "learning_rate": 6.104750419022348e-05, "loss": 0.0177, "step": 137290 }, { "epoch": 87.50796685787125, "grad_norm": 0.26936498284339905, "learning_rate": 6.104212877476403e-05, "loss": 0.0169, "step": 137300 }, { "epoch": 87.51434034416826, "grad_norm": 0.14850039780139923, "learning_rate": 6.103675322513228e-05, "loss": 0.0104, "step": 137310 }, { "epoch": 87.52071383046527, "grad_norm": 0.3515681028366089, "learning_rate": 6.1031377541393573e-05, "loss": 0.0086, "step": 137320 }, { "epoch": 87.52708731676226, "grad_norm": 0.5784274339675903, "learning_rate": 6.102600172361321e-05, "loss": 0.0159, "step": 137330 }, { "epoch": 87.53346080305927, "grad_norm": 0.22007283568382263, "learning_rate": 6.102062577185652e-05, "loss": 0.0208, "step": 137340 }, { "epoch": 87.53983428935628, "grad_norm": 0.3635593056678772, "learning_rate": 6.10152496861888e-05, "loss": 0.0167, "step": 137350 }, { "epoch": 87.54620777565329, "grad_norm": 0.33827078342437744, "learning_rate": 6.1009873466675424e-05, "loss": 0.0149, "step": 137360 }, { "epoch": 87.55258126195028, "grad_norm": 0.23772995173931122, "learning_rate": 6.100449711338169e-05, "loss": 0.0121, "step": 137370 }, { "epoch": 87.55895474824729, "grad_norm": 0.30862006545066833, "learning_rate": 6.0999120626372894e-05, "loss": 0.0171, "step": 137380 }, { "epoch": 87.5653282345443, "grad_norm": 0.384147971868515, "learning_rate": 6.099374400571443e-05, "loss": 0.0227, "step": 137390 }, { "epoch": 87.5717017208413, "grad_norm": 0.3602431118488312, "learning_rate": 6.0988367251471577e-05, "loss": 0.0124, "step": 137400 }, { "epoch": 87.5780752071383, "grad_norm": 0.27384674549102783, "learning_rate": 6.09829903637097e-05, "loss": 0.0175, "step": 137410 }, { "epoch": 87.5844486934353, "grad_norm": 0.5771077871322632, "learning_rate": 6.097761334249411e-05, "loss": 0.0132, "step": 137420 }, { "epoch": 87.59082217973231, "grad_norm": 0.28370943665504456, "learning_rate": 6.097223618789017e-05, "loss": 0.0163, "step": 137430 }, { "epoch": 87.59719566602932, "grad_norm": 0.2920137345790863, "learning_rate": 6.0966858899963175e-05, "loss": 0.024, "step": 137440 }, { "epoch": 87.60356915232632, "grad_norm": 0.2804521918296814, "learning_rate": 6.096148147877851e-05, "loss": 0.0176, "step": 137450 }, { "epoch": 87.60994263862332, "grad_norm": 0.26651501655578613, "learning_rate": 6.0956103924401484e-05, "loss": 0.0172, "step": 137460 }, { "epoch": 87.61631612492033, "grad_norm": 0.30589309334754944, "learning_rate": 6.095072623689746e-05, "loss": 0.0123, "step": 137470 }, { "epoch": 87.62268961121734, "grad_norm": 0.5025419592857361, "learning_rate": 6.094534841633176e-05, "loss": 0.0172, "step": 137480 }, { "epoch": 87.62906309751435, "grad_norm": 0.5468724966049194, "learning_rate": 6.0939970462769756e-05, "loss": 0.016, "step": 137490 }, { "epoch": 87.63543658381134, "grad_norm": 0.2551041841506958, "learning_rate": 6.093459237627678e-05, "loss": 0.0124, "step": 137500 }, { "epoch": 87.64181007010835, "grad_norm": 0.23530267179012299, "learning_rate": 6.092921415691818e-05, "loss": 0.0162, "step": 137510 }, { "epoch": 87.64818355640536, "grad_norm": 0.5390749573707581, "learning_rate": 6.092383580475931e-05, "loss": 0.0114, "step": 137520 }, { "epoch": 87.65455704270236, "grad_norm": 0.3395254611968994, "learning_rate": 6.091845731986552e-05, "loss": 0.0133, "step": 137530 }, { "epoch": 87.66093052899936, "grad_norm": 0.3853604197502136, "learning_rate": 6.091307870230216e-05, "loss": 0.0168, "step": 137540 }, { "epoch": 87.66730401529637, "grad_norm": 0.49970465898513794, "learning_rate": 6.09076999521346e-05, "loss": 0.0125, "step": 137550 }, { "epoch": 87.67367750159337, "grad_norm": 0.19122712314128876, "learning_rate": 6.090232106942818e-05, "loss": 0.0141, "step": 137560 }, { "epoch": 87.68005098789038, "grad_norm": 0.19156301021575928, "learning_rate": 6.089694205424828e-05, "loss": 0.0131, "step": 137570 }, { "epoch": 87.68642447418738, "grad_norm": 0.29423239827156067, "learning_rate": 6.0891562906660226e-05, "loss": 0.0137, "step": 137580 }, { "epoch": 87.69279796048438, "grad_norm": 0.4403967559337616, "learning_rate": 6.088618362672941e-05, "loss": 0.013, "step": 137590 }, { "epoch": 87.69917144678139, "grad_norm": 0.5550444722175598, "learning_rate": 6.088080421452118e-05, "loss": 0.0121, "step": 137600 }, { "epoch": 87.7055449330784, "grad_norm": 0.30383792519569397, "learning_rate": 6.087542467010091e-05, "loss": 0.0211, "step": 137610 }, { "epoch": 87.7119184193754, "grad_norm": 0.44512856006622314, "learning_rate": 6.087004499353396e-05, "loss": 0.0185, "step": 137620 }, { "epoch": 87.7182919056724, "grad_norm": 0.3710156977176666, "learning_rate": 6.0864665184885694e-05, "loss": 0.0137, "step": 137630 }, { "epoch": 87.72466539196941, "grad_norm": 0.5007081031799316, "learning_rate": 6.085928524422151e-05, "loss": 0.0097, "step": 137640 }, { "epoch": 87.73103887826642, "grad_norm": 0.5960485339164734, "learning_rate": 6.085390517160673e-05, "loss": 0.0176, "step": 137650 }, { "epoch": 87.73741236456341, "grad_norm": 0.44741541147232056, "learning_rate": 6.084852496710677e-05, "loss": 0.0103, "step": 137660 }, { "epoch": 87.74378585086042, "grad_norm": 0.2819337844848633, "learning_rate": 6.0843144630786975e-05, "loss": 0.0113, "step": 137670 }, { "epoch": 87.75015933715743, "grad_norm": 0.37044236063957214, "learning_rate": 6.083776416271275e-05, "loss": 0.0218, "step": 137680 }, { "epoch": 87.75653282345444, "grad_norm": 0.48306572437286377, "learning_rate": 6.0832383562949444e-05, "loss": 0.0304, "step": 137690 }, { "epoch": 87.76290630975143, "grad_norm": 0.28179633617401123, "learning_rate": 6.082700283156247e-05, "loss": 0.0105, "step": 137700 }, { "epoch": 87.76927979604844, "grad_norm": 0.26671460270881653, "learning_rate": 6.082162196861718e-05, "loss": 0.0088, "step": 137710 }, { "epoch": 87.77565328234544, "grad_norm": 0.23846182227134705, "learning_rate": 6.081624097417896e-05, "loss": 0.0192, "step": 137720 }, { "epoch": 87.78202676864245, "grad_norm": 0.280593603849411, "learning_rate": 6.0810859848313206e-05, "loss": 0.0196, "step": 137730 }, { "epoch": 87.78840025493945, "grad_norm": 0.369967520236969, "learning_rate": 6.08054785910853e-05, "loss": 0.013, "step": 137740 }, { "epoch": 87.79477374123645, "grad_norm": 0.3321811258792877, "learning_rate": 6.0800097202560624e-05, "loss": 0.0121, "step": 137750 }, { "epoch": 87.80114722753346, "grad_norm": 0.2858947813510895, "learning_rate": 6.079471568280456e-05, "loss": 0.034, "step": 137760 }, { "epoch": 87.80752071383047, "grad_norm": 0.2513634264469147, "learning_rate": 6.078933403188253e-05, "loss": 0.0098, "step": 137770 }, { "epoch": 87.81389420012746, "grad_norm": 0.46375611424446106, "learning_rate": 6.078395224985989e-05, "loss": 0.0158, "step": 137780 }, { "epoch": 87.82026768642447, "grad_norm": 0.2529365122318268, "learning_rate": 6.0778570336802044e-05, "loss": 0.0226, "step": 137790 }, { "epoch": 87.82664117272148, "grad_norm": 0.41690701246261597, "learning_rate": 6.077318829277441e-05, "loss": 0.0116, "step": 137800 }, { "epoch": 87.83301465901849, "grad_norm": 0.9054744839668274, "learning_rate": 6.076780611784234e-05, "loss": 0.0135, "step": 137810 }, { "epoch": 87.83938814531548, "grad_norm": 0.16648191213607788, "learning_rate": 6.076242381207129e-05, "loss": 0.0109, "step": 137820 }, { "epoch": 87.84576163161249, "grad_norm": 0.2318059504032135, "learning_rate": 6.07570413755266e-05, "loss": 0.0249, "step": 137830 }, { "epoch": 87.8521351179095, "grad_norm": 0.355350524187088, "learning_rate": 6.0751658808273724e-05, "loss": 0.0158, "step": 137840 }, { "epoch": 87.8585086042065, "grad_norm": 0.46908000111579895, "learning_rate": 6.074627611037802e-05, "loss": 0.0228, "step": 137850 }, { "epoch": 87.8648820905035, "grad_norm": 0.4624278247356415, "learning_rate": 6.074089328190493e-05, "loss": 0.0135, "step": 137860 }, { "epoch": 87.87125557680051, "grad_norm": 0.29126614332199097, "learning_rate": 6.0735510322919855e-05, "loss": 0.013, "step": 137870 }, { "epoch": 87.87762906309752, "grad_norm": 0.3741316497325897, "learning_rate": 6.073012723348818e-05, "loss": 0.0186, "step": 137880 }, { "epoch": 87.88400254939452, "grad_norm": 0.46564021706581116, "learning_rate": 6.0724744013675325e-05, "loss": 0.0169, "step": 137890 }, { "epoch": 87.89037603569152, "grad_norm": 0.28127071261405945, "learning_rate": 6.071936066354671e-05, "loss": 0.0085, "step": 137900 }, { "epoch": 87.89674952198853, "grad_norm": 0.26654067635536194, "learning_rate": 6.071397718316775e-05, "loss": 0.0132, "step": 137910 }, { "epoch": 87.90312300828553, "grad_norm": 0.32834935188293457, "learning_rate": 6.070859357260383e-05, "loss": 0.0215, "step": 137920 }, { "epoch": 87.90949649458254, "grad_norm": 0.3755296468734741, "learning_rate": 6.070320983192042e-05, "loss": 0.0165, "step": 137930 }, { "epoch": 87.91586998087953, "grad_norm": 0.596771240234375, "learning_rate": 6.069782596118288e-05, "loss": 0.0119, "step": 137940 }, { "epoch": 87.92224346717654, "grad_norm": 0.3252980411052704, "learning_rate": 6.0692441960456657e-05, "loss": 0.0107, "step": 137950 }, { "epoch": 87.92861695347355, "grad_norm": 0.441940039396286, "learning_rate": 6.068705782980717e-05, "loss": 0.014, "step": 137960 }, { "epoch": 87.93499043977056, "grad_norm": 0.5034086108207703, "learning_rate": 6.0681673569299834e-05, "loss": 0.0175, "step": 137970 }, { "epoch": 87.94136392606755, "grad_norm": 0.7978722453117371, "learning_rate": 6.06762891790001e-05, "loss": 0.029, "step": 137980 }, { "epoch": 87.94773741236456, "grad_norm": 0.461183100938797, "learning_rate": 6.067090465897335e-05, "loss": 0.0211, "step": 137990 }, { "epoch": 87.95411089866157, "grad_norm": 0.3494621515274048, "learning_rate": 6.066552000928505e-05, "loss": 0.0209, "step": 138000 }, { "epoch": 87.96048438495858, "grad_norm": 0.3269343078136444, "learning_rate": 6.0660135230000604e-05, "loss": 0.0136, "step": 138010 }, { "epoch": 87.96685787125557, "grad_norm": 0.3699321746826172, "learning_rate": 6.065475032118545e-05, "loss": 0.011, "step": 138020 }, { "epoch": 87.97323135755258, "grad_norm": 0.2585626244544983, "learning_rate": 6.064936528290503e-05, "loss": 0.015, "step": 138030 }, { "epoch": 87.97960484384959, "grad_norm": 0.18687231838703156, "learning_rate": 6.064398011522474e-05, "loss": 0.0185, "step": 138040 }, { "epoch": 87.9859783301466, "grad_norm": 0.6049677133560181, "learning_rate": 6.063859481821007e-05, "loss": 0.0149, "step": 138050 }, { "epoch": 87.99235181644359, "grad_norm": 0.49252572655677795, "learning_rate": 6.063320939192641e-05, "loss": 0.0108, "step": 138060 }, { "epoch": 87.9987253027406, "grad_norm": 0.41785725951194763, "learning_rate": 6.0627823836439234e-05, "loss": 0.009, "step": 138070 }, { "epoch": 88.0050987890376, "grad_norm": 0.21481095254421234, "learning_rate": 6.062243815181394e-05, "loss": 0.0112, "step": 138080 }, { "epoch": 88.01147227533461, "grad_norm": 0.20791058242321014, "learning_rate": 6.061705233811602e-05, "loss": 0.0174, "step": 138090 }, { "epoch": 88.01784576163162, "grad_norm": 0.4167209565639496, "learning_rate": 6.0611666395410885e-05, "loss": 0.0188, "step": 138100 }, { "epoch": 88.02421924792861, "grad_norm": 0.46749648451805115, "learning_rate": 6.0606280323763966e-05, "loss": 0.0209, "step": 138110 }, { "epoch": 88.03059273422562, "grad_norm": 0.2618454396724701, "learning_rate": 6.060089412324075e-05, "loss": 0.017, "step": 138120 }, { "epoch": 88.03696622052263, "grad_norm": 0.5788019895553589, "learning_rate": 6.059550779390664e-05, "loss": 0.0153, "step": 138130 }, { "epoch": 88.04333970681964, "grad_norm": 0.30578893423080444, "learning_rate": 6.059012133582712e-05, "loss": 0.016, "step": 138140 }, { "epoch": 88.04971319311663, "grad_norm": 0.42068061232566833, "learning_rate": 6.058473474906762e-05, "loss": 0.0141, "step": 138150 }, { "epoch": 88.05608667941364, "grad_norm": 0.5631097555160522, "learning_rate": 6.0579348033693604e-05, "loss": 0.0189, "step": 138160 }, { "epoch": 88.06246016571065, "grad_norm": 0.21441565454006195, "learning_rate": 6.0573961189770515e-05, "loss": 0.0148, "step": 138170 }, { "epoch": 88.06883365200765, "grad_norm": 0.5174910426139832, "learning_rate": 6.056857421736383e-05, "loss": 0.0125, "step": 138180 }, { "epoch": 88.07520713830465, "grad_norm": 0.48113512992858887, "learning_rate": 6.056318711653897e-05, "loss": 0.0076, "step": 138190 }, { "epoch": 88.08158062460166, "grad_norm": 0.5321341753005981, "learning_rate": 6.055779988736142e-05, "loss": 0.0161, "step": 138200 }, { "epoch": 88.08795411089866, "grad_norm": 0.5354197025299072, "learning_rate": 6.055241252989663e-05, "loss": 0.0154, "step": 138210 }, { "epoch": 88.09432759719567, "grad_norm": 0.4129907488822937, "learning_rate": 6.054702504421007e-05, "loss": 0.0163, "step": 138220 }, { "epoch": 88.10070108349267, "grad_norm": 0.30221715569496155, "learning_rate": 6.054163743036719e-05, "loss": 0.0166, "step": 138230 }, { "epoch": 88.10707456978967, "grad_norm": 0.3528828024864197, "learning_rate": 6.053624968843348e-05, "loss": 0.0128, "step": 138240 }, { "epoch": 88.11344805608668, "grad_norm": 0.2371608316898346, "learning_rate": 6.053086181847437e-05, "loss": 0.0171, "step": 138250 }, { "epoch": 88.11982154238369, "grad_norm": 0.32056957483291626, "learning_rate": 6.0525473820555356e-05, "loss": 0.0169, "step": 138260 }, { "epoch": 88.12619502868068, "grad_norm": 0.36592888832092285, "learning_rate": 6.052008569474189e-05, "loss": 0.0169, "step": 138270 }, { "epoch": 88.13256851497769, "grad_norm": 0.27562564611434937, "learning_rate": 6.0514697441099465e-05, "loss": 0.0206, "step": 138280 }, { "epoch": 88.1389420012747, "grad_norm": 0.4157578945159912, "learning_rate": 6.050930905969352e-05, "loss": 0.0085, "step": 138290 }, { "epoch": 88.14531548757171, "grad_norm": 0.22583967447280884, "learning_rate": 6.050392055058955e-05, "loss": 0.0092, "step": 138300 }, { "epoch": 88.1516889738687, "grad_norm": 0.3714357912540436, "learning_rate": 6.049853191385304e-05, "loss": 0.0169, "step": 138310 }, { "epoch": 88.15806246016571, "grad_norm": 0.35761114954948425, "learning_rate": 6.049314314954946e-05, "loss": 0.0105, "step": 138320 }, { "epoch": 88.16443594646272, "grad_norm": 0.5029021501541138, "learning_rate": 6.048775425774425e-05, "loss": 0.0176, "step": 138330 }, { "epoch": 88.17080943275973, "grad_norm": 0.41381770372390747, "learning_rate": 6.0482365238502956e-05, "loss": 0.0091, "step": 138340 }, { "epoch": 88.17718291905672, "grad_norm": 0.4459671378135681, "learning_rate": 6.0476976091891025e-05, "loss": 0.0136, "step": 138350 }, { "epoch": 88.18355640535373, "grad_norm": 0.5066105723381042, "learning_rate": 6.047158681797394e-05, "loss": 0.0091, "step": 138360 }, { "epoch": 88.18992989165073, "grad_norm": 0.2561478316783905, "learning_rate": 6.046619741681718e-05, "loss": 0.0134, "step": 138370 }, { "epoch": 88.19630337794774, "grad_norm": 0.9254859685897827, "learning_rate": 6.046080788848624e-05, "loss": 0.0197, "step": 138380 }, { "epoch": 88.20267686424474, "grad_norm": 0.6464645266532898, "learning_rate": 6.0455418233046614e-05, "loss": 0.0145, "step": 138390 }, { "epoch": 88.20905035054174, "grad_norm": 0.5584646463394165, "learning_rate": 6.0450028450563776e-05, "loss": 0.0283, "step": 138400 }, { "epoch": 88.21542383683875, "grad_norm": 0.32420289516448975, "learning_rate": 6.044463854110324e-05, "loss": 0.0211, "step": 138410 }, { "epoch": 88.22179732313576, "grad_norm": 0.353691428899765, "learning_rate": 6.043924850473048e-05, "loss": 0.0077, "step": 138420 }, { "epoch": 88.22817080943275, "grad_norm": 0.32140982151031494, "learning_rate": 6.043385834151099e-05, "loss": 0.0163, "step": 138430 }, { "epoch": 88.23454429572976, "grad_norm": 0.41910308599472046, "learning_rate": 6.042846805151027e-05, "loss": 0.0129, "step": 138440 }, { "epoch": 88.24091778202677, "grad_norm": 0.27970653772354126, "learning_rate": 6.042307763479381e-05, "loss": 0.0154, "step": 138450 }, { "epoch": 88.24729126832378, "grad_norm": 0.17582742869853973, "learning_rate": 6.0417687091427124e-05, "loss": 0.0191, "step": 138460 }, { "epoch": 88.25366475462077, "grad_norm": 0.42742201685905457, "learning_rate": 6.04122964214757e-05, "loss": 0.0121, "step": 138470 }, { "epoch": 88.26003824091778, "grad_norm": 0.29840150475502014, "learning_rate": 6.040690562500505e-05, "loss": 0.0145, "step": 138480 }, { "epoch": 88.26641172721479, "grad_norm": 0.24473686516284943, "learning_rate": 6.0401514702080666e-05, "loss": 0.0181, "step": 138490 }, { "epoch": 88.2727852135118, "grad_norm": 0.45903193950653076, "learning_rate": 6.039612365276807e-05, "loss": 0.0258, "step": 138500 }, { "epoch": 88.27915869980879, "grad_norm": 0.3691752254962921, "learning_rate": 6.0390732477132736e-05, "loss": 0.0197, "step": 138510 }, { "epoch": 88.2855321861058, "grad_norm": 0.3191525638103485, "learning_rate": 6.0385341175240205e-05, "loss": 0.0252, "step": 138520 }, { "epoch": 88.2919056724028, "grad_norm": 0.45346665382385254, "learning_rate": 6.0379949747155974e-05, "loss": 0.0203, "step": 138530 }, { "epoch": 88.29827915869981, "grad_norm": 0.788385808467865, "learning_rate": 6.037455819294554e-05, "loss": 0.0188, "step": 138540 }, { "epoch": 88.3046526449968, "grad_norm": 0.4732201099395752, "learning_rate": 6.036916651267445e-05, "loss": 0.0157, "step": 138550 }, { "epoch": 88.31102613129381, "grad_norm": 0.6238333582878113, "learning_rate": 6.036377470640817e-05, "loss": 0.0175, "step": 138560 }, { "epoch": 88.31739961759082, "grad_norm": 0.3412695825099945, "learning_rate": 6.035838277421226e-05, "loss": 0.0112, "step": 138570 }, { "epoch": 88.32377310388783, "grad_norm": 0.4377163350582123, "learning_rate": 6.0352990716152224e-05, "loss": 0.0193, "step": 138580 }, { "epoch": 88.33014659018482, "grad_norm": 0.2816741466522217, "learning_rate": 6.034759853229356e-05, "loss": 0.0116, "step": 138590 }, { "epoch": 88.33652007648183, "grad_norm": 0.5502408146858215, "learning_rate": 6.034220622270181e-05, "loss": 0.0137, "step": 138600 }, { "epoch": 88.34289356277884, "grad_norm": 0.39333054423332214, "learning_rate": 6.033681378744249e-05, "loss": 0.0106, "step": 138610 }, { "epoch": 88.34926704907585, "grad_norm": 0.21571218967437744, "learning_rate": 6.033142122658113e-05, "loss": 0.0166, "step": 138620 }, { "epoch": 88.35564053537284, "grad_norm": 0.38808146119117737, "learning_rate": 6.032602854018324e-05, "loss": 0.0247, "step": 138630 }, { "epoch": 88.36201402166985, "grad_norm": 0.27111512422561646, "learning_rate": 6.032063572831437e-05, "loss": 0.0095, "step": 138640 }, { "epoch": 88.36838750796686, "grad_norm": 0.6893284320831299, "learning_rate": 6.031524279104001e-05, "loss": 0.0212, "step": 138650 }, { "epoch": 88.37476099426387, "grad_norm": 0.20324267446994781, "learning_rate": 6.030984972842571e-05, "loss": 0.0138, "step": 138660 }, { "epoch": 88.38113448056086, "grad_norm": 0.2440280318260193, "learning_rate": 6.030445654053702e-05, "loss": 0.0195, "step": 138670 }, { "epoch": 88.38750796685787, "grad_norm": 0.25892162322998047, "learning_rate": 6.0299063227439445e-05, "loss": 0.0139, "step": 138680 }, { "epoch": 88.39388145315488, "grad_norm": 0.25729477405548096, "learning_rate": 6.0293669789198524e-05, "loss": 0.0164, "step": 138690 }, { "epoch": 88.40025493945188, "grad_norm": 0.2350446879863739, "learning_rate": 6.0288276225879805e-05, "loss": 0.0085, "step": 138700 }, { "epoch": 88.40662842574889, "grad_norm": 0.2607162296772003, "learning_rate": 6.0282882537548815e-05, "loss": 0.0133, "step": 138710 }, { "epoch": 88.41300191204589, "grad_norm": 0.5814048051834106, "learning_rate": 6.027748872427108e-05, "loss": 0.0171, "step": 138720 }, { "epoch": 88.4193753983429, "grad_norm": 0.608008861541748, "learning_rate": 6.027209478611217e-05, "loss": 0.0114, "step": 138730 }, { "epoch": 88.4257488846399, "grad_norm": 0.2913368046283722, "learning_rate": 6.026670072313759e-05, "loss": 0.0093, "step": 138740 }, { "epoch": 88.43212237093691, "grad_norm": 0.45396164059638977, "learning_rate": 6.026130653541292e-05, "loss": 0.0191, "step": 138750 }, { "epoch": 88.4384958572339, "grad_norm": 0.3579753637313843, "learning_rate": 6.025591222300368e-05, "loss": 0.0133, "step": 138760 }, { "epoch": 88.44486934353091, "grad_norm": 0.3730071187019348, "learning_rate": 6.025051778597543e-05, "loss": 0.0199, "step": 138770 }, { "epoch": 88.45124282982792, "grad_norm": 0.24438969790935516, "learning_rate": 6.0245123224393705e-05, "loss": 0.0128, "step": 138780 }, { "epoch": 88.45761631612493, "grad_norm": 0.3155497610569, "learning_rate": 6.023972853832406e-05, "loss": 0.0158, "step": 138790 }, { "epoch": 88.46398980242192, "grad_norm": 0.33482393622398376, "learning_rate": 6.023433372783205e-05, "loss": 0.0143, "step": 138800 }, { "epoch": 88.47036328871893, "grad_norm": 0.3560594618320465, "learning_rate": 6.022893879298321e-05, "loss": 0.0128, "step": 138810 }, { "epoch": 88.47673677501594, "grad_norm": 0.2742050886154175, "learning_rate": 6.022354373384312e-05, "loss": 0.012, "step": 138820 }, { "epoch": 88.48311026131294, "grad_norm": 0.4026924967765808, "learning_rate": 6.0218148550477324e-05, "loss": 0.0165, "step": 138830 }, { "epoch": 88.48948374760994, "grad_norm": 0.12349523603916168, "learning_rate": 6.021275324295137e-05, "loss": 0.0185, "step": 138840 }, { "epoch": 88.49585723390695, "grad_norm": 0.2741256058216095, "learning_rate": 6.0207357811330825e-05, "loss": 0.0142, "step": 138850 }, { "epoch": 88.50223072020395, "grad_norm": 0.3824670612812042, "learning_rate": 6.020196225568123e-05, "loss": 0.0115, "step": 138860 }, { "epoch": 88.50860420650096, "grad_norm": 0.3469003140926361, "learning_rate": 6.0196566576068184e-05, "loss": 0.0145, "step": 138870 }, { "epoch": 88.51497769279796, "grad_norm": 0.33676785230636597, "learning_rate": 6.01911707725572e-05, "loss": 0.0114, "step": 138880 }, { "epoch": 88.52135117909496, "grad_norm": 0.3251839578151703, "learning_rate": 6.0185774845213884e-05, "loss": 0.0161, "step": 138890 }, { "epoch": 88.52772466539197, "grad_norm": 0.33980852365493774, "learning_rate": 6.018037879410379e-05, "loss": 0.0156, "step": 138900 }, { "epoch": 88.53409815168898, "grad_norm": 0.36193445324897766, "learning_rate": 6.0174982619292476e-05, "loss": 0.0259, "step": 138910 }, { "epoch": 88.54047163798597, "grad_norm": 0.4486226737499237, "learning_rate": 6.016958632084552e-05, "loss": 0.0167, "step": 138920 }, { "epoch": 88.54684512428298, "grad_norm": 0.39785337448120117, "learning_rate": 6.016418989882848e-05, "loss": 0.0139, "step": 138930 }, { "epoch": 88.55321861057999, "grad_norm": 0.33903101086616516, "learning_rate": 6.015879335330694e-05, "loss": 0.0316, "step": 138940 }, { "epoch": 88.559592096877, "grad_norm": 0.46979644894599915, "learning_rate": 6.0153396684346475e-05, "loss": 0.0126, "step": 138950 }, { "epoch": 88.56596558317399, "grad_norm": 0.420225590467453, "learning_rate": 6.014799989201265e-05, "loss": 0.0138, "step": 138960 }, { "epoch": 88.572339069471, "grad_norm": 0.2955191433429718, "learning_rate": 6.0142602976371044e-05, "loss": 0.0154, "step": 138970 }, { "epoch": 88.578712555768, "grad_norm": 0.34962522983551025, "learning_rate": 6.013720593748724e-05, "loss": 0.0142, "step": 138980 }, { "epoch": 88.58508604206502, "grad_norm": 0.5609009861946106, "learning_rate": 6.013180877542681e-05, "loss": 0.0183, "step": 138990 }, { "epoch": 88.59145952836201, "grad_norm": 0.5953995585441589, "learning_rate": 6.012641149025533e-05, "loss": 0.0257, "step": 139000 }, { "epoch": 88.59783301465902, "grad_norm": 0.44574853777885437, "learning_rate": 6.012101408203841e-05, "loss": 0.0167, "step": 139010 }, { "epoch": 88.60420650095602, "grad_norm": 0.4414599537849426, "learning_rate": 6.0115616550841593e-05, "loss": 0.0163, "step": 139020 }, { "epoch": 88.61057998725303, "grad_norm": 0.23228652775287628, "learning_rate": 6.01102188967305e-05, "loss": 0.0074, "step": 139030 }, { "epoch": 88.61695347355003, "grad_norm": 0.3525727093219757, "learning_rate": 6.010482111977068e-05, "loss": 0.0108, "step": 139040 }, { "epoch": 88.62332695984703, "grad_norm": 0.2600214183330536, "learning_rate": 6.0099423220027764e-05, "loss": 0.0119, "step": 139050 }, { "epoch": 88.62970044614404, "grad_norm": 0.48309341073036194, "learning_rate": 6.0094025197567304e-05, "loss": 0.018, "step": 139060 }, { "epoch": 88.63607393244105, "grad_norm": 0.38054925203323364, "learning_rate": 6.0088627052454914e-05, "loss": 0.0092, "step": 139070 }, { "epoch": 88.64244741873804, "grad_norm": 0.2867327928543091, "learning_rate": 6.0083228784756196e-05, "loss": 0.0122, "step": 139080 }, { "epoch": 88.64882090503505, "grad_norm": 0.2530948519706726, "learning_rate": 6.00778303945367e-05, "loss": 0.0125, "step": 139090 }, { "epoch": 88.65519439133206, "grad_norm": 0.4281449317932129, "learning_rate": 6.007243188186207e-05, "loss": 0.0131, "step": 139100 }, { "epoch": 88.66156787762907, "grad_norm": 0.16592271625995636, "learning_rate": 6.006703324679788e-05, "loss": 0.0132, "step": 139110 }, { "epoch": 88.66794136392606, "grad_norm": 0.5478826761245728, "learning_rate": 6.006163448940972e-05, "loss": 0.0119, "step": 139120 }, { "epoch": 88.67431485022307, "grad_norm": 0.2892785668373108, "learning_rate": 6.0056235609763215e-05, "loss": 0.021, "step": 139130 }, { "epoch": 88.68068833652008, "grad_norm": 0.47925692796707153, "learning_rate": 6.0050836607923945e-05, "loss": 0.0186, "step": 139140 }, { "epoch": 88.68706182281709, "grad_norm": 0.4407724440097809, "learning_rate": 6.004543748395752e-05, "loss": 0.0158, "step": 139150 }, { "epoch": 88.69343530911408, "grad_norm": 0.6699056625366211, "learning_rate": 6.0040038237929554e-05, "loss": 0.0185, "step": 139160 }, { "epoch": 88.69980879541109, "grad_norm": 0.2031799852848053, "learning_rate": 6.0034638869905636e-05, "loss": 0.0154, "step": 139170 }, { "epoch": 88.7061822817081, "grad_norm": 0.274306982755661, "learning_rate": 6.002923937995137e-05, "loss": 0.0208, "step": 139180 }, { "epoch": 88.7125557680051, "grad_norm": 0.35753464698791504, "learning_rate": 6.00238397681324e-05, "loss": 0.0101, "step": 139190 }, { "epoch": 88.7189292543021, "grad_norm": 0.469097375869751, "learning_rate": 6.00184400345143e-05, "loss": 0.0184, "step": 139200 }, { "epoch": 88.7253027405991, "grad_norm": 0.5874054431915283, "learning_rate": 6.0013040179162694e-05, "loss": 0.0226, "step": 139210 }, { "epoch": 88.73167622689611, "grad_norm": 0.1944512128829956, "learning_rate": 6.00076402021432e-05, "loss": 0.0192, "step": 139220 }, { "epoch": 88.73804971319312, "grad_norm": 0.18699809908866882, "learning_rate": 6.000224010352142e-05, "loss": 0.0161, "step": 139230 }, { "epoch": 88.74442319949011, "grad_norm": 0.1878189742565155, "learning_rate": 5.9996839883363e-05, "loss": 0.0233, "step": 139240 }, { "epoch": 88.75079668578712, "grad_norm": 0.2968621551990509, "learning_rate": 5.999143954173352e-05, "loss": 0.0185, "step": 139250 }, { "epoch": 88.75717017208413, "grad_norm": 0.7470718026161194, "learning_rate": 5.998603907869862e-05, "loss": 0.0166, "step": 139260 }, { "epoch": 88.76354365838114, "grad_norm": 0.44994062185287476, "learning_rate": 5.998063849432391e-05, "loss": 0.0136, "step": 139270 }, { "epoch": 88.76991714467813, "grad_norm": 0.9254646301269531, "learning_rate": 5.997523778867503e-05, "loss": 0.0164, "step": 139280 }, { "epoch": 88.77629063097514, "grad_norm": 0.30074551701545715, "learning_rate": 5.996983696181758e-05, "loss": 0.0137, "step": 139290 }, { "epoch": 88.78266411727215, "grad_norm": 0.529801070690155, "learning_rate": 5.996443601381722e-05, "loss": 0.0101, "step": 139300 }, { "epoch": 88.78903760356916, "grad_norm": 0.4674713611602783, "learning_rate": 5.995903494473953e-05, "loss": 0.0191, "step": 139310 }, { "epoch": 88.79541108986615, "grad_norm": 0.6590204834938049, "learning_rate": 5.995363375465017e-05, "loss": 0.0127, "step": 139320 }, { "epoch": 88.80178457616316, "grad_norm": 0.3950510621070862, "learning_rate": 5.9948232443614774e-05, "loss": 0.0124, "step": 139330 }, { "epoch": 88.80815806246017, "grad_norm": 0.4036603569984436, "learning_rate": 5.9942831011698944e-05, "loss": 0.0092, "step": 139340 }, { "epoch": 88.81453154875717, "grad_norm": 0.5487477779388428, "learning_rate": 5.993742945896834e-05, "loss": 0.0129, "step": 139350 }, { "epoch": 88.82090503505418, "grad_norm": 0.7304756045341492, "learning_rate": 5.993202778548858e-05, "loss": 0.0161, "step": 139360 }, { "epoch": 88.82727852135118, "grad_norm": 0.3287215828895569, "learning_rate": 5.992662599132531e-05, "loss": 0.015, "step": 139370 }, { "epoch": 88.83365200764818, "grad_norm": 0.13390077650547028, "learning_rate": 5.9921224076544155e-05, "loss": 0.0169, "step": 139380 }, { "epoch": 88.84002549394519, "grad_norm": 0.2724994421005249, "learning_rate": 5.9915822041210775e-05, "loss": 0.0163, "step": 139390 }, { "epoch": 88.8463989802422, "grad_norm": 0.465359628200531, "learning_rate": 5.991041988539079e-05, "loss": 0.0141, "step": 139400 }, { "epoch": 88.8527724665392, "grad_norm": 0.30984529852867126, "learning_rate": 5.990501760914984e-05, "loss": 0.0125, "step": 139410 }, { "epoch": 88.8591459528362, "grad_norm": 0.3201258182525635, "learning_rate": 5.989961521255357e-05, "loss": 0.0161, "step": 139420 }, { "epoch": 88.86551943913321, "grad_norm": 0.4278925359249115, "learning_rate": 5.989421269566763e-05, "loss": 0.0216, "step": 139430 }, { "epoch": 88.87189292543022, "grad_norm": 0.3441429138183594, "learning_rate": 5.9888810058557674e-05, "loss": 0.0119, "step": 139440 }, { "epoch": 88.87826641172721, "grad_norm": 0.28052857518196106, "learning_rate": 5.988340730128933e-05, "loss": 0.02, "step": 139450 }, { "epoch": 88.88463989802422, "grad_norm": 0.34180745482444763, "learning_rate": 5.9878004423928256e-05, "loss": 0.0096, "step": 139460 }, { "epoch": 88.89101338432123, "grad_norm": 0.4202592670917511, "learning_rate": 5.987260142654012e-05, "loss": 0.0118, "step": 139470 }, { "epoch": 88.89738687061823, "grad_norm": 0.7288104891777039, "learning_rate": 5.986719830919053e-05, "loss": 0.0122, "step": 139480 }, { "epoch": 88.90376035691523, "grad_norm": 0.5493438839912415, "learning_rate": 5.986179507194518e-05, "loss": 0.0156, "step": 139490 }, { "epoch": 88.91013384321224, "grad_norm": 0.17476196587085724, "learning_rate": 5.98563917148697e-05, "loss": 0.0114, "step": 139500 }, { "epoch": 88.91650732950924, "grad_norm": 0.3440963625907898, "learning_rate": 5.985098823802976e-05, "loss": 0.0179, "step": 139510 }, { "epoch": 88.92288081580625, "grad_norm": 0.3376140892505646, "learning_rate": 5.9845584641491e-05, "loss": 0.011, "step": 139520 }, { "epoch": 88.92925430210325, "grad_norm": 0.5809616446495056, "learning_rate": 5.984018092531911e-05, "loss": 0.0137, "step": 139530 }, { "epoch": 88.93562778840025, "grad_norm": 0.5299781560897827, "learning_rate": 5.983477708957972e-05, "loss": 0.0168, "step": 139540 }, { "epoch": 88.94200127469726, "grad_norm": 0.3779919743537903, "learning_rate": 5.9829373134338503e-05, "loss": 0.0104, "step": 139550 }, { "epoch": 88.94837476099427, "grad_norm": 0.27828145027160645, "learning_rate": 5.9823969059661135e-05, "loss": 0.0094, "step": 139560 }, { "epoch": 88.95474824729126, "grad_norm": 0.30722352862358093, "learning_rate": 5.981856486561325e-05, "loss": 0.0182, "step": 139570 }, { "epoch": 88.96112173358827, "grad_norm": 0.32641685009002686, "learning_rate": 5.981316055226054e-05, "loss": 0.0071, "step": 139580 }, { "epoch": 88.96749521988528, "grad_norm": 0.21725726127624512, "learning_rate": 5.980775611966866e-05, "loss": 0.0129, "step": 139590 }, { "epoch": 88.97386870618229, "grad_norm": 0.4590458273887634, "learning_rate": 5.980235156790329e-05, "loss": 0.0105, "step": 139600 }, { "epoch": 88.98024219247928, "grad_norm": 0.5501264929771423, "learning_rate": 5.979694689703008e-05, "loss": 0.0159, "step": 139610 }, { "epoch": 88.98661567877629, "grad_norm": 0.3600703179836273, "learning_rate": 5.979154210711473e-05, "loss": 0.008, "step": 139620 }, { "epoch": 88.9929891650733, "grad_norm": 0.23298178613185883, "learning_rate": 5.978613719822289e-05, "loss": 0.0133, "step": 139630 }, { "epoch": 88.9993626513703, "grad_norm": 0.16055364906787872, "learning_rate": 5.978073217042025e-05, "loss": 0.0063, "step": 139640 }, { "epoch": 89.0057361376673, "grad_norm": 0.34845900535583496, "learning_rate": 5.9775327023772465e-05, "loss": 0.0193, "step": 139650 }, { "epoch": 89.0121096239643, "grad_norm": 0.5981891751289368, "learning_rate": 5.9769921758345246e-05, "loss": 0.0142, "step": 139660 }, { "epoch": 89.01848311026131, "grad_norm": 0.22142700850963593, "learning_rate": 5.976451637420424e-05, "loss": 0.0131, "step": 139670 }, { "epoch": 89.02485659655832, "grad_norm": 0.12764482200145721, "learning_rate": 5.975911087141516e-05, "loss": 0.016, "step": 139680 }, { "epoch": 89.03123008285532, "grad_norm": 0.3739071190357208, "learning_rate": 5.975370525004365e-05, "loss": 0.0181, "step": 139690 }, { "epoch": 89.03760356915232, "grad_norm": 0.2990763783454895, "learning_rate": 5.974829951015542e-05, "loss": 0.0101, "step": 139700 }, { "epoch": 89.04397705544933, "grad_norm": 0.4330497682094574, "learning_rate": 5.974289365181613e-05, "loss": 0.0126, "step": 139710 }, { "epoch": 89.05035054174634, "grad_norm": 0.312810480594635, "learning_rate": 5.9737487675091506e-05, "loss": 0.0227, "step": 139720 }, { "epoch": 89.05672402804333, "grad_norm": 0.34370365738868713, "learning_rate": 5.973208158004719e-05, "loss": 0.0159, "step": 139730 }, { "epoch": 89.06309751434034, "grad_norm": 0.2227131724357605, "learning_rate": 5.972667536674892e-05, "loss": 0.0231, "step": 139740 }, { "epoch": 89.06947100063735, "grad_norm": 0.3727726638317108, "learning_rate": 5.972126903526235e-05, "loss": 0.0182, "step": 139750 }, { "epoch": 89.07584448693436, "grad_norm": 0.21760053932666779, "learning_rate": 5.971586258565318e-05, "loss": 0.0108, "step": 139760 }, { "epoch": 89.08221797323135, "grad_norm": 0.4540652632713318, "learning_rate": 5.971045601798711e-05, "loss": 0.0131, "step": 139770 }, { "epoch": 89.08859145952836, "grad_norm": 0.35886117815971375, "learning_rate": 5.970504933232983e-05, "loss": 0.0101, "step": 139780 }, { "epoch": 89.09496494582537, "grad_norm": 0.4414829909801483, "learning_rate": 5.9699642528747044e-05, "loss": 0.0179, "step": 139790 }, { "epoch": 89.10133843212238, "grad_norm": 0.18439535796642303, "learning_rate": 5.9694235607304426e-05, "loss": 0.0086, "step": 139800 }, { "epoch": 89.10771191841937, "grad_norm": 0.321682333946228, "learning_rate": 5.9688828568067715e-05, "loss": 0.0203, "step": 139810 }, { "epoch": 89.11408540471638, "grad_norm": 0.2979534864425659, "learning_rate": 5.968342141110257e-05, "loss": 0.0097, "step": 139820 }, { "epoch": 89.12045889101339, "grad_norm": 0.28446081280708313, "learning_rate": 5.9678014136474726e-05, "loss": 0.0115, "step": 139830 }, { "epoch": 89.1268323773104, "grad_norm": 0.4428972601890564, "learning_rate": 5.967260674424986e-05, "loss": 0.0149, "step": 139840 }, { "epoch": 89.13320586360739, "grad_norm": 0.23701967298984528, "learning_rate": 5.96671992344937e-05, "loss": 0.0154, "step": 139850 }, { "epoch": 89.1395793499044, "grad_norm": 0.17305836081504822, "learning_rate": 5.966179160727193e-05, "loss": 0.0151, "step": 139860 }, { "epoch": 89.1459528362014, "grad_norm": 0.24773304164409637, "learning_rate": 5.965638386265028e-05, "loss": 0.0127, "step": 139870 }, { "epoch": 89.15232632249841, "grad_norm": 0.6708719730377197, "learning_rate": 5.965097600069446e-05, "loss": 0.0196, "step": 139880 }, { "epoch": 89.1586998087954, "grad_norm": 0.47111719846725464, "learning_rate": 5.964556802147015e-05, "loss": 0.0174, "step": 139890 }, { "epoch": 89.16507329509241, "grad_norm": 0.1898910403251648, "learning_rate": 5.964015992504308e-05, "loss": 0.0177, "step": 139900 }, { "epoch": 89.17144678138942, "grad_norm": 0.21255114674568176, "learning_rate": 5.963475171147898e-05, "loss": 0.0084, "step": 139910 }, { "epoch": 89.17782026768643, "grad_norm": 0.2475043386220932, "learning_rate": 5.9629343380843536e-05, "loss": 0.0173, "step": 139920 }, { "epoch": 89.18419375398342, "grad_norm": 0.23626920580863953, "learning_rate": 5.9623934933202486e-05, "loss": 0.0157, "step": 139930 }, { "epoch": 89.19056724028043, "grad_norm": 0.24922537803649902, "learning_rate": 5.961852636862154e-05, "loss": 0.0226, "step": 139940 }, { "epoch": 89.19694072657744, "grad_norm": 0.46172213554382324, "learning_rate": 5.961311768716641e-05, "loss": 0.0094, "step": 139950 }, { "epoch": 89.20331421287445, "grad_norm": 0.45166119933128357, "learning_rate": 5.960770888890283e-05, "loss": 0.0185, "step": 139960 }, { "epoch": 89.20968769917145, "grad_norm": 0.663911759853363, "learning_rate": 5.960229997389652e-05, "loss": 0.0112, "step": 139970 }, { "epoch": 89.21606118546845, "grad_norm": 0.2624914348125458, "learning_rate": 5.9596890942213204e-05, "loss": 0.0102, "step": 139980 }, { "epoch": 89.22243467176546, "grad_norm": 0.8327314853668213, "learning_rate": 5.959148179391859e-05, "loss": 0.0126, "step": 139990 }, { "epoch": 89.22880815806246, "grad_norm": 0.3748314678668976, "learning_rate": 5.958607252907842e-05, "loss": 0.0115, "step": 140000 }, { "epoch": 89.23518164435947, "grad_norm": 0.5831470489501953, "learning_rate": 5.958066314775842e-05, "loss": 0.0219, "step": 140010 }, { "epoch": 89.24155513065647, "grad_norm": 0.24455757439136505, "learning_rate": 5.957525365002431e-05, "loss": 0.0165, "step": 140020 }, { "epoch": 89.24792861695347, "grad_norm": 0.35460546612739563, "learning_rate": 5.956984403594184e-05, "loss": 0.014, "step": 140030 }, { "epoch": 89.25430210325048, "grad_norm": 0.3716607689857483, "learning_rate": 5.956443430557672e-05, "loss": 0.0112, "step": 140040 }, { "epoch": 89.26067558954749, "grad_norm": 0.40056151151657104, "learning_rate": 5.9559024458994695e-05, "loss": 0.012, "step": 140050 }, { "epoch": 89.26704907584448, "grad_norm": 0.509259819984436, "learning_rate": 5.95536144962615e-05, "loss": 0.0153, "step": 140060 }, { "epoch": 89.27342256214149, "grad_norm": 0.31718045473098755, "learning_rate": 5.954820441744285e-05, "loss": 0.0145, "step": 140070 }, { "epoch": 89.2797960484385, "grad_norm": 0.3457518517971039, "learning_rate": 5.9542794222604526e-05, "loss": 0.0167, "step": 140080 }, { "epoch": 89.2861695347355, "grad_norm": 0.2683109939098358, "learning_rate": 5.953738391181223e-05, "loss": 0.011, "step": 140090 }, { "epoch": 89.2925430210325, "grad_norm": 0.224969282746315, "learning_rate": 5.953197348513172e-05, "loss": 0.0133, "step": 140100 }, { "epoch": 89.29891650732951, "grad_norm": 0.3990812599658966, "learning_rate": 5.952656294262873e-05, "loss": 0.0179, "step": 140110 }, { "epoch": 89.30528999362652, "grad_norm": 0.2900805175304413, "learning_rate": 5.952115228436901e-05, "loss": 0.0115, "step": 140120 }, { "epoch": 89.31166347992352, "grad_norm": 0.34667378664016724, "learning_rate": 5.95157415104183e-05, "loss": 0.0181, "step": 140130 }, { "epoch": 89.31803696622052, "grad_norm": 0.2615636885166168, "learning_rate": 5.9510330620842333e-05, "loss": 0.013, "step": 140140 }, { "epoch": 89.32441045251753, "grad_norm": 0.20190392434597015, "learning_rate": 5.9504919615706886e-05, "loss": 0.018, "step": 140150 }, { "epoch": 89.33078393881453, "grad_norm": 0.3678366243839264, "learning_rate": 5.949950849507768e-05, "loss": 0.0226, "step": 140160 }, { "epoch": 89.33715742511154, "grad_norm": 0.3506247103214264, "learning_rate": 5.949409725902049e-05, "loss": 0.0138, "step": 140170 }, { "epoch": 89.34353091140854, "grad_norm": 0.231230229139328, "learning_rate": 5.948868590760104e-05, "loss": 0.0101, "step": 140180 }, { "epoch": 89.34990439770554, "grad_norm": 0.4461985230445862, "learning_rate": 5.9483274440885103e-05, "loss": 0.0149, "step": 140190 }, { "epoch": 89.35627788400255, "grad_norm": 0.6397733092308044, "learning_rate": 5.947786285893844e-05, "loss": 0.0125, "step": 140200 }, { "epoch": 89.36265137029956, "grad_norm": 0.37002941966056824, "learning_rate": 5.947245116182678e-05, "loss": 0.0231, "step": 140210 }, { "epoch": 89.36902485659655, "grad_norm": 0.5077957510948181, "learning_rate": 5.9467039349615903e-05, "loss": 0.0202, "step": 140220 }, { "epoch": 89.37539834289356, "grad_norm": 0.5691283345222473, "learning_rate": 5.946162742237154e-05, "loss": 0.0241, "step": 140230 }, { "epoch": 89.38177182919057, "grad_norm": 0.3201838433742523, "learning_rate": 5.9456215380159496e-05, "loss": 0.0147, "step": 140240 }, { "epoch": 89.38814531548758, "grad_norm": 0.41356635093688965, "learning_rate": 5.945080322304549e-05, "loss": 0.0116, "step": 140250 }, { "epoch": 89.39451880178457, "grad_norm": 0.2714832127094269, "learning_rate": 5.944539095109531e-05, "loss": 0.0204, "step": 140260 }, { "epoch": 89.40089228808158, "grad_norm": 0.16562718152999878, "learning_rate": 5.943997856437471e-05, "loss": 0.0166, "step": 140270 }, { "epoch": 89.40726577437859, "grad_norm": 0.2938501536846161, "learning_rate": 5.943456606294947e-05, "loss": 0.0131, "step": 140280 }, { "epoch": 89.4136392606756, "grad_norm": 0.32314610481262207, "learning_rate": 5.9429153446885335e-05, "loss": 0.0118, "step": 140290 }, { "epoch": 89.42001274697259, "grad_norm": 0.198868066072464, "learning_rate": 5.942374071624808e-05, "loss": 0.0217, "step": 140300 }, { "epoch": 89.4263862332696, "grad_norm": 0.44732466340065, "learning_rate": 5.9418327871103476e-05, "loss": 0.0214, "step": 140310 }, { "epoch": 89.4327597195666, "grad_norm": 0.3951016366481781, "learning_rate": 5.9412914911517294e-05, "loss": 0.014, "step": 140320 }, { "epoch": 89.43913320586361, "grad_norm": 0.20579995214939117, "learning_rate": 5.940750183755531e-05, "loss": 0.0139, "step": 140330 }, { "epoch": 89.4455066921606, "grad_norm": 0.26660069823265076, "learning_rate": 5.9402088649283306e-05, "loss": 0.0167, "step": 140340 }, { "epoch": 89.45188017845761, "grad_norm": 0.28563156723976135, "learning_rate": 5.939667534676704e-05, "loss": 0.0179, "step": 140350 }, { "epoch": 89.45825366475462, "grad_norm": 0.2541399598121643, "learning_rate": 5.93912619300723e-05, "loss": 0.0168, "step": 140360 }, { "epoch": 89.46462715105163, "grad_norm": 0.22344167530536652, "learning_rate": 5.938584839926485e-05, "loss": 0.0109, "step": 140370 }, { "epoch": 89.47100063734862, "grad_norm": 0.3174265921115875, "learning_rate": 5.938043475441049e-05, "loss": 0.0154, "step": 140380 }, { "epoch": 89.47737412364563, "grad_norm": 0.45084941387176514, "learning_rate": 5.9375020995574996e-05, "loss": 0.0151, "step": 140390 }, { "epoch": 89.48374760994264, "grad_norm": 0.2869286835193634, "learning_rate": 5.9369607122824135e-05, "loss": 0.012, "step": 140400 }, { "epoch": 89.49012109623965, "grad_norm": 0.29156023263931274, "learning_rate": 5.936419313622371e-05, "loss": 0.0122, "step": 140410 }, { "epoch": 89.49649458253664, "grad_norm": 0.45514580607414246, "learning_rate": 5.935877903583948e-05, "loss": 0.0196, "step": 140420 }, { "epoch": 89.50286806883365, "grad_norm": 0.4401954710483551, "learning_rate": 5.9353364821737265e-05, "loss": 0.0195, "step": 140430 }, { "epoch": 89.50924155513066, "grad_norm": 0.3170454800128937, "learning_rate": 5.934795049398283e-05, "loss": 0.0157, "step": 140440 }, { "epoch": 89.51561504142767, "grad_norm": 0.26170599460601807, "learning_rate": 5.934253605264198e-05, "loss": 0.0148, "step": 140450 }, { "epoch": 89.52198852772466, "grad_norm": 0.27041780948638916, "learning_rate": 5.9337121497780476e-05, "loss": 0.0235, "step": 140460 }, { "epoch": 89.52836201402167, "grad_norm": 0.5133333206176758, "learning_rate": 5.933170682946416e-05, "loss": 0.0209, "step": 140470 }, { "epoch": 89.53473550031867, "grad_norm": 0.5285617709159851, "learning_rate": 5.9326292047758766e-05, "loss": 0.0123, "step": 140480 }, { "epoch": 89.54110898661568, "grad_norm": 0.3402102589607239, "learning_rate": 5.932087715273014e-05, "loss": 0.0113, "step": 140490 }, { "epoch": 89.54748247291268, "grad_norm": 0.44394609332084656, "learning_rate": 5.931546214444405e-05, "loss": 0.0173, "step": 140500 }, { "epoch": 89.55385595920968, "grad_norm": 0.22288595139980316, "learning_rate": 5.93100470229663e-05, "loss": 0.0172, "step": 140510 }, { "epoch": 89.56022944550669, "grad_norm": 0.48308584094047546, "learning_rate": 5.930463178836269e-05, "loss": 0.0159, "step": 140520 }, { "epoch": 89.5666029318037, "grad_norm": 0.24752068519592285, "learning_rate": 5.9299216440699005e-05, "loss": 0.0288, "step": 140530 }, { "epoch": 89.5729764181007, "grad_norm": 0.29382896423339844, "learning_rate": 5.929380098004107e-05, "loss": 0.0117, "step": 140540 }, { "epoch": 89.5793499043977, "grad_norm": 0.5225654244422913, "learning_rate": 5.9288385406454674e-05, "loss": 0.0122, "step": 140550 }, { "epoch": 89.58572339069471, "grad_norm": 0.30615851283073425, "learning_rate": 5.9282969720005635e-05, "loss": 0.0072, "step": 140560 }, { "epoch": 89.59209687699172, "grad_norm": 0.5407669544219971, "learning_rate": 5.927755392075973e-05, "loss": 0.0253, "step": 140570 }, { "epoch": 89.59847036328873, "grad_norm": 0.31899207830429077, "learning_rate": 5.9272138008782805e-05, "loss": 0.0171, "step": 140580 }, { "epoch": 89.60484384958572, "grad_norm": 0.22335997223854065, "learning_rate": 5.9266721984140636e-05, "loss": 0.0118, "step": 140590 }, { "epoch": 89.61121733588273, "grad_norm": 0.17497819662094116, "learning_rate": 5.9261305846899063e-05, "loss": 0.0185, "step": 140600 }, { "epoch": 89.61759082217974, "grad_norm": 0.19810721278190613, "learning_rate": 5.925588959712387e-05, "loss": 0.0147, "step": 140610 }, { "epoch": 89.62396430847674, "grad_norm": 0.5289632678031921, "learning_rate": 5.925047323488088e-05, "loss": 0.0228, "step": 140620 }, { "epoch": 89.63033779477374, "grad_norm": 0.3366800844669342, "learning_rate": 5.924505676023591e-05, "loss": 0.0121, "step": 140630 }, { "epoch": 89.63671128107075, "grad_norm": 0.3393937647342682, "learning_rate": 5.9239640173254766e-05, "loss": 0.0084, "step": 140640 }, { "epoch": 89.64308476736775, "grad_norm": 0.4404216706752777, "learning_rate": 5.923422347400327e-05, "loss": 0.014, "step": 140650 }, { "epoch": 89.64945825366476, "grad_norm": 0.4747549891471863, "learning_rate": 5.9228806662547234e-05, "loss": 0.0126, "step": 140660 }, { "epoch": 89.65583173996176, "grad_norm": 0.5650514960289001, "learning_rate": 5.9223389738952495e-05, "loss": 0.0171, "step": 140670 }, { "epoch": 89.66220522625876, "grad_norm": 0.47267434000968933, "learning_rate": 5.921797270328486e-05, "loss": 0.0123, "step": 140680 }, { "epoch": 89.66857871255577, "grad_norm": 0.4096846878528595, "learning_rate": 5.921255555561015e-05, "loss": 0.0228, "step": 140690 }, { "epoch": 89.67495219885278, "grad_norm": 0.6346778869628906, "learning_rate": 5.92071382959942e-05, "loss": 0.018, "step": 140700 }, { "epoch": 89.68132568514977, "grad_norm": 0.42007216811180115, "learning_rate": 5.920172092450281e-05, "loss": 0.0164, "step": 140710 }, { "epoch": 89.68769917144678, "grad_norm": 0.2826941907405853, "learning_rate": 5.919630344120184e-05, "loss": 0.0145, "step": 140720 }, { "epoch": 89.69407265774379, "grad_norm": 0.33132630586624146, "learning_rate": 5.919088584615708e-05, "loss": 0.0084, "step": 140730 }, { "epoch": 89.7004461440408, "grad_norm": 0.344742089509964, "learning_rate": 5.918546813943441e-05, "loss": 0.01, "step": 140740 }, { "epoch": 89.70681963033779, "grad_norm": 0.19326291978359222, "learning_rate": 5.91800503210996e-05, "loss": 0.0108, "step": 140750 }, { "epoch": 89.7131931166348, "grad_norm": 0.5279132723808289, "learning_rate": 5.917463239121853e-05, "loss": 0.0292, "step": 140760 }, { "epoch": 89.7195666029318, "grad_norm": 0.1725301593542099, "learning_rate": 5.916921434985702e-05, "loss": 0.012, "step": 140770 }, { "epoch": 89.72594008922881, "grad_norm": 0.2936982214450836, "learning_rate": 5.916379619708088e-05, "loss": 0.0092, "step": 140780 }, { "epoch": 89.73231357552581, "grad_norm": 0.2299092561006546, "learning_rate": 5.915837793295598e-05, "loss": 0.0318, "step": 140790 }, { "epoch": 89.73868706182282, "grad_norm": 0.3931572139263153, "learning_rate": 5.9152959557548117e-05, "loss": 0.0182, "step": 140800 }, { "epoch": 89.74506054811982, "grad_norm": 0.26847386360168457, "learning_rate": 5.9147541070923174e-05, "loss": 0.0095, "step": 140810 }, { "epoch": 89.75143403441683, "grad_norm": 0.5482473373413086, "learning_rate": 5.914212247314696e-05, "loss": 0.0208, "step": 140820 }, { "epoch": 89.75780752071383, "grad_norm": 0.39805102348327637, "learning_rate": 5.913670376428533e-05, "loss": 0.0146, "step": 140830 }, { "epoch": 89.76418100701083, "grad_norm": 0.570382297039032, "learning_rate": 5.913128494440413e-05, "loss": 0.0158, "step": 140840 }, { "epoch": 89.77055449330784, "grad_norm": 0.4761384427547455, "learning_rate": 5.9125866013569186e-05, "loss": 0.0184, "step": 140850 }, { "epoch": 89.77692797960485, "grad_norm": 0.5653547048568726, "learning_rate": 5.9120446971846366e-05, "loss": 0.0154, "step": 140860 }, { "epoch": 89.78330146590184, "grad_norm": 0.2734260857105255, "learning_rate": 5.911502781930148e-05, "loss": 0.0101, "step": 140870 }, { "epoch": 89.78967495219885, "grad_norm": 0.3417934775352478, "learning_rate": 5.910960855600043e-05, "loss": 0.0121, "step": 140880 }, { "epoch": 89.79604843849586, "grad_norm": 0.2784649729728699, "learning_rate": 5.9104189182009016e-05, "loss": 0.0117, "step": 140890 }, { "epoch": 89.80242192479287, "grad_norm": 0.2368330955505371, "learning_rate": 5.909876969739312e-05, "loss": 0.013, "step": 140900 }, { "epoch": 89.80879541108986, "grad_norm": 0.4651283025741577, "learning_rate": 5.909335010221857e-05, "loss": 0.0124, "step": 140910 }, { "epoch": 89.81516889738687, "grad_norm": 0.3086391091346741, "learning_rate": 5.9087930396551237e-05, "loss": 0.0108, "step": 140920 }, { "epoch": 89.82154238368388, "grad_norm": 0.329018771648407, "learning_rate": 5.908251058045697e-05, "loss": 0.0133, "step": 140930 }, { "epoch": 89.82791586998088, "grad_norm": 0.3858122229576111, "learning_rate": 5.907709065400162e-05, "loss": 0.0124, "step": 140940 }, { "epoch": 89.83428935627788, "grad_norm": 0.25778019428253174, "learning_rate": 5.907167061725105e-05, "loss": 0.0118, "step": 140950 }, { "epoch": 89.84066284257489, "grad_norm": 0.424681693315506, "learning_rate": 5.906625047027111e-05, "loss": 0.0137, "step": 140960 }, { "epoch": 89.8470363288719, "grad_norm": 0.15689992904663086, "learning_rate": 5.906083021312768e-05, "loss": 0.0093, "step": 140970 }, { "epoch": 89.8534098151689, "grad_norm": 0.5167582631111145, "learning_rate": 5.90554098458866e-05, "loss": 0.0138, "step": 140980 }, { "epoch": 89.8597833014659, "grad_norm": 0.3342542350292206, "learning_rate": 5.9049989368613745e-05, "loss": 0.0107, "step": 140990 }, { "epoch": 89.8661567877629, "grad_norm": 0.5277170538902283, "learning_rate": 5.904456878137498e-05, "loss": 0.0122, "step": 141000 }, { "epoch": 89.87253027405991, "grad_norm": 0.26743364334106445, "learning_rate": 5.903914808423614e-05, "loss": 0.0195, "step": 141010 }, { "epoch": 89.87890376035692, "grad_norm": 0.24508143961429596, "learning_rate": 5.903372727726314e-05, "loss": 0.0133, "step": 141020 }, { "epoch": 89.88527724665391, "grad_norm": 0.2685932219028473, "learning_rate": 5.9028306360521815e-05, "loss": 0.0211, "step": 141030 }, { "epoch": 89.89165073295092, "grad_norm": 0.161499485373497, "learning_rate": 5.902288533407805e-05, "loss": 0.0149, "step": 141040 }, { "epoch": 89.89802421924793, "grad_norm": 0.43055227398872375, "learning_rate": 5.9017464197997695e-05, "loss": 0.0129, "step": 141050 }, { "epoch": 89.90439770554494, "grad_norm": 0.705884575843811, "learning_rate": 5.901204295234666e-05, "loss": 0.0156, "step": 141060 }, { "epoch": 89.91077119184193, "grad_norm": 0.3250654637813568, "learning_rate": 5.900662159719077e-05, "loss": 0.0218, "step": 141070 }, { "epoch": 89.91714467813894, "grad_norm": 0.25999534130096436, "learning_rate": 5.9001200132595934e-05, "loss": 0.0188, "step": 141080 }, { "epoch": 89.92351816443595, "grad_norm": 0.4639572203159332, "learning_rate": 5.899577855862802e-05, "loss": 0.0185, "step": 141090 }, { "epoch": 89.92989165073296, "grad_norm": 0.27911245822906494, "learning_rate": 5.89903568753529e-05, "loss": 0.0145, "step": 141100 }, { "epoch": 89.93626513702995, "grad_norm": 0.46071237325668335, "learning_rate": 5.8984935082836455e-05, "loss": 0.013, "step": 141110 }, { "epoch": 89.94263862332696, "grad_norm": 0.2991214394569397, "learning_rate": 5.897951318114458e-05, "loss": 0.018, "step": 141120 }, { "epoch": 89.94901210962396, "grad_norm": 0.22761650383472443, "learning_rate": 5.8974091170343114e-05, "loss": 0.0096, "step": 141130 }, { "epoch": 89.95538559592097, "grad_norm": 0.24365517497062683, "learning_rate": 5.8968669050497995e-05, "loss": 0.0121, "step": 141140 }, { "epoch": 89.96175908221797, "grad_norm": 0.2850628197193146, "learning_rate": 5.8963246821675055e-05, "loss": 0.0223, "step": 141150 }, { "epoch": 89.96813256851497, "grad_norm": 0.5716148614883423, "learning_rate": 5.8957824483940225e-05, "loss": 0.0203, "step": 141160 }, { "epoch": 89.97450605481198, "grad_norm": 0.3009267747402191, "learning_rate": 5.895240203735936e-05, "loss": 0.0127, "step": 141170 }, { "epoch": 89.98087954110899, "grad_norm": 0.7312164306640625, "learning_rate": 5.894697948199835e-05, "loss": 0.0154, "step": 141180 }, { "epoch": 89.98725302740598, "grad_norm": 0.5479052662849426, "learning_rate": 5.8941556817923106e-05, "loss": 0.0203, "step": 141190 }, { "epoch": 89.99362651370299, "grad_norm": 0.28416159749031067, "learning_rate": 5.89361340451995e-05, "loss": 0.0177, "step": 141200 }, { "epoch": 90.0, "grad_norm": 0.8861802220344543, "learning_rate": 5.893071116389341e-05, "loss": 0.0233, "step": 141210 }, { "epoch": 90.00637348629701, "grad_norm": 0.4462067782878876, "learning_rate": 5.8925288174070783e-05, "loss": 0.0195, "step": 141220 }, { "epoch": 90.01274697259402, "grad_norm": 0.18536436557769775, "learning_rate": 5.891986507579745e-05, "loss": 0.0129, "step": 141230 }, { "epoch": 90.01912045889101, "grad_norm": 0.6052350997924805, "learning_rate": 5.891444186913935e-05, "loss": 0.0247, "step": 141240 }, { "epoch": 90.02549394518802, "grad_norm": 0.21407094597816467, "learning_rate": 5.890901855416235e-05, "loss": 0.015, "step": 141250 }, { "epoch": 90.03186743148503, "grad_norm": 0.3702368140220642, "learning_rate": 5.890359513093238e-05, "loss": 0.0085, "step": 141260 }, { "epoch": 90.03824091778203, "grad_norm": 0.3234679698944092, "learning_rate": 5.8898171599515316e-05, "loss": 0.0189, "step": 141270 }, { "epoch": 90.04461440407903, "grad_norm": 0.42784059047698975, "learning_rate": 5.889274795997706e-05, "loss": 0.0213, "step": 141280 }, { "epoch": 90.05098789037604, "grad_norm": 0.25479423999786377, "learning_rate": 5.8887324212383523e-05, "loss": 0.0136, "step": 141290 }, { "epoch": 90.05736137667304, "grad_norm": 0.37001538276672363, "learning_rate": 5.8881900356800604e-05, "loss": 0.0085, "step": 141300 }, { "epoch": 90.06373486297005, "grad_norm": 0.19190993905067444, "learning_rate": 5.887647639329421e-05, "loss": 0.0107, "step": 141310 }, { "epoch": 90.07010834926704, "grad_norm": 0.18902304768562317, "learning_rate": 5.8871052321930255e-05, "loss": 0.0192, "step": 141320 }, { "epoch": 90.07648183556405, "grad_norm": 0.17648647725582123, "learning_rate": 5.886562814277463e-05, "loss": 0.0146, "step": 141330 }, { "epoch": 90.08285532186106, "grad_norm": 0.3028877377510071, "learning_rate": 5.886020385589326e-05, "loss": 0.0105, "step": 141340 }, { "epoch": 90.08922880815807, "grad_norm": 0.18507160246372223, "learning_rate": 5.885477946135204e-05, "loss": 0.0111, "step": 141350 }, { "epoch": 90.09560229445506, "grad_norm": 0.34752485156059265, "learning_rate": 5.88493549592169e-05, "loss": 0.0107, "step": 141360 }, { "epoch": 90.10197578075207, "grad_norm": 0.2544264793395996, "learning_rate": 5.8843930349553736e-05, "loss": 0.0131, "step": 141370 }, { "epoch": 90.10834926704908, "grad_norm": 0.2853991389274597, "learning_rate": 5.8838505632428466e-05, "loss": 0.0117, "step": 141380 }, { "epoch": 90.11472275334609, "grad_norm": 0.3517368733882904, "learning_rate": 5.8833080807907e-05, "loss": 0.0124, "step": 141390 }, { "epoch": 90.12109623964308, "grad_norm": 0.25236478447914124, "learning_rate": 5.8827655876055274e-05, "loss": 0.0134, "step": 141400 }, { "epoch": 90.12746972594009, "grad_norm": 0.35362616181373596, "learning_rate": 5.88222308369392e-05, "loss": 0.0092, "step": 141410 }, { "epoch": 90.1338432122371, "grad_norm": 0.4418635666370392, "learning_rate": 5.881680569062468e-05, "loss": 0.0159, "step": 141420 }, { "epoch": 90.1402166985341, "grad_norm": 0.46173015236854553, "learning_rate": 5.8811380437177654e-05, "loss": 0.0097, "step": 141430 }, { "epoch": 90.1465901848311, "grad_norm": 0.2991622984409332, "learning_rate": 5.880595507666402e-05, "loss": 0.011, "step": 141440 }, { "epoch": 90.1529636711281, "grad_norm": 0.40630093216896057, "learning_rate": 5.8800529609149734e-05, "loss": 0.018, "step": 141450 }, { "epoch": 90.15933715742511, "grad_norm": 0.2692103981971741, "learning_rate": 5.87951040347007e-05, "loss": 0.0093, "step": 141460 }, { "epoch": 90.16571064372212, "grad_norm": 0.38832351565361023, "learning_rate": 5.878967835338285e-05, "loss": 0.0189, "step": 141470 }, { "epoch": 90.17208413001912, "grad_norm": 0.31903260946273804, "learning_rate": 5.87842525652621e-05, "loss": 0.0178, "step": 141480 }, { "epoch": 90.17845761631612, "grad_norm": 0.6249162554740906, "learning_rate": 5.87788266704044e-05, "loss": 0.0155, "step": 141490 }, { "epoch": 90.18483110261313, "grad_norm": 0.5182027816772461, "learning_rate": 5.877340066887567e-05, "loss": 0.0284, "step": 141500 }, { "epoch": 90.19120458891014, "grad_norm": 0.5540375113487244, "learning_rate": 5.876797456074182e-05, "loss": 0.0228, "step": 141510 }, { "epoch": 90.19757807520713, "grad_norm": 0.40690553188323975, "learning_rate": 5.8762548346068826e-05, "loss": 0.0124, "step": 141520 }, { "epoch": 90.20395156150414, "grad_norm": 0.6415233016014099, "learning_rate": 5.8757122024922576e-05, "loss": 0.0241, "step": 141530 }, { "epoch": 90.21032504780115, "grad_norm": 0.46244391798973083, "learning_rate": 5.8751695597369025e-05, "loss": 0.0122, "step": 141540 }, { "epoch": 90.21669853409816, "grad_norm": 0.2667366862297058, "learning_rate": 5.8746269063474114e-05, "loss": 0.0164, "step": 141550 }, { "epoch": 90.22307202039515, "grad_norm": 0.3400197923183441, "learning_rate": 5.8740842423303773e-05, "loss": 0.0093, "step": 141560 }, { "epoch": 90.22944550669216, "grad_norm": 0.32450181245803833, "learning_rate": 5.8735415676923965e-05, "loss": 0.0073, "step": 141570 }, { "epoch": 90.23581899298917, "grad_norm": 0.608775794506073, "learning_rate": 5.872998882440058e-05, "loss": 0.0257, "step": 141580 }, { "epoch": 90.24219247928617, "grad_norm": 0.6961585283279419, "learning_rate": 5.872456186579961e-05, "loss": 0.0108, "step": 141590 }, { "epoch": 90.24856596558317, "grad_norm": 0.39716508984565735, "learning_rate": 5.871913480118696e-05, "loss": 0.0148, "step": 141600 }, { "epoch": 90.25493945188018, "grad_norm": 0.4208233952522278, "learning_rate": 5.87137076306286e-05, "loss": 0.0099, "step": 141610 }, { "epoch": 90.26131293817718, "grad_norm": 0.2669081687927246, "learning_rate": 5.870828035419046e-05, "loss": 0.0096, "step": 141620 }, { "epoch": 90.26768642447419, "grad_norm": 0.34281426668167114, "learning_rate": 5.870285297193849e-05, "loss": 0.0095, "step": 141630 }, { "epoch": 90.27405991077119, "grad_norm": 0.6416864395141602, "learning_rate": 5.869742548393864e-05, "loss": 0.0177, "step": 141640 }, { "epoch": 90.2804333970682, "grad_norm": 0.3858131468296051, "learning_rate": 5.869199789025687e-05, "loss": 0.0127, "step": 141650 }, { "epoch": 90.2868068833652, "grad_norm": 0.2078344076871872, "learning_rate": 5.868657019095911e-05, "loss": 0.0087, "step": 141660 }, { "epoch": 90.29318036966221, "grad_norm": 0.2613237798213959, "learning_rate": 5.868114238611132e-05, "loss": 0.0089, "step": 141670 }, { "epoch": 90.2995538559592, "grad_norm": 0.31048911809921265, "learning_rate": 5.867571447577945e-05, "loss": 0.0115, "step": 141680 }, { "epoch": 90.30592734225621, "grad_norm": 0.5604158639907837, "learning_rate": 5.867028646002946e-05, "loss": 0.0135, "step": 141690 }, { "epoch": 90.31230082855322, "grad_norm": 0.3619856834411621, "learning_rate": 5.866485833892732e-05, "loss": 0.0168, "step": 141700 }, { "epoch": 90.31867431485023, "grad_norm": 0.26110339164733887, "learning_rate": 5.865943011253895e-05, "loss": 0.0099, "step": 141710 }, { "epoch": 90.32504780114722, "grad_norm": 0.601672887802124, "learning_rate": 5.8654001780930334e-05, "loss": 0.0083, "step": 141720 }, { "epoch": 90.33142128744423, "grad_norm": 0.4413936734199524, "learning_rate": 5.8648573344167436e-05, "loss": 0.0114, "step": 141730 }, { "epoch": 90.33779477374124, "grad_norm": 0.22006887197494507, "learning_rate": 5.864314480231619e-05, "loss": 0.0109, "step": 141740 }, { "epoch": 90.34416826003824, "grad_norm": 0.41744983196258545, "learning_rate": 5.863771615544258e-05, "loss": 0.0163, "step": 141750 }, { "epoch": 90.35054174633524, "grad_norm": 0.34062570333480835, "learning_rate": 5.8632287403612564e-05, "loss": 0.0155, "step": 141760 }, { "epoch": 90.35691523263225, "grad_norm": 0.5805906653404236, "learning_rate": 5.8626858546892106e-05, "loss": 0.013, "step": 141770 }, { "epoch": 90.36328871892925, "grad_norm": 0.5107944011688232, "learning_rate": 5.862142958534717e-05, "loss": 0.0176, "step": 141780 }, { "epoch": 90.36966220522626, "grad_norm": 0.5608083605766296, "learning_rate": 5.861600051904372e-05, "loss": 0.0163, "step": 141790 }, { "epoch": 90.37603569152326, "grad_norm": 0.2686828374862671, "learning_rate": 5.861057134804774e-05, "loss": 0.0096, "step": 141800 }, { "epoch": 90.38240917782026, "grad_norm": 0.3189029395580292, "learning_rate": 5.860514207242518e-05, "loss": 0.0163, "step": 141810 }, { "epoch": 90.38878266411727, "grad_norm": 0.28966790437698364, "learning_rate": 5.859971269224203e-05, "loss": 0.015, "step": 141820 }, { "epoch": 90.39515615041428, "grad_norm": 0.2575758397579193, "learning_rate": 5.859428320756424e-05, "loss": 0.0212, "step": 141830 }, { "epoch": 90.40152963671127, "grad_norm": 0.24366039037704468, "learning_rate": 5.8588853618457795e-05, "loss": 0.0108, "step": 141840 }, { "epoch": 90.40790312300828, "grad_norm": 0.4146634638309479, "learning_rate": 5.858342392498868e-05, "loss": 0.0174, "step": 141850 }, { "epoch": 90.41427660930529, "grad_norm": 0.33235785365104675, "learning_rate": 5.857799412722285e-05, "loss": 0.0099, "step": 141860 }, { "epoch": 90.4206500956023, "grad_norm": 0.5032908916473389, "learning_rate": 5.857256422522629e-05, "loss": 0.018, "step": 141870 }, { "epoch": 90.4270235818993, "grad_norm": 0.39025989174842834, "learning_rate": 5.856713421906498e-05, "loss": 0.0114, "step": 141880 }, { "epoch": 90.4333970681963, "grad_norm": 0.5562689900398254, "learning_rate": 5.8561704108804916e-05, "loss": 0.0109, "step": 141890 }, { "epoch": 90.43977055449331, "grad_norm": 0.9375810027122498, "learning_rate": 5.855627389451205e-05, "loss": 0.0192, "step": 141900 }, { "epoch": 90.44614404079032, "grad_norm": 0.335371732711792, "learning_rate": 5.855084357625238e-05, "loss": 0.014, "step": 141910 }, { "epoch": 90.45251752708732, "grad_norm": 0.39729705452919006, "learning_rate": 5.854541315409189e-05, "loss": 0.0101, "step": 141920 }, { "epoch": 90.45889101338432, "grad_norm": 0.3047296404838562, "learning_rate": 5.8539982628096556e-05, "loss": 0.0126, "step": 141930 }, { "epoch": 90.46526449968133, "grad_norm": 0.21051102876663208, "learning_rate": 5.853455199833238e-05, "loss": 0.0158, "step": 141940 }, { "epoch": 90.47163798597833, "grad_norm": 0.3057529926300049, "learning_rate": 5.852912126486534e-05, "loss": 0.0098, "step": 141950 }, { "epoch": 90.47801147227534, "grad_norm": 0.3566526770591736, "learning_rate": 5.85236904277614e-05, "loss": 0.0123, "step": 141960 }, { "epoch": 90.48438495857233, "grad_norm": 0.28281307220458984, "learning_rate": 5.85182594870866e-05, "loss": 0.0078, "step": 141970 }, { "epoch": 90.49075844486934, "grad_norm": 0.25493302941322327, "learning_rate": 5.851282844290691e-05, "loss": 0.0213, "step": 141980 }, { "epoch": 90.49713193116635, "grad_norm": 0.2636866867542267, "learning_rate": 5.850739729528829e-05, "loss": 0.0173, "step": 141990 }, { "epoch": 90.50350541746336, "grad_norm": 0.13784508407115936, "learning_rate": 5.8501966044296784e-05, "loss": 0.02, "step": 142000 }, { "epoch": 90.50987890376035, "grad_norm": 0.18250297009944916, "learning_rate": 5.849653468999834e-05, "loss": 0.0107, "step": 142010 }, { "epoch": 90.51625239005736, "grad_norm": 0.4454967677593231, "learning_rate": 5.8491103232459e-05, "loss": 0.0146, "step": 142020 }, { "epoch": 90.52262587635437, "grad_norm": 0.511841356754303, "learning_rate": 5.8485671671744726e-05, "loss": 0.0169, "step": 142030 }, { "epoch": 90.52899936265138, "grad_norm": 0.36742061376571655, "learning_rate": 5.848024000792154e-05, "loss": 0.0097, "step": 142040 }, { "epoch": 90.53537284894837, "grad_norm": 0.45918187499046326, "learning_rate": 5.8474808241055426e-05, "loss": 0.0194, "step": 142050 }, { "epoch": 90.54174633524538, "grad_norm": 0.32693660259246826, "learning_rate": 5.846937637121238e-05, "loss": 0.0141, "step": 142060 }, { "epoch": 90.54811982154239, "grad_norm": 0.22170774638652802, "learning_rate": 5.846394439845843e-05, "loss": 0.0139, "step": 142070 }, { "epoch": 90.5544933078394, "grad_norm": 0.23606182634830475, "learning_rate": 5.8458512322859546e-05, "loss": 0.0106, "step": 142080 }, { "epoch": 90.56086679413639, "grad_norm": 0.3069648742675781, "learning_rate": 5.8453080144481774e-05, "loss": 0.0072, "step": 142090 }, { "epoch": 90.5672402804334, "grad_norm": 0.1525389552116394, "learning_rate": 5.844764786339107e-05, "loss": 0.0118, "step": 142100 }, { "epoch": 90.5736137667304, "grad_norm": 0.22297807037830353, "learning_rate": 5.8442215479653496e-05, "loss": 0.0112, "step": 142110 }, { "epoch": 90.57998725302741, "grad_norm": 0.6187921762466431, "learning_rate": 5.843678299333501e-05, "loss": 0.0136, "step": 142120 }, { "epoch": 90.5863607393244, "grad_norm": 0.46292364597320557, "learning_rate": 5.843135040450167e-05, "loss": 0.012, "step": 142130 }, { "epoch": 90.59273422562141, "grad_norm": 0.6930348873138428, "learning_rate": 5.842591771321945e-05, "loss": 0.0134, "step": 142140 }, { "epoch": 90.59910771191842, "grad_norm": 0.2729194760322571, "learning_rate": 5.842048491955436e-05, "loss": 0.0092, "step": 142150 }, { "epoch": 90.60548119821543, "grad_norm": 0.5713133811950684, "learning_rate": 5.841505202357245e-05, "loss": 0.0192, "step": 142160 }, { "epoch": 90.61185468451242, "grad_norm": 0.3286181092262268, "learning_rate": 5.84096190253397e-05, "loss": 0.0084, "step": 142170 }, { "epoch": 90.61822817080943, "grad_norm": 0.4007602035999298, "learning_rate": 5.840418592492215e-05, "loss": 0.0174, "step": 142180 }, { "epoch": 90.62460165710644, "grad_norm": 0.25987479090690613, "learning_rate": 5.83987527223858e-05, "loss": 0.0112, "step": 142190 }, { "epoch": 90.63097514340345, "grad_norm": 0.42849498987197876, "learning_rate": 5.8393319417796664e-05, "loss": 0.0111, "step": 142200 }, { "epoch": 90.63734862970044, "grad_norm": 0.41704168915748596, "learning_rate": 5.838788601122078e-05, "loss": 0.0122, "step": 142210 }, { "epoch": 90.64372211599745, "grad_norm": 0.2689794898033142, "learning_rate": 5.838245250272417e-05, "loss": 0.0143, "step": 142220 }, { "epoch": 90.65009560229446, "grad_norm": 0.2969506084918976, "learning_rate": 5.837701889237285e-05, "loss": 0.0143, "step": 142230 }, { "epoch": 90.65646908859146, "grad_norm": 0.23657868802547455, "learning_rate": 5.8371585180232823e-05, "loss": 0.014, "step": 142240 }, { "epoch": 90.66284257488846, "grad_norm": 0.36705294251441956, "learning_rate": 5.8366151366370145e-05, "loss": 0.0126, "step": 142250 }, { "epoch": 90.66921606118547, "grad_norm": 0.48882240056991577, "learning_rate": 5.836071745085082e-05, "loss": 0.0148, "step": 142260 }, { "epoch": 90.67558954748247, "grad_norm": 0.5141257047653198, "learning_rate": 5.8355283433740893e-05, "loss": 0.0154, "step": 142270 }, { "epoch": 90.68196303377948, "grad_norm": 0.3155660331249237, "learning_rate": 5.8349849315106376e-05, "loss": 0.0131, "step": 142280 }, { "epoch": 90.68833652007648, "grad_norm": 0.26177123188972473, "learning_rate": 5.8344415095013316e-05, "loss": 0.0108, "step": 142290 }, { "epoch": 90.69471000637348, "grad_norm": 0.17189887166023254, "learning_rate": 5.833898077352773e-05, "loss": 0.0093, "step": 142300 }, { "epoch": 90.70108349267049, "grad_norm": 0.1409689486026764, "learning_rate": 5.833354635071565e-05, "loss": 0.0205, "step": 142310 }, { "epoch": 90.7074569789675, "grad_norm": 0.23328454792499542, "learning_rate": 5.8328111826643114e-05, "loss": 0.0277, "step": 142320 }, { "epoch": 90.7138304652645, "grad_norm": 0.3241676986217499, "learning_rate": 5.832267720137616e-05, "loss": 0.0128, "step": 142330 }, { "epoch": 90.7202039515615, "grad_norm": 0.2147231549024582, "learning_rate": 5.8317242474980826e-05, "loss": 0.0117, "step": 142340 }, { "epoch": 90.72657743785851, "grad_norm": 0.2729504406452179, "learning_rate": 5.831180764752313e-05, "loss": 0.0195, "step": 142350 }, { "epoch": 90.73295092415552, "grad_norm": 0.2550337314605713, "learning_rate": 5.830637271906914e-05, "loss": 0.0126, "step": 142360 }, { "epoch": 90.73932441045251, "grad_norm": 0.3214569091796875, "learning_rate": 5.830093768968485e-05, "loss": 0.012, "step": 142370 }, { "epoch": 90.74569789674952, "grad_norm": 0.22962884604930878, "learning_rate": 5.829550255943636e-05, "loss": 0.0094, "step": 142380 }, { "epoch": 90.75207138304653, "grad_norm": 0.39214545488357544, "learning_rate": 5.829006732838966e-05, "loss": 0.0129, "step": 142390 }, { "epoch": 90.75844486934353, "grad_norm": 0.5053321123123169, "learning_rate": 5.8284631996610827e-05, "loss": 0.0144, "step": 142400 }, { "epoch": 90.76481835564053, "grad_norm": 0.45925453305244446, "learning_rate": 5.827919656416589e-05, "loss": 0.0238, "step": 142410 }, { "epoch": 90.77119184193754, "grad_norm": 0.5793464183807373, "learning_rate": 5.8273761031120896e-05, "loss": 0.015, "step": 142420 }, { "epoch": 90.77756532823454, "grad_norm": 0.37298378348350525, "learning_rate": 5.82683253975419e-05, "loss": 0.0117, "step": 142430 }, { "epoch": 90.78393881453155, "grad_norm": 0.2786058187484741, "learning_rate": 5.826288966349493e-05, "loss": 0.0106, "step": 142440 }, { "epoch": 90.79031230082855, "grad_norm": 0.21934247016906738, "learning_rate": 5.8257453829046074e-05, "loss": 0.011, "step": 142450 }, { "epoch": 90.79668578712555, "grad_norm": 0.39184579253196716, "learning_rate": 5.825201789426135e-05, "loss": 0.0191, "step": 142460 }, { "epoch": 90.80305927342256, "grad_norm": 0.20093168318271637, "learning_rate": 5.8246581859206796e-05, "loss": 0.0199, "step": 142470 }, { "epoch": 90.80943275971957, "grad_norm": 0.3172554671764374, "learning_rate": 5.8241145723948506e-05, "loss": 0.0226, "step": 142480 }, { "epoch": 90.81580624601658, "grad_norm": 0.41148945689201355, "learning_rate": 5.8235709488552504e-05, "loss": 0.0127, "step": 142490 }, { "epoch": 90.82217973231357, "grad_norm": 0.28104740381240845, "learning_rate": 5.823027315308486e-05, "loss": 0.0119, "step": 142500 }, { "epoch": 90.82855321861058, "grad_norm": 0.5401007533073425, "learning_rate": 5.822483671761163e-05, "loss": 0.0134, "step": 142510 }, { "epoch": 90.83492670490759, "grad_norm": 0.36608394980430603, "learning_rate": 5.821940018219886e-05, "loss": 0.0108, "step": 142520 }, { "epoch": 90.8413001912046, "grad_norm": 0.5117450952529907, "learning_rate": 5.8213963546912634e-05, "loss": 0.0162, "step": 142530 }, { "epoch": 90.84767367750159, "grad_norm": 0.4670466184616089, "learning_rate": 5.820852681181898e-05, "loss": 0.0133, "step": 142540 }, { "epoch": 90.8540471637986, "grad_norm": 0.3224651515483856, "learning_rate": 5.8203089976983985e-05, "loss": 0.0115, "step": 142550 }, { "epoch": 90.8604206500956, "grad_norm": 0.3221488296985626, "learning_rate": 5.8197653042473686e-05, "loss": 0.0115, "step": 142560 }, { "epoch": 90.86679413639261, "grad_norm": 0.33710774779319763, "learning_rate": 5.819221600835418e-05, "loss": 0.0137, "step": 142570 }, { "epoch": 90.8731676226896, "grad_norm": 0.40489786863327026, "learning_rate": 5.8186778874691504e-05, "loss": 0.0247, "step": 142580 }, { "epoch": 90.87954110898661, "grad_norm": 0.15585626661777496, "learning_rate": 5.8181341641551736e-05, "loss": 0.024, "step": 142590 }, { "epoch": 90.88591459528362, "grad_norm": 0.47592276334762573, "learning_rate": 5.817590430900094e-05, "loss": 0.016, "step": 142600 }, { "epoch": 90.89228808158063, "grad_norm": 0.27935731410980225, "learning_rate": 5.81704668771052e-05, "loss": 0.0124, "step": 142610 }, { "epoch": 90.89866156787762, "grad_norm": 0.3360534608364105, "learning_rate": 5.8165029345930575e-05, "loss": 0.0147, "step": 142620 }, { "epoch": 90.90503505417463, "grad_norm": 0.44825997948646545, "learning_rate": 5.815959171554312e-05, "loss": 0.0128, "step": 142630 }, { "epoch": 90.91140854047164, "grad_norm": 0.4470396041870117, "learning_rate": 5.815415398600893e-05, "loss": 0.0192, "step": 142640 }, { "epoch": 90.91778202676865, "grad_norm": 0.17037133872509003, "learning_rate": 5.814871615739407e-05, "loss": 0.0132, "step": 142650 }, { "epoch": 90.92415551306564, "grad_norm": 0.5636090636253357, "learning_rate": 5.814327822976461e-05, "loss": 0.0126, "step": 142660 }, { "epoch": 90.93052899936265, "grad_norm": 0.36405283212661743, "learning_rate": 5.8137840203186636e-05, "loss": 0.0096, "step": 142670 }, { "epoch": 90.93690248565966, "grad_norm": 0.2567422091960907, "learning_rate": 5.8132402077726215e-05, "loss": 0.0128, "step": 142680 }, { "epoch": 90.94327597195667, "grad_norm": 0.31094878911972046, "learning_rate": 5.8126963853449425e-05, "loss": 0.0123, "step": 142690 }, { "epoch": 90.94964945825366, "grad_norm": 0.4025634229183197, "learning_rate": 5.812152553042237e-05, "loss": 0.0199, "step": 142700 }, { "epoch": 90.95602294455067, "grad_norm": 0.5137103796005249, "learning_rate": 5.8116087108711104e-05, "loss": 0.0264, "step": 142710 }, { "epoch": 90.96239643084768, "grad_norm": 0.48208653926849365, "learning_rate": 5.811064858838171e-05, "loss": 0.0102, "step": 142720 }, { "epoch": 90.96876991714468, "grad_norm": 0.4193313419818878, "learning_rate": 5.810520996950028e-05, "loss": 0.0271, "step": 142730 }, { "epoch": 90.97514340344168, "grad_norm": 0.3766099512577057, "learning_rate": 5.809977125213289e-05, "loss": 0.0166, "step": 142740 }, { "epoch": 90.98151688973869, "grad_norm": 0.23898516595363617, "learning_rate": 5.809433243634565e-05, "loss": 0.0138, "step": 142750 }, { "epoch": 90.9878903760357, "grad_norm": 0.6055013537406921, "learning_rate": 5.8088893522204614e-05, "loss": 0.0221, "step": 142760 }, { "epoch": 90.9942638623327, "grad_norm": 0.3679284155368805, "learning_rate": 5.808345450977589e-05, "loss": 0.0161, "step": 142770 }, { "epoch": 91.0006373486297, "grad_norm": 0.35362154245376587, "learning_rate": 5.8078015399125573e-05, "loss": 0.0123, "step": 142780 }, { "epoch": 91.0070108349267, "grad_norm": 0.2732907235622406, "learning_rate": 5.8072576190319716e-05, "loss": 0.0122, "step": 142790 }, { "epoch": 91.01338432122371, "grad_norm": 0.5536474585533142, "learning_rate": 5.806713688342447e-05, "loss": 0.0158, "step": 142800 }, { "epoch": 91.01975780752072, "grad_norm": 0.6747725605964661, "learning_rate": 5.806169747850586e-05, "loss": 0.0151, "step": 142810 }, { "epoch": 91.02613129381771, "grad_norm": 0.5221205949783325, "learning_rate": 5.8056257975630036e-05, "loss": 0.0175, "step": 142820 }, { "epoch": 91.03250478011472, "grad_norm": 0.5564306378364563, "learning_rate": 5.8050818374863056e-05, "loss": 0.0253, "step": 142830 }, { "epoch": 91.03887826641173, "grad_norm": 0.36881595849990845, "learning_rate": 5.804537867627104e-05, "loss": 0.0124, "step": 142840 }, { "epoch": 91.04525175270874, "grad_norm": 0.16863568127155304, "learning_rate": 5.8039938879920076e-05, "loss": 0.0145, "step": 142850 }, { "epoch": 91.05162523900573, "grad_norm": 0.5783346891403198, "learning_rate": 5.803449898587627e-05, "loss": 0.0152, "step": 142860 }, { "epoch": 91.05799872530274, "grad_norm": 0.2659682333469391, "learning_rate": 5.802905899420571e-05, "loss": 0.013, "step": 142870 }, { "epoch": 91.06437221159975, "grad_norm": 0.4924461245536804, "learning_rate": 5.8023618904974494e-05, "loss": 0.0135, "step": 142880 }, { "epoch": 91.07074569789675, "grad_norm": 0.28429359197616577, "learning_rate": 5.801817871824875e-05, "loss": 0.0119, "step": 142890 }, { "epoch": 91.07711918419375, "grad_norm": 0.543844997882843, "learning_rate": 5.801273843409455e-05, "loss": 0.0135, "step": 142900 }, { "epoch": 91.08349267049076, "grad_norm": 0.4478318691253662, "learning_rate": 5.800729805257803e-05, "loss": 0.0173, "step": 142910 }, { "epoch": 91.08986615678776, "grad_norm": 0.33464542031288147, "learning_rate": 5.800185757376525e-05, "loss": 0.0135, "step": 142920 }, { "epoch": 91.09623964308477, "grad_norm": 0.24073347449302673, "learning_rate": 5.799641699772237e-05, "loss": 0.0148, "step": 142930 }, { "epoch": 91.10261312938177, "grad_norm": 0.5524978041648865, "learning_rate": 5.799097632451547e-05, "loss": 0.019, "step": 142940 }, { "epoch": 91.10898661567877, "grad_norm": 0.47264567017555237, "learning_rate": 5.798553555421066e-05, "loss": 0.0129, "step": 142950 }, { "epoch": 91.11536010197578, "grad_norm": 0.27267247438430786, "learning_rate": 5.7980094686874054e-05, "loss": 0.0156, "step": 142960 }, { "epoch": 91.12173358827279, "grad_norm": 0.40088367462158203, "learning_rate": 5.7974653722571746e-05, "loss": 0.0115, "step": 142970 }, { "epoch": 91.12810707456978, "grad_norm": 0.4748351275920868, "learning_rate": 5.7969212661369885e-05, "loss": 0.0135, "step": 142980 }, { "epoch": 91.13448056086679, "grad_norm": 0.3378336727619171, "learning_rate": 5.796377150333456e-05, "loss": 0.0087, "step": 142990 }, { "epoch": 91.1408540471638, "grad_norm": 0.27643996477127075, "learning_rate": 5.79583302485319e-05, "loss": 0.0117, "step": 143000 }, { "epoch": 91.1472275334608, "grad_norm": 0.2982424199581146, "learning_rate": 5.7952888897028e-05, "loss": 0.0078, "step": 143010 }, { "epoch": 91.1536010197578, "grad_norm": 0.14035022258758545, "learning_rate": 5.7947447448889e-05, "loss": 0.01, "step": 143020 }, { "epoch": 91.15997450605481, "grad_norm": 0.3021611273288727, "learning_rate": 5.7942005904181016e-05, "loss": 0.0193, "step": 143030 }, { "epoch": 91.16634799235182, "grad_norm": 0.3575431704521179, "learning_rate": 5.793656426297015e-05, "loss": 0.0106, "step": 143040 }, { "epoch": 91.17272147864882, "grad_norm": 0.8330222368240356, "learning_rate": 5.793112252532255e-05, "loss": 0.025, "step": 143050 }, { "epoch": 91.17909496494582, "grad_norm": 0.3998124897480011, "learning_rate": 5.7925680691304305e-05, "loss": 0.015, "step": 143060 }, { "epoch": 91.18546845124283, "grad_norm": 0.4637693464756012, "learning_rate": 5.792023876098157e-05, "loss": 0.014, "step": 143070 }, { "epoch": 91.19184193753983, "grad_norm": 0.2698320150375366, "learning_rate": 5.791479673442044e-05, "loss": 0.0073, "step": 143080 }, { "epoch": 91.19821542383684, "grad_norm": 0.3761422038078308, "learning_rate": 5.790935461168707e-05, "loss": 0.0212, "step": 143090 }, { "epoch": 91.20458891013385, "grad_norm": 0.40735912322998047, "learning_rate": 5.790391239284758e-05, "loss": 0.0108, "step": 143100 }, { "epoch": 91.21096239643084, "grad_norm": 0.43654683232307434, "learning_rate": 5.789847007796807e-05, "loss": 0.0228, "step": 143110 }, { "epoch": 91.21733588272785, "grad_norm": 0.22735877335071564, "learning_rate": 5.789302766711471e-05, "loss": 0.0099, "step": 143120 }, { "epoch": 91.22370936902486, "grad_norm": 0.3028746247291565, "learning_rate": 5.78875851603536e-05, "loss": 0.0129, "step": 143130 }, { "epoch": 91.23008285532187, "grad_norm": 0.7635648250579834, "learning_rate": 5.7882142557750895e-05, "loss": 0.0131, "step": 143140 }, { "epoch": 91.23645634161886, "grad_norm": 0.26345154643058777, "learning_rate": 5.7876699859372704e-05, "loss": 0.0102, "step": 143150 }, { "epoch": 91.24282982791587, "grad_norm": 0.531442403793335, "learning_rate": 5.787125706528518e-05, "loss": 0.0156, "step": 143160 }, { "epoch": 91.24920331421288, "grad_norm": 0.36445334553718567, "learning_rate": 5.7865814175554445e-05, "loss": 0.0099, "step": 143170 }, { "epoch": 91.25557680050989, "grad_norm": 0.3424485921859741, "learning_rate": 5.7860371190246656e-05, "loss": 0.0121, "step": 143180 }, { "epoch": 91.26195028680688, "grad_norm": 0.5669740438461304, "learning_rate": 5.785492810942792e-05, "loss": 0.0185, "step": 143190 }, { "epoch": 91.26832377310389, "grad_norm": 0.5091249942779541, "learning_rate": 5.78494849331644e-05, "loss": 0.0114, "step": 143200 }, { "epoch": 91.2746972594009, "grad_norm": 0.38268426060676575, "learning_rate": 5.784404166152222e-05, "loss": 0.0082, "step": 143210 }, { "epoch": 91.2810707456979, "grad_norm": 0.21912887692451477, "learning_rate": 5.7838598294567525e-05, "loss": 0.0207, "step": 143220 }, { "epoch": 91.2874442319949, "grad_norm": 0.47931501269340515, "learning_rate": 5.7833154832366465e-05, "loss": 0.0169, "step": 143230 }, { "epoch": 91.2938177182919, "grad_norm": 0.47381630539894104, "learning_rate": 5.7827711274985186e-05, "loss": 0.015, "step": 143240 }, { "epoch": 91.30019120458891, "grad_norm": 0.2243773490190506, "learning_rate": 5.782226762248981e-05, "loss": 0.0126, "step": 143250 }, { "epoch": 91.30656469088592, "grad_norm": 0.3762310743331909, "learning_rate": 5.781682387494651e-05, "loss": 0.0156, "step": 143260 }, { "epoch": 91.31293817718291, "grad_norm": 0.32886582612991333, "learning_rate": 5.7811380032421405e-05, "loss": 0.0171, "step": 143270 }, { "epoch": 91.31931166347992, "grad_norm": 0.3300354480743408, "learning_rate": 5.780593609498067e-05, "loss": 0.011, "step": 143280 }, { "epoch": 91.32568514977693, "grad_norm": 0.44899001717567444, "learning_rate": 5.7800492062690435e-05, "loss": 0.016, "step": 143290 }, { "epoch": 91.33205863607394, "grad_norm": 0.6902531385421753, "learning_rate": 5.7795047935616866e-05, "loss": 0.0218, "step": 143300 }, { "epoch": 91.33843212237093, "grad_norm": 0.39714664220809937, "learning_rate": 5.778960371382609e-05, "loss": 0.0129, "step": 143310 }, { "epoch": 91.34480560866794, "grad_norm": 0.5821247100830078, "learning_rate": 5.7784159397384283e-05, "loss": 0.0082, "step": 143320 }, { "epoch": 91.35117909496495, "grad_norm": 0.5112342834472656, "learning_rate": 5.777871498635758e-05, "loss": 0.0271, "step": 143330 }, { "epoch": 91.35755258126196, "grad_norm": 0.23704199492931366, "learning_rate": 5.777327048081217e-05, "loss": 0.0189, "step": 143340 }, { "epoch": 91.36392606755895, "grad_norm": 0.3783239722251892, "learning_rate": 5.7767825880814174e-05, "loss": 0.0175, "step": 143350 }, { "epoch": 91.37029955385596, "grad_norm": 0.294452041387558, "learning_rate": 5.7762381186429745e-05, "loss": 0.0219, "step": 143360 }, { "epoch": 91.37667304015297, "grad_norm": 0.3146178126335144, "learning_rate": 5.775693639772507e-05, "loss": 0.012, "step": 143370 }, { "epoch": 91.38304652644997, "grad_norm": 0.20632924139499664, "learning_rate": 5.775149151476629e-05, "loss": 0.0121, "step": 143380 }, { "epoch": 91.38942001274697, "grad_norm": 0.4169023036956787, "learning_rate": 5.774604653761957e-05, "loss": 0.014, "step": 143390 }, { "epoch": 91.39579349904398, "grad_norm": 0.32090768218040466, "learning_rate": 5.774060146635106e-05, "loss": 0.0291, "step": 143400 }, { "epoch": 91.40216698534098, "grad_norm": 0.26907211542129517, "learning_rate": 5.7735156301026946e-05, "loss": 0.0123, "step": 143410 }, { "epoch": 91.40854047163799, "grad_norm": 0.30807265639305115, "learning_rate": 5.772971104171337e-05, "loss": 0.026, "step": 143420 }, { "epoch": 91.41491395793498, "grad_norm": 0.1753760725259781, "learning_rate": 5.772426568847653e-05, "loss": 0.0114, "step": 143430 }, { "epoch": 91.421287444232, "grad_norm": 0.23186221718788147, "learning_rate": 5.771882024138255e-05, "loss": 0.0139, "step": 143440 }, { "epoch": 91.427660930529, "grad_norm": 0.23614908754825592, "learning_rate": 5.771337470049763e-05, "loss": 0.0118, "step": 143450 }, { "epoch": 91.43403441682601, "grad_norm": 0.5569922924041748, "learning_rate": 5.770792906588791e-05, "loss": 0.0165, "step": 143460 }, { "epoch": 91.440407903123, "grad_norm": 0.19641266763210297, "learning_rate": 5.770248333761957e-05, "loss": 0.0111, "step": 143470 }, { "epoch": 91.44678138942001, "grad_norm": 0.2690991461277008, "learning_rate": 5.76970375157588e-05, "loss": 0.0113, "step": 143480 }, { "epoch": 91.45315487571702, "grad_norm": 0.24708661437034607, "learning_rate": 5.769159160037174e-05, "loss": 0.0152, "step": 143490 }, { "epoch": 91.45952836201403, "grad_norm": 0.2555195093154907, "learning_rate": 5.76861455915246e-05, "loss": 0.0169, "step": 143500 }, { "epoch": 91.46590184831102, "grad_norm": 0.3025877773761749, "learning_rate": 5.768069948928353e-05, "loss": 0.0097, "step": 143510 }, { "epoch": 91.47227533460803, "grad_norm": 0.4445381462574005, "learning_rate": 5.76752532937147e-05, "loss": 0.0099, "step": 143520 }, { "epoch": 91.47864882090504, "grad_norm": 0.428090900182724, "learning_rate": 5.7669807004884304e-05, "loss": 0.02, "step": 143530 }, { "epoch": 91.48502230720204, "grad_norm": 0.23505356907844543, "learning_rate": 5.76643606228585e-05, "loss": 0.0118, "step": 143540 }, { "epoch": 91.49139579349904, "grad_norm": 0.15751467645168304, "learning_rate": 5.7658914147703494e-05, "loss": 0.0171, "step": 143550 }, { "epoch": 91.49776927979605, "grad_norm": 0.173712357878685, "learning_rate": 5.765346757948543e-05, "loss": 0.0173, "step": 143560 }, { "epoch": 91.50414276609305, "grad_norm": 0.31655415892601013, "learning_rate": 5.764802091827053e-05, "loss": 0.0198, "step": 143570 }, { "epoch": 91.51051625239006, "grad_norm": 0.19907057285308838, "learning_rate": 5.764257416412494e-05, "loss": 0.0127, "step": 143580 }, { "epoch": 91.51688973868706, "grad_norm": 0.26132693886756897, "learning_rate": 5.763712731711487e-05, "loss": 0.0073, "step": 143590 }, { "epoch": 91.52326322498406, "grad_norm": 0.41726207733154297, "learning_rate": 5.763168037730649e-05, "loss": 0.0124, "step": 143600 }, { "epoch": 91.52963671128107, "grad_norm": 0.5490280985832214, "learning_rate": 5.762623334476598e-05, "loss": 0.0142, "step": 143610 }, { "epoch": 91.53601019757808, "grad_norm": 0.33242860436439514, "learning_rate": 5.762078621955954e-05, "loss": 0.02, "step": 143620 }, { "epoch": 91.54238368387507, "grad_norm": 0.34122318029403687, "learning_rate": 5.761533900175334e-05, "loss": 0.0143, "step": 143630 }, { "epoch": 91.54875717017208, "grad_norm": 0.6027663946151733, "learning_rate": 5.7609891691413595e-05, "loss": 0.011, "step": 143640 }, { "epoch": 91.55513065646909, "grad_norm": 0.3250682055950165, "learning_rate": 5.7604444288606474e-05, "loss": 0.0126, "step": 143650 }, { "epoch": 91.5615041427661, "grad_norm": 0.4440905749797821, "learning_rate": 5.759899679339819e-05, "loss": 0.0146, "step": 143660 }, { "epoch": 91.56787762906309, "grad_norm": 0.5600935816764832, "learning_rate": 5.7593549205854916e-05, "loss": 0.0208, "step": 143670 }, { "epoch": 91.5742511153601, "grad_norm": 0.25952187180519104, "learning_rate": 5.758810152604284e-05, "loss": 0.0098, "step": 143680 }, { "epoch": 91.5806246016571, "grad_norm": 0.4033315181732178, "learning_rate": 5.758265375402817e-05, "loss": 0.0169, "step": 143690 }, { "epoch": 91.58699808795411, "grad_norm": 0.3822246789932251, "learning_rate": 5.757720588987708e-05, "loss": 0.0139, "step": 143700 }, { "epoch": 91.59337157425111, "grad_norm": 0.27464914321899414, "learning_rate": 5.7571757933655814e-05, "loss": 0.0153, "step": 143710 }, { "epoch": 91.59974506054812, "grad_norm": 0.334176629781723, "learning_rate": 5.7566309885430526e-05, "loss": 0.0169, "step": 143720 }, { "epoch": 91.60611854684512, "grad_norm": 0.43456992506980896, "learning_rate": 5.756086174526743e-05, "loss": 0.0189, "step": 143730 }, { "epoch": 91.61249203314213, "grad_norm": 0.3372412919998169, "learning_rate": 5.755541351323272e-05, "loss": 0.0096, "step": 143740 }, { "epoch": 91.61886551943914, "grad_norm": 0.17508728802204132, "learning_rate": 5.754996518939262e-05, "loss": 0.017, "step": 143750 }, { "epoch": 91.62523900573613, "grad_norm": 0.1819859892129898, "learning_rate": 5.754451677381329e-05, "loss": 0.0259, "step": 143760 }, { "epoch": 91.63161249203314, "grad_norm": 0.34899598360061646, "learning_rate": 5.753906826656097e-05, "loss": 0.0182, "step": 143770 }, { "epoch": 91.63798597833015, "grad_norm": 0.21936041116714478, "learning_rate": 5.753361966770187e-05, "loss": 0.0069, "step": 143780 }, { "epoch": 91.64435946462716, "grad_norm": 0.550836980342865, "learning_rate": 5.752817097730214e-05, "loss": 0.0211, "step": 143790 }, { "epoch": 91.65073295092415, "grad_norm": 0.31838512420654297, "learning_rate": 5.7522722195428056e-05, "loss": 0.0115, "step": 143800 }, { "epoch": 91.65710643722116, "grad_norm": 0.15074428915977478, "learning_rate": 5.7517273322145785e-05, "loss": 0.0088, "step": 143810 }, { "epoch": 91.66347992351817, "grad_norm": 0.22498367726802826, "learning_rate": 5.751182435752155e-05, "loss": 0.022, "step": 143820 }, { "epoch": 91.66985340981518, "grad_norm": 0.21611127257347107, "learning_rate": 5.750637530162155e-05, "loss": 0.0095, "step": 143830 }, { "epoch": 91.67622689611217, "grad_norm": 0.31902602314949036, "learning_rate": 5.7500926154512004e-05, "loss": 0.0114, "step": 143840 }, { "epoch": 91.68260038240918, "grad_norm": 0.1984030306339264, "learning_rate": 5.749547691625913e-05, "loss": 0.0077, "step": 143850 }, { "epoch": 91.68897386870619, "grad_norm": 0.6419076323509216, "learning_rate": 5.749002758692912e-05, "loss": 0.0119, "step": 143860 }, { "epoch": 91.6953473550032, "grad_norm": 0.34690120816230774, "learning_rate": 5.748457816658821e-05, "loss": 0.0125, "step": 143870 }, { "epoch": 91.70172084130019, "grad_norm": 0.23911719024181366, "learning_rate": 5.747912865530261e-05, "loss": 0.0135, "step": 143880 }, { "epoch": 91.7080943275972, "grad_norm": 0.271011620759964, "learning_rate": 5.747367905313854e-05, "loss": 0.0083, "step": 143890 }, { "epoch": 91.7144678138942, "grad_norm": 0.47997817397117615, "learning_rate": 5.746822936016221e-05, "loss": 0.0149, "step": 143900 }, { "epoch": 91.72084130019121, "grad_norm": 0.5543095469474792, "learning_rate": 5.746277957643984e-05, "loss": 0.0138, "step": 143910 }, { "epoch": 91.7272147864882, "grad_norm": 0.18243597447872162, "learning_rate": 5.7457329702037646e-05, "loss": 0.013, "step": 143920 }, { "epoch": 91.73358827278521, "grad_norm": 0.2860872149467468, "learning_rate": 5.745187973702186e-05, "loss": 0.0146, "step": 143930 }, { "epoch": 91.73996175908222, "grad_norm": 0.23381422460079193, "learning_rate": 5.74464296814587e-05, "loss": 0.0203, "step": 143940 }, { "epoch": 91.74633524537923, "grad_norm": 0.38907214999198914, "learning_rate": 5.7440979535414395e-05, "loss": 0.0117, "step": 143950 }, { "epoch": 91.75270873167622, "grad_norm": 0.24458523094654083, "learning_rate": 5.743552929895515e-05, "loss": 0.0175, "step": 143960 }, { "epoch": 91.75908221797323, "grad_norm": 0.4146716892719269, "learning_rate": 5.7430078972147226e-05, "loss": 0.011, "step": 143970 }, { "epoch": 91.76545570427024, "grad_norm": 0.2958703339099884, "learning_rate": 5.74246285550568e-05, "loss": 0.0145, "step": 143980 }, { "epoch": 91.77182919056725, "grad_norm": 0.24257995188236237, "learning_rate": 5.741917804775014e-05, "loss": 0.0168, "step": 143990 }, { "epoch": 91.77820267686424, "grad_norm": 0.2928883731365204, "learning_rate": 5.7413727450293465e-05, "loss": 0.0161, "step": 144000 }, { "epoch": 91.78457616316125, "grad_norm": 1.1435242891311646, "learning_rate": 5.7408276762753e-05, "loss": 0.0093, "step": 144010 }, { "epoch": 91.79094964945826, "grad_norm": 0.420482873916626, "learning_rate": 5.740282598519497e-05, "loss": 0.0208, "step": 144020 }, { "epoch": 91.79732313575526, "grad_norm": 0.49920010566711426, "learning_rate": 5.739737511768564e-05, "loss": 0.0181, "step": 144030 }, { "epoch": 91.80369662205226, "grad_norm": 0.30887657403945923, "learning_rate": 5.739192416029119e-05, "loss": 0.0134, "step": 144040 }, { "epoch": 91.81007010834927, "grad_norm": 0.25943177938461304, "learning_rate": 5.73864731130779e-05, "loss": 0.0158, "step": 144050 }, { "epoch": 91.81644359464627, "grad_norm": 0.46177148818969727, "learning_rate": 5.738102197611197e-05, "loss": 0.0226, "step": 144060 }, { "epoch": 91.82281708094328, "grad_norm": 0.15177281200885773, "learning_rate": 5.7375570749459676e-05, "loss": 0.0127, "step": 144070 }, { "epoch": 91.82919056724027, "grad_norm": 0.41283848881721497, "learning_rate": 5.737011943318722e-05, "loss": 0.0116, "step": 144080 }, { "epoch": 91.83556405353728, "grad_norm": 0.280236154794693, "learning_rate": 5.736466802736085e-05, "loss": 0.0084, "step": 144090 }, { "epoch": 91.84193753983429, "grad_norm": 0.3123399317264557, "learning_rate": 5.735921653204682e-05, "loss": 0.0237, "step": 144100 }, { "epoch": 91.8483110261313, "grad_norm": 0.29268351197242737, "learning_rate": 5.735376494731135e-05, "loss": 0.0116, "step": 144110 }, { "epoch": 91.85468451242829, "grad_norm": 0.47951367497444153, "learning_rate": 5.73483132732207e-05, "loss": 0.0182, "step": 144120 }, { "epoch": 91.8610579987253, "grad_norm": 0.43992215394973755, "learning_rate": 5.734286150984111e-05, "loss": 0.0141, "step": 144130 }, { "epoch": 91.86743148502231, "grad_norm": 0.3784279525279999, "learning_rate": 5.733740965723882e-05, "loss": 0.0133, "step": 144140 }, { "epoch": 91.87380497131932, "grad_norm": 0.2926246225833893, "learning_rate": 5.7331957715480065e-05, "loss": 0.0202, "step": 144150 }, { "epoch": 91.88017845761631, "grad_norm": 0.4754764437675476, "learning_rate": 5.7326505684631105e-05, "loss": 0.0163, "step": 144160 }, { "epoch": 91.88655194391332, "grad_norm": 0.3183499574661255, "learning_rate": 5.732105356475819e-05, "loss": 0.0165, "step": 144170 }, { "epoch": 91.89292543021033, "grad_norm": 0.8661971688270569, "learning_rate": 5.731560135592755e-05, "loss": 0.0111, "step": 144180 }, { "epoch": 91.89929891650733, "grad_norm": 0.18645060062408447, "learning_rate": 5.731014905820547e-05, "loss": 0.0104, "step": 144190 }, { "epoch": 91.90567240280433, "grad_norm": 0.28012487292289734, "learning_rate": 5.730469667165814e-05, "loss": 0.0175, "step": 144200 }, { "epoch": 91.91204588910134, "grad_norm": 0.25844624638557434, "learning_rate": 5.729924419635189e-05, "loss": 0.0112, "step": 144210 }, { "epoch": 91.91841937539834, "grad_norm": 0.3714515268802643, "learning_rate": 5.72937916323529e-05, "loss": 0.0149, "step": 144220 }, { "epoch": 91.92479286169535, "grad_norm": 0.33458569645881653, "learning_rate": 5.728833897972747e-05, "loss": 0.0153, "step": 144230 }, { "epoch": 91.93116634799235, "grad_norm": 0.29994329810142517, "learning_rate": 5.7282886238541834e-05, "loss": 0.0118, "step": 144240 }, { "epoch": 91.93753983428935, "grad_norm": 0.7645736336708069, "learning_rate": 5.727743340886226e-05, "loss": 0.0125, "step": 144250 }, { "epoch": 91.94391332058636, "grad_norm": 0.3496071696281433, "learning_rate": 5.7271980490754994e-05, "loss": 0.0195, "step": 144260 }, { "epoch": 91.95028680688337, "grad_norm": 0.24165931344032288, "learning_rate": 5.7266527484286295e-05, "loss": 0.0121, "step": 144270 }, { "epoch": 91.95666029318036, "grad_norm": 0.5784472227096558, "learning_rate": 5.726107438952244e-05, "loss": 0.0175, "step": 144280 }, { "epoch": 91.96303377947737, "grad_norm": 0.5242268443107605, "learning_rate": 5.725562120652965e-05, "loss": 0.0191, "step": 144290 }, { "epoch": 91.96940726577438, "grad_norm": 0.492284893989563, "learning_rate": 5.725016793537424e-05, "loss": 0.0143, "step": 144300 }, { "epoch": 91.97578075207139, "grad_norm": 0.39136508107185364, "learning_rate": 5.724471457612244e-05, "loss": 0.0159, "step": 144310 }, { "epoch": 91.98215423836838, "grad_norm": 0.35635054111480713, "learning_rate": 5.723926112884051e-05, "loss": 0.0096, "step": 144320 }, { "epoch": 91.98852772466539, "grad_norm": 0.6183533072471619, "learning_rate": 5.7233807593594726e-05, "loss": 0.0169, "step": 144330 }, { "epoch": 91.9949012109624, "grad_norm": 0.16442865133285522, "learning_rate": 5.722835397045134e-05, "loss": 0.0114, "step": 144340 }, { "epoch": 92.0012746972594, "grad_norm": 0.23481552302837372, "learning_rate": 5.722290025947664e-05, "loss": 0.0157, "step": 144350 }, { "epoch": 92.00764818355641, "grad_norm": 0.5103223323822021, "learning_rate": 5.721744646073688e-05, "loss": 0.0201, "step": 144360 }, { "epoch": 92.0140216698534, "grad_norm": 0.33418670296669006, "learning_rate": 5.7211992574298325e-05, "loss": 0.0136, "step": 144370 }, { "epoch": 92.02039515615041, "grad_norm": 0.19295544922351837, "learning_rate": 5.720653860022726e-05, "loss": 0.0174, "step": 144380 }, { "epoch": 92.02676864244742, "grad_norm": 0.42627185583114624, "learning_rate": 5.7201084538589945e-05, "loss": 0.0222, "step": 144390 }, { "epoch": 92.03314212874443, "grad_norm": 0.2236996740102768, "learning_rate": 5.719563038945265e-05, "loss": 0.0105, "step": 144400 }, { "epoch": 92.03951561504142, "grad_norm": 0.23806728422641754, "learning_rate": 5.719017615288165e-05, "loss": 0.023, "step": 144410 }, { "epoch": 92.04588910133843, "grad_norm": 0.40482738614082336, "learning_rate": 5.718472182894322e-05, "loss": 0.0102, "step": 144420 }, { "epoch": 92.05226258763544, "grad_norm": 0.18808086216449738, "learning_rate": 5.717926741770364e-05, "loss": 0.0068, "step": 144430 }, { "epoch": 92.05863607393245, "grad_norm": 0.2935611605644226, "learning_rate": 5.717381291922919e-05, "loss": 0.0137, "step": 144440 }, { "epoch": 92.06500956022944, "grad_norm": 0.34073376655578613, "learning_rate": 5.716835833358612e-05, "loss": 0.0105, "step": 144450 }, { "epoch": 92.07138304652645, "grad_norm": 0.4030036926269531, "learning_rate": 5.716290366084075e-05, "loss": 0.0169, "step": 144460 }, { "epoch": 92.07775653282346, "grad_norm": 0.9832319021224976, "learning_rate": 5.715744890105933e-05, "loss": 0.0144, "step": 144470 }, { "epoch": 92.08413001912047, "grad_norm": 0.19579362869262695, "learning_rate": 5.715199405430814e-05, "loss": 0.0065, "step": 144480 }, { "epoch": 92.09050350541746, "grad_norm": 0.16694028675556183, "learning_rate": 5.7146539120653486e-05, "loss": 0.0108, "step": 144490 }, { "epoch": 92.09687699171447, "grad_norm": 0.3932695686817169, "learning_rate": 5.7141084100161616e-05, "loss": 0.0139, "step": 144500 }, { "epoch": 92.10325047801147, "grad_norm": 0.35489794611930847, "learning_rate": 5.713562899289885e-05, "loss": 0.0117, "step": 144510 }, { "epoch": 92.10962396430848, "grad_norm": 0.3094278872013092, "learning_rate": 5.713017379893144e-05, "loss": 0.0079, "step": 144520 }, { "epoch": 92.11599745060548, "grad_norm": 0.2102876454591751, "learning_rate": 5.712471851832569e-05, "loss": 0.0214, "step": 144530 }, { "epoch": 92.12237093690248, "grad_norm": 0.6374598145484924, "learning_rate": 5.7119263151147885e-05, "loss": 0.0148, "step": 144540 }, { "epoch": 92.12874442319949, "grad_norm": 0.39451780915260315, "learning_rate": 5.711380769746432e-05, "loss": 0.0113, "step": 144550 }, { "epoch": 92.1351179094965, "grad_norm": 0.29130491614341736, "learning_rate": 5.710835215734126e-05, "loss": 0.0158, "step": 144560 }, { "epoch": 92.1414913957935, "grad_norm": 0.3967180848121643, "learning_rate": 5.7102896530845015e-05, "loss": 0.014, "step": 144570 }, { "epoch": 92.1478648820905, "grad_norm": 0.2926569879055023, "learning_rate": 5.709744081804187e-05, "loss": 0.0162, "step": 144580 }, { "epoch": 92.15423836838751, "grad_norm": 0.43795520067214966, "learning_rate": 5.7091985018998116e-05, "loss": 0.01, "step": 144590 }, { "epoch": 92.16061185468452, "grad_norm": 0.3280838131904602, "learning_rate": 5.7086529133780054e-05, "loss": 0.0196, "step": 144600 }, { "epoch": 92.16698534098151, "grad_norm": 0.3483107089996338, "learning_rate": 5.708107316245397e-05, "loss": 0.0124, "step": 144610 }, { "epoch": 92.17335882727852, "grad_norm": 0.530800998210907, "learning_rate": 5.7075617105086166e-05, "loss": 0.0183, "step": 144620 }, { "epoch": 92.17973231357553, "grad_norm": 0.39673465490341187, "learning_rate": 5.707016096174291e-05, "loss": 0.0196, "step": 144630 }, { "epoch": 92.18610579987254, "grad_norm": 0.2795568108558655, "learning_rate": 5.706470473249056e-05, "loss": 0.0157, "step": 144640 }, { "epoch": 92.19247928616953, "grad_norm": 0.46249303221702576, "learning_rate": 5.705924841739535e-05, "loss": 0.0149, "step": 144650 }, { "epoch": 92.19885277246654, "grad_norm": 0.2643163204193115, "learning_rate": 5.7053792016523625e-05, "loss": 0.0118, "step": 144660 }, { "epoch": 92.20522625876355, "grad_norm": 0.2976883351802826, "learning_rate": 5.704833552994164e-05, "loss": 0.0158, "step": 144670 }, { "epoch": 92.21159974506055, "grad_norm": 0.4826933741569519, "learning_rate": 5.704287895771575e-05, "loss": 0.0209, "step": 144680 }, { "epoch": 92.21797323135755, "grad_norm": 0.19507089257240295, "learning_rate": 5.7037422299912226e-05, "loss": 0.0109, "step": 144690 }, { "epoch": 92.22434671765456, "grad_norm": 0.22496354579925537, "learning_rate": 5.703196555659737e-05, "loss": 0.0102, "step": 144700 }, { "epoch": 92.23072020395156, "grad_norm": 0.19993332028388977, "learning_rate": 5.702650872783749e-05, "loss": 0.0138, "step": 144710 }, { "epoch": 92.23709369024857, "grad_norm": 0.40232613682746887, "learning_rate": 5.702105181369891e-05, "loss": 0.0123, "step": 144720 }, { "epoch": 92.24346717654556, "grad_norm": 0.1777331531047821, "learning_rate": 5.701559481424791e-05, "loss": 0.0108, "step": 144730 }, { "epoch": 92.24984066284257, "grad_norm": 0.2166650891304016, "learning_rate": 5.701013772955082e-05, "loss": 0.0172, "step": 144740 }, { "epoch": 92.25621414913958, "grad_norm": 0.25684210658073425, "learning_rate": 5.7004680559673915e-05, "loss": 0.0192, "step": 144750 }, { "epoch": 92.26258763543659, "grad_norm": 0.19950206577777863, "learning_rate": 5.699922330468355e-05, "loss": 0.0119, "step": 144760 }, { "epoch": 92.26896112173358, "grad_norm": 0.5421521067619324, "learning_rate": 5.6993765964645993e-05, "loss": 0.0148, "step": 144770 }, { "epoch": 92.27533460803059, "grad_norm": 0.19173504412174225, "learning_rate": 5.69883085396276e-05, "loss": 0.0143, "step": 144780 }, { "epoch": 92.2817080943276, "grad_norm": 0.21388083696365356, "learning_rate": 5.6982851029694636e-05, "loss": 0.0168, "step": 144790 }, { "epoch": 92.2880815806246, "grad_norm": 0.2538701891899109, "learning_rate": 5.697739343491346e-05, "loss": 0.0125, "step": 144800 }, { "epoch": 92.2944550669216, "grad_norm": 0.24677500128746033, "learning_rate": 5.697193575535037e-05, "loss": 0.0146, "step": 144810 }, { "epoch": 92.30082855321861, "grad_norm": 0.20921340584754944, "learning_rate": 5.696647799107165e-05, "loss": 0.0103, "step": 144820 }, { "epoch": 92.30720203951562, "grad_norm": 0.22500020265579224, "learning_rate": 5.6961020142143676e-05, "loss": 0.0101, "step": 144830 }, { "epoch": 92.31357552581262, "grad_norm": 0.23651817440986633, "learning_rate": 5.695556220863271e-05, "loss": 0.0118, "step": 144840 }, { "epoch": 92.31994901210962, "grad_norm": 0.3561020493507385, "learning_rate": 5.695010419060511e-05, "loss": 0.0176, "step": 144850 }, { "epoch": 92.32632249840663, "grad_norm": 0.4764436185359955, "learning_rate": 5.6944646088127174e-05, "loss": 0.0217, "step": 144860 }, { "epoch": 92.33269598470363, "grad_norm": 0.411953330039978, "learning_rate": 5.6939187901265244e-05, "loss": 0.0154, "step": 144870 }, { "epoch": 92.33906947100064, "grad_norm": 0.5255167484283447, "learning_rate": 5.6933729630085617e-05, "loss": 0.0098, "step": 144880 }, { "epoch": 92.34544295729764, "grad_norm": 0.17044448852539062, "learning_rate": 5.692827127465463e-05, "loss": 0.0101, "step": 144890 }, { "epoch": 92.35181644359464, "grad_norm": 0.38473448157310486, "learning_rate": 5.692281283503862e-05, "loss": 0.0115, "step": 144900 }, { "epoch": 92.35818992989165, "grad_norm": 0.40604695677757263, "learning_rate": 5.691735431130387e-05, "loss": 0.0221, "step": 144910 }, { "epoch": 92.36456341618866, "grad_norm": 0.24267427623271942, "learning_rate": 5.691189570351676e-05, "loss": 0.0099, "step": 144920 }, { "epoch": 92.37093690248565, "grad_norm": 0.26977965235710144, "learning_rate": 5.690643701174357e-05, "loss": 0.0149, "step": 144930 }, { "epoch": 92.37731038878266, "grad_norm": 0.4230418801307678, "learning_rate": 5.690097823605067e-05, "loss": 0.0104, "step": 144940 }, { "epoch": 92.38368387507967, "grad_norm": 0.25527092814445496, "learning_rate": 5.689551937650436e-05, "loss": 0.0127, "step": 144950 }, { "epoch": 92.39005736137668, "grad_norm": 0.12630251049995422, "learning_rate": 5.689006043317098e-05, "loss": 0.012, "step": 144960 }, { "epoch": 92.39643084767367, "grad_norm": 0.17016585171222687, "learning_rate": 5.688460140611687e-05, "loss": 0.0229, "step": 144970 }, { "epoch": 92.40280433397068, "grad_norm": 0.4182908535003662, "learning_rate": 5.687914229540833e-05, "loss": 0.0171, "step": 144980 }, { "epoch": 92.40917782026769, "grad_norm": 0.3214331865310669, "learning_rate": 5.687368310111173e-05, "loss": 0.0104, "step": 144990 }, { "epoch": 92.4155513065647, "grad_norm": 0.4221337139606476, "learning_rate": 5.686822382329339e-05, "loss": 0.0149, "step": 145000 }, { "epoch": 92.4219247928617, "grad_norm": 0.18912814557552338, "learning_rate": 5.6862764462019646e-05, "loss": 0.0104, "step": 145010 }, { "epoch": 92.4282982791587, "grad_norm": 0.22305582463741302, "learning_rate": 5.685730501735682e-05, "loss": 0.0099, "step": 145020 }, { "epoch": 92.4346717654557, "grad_norm": 0.317444384098053, "learning_rate": 5.685184548937129e-05, "loss": 0.0084, "step": 145030 }, { "epoch": 92.44104525175271, "grad_norm": 0.285159170627594, "learning_rate": 5.684638587812936e-05, "loss": 0.013, "step": 145040 }, { "epoch": 92.44741873804972, "grad_norm": 0.15300928056240082, "learning_rate": 5.684092618369736e-05, "loss": 0.0107, "step": 145050 }, { "epoch": 92.45379222434671, "grad_norm": 0.40498650074005127, "learning_rate": 5.683546640614167e-05, "loss": 0.0213, "step": 145060 }, { "epoch": 92.46016571064372, "grad_norm": 0.5655746459960938, "learning_rate": 5.6830006545528594e-05, "loss": 0.0141, "step": 145070 }, { "epoch": 92.46653919694073, "grad_norm": 1.1130366325378418, "learning_rate": 5.68245466019245e-05, "loss": 0.0164, "step": 145080 }, { "epoch": 92.47291268323774, "grad_norm": 0.27705803513526917, "learning_rate": 5.681908657539571e-05, "loss": 0.0231, "step": 145090 }, { "epoch": 92.47928616953473, "grad_norm": 0.5756409168243408, "learning_rate": 5.6813626466008585e-05, "loss": 0.0178, "step": 145100 }, { "epoch": 92.48565965583174, "grad_norm": 0.6108702421188354, "learning_rate": 5.6808166273829465e-05, "loss": 0.012, "step": 145110 }, { "epoch": 92.49203314212875, "grad_norm": 0.37920907139778137, "learning_rate": 5.68027059989247e-05, "loss": 0.0312, "step": 145120 }, { "epoch": 92.49840662842576, "grad_norm": 0.6271448135375977, "learning_rate": 5.6797245641360644e-05, "loss": 0.02, "step": 145130 }, { "epoch": 92.50478011472275, "grad_norm": 0.5343686938285828, "learning_rate": 5.6791785201203616e-05, "loss": 0.026, "step": 145140 }, { "epoch": 92.51115360101976, "grad_norm": 0.5083519220352173, "learning_rate": 5.678632467852e-05, "loss": 0.0339, "step": 145150 }, { "epoch": 92.51752708731676, "grad_norm": 0.5379785895347595, "learning_rate": 5.678086407337612e-05, "loss": 0.0093, "step": 145160 }, { "epoch": 92.52390057361377, "grad_norm": 0.37798309326171875, "learning_rate": 5.6775403385838354e-05, "loss": 0.0128, "step": 145170 }, { "epoch": 92.53027405991077, "grad_norm": 0.3203207552433014, "learning_rate": 5.676994261597301e-05, "loss": 0.0097, "step": 145180 }, { "epoch": 92.53664754620777, "grad_norm": 0.21205225586891174, "learning_rate": 5.676448176384651e-05, "loss": 0.0125, "step": 145190 }, { "epoch": 92.54302103250478, "grad_norm": 0.20712727308273315, "learning_rate": 5.6759020829525154e-05, "loss": 0.0141, "step": 145200 }, { "epoch": 92.54939451880179, "grad_norm": 0.1966904103755951, "learning_rate": 5.675355981307531e-05, "loss": 0.008, "step": 145210 }, { "epoch": 92.55576800509878, "grad_norm": 0.22146159410476685, "learning_rate": 5.6748098714563345e-05, "loss": 0.0126, "step": 145220 }, { "epoch": 92.56214149139579, "grad_norm": 1.6382886171340942, "learning_rate": 5.67426375340556e-05, "loss": 0.0095, "step": 145230 }, { "epoch": 92.5685149776928, "grad_norm": 0.27126505970954895, "learning_rate": 5.673717627161844e-05, "loss": 0.0192, "step": 145240 }, { "epoch": 92.57488846398981, "grad_norm": 0.2915521264076233, "learning_rate": 5.6731714927318236e-05, "loss": 0.0129, "step": 145250 }, { "epoch": 92.5812619502868, "grad_norm": 0.31756970286369324, "learning_rate": 5.6726253501221336e-05, "loss": 0.0147, "step": 145260 }, { "epoch": 92.58763543658381, "grad_norm": 0.25484082102775574, "learning_rate": 5.6720791993394104e-05, "loss": 0.0092, "step": 145270 }, { "epoch": 92.59400892288082, "grad_norm": 0.5886210799217224, "learning_rate": 5.671533040390291e-05, "loss": 0.0156, "step": 145280 }, { "epoch": 92.60038240917783, "grad_norm": 0.282459020614624, "learning_rate": 5.670986873281412e-05, "loss": 0.0104, "step": 145290 }, { "epoch": 92.60675589547482, "grad_norm": 0.22436954081058502, "learning_rate": 5.670440698019407e-05, "loss": 0.0179, "step": 145300 }, { "epoch": 92.61312938177183, "grad_norm": 0.4374842345714569, "learning_rate": 5.669894514610916e-05, "loss": 0.0109, "step": 145310 }, { "epoch": 92.61950286806884, "grad_norm": 0.53550124168396, "learning_rate": 5.669348323062572e-05, "loss": 0.0101, "step": 145320 }, { "epoch": 92.62587635436584, "grad_norm": 0.7441917061805725, "learning_rate": 5.6688021233810164e-05, "loss": 0.0306, "step": 145330 }, { "epoch": 92.63224984066284, "grad_norm": 0.4227094352245331, "learning_rate": 5.668255915572882e-05, "loss": 0.0187, "step": 145340 }, { "epoch": 92.63862332695984, "grad_norm": 0.41193634271621704, "learning_rate": 5.667709699644808e-05, "loss": 0.0112, "step": 145350 }, { "epoch": 92.64499681325685, "grad_norm": 0.2775179445743561, "learning_rate": 5.6671634756034295e-05, "loss": 0.0237, "step": 145360 }, { "epoch": 92.65137029955386, "grad_norm": 0.2522629499435425, "learning_rate": 5.666617243455387e-05, "loss": 0.0186, "step": 145370 }, { "epoch": 92.65774378585085, "grad_norm": 0.40172287821769714, "learning_rate": 5.666071003207314e-05, "loss": 0.0076, "step": 145380 }, { "epoch": 92.66411727214786, "grad_norm": 0.18423353135585785, "learning_rate": 5.66552475486585e-05, "loss": 0.0207, "step": 145390 }, { "epoch": 92.67049075844487, "grad_norm": 0.5866109132766724, "learning_rate": 5.664978498437632e-05, "loss": 0.0105, "step": 145400 }, { "epoch": 92.67686424474188, "grad_norm": 0.4495161473751068, "learning_rate": 5.6644322339292975e-05, "loss": 0.0213, "step": 145410 }, { "epoch": 92.68323773103887, "grad_norm": 0.23948319256305695, "learning_rate": 5.6638859613474825e-05, "loss": 0.0111, "step": 145420 }, { "epoch": 92.68961121733588, "grad_norm": 0.48597249388694763, "learning_rate": 5.663339680698828e-05, "loss": 0.0153, "step": 145430 }, { "epoch": 92.69598470363289, "grad_norm": 0.8170632123947144, "learning_rate": 5.662793391989969e-05, "loss": 0.0174, "step": 145440 }, { "epoch": 92.7023581899299, "grad_norm": 0.26335737109184265, "learning_rate": 5.662247095227546e-05, "loss": 0.0227, "step": 145450 }, { "epoch": 92.70873167622689, "grad_norm": 0.24948248267173767, "learning_rate": 5.661700790418194e-05, "loss": 0.0066, "step": 145460 }, { "epoch": 92.7151051625239, "grad_norm": 0.28452831506729126, "learning_rate": 5.661154477568553e-05, "loss": 0.009, "step": 145470 }, { "epoch": 92.7214786488209, "grad_norm": 0.6214900016784668, "learning_rate": 5.660608156685261e-05, "loss": 0.0118, "step": 145480 }, { "epoch": 92.72785213511791, "grad_norm": 0.45220911502838135, "learning_rate": 5.6600618277749575e-05, "loss": 0.0209, "step": 145490 }, { "epoch": 92.73422562141491, "grad_norm": 0.495716392993927, "learning_rate": 5.659515490844277e-05, "loss": 0.0123, "step": 145500 }, { "epoch": 92.74059910771192, "grad_norm": 0.3224101960659027, "learning_rate": 5.658969145899863e-05, "loss": 0.0117, "step": 145510 }, { "epoch": 92.74697259400892, "grad_norm": 0.22997558116912842, "learning_rate": 5.6584227929483514e-05, "loss": 0.0159, "step": 145520 }, { "epoch": 92.75334608030593, "grad_norm": 0.7531261444091797, "learning_rate": 5.65787643199638e-05, "loss": 0.0108, "step": 145530 }, { "epoch": 92.75971956660293, "grad_norm": 0.21900245547294617, "learning_rate": 5.657330063050589e-05, "loss": 0.0098, "step": 145540 }, { "epoch": 92.76609305289993, "grad_norm": 0.4711115062236786, "learning_rate": 5.656783686117617e-05, "loss": 0.0168, "step": 145550 }, { "epoch": 92.77246653919694, "grad_norm": 0.5555914044380188, "learning_rate": 5.656237301204104e-05, "loss": 0.0136, "step": 145560 }, { "epoch": 92.77884002549395, "grad_norm": 0.3505522310733795, "learning_rate": 5.655690908316688e-05, "loss": 0.0113, "step": 145570 }, { "epoch": 92.78521351179094, "grad_norm": 0.292263388633728, "learning_rate": 5.655144507462008e-05, "loss": 0.0135, "step": 145580 }, { "epoch": 92.79158699808795, "grad_norm": 0.2468748241662979, "learning_rate": 5.654598098646703e-05, "loss": 0.0132, "step": 145590 }, { "epoch": 92.79796048438496, "grad_norm": 0.2356233149766922, "learning_rate": 5.654051681877415e-05, "loss": 0.0068, "step": 145600 }, { "epoch": 92.80433397068197, "grad_norm": 0.44373148679733276, "learning_rate": 5.6535052571607806e-05, "loss": 0.0101, "step": 145610 }, { "epoch": 92.81070745697897, "grad_norm": 0.5234602689743042, "learning_rate": 5.6529588245034396e-05, "loss": 0.0112, "step": 145620 }, { "epoch": 92.81708094327597, "grad_norm": 0.38471031188964844, "learning_rate": 5.652412383912034e-05, "loss": 0.014, "step": 145630 }, { "epoch": 92.82345442957298, "grad_norm": 0.22267208993434906, "learning_rate": 5.651865935393199e-05, "loss": 0.0196, "step": 145640 }, { "epoch": 92.82982791586998, "grad_norm": 0.5239781141281128, "learning_rate": 5.6513194789535805e-05, "loss": 0.0131, "step": 145650 }, { "epoch": 92.83620140216699, "grad_norm": 0.56694096326828, "learning_rate": 5.6507730145998126e-05, "loss": 0.0203, "step": 145660 }, { "epoch": 92.84257488846399, "grad_norm": 0.540851354598999, "learning_rate": 5.65022654233854e-05, "loss": 0.0109, "step": 145670 }, { "epoch": 92.848948374761, "grad_norm": 0.3921886384487152, "learning_rate": 5.649680062176401e-05, "loss": 0.0148, "step": 145680 }, { "epoch": 92.855321861058, "grad_norm": 0.32475823163986206, "learning_rate": 5.6491335741200356e-05, "loss": 0.0132, "step": 145690 }, { "epoch": 92.86169534735501, "grad_norm": 0.24148790538311005, "learning_rate": 5.648587078176085e-05, "loss": 0.0182, "step": 145700 }, { "epoch": 92.868068833652, "grad_norm": 0.22038544714450836, "learning_rate": 5.6480405743511876e-05, "loss": 0.0068, "step": 145710 }, { "epoch": 92.87444231994901, "grad_norm": 0.6419162750244141, "learning_rate": 5.6474940626519866e-05, "loss": 0.024, "step": 145720 }, { "epoch": 92.88081580624602, "grad_norm": 0.23162826895713806, "learning_rate": 5.646947543085121e-05, "loss": 0.0169, "step": 145730 }, { "epoch": 92.88718929254303, "grad_norm": 0.1890372931957245, "learning_rate": 5.646401015657232e-05, "loss": 0.0127, "step": 145740 }, { "epoch": 92.89356277884002, "grad_norm": 0.49108946323394775, "learning_rate": 5.645854480374959e-05, "loss": 0.0154, "step": 145750 }, { "epoch": 92.89993626513703, "grad_norm": 1.136120319366455, "learning_rate": 5.6453079372449466e-05, "loss": 0.0216, "step": 145760 }, { "epoch": 92.90630975143404, "grad_norm": 0.5663464665412903, "learning_rate": 5.644761386273833e-05, "loss": 0.0129, "step": 145770 }, { "epoch": 92.91268323773104, "grad_norm": 0.32834672927856445, "learning_rate": 5.6442148274682584e-05, "loss": 0.0117, "step": 145780 }, { "epoch": 92.91905672402804, "grad_norm": 0.25218522548675537, "learning_rate": 5.643668260834867e-05, "loss": 0.0141, "step": 145790 }, { "epoch": 92.92543021032505, "grad_norm": 0.3194790184497833, "learning_rate": 5.643121686380297e-05, "loss": 0.0169, "step": 145800 }, { "epoch": 92.93180369662205, "grad_norm": 0.4805222153663635, "learning_rate": 5.642575104111192e-05, "loss": 0.0267, "step": 145810 }, { "epoch": 92.93817718291906, "grad_norm": 0.2933099865913391, "learning_rate": 5.642028514034192e-05, "loss": 0.0104, "step": 145820 }, { "epoch": 92.94455066921606, "grad_norm": 0.3519473671913147, "learning_rate": 5.641481916155942e-05, "loss": 0.0128, "step": 145830 }, { "epoch": 92.95092415551306, "grad_norm": 0.4094841182231903, "learning_rate": 5.640935310483078e-05, "loss": 0.0107, "step": 145840 }, { "epoch": 92.95729764181007, "grad_norm": 0.34233465790748596, "learning_rate": 5.6403886970222463e-05, "loss": 0.0155, "step": 145850 }, { "epoch": 92.96367112810708, "grad_norm": 0.30390235781669617, "learning_rate": 5.639842075780087e-05, "loss": 0.0098, "step": 145860 }, { "epoch": 92.97004461440407, "grad_norm": 0.17502020299434662, "learning_rate": 5.639295446763242e-05, "loss": 0.0142, "step": 145870 }, { "epoch": 92.97641810070108, "grad_norm": 0.4761334955692291, "learning_rate": 5.6387488099783546e-05, "loss": 0.0183, "step": 145880 }, { "epoch": 92.98279158699809, "grad_norm": 0.25022220611572266, "learning_rate": 5.638202165432066e-05, "loss": 0.0132, "step": 145890 }, { "epoch": 92.9891650732951, "grad_norm": 0.1968190222978592, "learning_rate": 5.637655513131018e-05, "loss": 0.0173, "step": 145900 }, { "epoch": 92.99553855959209, "grad_norm": 0.32935282588005066, "learning_rate": 5.637108853081854e-05, "loss": 0.0125, "step": 145910 }, { "epoch": 93.0019120458891, "grad_norm": 0.3159519135951996, "learning_rate": 5.6365621852912154e-05, "loss": 0.0093, "step": 145920 }, { "epoch": 93.00828553218611, "grad_norm": 0.28224751353263855, "learning_rate": 5.6360155097657464e-05, "loss": 0.0121, "step": 145930 }, { "epoch": 93.01465901848312, "grad_norm": 0.3833948075771332, "learning_rate": 5.635468826512087e-05, "loss": 0.0138, "step": 145940 }, { "epoch": 93.02103250478011, "grad_norm": 0.2712289094924927, "learning_rate": 5.634922135536882e-05, "loss": 0.0115, "step": 145950 }, { "epoch": 93.02740599107712, "grad_norm": 0.17850737273693085, "learning_rate": 5.634375436846773e-05, "loss": 0.0157, "step": 145960 }, { "epoch": 93.03377947737413, "grad_norm": 0.35803720355033875, "learning_rate": 5.633828730448404e-05, "loss": 0.0102, "step": 145970 }, { "epoch": 93.04015296367113, "grad_norm": 0.3269091546535492, "learning_rate": 5.633282016348417e-05, "loss": 0.008, "step": 145980 }, { "epoch": 93.04652644996813, "grad_norm": 0.7712300419807434, "learning_rate": 5.632735294553456e-05, "loss": 0.0294, "step": 145990 }, { "epoch": 93.05289993626513, "grad_norm": 0.37199366092681885, "learning_rate": 5.632188565070163e-05, "loss": 0.0159, "step": 146000 }, { "epoch": 93.05927342256214, "grad_norm": 0.3492395579814911, "learning_rate": 5.6316418279051844e-05, "loss": 0.0114, "step": 146010 }, { "epoch": 93.06564690885915, "grad_norm": 0.359671950340271, "learning_rate": 5.63109508306516e-05, "loss": 0.0135, "step": 146020 }, { "epoch": 93.07202039515614, "grad_norm": 0.457660973072052, "learning_rate": 5.630548330556733e-05, "loss": 0.0113, "step": 146030 }, { "epoch": 93.07839388145315, "grad_norm": 0.2951827645301819, "learning_rate": 5.6300015703865494e-05, "loss": 0.0121, "step": 146040 }, { "epoch": 93.08476736775016, "grad_norm": 0.7867652177810669, "learning_rate": 5.6294548025612514e-05, "loss": 0.0144, "step": 146050 }, { "epoch": 93.09114085404717, "grad_norm": 0.3991543650627136, "learning_rate": 5.628908027087485e-05, "loss": 0.0124, "step": 146060 }, { "epoch": 93.09751434034416, "grad_norm": 0.31761133670806885, "learning_rate": 5.62836124397189e-05, "loss": 0.0186, "step": 146070 }, { "epoch": 93.10388782664117, "grad_norm": 0.3676363229751587, "learning_rate": 5.627814453221115e-05, "loss": 0.0108, "step": 146080 }, { "epoch": 93.11026131293818, "grad_norm": 0.5715415477752686, "learning_rate": 5.6272676548418e-05, "loss": 0.0199, "step": 146090 }, { "epoch": 93.11663479923519, "grad_norm": 0.27619993686676025, "learning_rate": 5.6267208488405924e-05, "loss": 0.0112, "step": 146100 }, { "epoch": 93.12300828553218, "grad_norm": 0.3095812499523163, "learning_rate": 5.626174035224132e-05, "loss": 0.0093, "step": 146110 }, { "epoch": 93.12938177182919, "grad_norm": 0.3883094787597656, "learning_rate": 5.6256272139990676e-05, "loss": 0.015, "step": 146120 }, { "epoch": 93.1357552581262, "grad_norm": 0.3421034514904022, "learning_rate": 5.6250803851720416e-05, "loss": 0.0147, "step": 146130 }, { "epoch": 93.1421287444232, "grad_norm": 0.19505247473716736, "learning_rate": 5.624533548749699e-05, "loss": 0.0132, "step": 146140 }, { "epoch": 93.1485022307202, "grad_norm": 0.7759355902671814, "learning_rate": 5.623986704738684e-05, "loss": 0.0193, "step": 146150 }, { "epoch": 93.1548757170172, "grad_norm": 0.3946567177772522, "learning_rate": 5.623439853145641e-05, "loss": 0.0147, "step": 146160 }, { "epoch": 93.16124920331421, "grad_norm": 0.560217559337616, "learning_rate": 5.622892993977216e-05, "loss": 0.0153, "step": 146170 }, { "epoch": 93.16762268961122, "grad_norm": 0.35469695925712585, "learning_rate": 5.6223461272400525e-05, "loss": 0.0166, "step": 146180 }, { "epoch": 93.17399617590821, "grad_norm": 0.23648107051849365, "learning_rate": 5.6217992529407946e-05, "loss": 0.0132, "step": 146190 }, { "epoch": 93.18036966220522, "grad_norm": 0.32512709498405457, "learning_rate": 5.621252371086091e-05, "loss": 0.0101, "step": 146200 }, { "epoch": 93.18674314850223, "grad_norm": 0.20987388491630554, "learning_rate": 5.6207054816825835e-05, "loss": 0.0102, "step": 146210 }, { "epoch": 93.19311663479924, "grad_norm": 0.682107150554657, "learning_rate": 5.6201585847369185e-05, "loss": 0.0119, "step": 146220 }, { "epoch": 93.19949012109623, "grad_norm": 0.3830803334712982, "learning_rate": 5.619611680255741e-05, "loss": 0.0112, "step": 146230 }, { "epoch": 93.20586360739324, "grad_norm": 1.043789029121399, "learning_rate": 5.6190647682456965e-05, "loss": 0.0294, "step": 146240 }, { "epoch": 93.21223709369025, "grad_norm": 0.5551419258117676, "learning_rate": 5.618517848713432e-05, "loss": 0.0137, "step": 146250 }, { "epoch": 93.21861057998726, "grad_norm": 0.24493442475795746, "learning_rate": 5.6179709216655895e-05, "loss": 0.0143, "step": 146260 }, { "epoch": 93.22498406628426, "grad_norm": 0.6193442940711975, "learning_rate": 5.617423987108818e-05, "loss": 0.0102, "step": 146270 }, { "epoch": 93.23135755258126, "grad_norm": 0.48596128821372986, "learning_rate": 5.616877045049762e-05, "loss": 0.0194, "step": 146280 }, { "epoch": 93.23773103887827, "grad_norm": 0.4406862258911133, "learning_rate": 5.616330095495067e-05, "loss": 0.021, "step": 146290 }, { "epoch": 93.24410452517527, "grad_norm": 0.2438385933637619, "learning_rate": 5.6157831384513796e-05, "loss": 0.0101, "step": 146300 }, { "epoch": 93.25047801147228, "grad_norm": 0.4448620676994324, "learning_rate": 5.615236173925347e-05, "loss": 0.0116, "step": 146310 }, { "epoch": 93.25685149776928, "grad_norm": 0.6925949454307556, "learning_rate": 5.614689201923612e-05, "loss": 0.0211, "step": 146320 }, { "epoch": 93.26322498406628, "grad_norm": 0.5809403657913208, "learning_rate": 5.614142222452824e-05, "loss": 0.0095, "step": 146330 }, { "epoch": 93.26959847036329, "grad_norm": 0.23931095004081726, "learning_rate": 5.613595235519629e-05, "loss": 0.0089, "step": 146340 }, { "epoch": 93.2759719566603, "grad_norm": 0.46331050992012024, "learning_rate": 5.613048241130671e-05, "loss": 0.0227, "step": 146350 }, { "epoch": 93.2823454429573, "grad_norm": 0.23879273235797882, "learning_rate": 5.612501239292598e-05, "loss": 0.0151, "step": 146360 }, { "epoch": 93.2887189292543, "grad_norm": 0.31689977645874023, "learning_rate": 5.611954230012056e-05, "loss": 0.0176, "step": 146370 }, { "epoch": 93.29509241555131, "grad_norm": 0.27960047125816345, "learning_rate": 5.611407213295694e-05, "loss": 0.011, "step": 146380 }, { "epoch": 93.30146590184832, "grad_norm": 0.46159446239471436, "learning_rate": 5.610860189150156e-05, "loss": 0.0247, "step": 146390 }, { "epoch": 93.30783938814531, "grad_norm": 0.8031337857246399, "learning_rate": 5.610313157582091e-05, "loss": 0.0126, "step": 146400 }, { "epoch": 93.31421287444232, "grad_norm": 0.48176252841949463, "learning_rate": 5.6097661185981455e-05, "loss": 0.0148, "step": 146410 }, { "epoch": 93.32058636073933, "grad_norm": 0.5185056328773499, "learning_rate": 5.609219072204964e-05, "loss": 0.0117, "step": 146420 }, { "epoch": 93.32695984703633, "grad_norm": 0.26833078265190125, "learning_rate": 5.6086720184091967e-05, "loss": 0.0075, "step": 146430 }, { "epoch": 93.33333333333333, "grad_norm": 0.4677334129810333, "learning_rate": 5.6081249572174884e-05, "loss": 0.0089, "step": 146440 }, { "epoch": 93.33970681963034, "grad_norm": 0.35330575704574585, "learning_rate": 5.6075778886364904e-05, "loss": 0.0149, "step": 146450 }, { "epoch": 93.34608030592734, "grad_norm": 0.31058207154273987, "learning_rate": 5.607030812672844e-05, "loss": 0.0127, "step": 146460 }, { "epoch": 93.35245379222435, "grad_norm": 0.2615984380245209, "learning_rate": 5.606483729333203e-05, "loss": 0.0136, "step": 146470 }, { "epoch": 93.35882727852135, "grad_norm": 0.18231306970119476, "learning_rate": 5.6059366386242105e-05, "loss": 0.0133, "step": 146480 }, { "epoch": 93.36520076481835, "grad_norm": 0.26447466015815735, "learning_rate": 5.605389540552517e-05, "loss": 0.021, "step": 146490 }, { "epoch": 93.37157425111536, "grad_norm": 0.46362578868865967, "learning_rate": 5.6048424351247686e-05, "loss": 0.0188, "step": 146500 }, { "epoch": 93.37794773741237, "grad_norm": 0.5507728457450867, "learning_rate": 5.604295322347613e-05, "loss": 0.0143, "step": 146510 }, { "epoch": 93.38432122370936, "grad_norm": 0.3092469871044159, "learning_rate": 5.603748202227699e-05, "loss": 0.0159, "step": 146520 }, { "epoch": 93.39069471000637, "grad_norm": 0.2868509590625763, "learning_rate": 5.603201074771674e-05, "loss": 0.0167, "step": 146530 }, { "epoch": 93.39706819630338, "grad_norm": 0.668409526348114, "learning_rate": 5.6026539399861855e-05, "loss": 0.0143, "step": 146540 }, { "epoch": 93.40344168260039, "grad_norm": 0.2760385572910309, "learning_rate": 5.602106797877883e-05, "loss": 0.0109, "step": 146550 }, { "epoch": 93.40981516889738, "grad_norm": 0.22825536131858826, "learning_rate": 5.601559648453416e-05, "loss": 0.0123, "step": 146560 }, { "epoch": 93.41618865519439, "grad_norm": 0.3017338216304779, "learning_rate": 5.6010124917194306e-05, "loss": 0.0172, "step": 146570 }, { "epoch": 93.4225621414914, "grad_norm": 0.2905019521713257, "learning_rate": 5.600465327682576e-05, "loss": 0.0241, "step": 146580 }, { "epoch": 93.4289356277884, "grad_norm": 0.21306301653385162, "learning_rate": 5.599918156349501e-05, "loss": 0.0157, "step": 146590 }, { "epoch": 93.4353091140854, "grad_norm": 0.37282463908195496, "learning_rate": 5.599370977726852e-05, "loss": 0.0119, "step": 146600 }, { "epoch": 93.4416826003824, "grad_norm": 0.19169357419013977, "learning_rate": 5.5988237918212814e-05, "loss": 0.0162, "step": 146610 }, { "epoch": 93.44805608667941, "grad_norm": 0.2760481536388397, "learning_rate": 5.5982765986394356e-05, "loss": 0.0109, "step": 146620 }, { "epoch": 93.45442957297642, "grad_norm": 0.3691377639770508, "learning_rate": 5.5977293981879655e-05, "loss": 0.0267, "step": 146630 }, { "epoch": 93.46080305927342, "grad_norm": 0.26188427209854126, "learning_rate": 5.597182190473518e-05, "loss": 0.0108, "step": 146640 }, { "epoch": 93.46717654557042, "grad_norm": 0.6165068745613098, "learning_rate": 5.5966349755027436e-05, "loss": 0.0154, "step": 146650 }, { "epoch": 93.47355003186743, "grad_norm": 0.2852493226528168, "learning_rate": 5.596087753282291e-05, "loss": 0.0123, "step": 146660 }, { "epoch": 93.47992351816444, "grad_norm": 0.16620545089244843, "learning_rate": 5.595540523818809e-05, "loss": 0.0172, "step": 146670 }, { "epoch": 93.48629700446143, "grad_norm": 0.46186888217926025, "learning_rate": 5.594993287118947e-05, "loss": 0.0234, "step": 146680 }, { "epoch": 93.49267049075844, "grad_norm": 0.5115196108818054, "learning_rate": 5.5944460431893545e-05, "loss": 0.0109, "step": 146690 }, { "epoch": 93.49904397705545, "grad_norm": 0.41353029012680054, "learning_rate": 5.593898792036682e-05, "loss": 0.0189, "step": 146700 }, { "epoch": 93.50541746335246, "grad_norm": 0.5312375426292419, "learning_rate": 5.593351533667578e-05, "loss": 0.0188, "step": 146710 }, { "epoch": 93.51179094964945, "grad_norm": 0.6242886185646057, "learning_rate": 5.592804268088694e-05, "loss": 0.0148, "step": 146720 }, { "epoch": 93.51816443594646, "grad_norm": 0.3480576276779175, "learning_rate": 5.5922569953066775e-05, "loss": 0.0246, "step": 146730 }, { "epoch": 93.52453792224347, "grad_norm": 0.5602370500564575, "learning_rate": 5.5917097153281785e-05, "loss": 0.0177, "step": 146740 }, { "epoch": 93.53091140854048, "grad_norm": 0.303656667470932, "learning_rate": 5.5911624281598496e-05, "loss": 0.0151, "step": 146750 }, { "epoch": 93.53728489483747, "grad_norm": 0.27307626605033875, "learning_rate": 5.5906151338083365e-05, "loss": 0.0118, "step": 146760 }, { "epoch": 93.54365838113448, "grad_norm": 0.492296040058136, "learning_rate": 5.5900678322802946e-05, "loss": 0.016, "step": 146770 }, { "epoch": 93.55003186743149, "grad_norm": 0.2231941819190979, "learning_rate": 5.589520523582369e-05, "loss": 0.0109, "step": 146780 }, { "epoch": 93.5564053537285, "grad_norm": 0.2319934070110321, "learning_rate": 5.5889732077212144e-05, "loss": 0.0125, "step": 146790 }, { "epoch": 93.56277884002549, "grad_norm": 0.33493542671203613, "learning_rate": 5.588425884703478e-05, "loss": 0.0218, "step": 146800 }, { "epoch": 93.5691523263225, "grad_norm": 0.204631507396698, "learning_rate": 5.587878554535813e-05, "loss": 0.0093, "step": 146810 }, { "epoch": 93.5755258126195, "grad_norm": 0.26580896973609924, "learning_rate": 5.587331217224867e-05, "loss": 0.0133, "step": 146820 }, { "epoch": 93.58189929891651, "grad_norm": 0.3853384852409363, "learning_rate": 5.586783872777292e-05, "loss": 0.0125, "step": 146830 }, { "epoch": 93.5882727852135, "grad_norm": 0.2811070382595062, "learning_rate": 5.586236521199739e-05, "loss": 0.0077, "step": 146840 }, { "epoch": 93.59464627151051, "grad_norm": 0.19772961735725403, "learning_rate": 5.5856891624988606e-05, "loss": 0.0141, "step": 146850 }, { "epoch": 93.60101975780752, "grad_norm": 0.27188682556152344, "learning_rate": 5.585141796681305e-05, "loss": 0.0147, "step": 146860 }, { "epoch": 93.60739324410453, "grad_norm": 0.4893603026866913, "learning_rate": 5.584594423753724e-05, "loss": 0.0307, "step": 146870 }, { "epoch": 93.61376673040154, "grad_norm": 0.3177490234375, "learning_rate": 5.5840470437227673e-05, "loss": 0.009, "step": 146880 }, { "epoch": 93.62014021669853, "grad_norm": 0.5333943963050842, "learning_rate": 5.583499656595089e-05, "loss": 0.008, "step": 146890 }, { "epoch": 93.62651370299554, "grad_norm": 0.5907066464424133, "learning_rate": 5.582952262377339e-05, "loss": 0.0128, "step": 146900 }, { "epoch": 93.63288718929255, "grad_norm": 0.3174366354942322, "learning_rate": 5.582404861076169e-05, "loss": 0.0153, "step": 146910 }, { "epoch": 93.63926067558955, "grad_norm": 0.44515666365623474, "learning_rate": 5.581857452698229e-05, "loss": 0.018, "step": 146920 }, { "epoch": 93.64563416188655, "grad_norm": 0.324459969997406, "learning_rate": 5.581310037250173e-05, "loss": 0.014, "step": 146930 }, { "epoch": 93.65200764818356, "grad_norm": 0.2650090754032135, "learning_rate": 5.580762614738649e-05, "loss": 0.0113, "step": 146940 }, { "epoch": 93.65838113448056, "grad_norm": 0.22670666873455048, "learning_rate": 5.580215185170313e-05, "loss": 0.0203, "step": 146950 }, { "epoch": 93.66475462077757, "grad_norm": 0.3930833041667938, "learning_rate": 5.579667748551813e-05, "loss": 0.0096, "step": 146960 }, { "epoch": 93.67112810707457, "grad_norm": 0.3619782328605652, "learning_rate": 5.5791203048898043e-05, "loss": 0.0137, "step": 146970 }, { "epoch": 93.67750159337157, "grad_norm": 0.39673498272895813, "learning_rate": 5.5785728541909366e-05, "loss": 0.0194, "step": 146980 }, { "epoch": 93.68387507966858, "grad_norm": 0.4701694846153259, "learning_rate": 5.578025396461861e-05, "loss": 0.0148, "step": 146990 }, { "epoch": 93.69024856596559, "grad_norm": 0.19050949811935425, "learning_rate": 5.577477931709233e-05, "loss": 0.0132, "step": 147000 }, { "epoch": 93.69662205226258, "grad_norm": 0.35949671268463135, "learning_rate": 5.576930459939702e-05, "loss": 0.0201, "step": 147010 }, { "epoch": 93.70299553855959, "grad_norm": 0.19395531713962555, "learning_rate": 5.576382981159921e-05, "loss": 0.0159, "step": 147020 }, { "epoch": 93.7093690248566, "grad_norm": 0.4551393687725067, "learning_rate": 5.5758354953765426e-05, "loss": 0.0269, "step": 147030 }, { "epoch": 93.7157425111536, "grad_norm": 0.5933775305747986, "learning_rate": 5.575288002596221e-05, "loss": 0.0144, "step": 147040 }, { "epoch": 93.7221159974506, "grad_norm": 0.40691158175468445, "learning_rate": 5.574740502825604e-05, "loss": 0.0161, "step": 147050 }, { "epoch": 93.72848948374761, "grad_norm": 0.3654322028160095, "learning_rate": 5.57419299607135e-05, "loss": 0.0125, "step": 147060 }, { "epoch": 93.73486297004462, "grad_norm": 0.33925312757492065, "learning_rate": 5.573645482340107e-05, "loss": 0.0199, "step": 147070 }, { "epoch": 93.74123645634162, "grad_norm": 0.3975124955177307, "learning_rate": 5.57309796163853e-05, "loss": 0.0196, "step": 147080 }, { "epoch": 93.74760994263862, "grad_norm": 0.280232310295105, "learning_rate": 5.5725504339732736e-05, "loss": 0.0111, "step": 147090 }, { "epoch": 93.75398342893563, "grad_norm": 0.23695294559001923, "learning_rate": 5.5720028993509865e-05, "loss": 0.0106, "step": 147100 }, { "epoch": 93.76035691523263, "grad_norm": 0.23087525367736816, "learning_rate": 5.571455357778326e-05, "loss": 0.0105, "step": 147110 }, { "epoch": 93.76673040152964, "grad_norm": 0.43688225746154785, "learning_rate": 5.570907809261942e-05, "loss": 0.0114, "step": 147120 }, { "epoch": 93.77310388782664, "grad_norm": 0.3735135793685913, "learning_rate": 5.5703602538084895e-05, "loss": 0.0199, "step": 147130 }, { "epoch": 93.77947737412364, "grad_norm": 0.2742200493812561, "learning_rate": 5.569812691424622e-05, "loss": 0.009, "step": 147140 }, { "epoch": 93.78585086042065, "grad_norm": 0.163246750831604, "learning_rate": 5.5692651221169914e-05, "loss": 0.0106, "step": 147150 }, { "epoch": 93.79222434671766, "grad_norm": 0.2281038612127304, "learning_rate": 5.568717545892253e-05, "loss": 0.0173, "step": 147160 }, { "epoch": 93.79859783301465, "grad_norm": 0.2654304504394531, "learning_rate": 5.568169962757058e-05, "loss": 0.0113, "step": 147170 }, { "epoch": 93.80497131931166, "grad_norm": 0.2713688313961029, "learning_rate": 5.5676223727180635e-05, "loss": 0.0082, "step": 147180 }, { "epoch": 93.81134480560867, "grad_norm": 0.42497652769088745, "learning_rate": 5.56707477578192e-05, "loss": 0.01, "step": 147190 }, { "epoch": 93.81771829190568, "grad_norm": 0.2189909666776657, "learning_rate": 5.5665271719552826e-05, "loss": 0.0177, "step": 147200 }, { "epoch": 93.82409177820267, "grad_norm": 0.1999620646238327, "learning_rate": 5.565979561244805e-05, "loss": 0.0178, "step": 147210 }, { "epoch": 93.83046526449968, "grad_norm": 0.30896949768066406, "learning_rate": 5.5654319436571425e-05, "loss": 0.0122, "step": 147220 }, { "epoch": 93.83683875079669, "grad_norm": 0.13463763892650604, "learning_rate": 5.5648843191989476e-05, "loss": 0.0142, "step": 147230 }, { "epoch": 93.8432122370937, "grad_norm": 0.36026543378829956, "learning_rate": 5.5643366878768734e-05, "loss": 0.0213, "step": 147240 }, { "epoch": 93.84958572339069, "grad_norm": 0.26850783824920654, "learning_rate": 5.5637890496975776e-05, "loss": 0.015, "step": 147250 }, { "epoch": 93.8559592096877, "grad_norm": 0.5835054516792297, "learning_rate": 5.5632414046677104e-05, "loss": 0.0162, "step": 147260 }, { "epoch": 93.8623326959847, "grad_norm": 0.2554718852043152, "learning_rate": 5.562693752793931e-05, "loss": 0.0179, "step": 147270 }, { "epoch": 93.86870618228171, "grad_norm": 0.22960048913955688, "learning_rate": 5.562146094082888e-05, "loss": 0.0223, "step": 147280 }, { "epoch": 93.8750796685787, "grad_norm": 0.4155656099319458, "learning_rate": 5.5615984285412414e-05, "loss": 0.0252, "step": 147290 }, { "epoch": 93.88145315487571, "grad_norm": 0.3126201629638672, "learning_rate": 5.561050756175644e-05, "loss": 0.0128, "step": 147300 }, { "epoch": 93.88782664117272, "grad_norm": 0.4329872727394104, "learning_rate": 5.560503076992749e-05, "loss": 0.0155, "step": 147310 }, { "epoch": 93.89420012746973, "grad_norm": 0.3621690273284912, "learning_rate": 5.559955390999213e-05, "loss": 0.0121, "step": 147320 }, { "epoch": 93.90057361376672, "grad_norm": 0.3472231328487396, "learning_rate": 5.5594076982016895e-05, "loss": 0.0103, "step": 147330 }, { "epoch": 93.90694710006373, "grad_norm": 0.26024600863456726, "learning_rate": 5.558859998606836e-05, "loss": 0.0133, "step": 147340 }, { "epoch": 93.91332058636074, "grad_norm": 0.2420089989900589, "learning_rate": 5.558312292221304e-05, "loss": 0.0101, "step": 147350 }, { "epoch": 93.91969407265775, "grad_norm": 0.5097218155860901, "learning_rate": 5.557764579051752e-05, "loss": 0.0235, "step": 147360 }, { "epoch": 93.92606755895474, "grad_norm": 0.3041343688964844, "learning_rate": 5.5572168591048324e-05, "loss": 0.0125, "step": 147370 }, { "epoch": 93.93244104525175, "grad_norm": 0.4220650792121887, "learning_rate": 5.556669132387203e-05, "loss": 0.0155, "step": 147380 }, { "epoch": 93.93881453154876, "grad_norm": 0.3493608832359314, "learning_rate": 5.556121398905518e-05, "loss": 0.0163, "step": 147390 }, { "epoch": 93.94518801784577, "grad_norm": 0.33359768986701965, "learning_rate": 5.555573658666432e-05, "loss": 0.0133, "step": 147400 }, { "epoch": 93.95156150414276, "grad_norm": 0.415194571018219, "learning_rate": 5.5550259116766037e-05, "loss": 0.0152, "step": 147410 }, { "epoch": 93.95793499043977, "grad_norm": 0.4668678939342499, "learning_rate": 5.554478157942684e-05, "loss": 0.0255, "step": 147420 }, { "epoch": 93.96430847673678, "grad_norm": 0.28652423620224, "learning_rate": 5.553930397471332e-05, "loss": 0.011, "step": 147430 }, { "epoch": 93.97068196303378, "grad_norm": 0.3434980809688568, "learning_rate": 5.5533826302692026e-05, "loss": 0.0158, "step": 147440 }, { "epoch": 93.97705544933078, "grad_norm": 0.34546107053756714, "learning_rate": 5.5528348563429524e-05, "loss": 0.0134, "step": 147450 }, { "epoch": 93.98342893562778, "grad_norm": 0.26357752084732056, "learning_rate": 5.5522870756992364e-05, "loss": 0.0089, "step": 147460 }, { "epoch": 93.9898024219248, "grad_norm": 0.35534146428108215, "learning_rate": 5.551739288344711e-05, "loss": 0.0127, "step": 147470 }, { "epoch": 93.9961759082218, "grad_norm": 0.3244878947734833, "learning_rate": 5.5511914942860323e-05, "loss": 0.0243, "step": 147480 }, { "epoch": 94.00254939451881, "grad_norm": 0.19800378382205963, "learning_rate": 5.550643693529856e-05, "loss": 0.0055, "step": 147490 }, { "epoch": 94.0089228808158, "grad_norm": 0.23249919712543488, "learning_rate": 5.550095886082839e-05, "loss": 0.0103, "step": 147500 }, { "epoch": 94.01529636711281, "grad_norm": 0.1489308774471283, "learning_rate": 5.549548071951638e-05, "loss": 0.0109, "step": 147510 }, { "epoch": 94.02166985340982, "grad_norm": 0.2706194221973419, "learning_rate": 5.549000251142909e-05, "loss": 0.0188, "step": 147520 }, { "epoch": 94.02804333970683, "grad_norm": 0.25989216566085815, "learning_rate": 5.548452423663309e-05, "loss": 0.0117, "step": 147530 }, { "epoch": 94.03441682600382, "grad_norm": 0.4431345462799072, "learning_rate": 5.5479045895194934e-05, "loss": 0.0165, "step": 147540 }, { "epoch": 94.04079031230083, "grad_norm": 0.22554750740528107, "learning_rate": 5.54735674871812e-05, "loss": 0.0087, "step": 147550 }, { "epoch": 94.04716379859784, "grad_norm": 0.20739416778087616, "learning_rate": 5.5468089012658455e-05, "loss": 0.0109, "step": 147560 }, { "epoch": 94.05353728489484, "grad_norm": 0.3964749872684479, "learning_rate": 5.5462610471693255e-05, "loss": 0.0091, "step": 147570 }, { "epoch": 94.05991077119184, "grad_norm": 0.3225475251674652, "learning_rate": 5.5457131864352196e-05, "loss": 0.017, "step": 147580 }, { "epoch": 94.06628425748885, "grad_norm": 0.19156919419765472, "learning_rate": 5.5451653190701826e-05, "loss": 0.0245, "step": 147590 }, { "epoch": 94.07265774378585, "grad_norm": 0.3057920038700104, "learning_rate": 5.544617445080873e-05, "loss": 0.0127, "step": 147600 }, { "epoch": 94.07903123008286, "grad_norm": 0.2883702218532562, "learning_rate": 5.544069564473946e-05, "loss": 0.0065, "step": 147610 }, { "epoch": 94.08540471637986, "grad_norm": 0.22612227499485016, "learning_rate": 5.54352167725606e-05, "loss": 0.0113, "step": 147620 }, { "epoch": 94.09177820267686, "grad_norm": 0.4367998242378235, "learning_rate": 5.542973783433873e-05, "loss": 0.0109, "step": 147630 }, { "epoch": 94.09815168897387, "grad_norm": 0.26230156421661377, "learning_rate": 5.542425883014043e-05, "loss": 0.0154, "step": 147640 }, { "epoch": 94.10452517527088, "grad_norm": 0.2477300465106964, "learning_rate": 5.541877976003225e-05, "loss": 0.0152, "step": 147650 }, { "epoch": 94.11089866156787, "grad_norm": 0.32599782943725586, "learning_rate": 5.541330062408079e-05, "loss": 0.0171, "step": 147660 }, { "epoch": 94.11727214786488, "grad_norm": 0.3691728413105011, "learning_rate": 5.54078214223526e-05, "loss": 0.0117, "step": 147670 }, { "epoch": 94.12364563416189, "grad_norm": 0.2239648550748825, "learning_rate": 5.540234215491429e-05, "loss": 0.017, "step": 147680 }, { "epoch": 94.1300191204589, "grad_norm": 0.39215850830078125, "learning_rate": 5.539686282183242e-05, "loss": 0.014, "step": 147690 }, { "epoch": 94.13639260675589, "grad_norm": 0.3220116198062897, "learning_rate": 5.539138342317357e-05, "loss": 0.0158, "step": 147700 }, { "epoch": 94.1427660930529, "grad_norm": 0.35392430424690247, "learning_rate": 5.5385903959004316e-05, "loss": 0.0161, "step": 147710 }, { "epoch": 94.1491395793499, "grad_norm": 1.4880084991455078, "learning_rate": 5.5380424429391245e-05, "loss": 0.0182, "step": 147720 }, { "epoch": 94.15551306564691, "grad_norm": 0.7110500335693359, "learning_rate": 5.5374944834400946e-05, "loss": 0.03, "step": 147730 }, { "epoch": 94.16188655194391, "grad_norm": 0.27056410908699036, "learning_rate": 5.5369465174099975e-05, "loss": 0.0086, "step": 147740 }, { "epoch": 94.16826003824092, "grad_norm": 0.24239632487297058, "learning_rate": 5.536398544855494e-05, "loss": 0.017, "step": 147750 }, { "epoch": 94.17463352453792, "grad_norm": 0.21519123017787933, "learning_rate": 5.535850565783243e-05, "loss": 0.0139, "step": 147760 }, { "epoch": 94.18100701083493, "grad_norm": 0.36156922578811646, "learning_rate": 5.5353025801999e-05, "loss": 0.0099, "step": 147770 }, { "epoch": 94.18738049713193, "grad_norm": 0.41914102435112, "learning_rate": 5.534754588112127e-05, "loss": 0.0129, "step": 147780 }, { "epoch": 94.19375398342893, "grad_norm": 0.2641972303390503, "learning_rate": 5.534206589526579e-05, "loss": 0.0102, "step": 147790 }, { "epoch": 94.20012746972594, "grad_norm": 0.4788120985031128, "learning_rate": 5.533658584449918e-05, "loss": 0.0211, "step": 147800 }, { "epoch": 94.20650095602295, "grad_norm": 0.7230458855628967, "learning_rate": 5.5331105728888e-05, "loss": 0.0083, "step": 147810 }, { "epoch": 94.21287444231994, "grad_norm": 0.25670725107192993, "learning_rate": 5.532562554849886e-05, "loss": 0.0137, "step": 147820 }, { "epoch": 94.21924792861695, "grad_norm": 0.27952322363853455, "learning_rate": 5.532014530339834e-05, "loss": 0.0142, "step": 147830 }, { "epoch": 94.22562141491396, "grad_norm": 0.3252853751182556, "learning_rate": 5.531466499365303e-05, "loss": 0.0115, "step": 147840 }, { "epoch": 94.23199490121097, "grad_norm": 0.130824014544487, "learning_rate": 5.5309184619329524e-05, "loss": 0.0151, "step": 147850 }, { "epoch": 94.23836838750796, "grad_norm": 1.048818588256836, "learning_rate": 5.530370418049442e-05, "loss": 0.0129, "step": 147860 }, { "epoch": 94.24474187380497, "grad_norm": 0.28404906392097473, "learning_rate": 5.52982236772143e-05, "loss": 0.0143, "step": 147870 }, { "epoch": 94.25111536010198, "grad_norm": 0.4501838684082031, "learning_rate": 5.529274310955575e-05, "loss": 0.014, "step": 147880 }, { "epoch": 94.25748884639899, "grad_norm": 0.28795942664146423, "learning_rate": 5.5287262477585376e-05, "loss": 0.0115, "step": 147890 }, { "epoch": 94.26386233269598, "grad_norm": 0.3622882068157196, "learning_rate": 5.528178178136978e-05, "loss": 0.0103, "step": 147900 }, { "epoch": 94.27023581899299, "grad_norm": 0.3344901204109192, "learning_rate": 5.527630102097554e-05, "loss": 0.0179, "step": 147910 }, { "epoch": 94.27660930529, "grad_norm": 0.5198674201965332, "learning_rate": 5.5270820196469266e-05, "loss": 0.0235, "step": 147920 }, { "epoch": 94.282982791587, "grad_norm": 0.7023051977157593, "learning_rate": 5.526533930791755e-05, "loss": 0.0125, "step": 147930 }, { "epoch": 94.289356277884, "grad_norm": 0.5925998687744141, "learning_rate": 5.525985835538699e-05, "loss": 0.0137, "step": 147940 }, { "epoch": 94.295729764181, "grad_norm": 0.20604760944843292, "learning_rate": 5.525437733894417e-05, "loss": 0.0185, "step": 147950 }, { "epoch": 94.30210325047801, "grad_norm": 0.1374204456806183, "learning_rate": 5.524889625865572e-05, "loss": 0.0128, "step": 147960 }, { "epoch": 94.30847673677502, "grad_norm": 0.26614484190940857, "learning_rate": 5.5243415114588224e-05, "loss": 0.016, "step": 147970 }, { "epoch": 94.31485022307201, "grad_norm": 0.5109741687774658, "learning_rate": 5.523793390680829e-05, "loss": 0.0198, "step": 147980 }, { "epoch": 94.32122370936902, "grad_norm": 0.44019779562950134, "learning_rate": 5.523245263538249e-05, "loss": 0.0077, "step": 147990 }, { "epoch": 94.32759719566603, "grad_norm": 0.7418987154960632, "learning_rate": 5.5226971300377486e-05, "loss": 0.0116, "step": 148000 }, { "epoch": 94.33397068196304, "grad_norm": 0.5081104636192322, "learning_rate": 5.5221489901859814e-05, "loss": 0.0207, "step": 148010 }, { "epoch": 94.34034416826003, "grad_norm": 0.19987323880195618, "learning_rate": 5.521600843989613e-05, "loss": 0.0111, "step": 148020 }, { "epoch": 94.34671765455704, "grad_norm": 0.33192336559295654, "learning_rate": 5.521052691455301e-05, "loss": 0.0125, "step": 148030 }, { "epoch": 94.35309114085405, "grad_norm": 0.4478926360607147, "learning_rate": 5.5205045325897065e-05, "loss": 0.0138, "step": 148040 }, { "epoch": 94.35946462715106, "grad_norm": 0.22463195025920868, "learning_rate": 5.51995636739949e-05, "loss": 0.0205, "step": 148050 }, { "epoch": 94.36583811344805, "grad_norm": 0.20886953175067902, "learning_rate": 5.5194081958913134e-05, "loss": 0.0125, "step": 148060 }, { "epoch": 94.37221159974506, "grad_norm": 0.19657589495182037, "learning_rate": 5.518860018071838e-05, "loss": 0.0118, "step": 148070 }, { "epoch": 94.37858508604207, "grad_norm": 0.385936975479126, "learning_rate": 5.518311833947721e-05, "loss": 0.0132, "step": 148080 }, { "epoch": 94.38495857233907, "grad_norm": 0.5278751850128174, "learning_rate": 5.5177636435256286e-05, "loss": 0.0166, "step": 148090 }, { "epoch": 94.39133205863607, "grad_norm": 0.21838858723640442, "learning_rate": 5.517215446812217e-05, "loss": 0.0079, "step": 148100 }, { "epoch": 94.39770554493307, "grad_norm": 0.2558227777481079, "learning_rate": 5.516667243814151e-05, "loss": 0.0196, "step": 148110 }, { "epoch": 94.40407903123008, "grad_norm": 0.4287673234939575, "learning_rate": 5.51611903453809e-05, "loss": 0.0192, "step": 148120 }, { "epoch": 94.41045251752709, "grad_norm": 0.3997235894203186, "learning_rate": 5.515570818990693e-05, "loss": 0.011, "step": 148130 }, { "epoch": 94.4168260038241, "grad_norm": 0.30725377798080444, "learning_rate": 5.5150225971786265e-05, "loss": 0.0234, "step": 148140 }, { "epoch": 94.42319949012109, "grad_norm": 0.397394597530365, "learning_rate": 5.514474369108548e-05, "loss": 0.0095, "step": 148150 }, { "epoch": 94.4295729764181, "grad_norm": 0.14131593704223633, "learning_rate": 5.513926134787121e-05, "loss": 0.01, "step": 148160 }, { "epoch": 94.43594646271511, "grad_norm": 0.34349629282951355, "learning_rate": 5.5133778942210045e-05, "loss": 0.0121, "step": 148170 }, { "epoch": 94.44231994901212, "grad_norm": 0.469112753868103, "learning_rate": 5.512829647416864e-05, "loss": 0.0159, "step": 148180 }, { "epoch": 94.44869343530911, "grad_norm": 0.3486471474170685, "learning_rate": 5.512281394381358e-05, "loss": 0.0215, "step": 148190 }, { "epoch": 94.45506692160612, "grad_norm": 0.2198377251625061, "learning_rate": 5.5117331351211485e-05, "loss": 0.0072, "step": 148200 }, { "epoch": 94.46144040790313, "grad_norm": 0.46365290880203247, "learning_rate": 5.5111848696429e-05, "loss": 0.0193, "step": 148210 }, { "epoch": 94.46781389420013, "grad_norm": 0.2521294355392456, "learning_rate": 5.5106365979532714e-05, "loss": 0.0113, "step": 148220 }, { "epoch": 94.47418738049713, "grad_norm": 0.49250829219818115, "learning_rate": 5.5100883200589274e-05, "loss": 0.0123, "step": 148230 }, { "epoch": 94.48056086679414, "grad_norm": 0.2167847901582718, "learning_rate": 5.5095400359665274e-05, "loss": 0.007, "step": 148240 }, { "epoch": 94.48693435309114, "grad_norm": 0.1605955809354782, "learning_rate": 5.5089917456827355e-05, "loss": 0.0225, "step": 148250 }, { "epoch": 94.49330783938815, "grad_norm": 0.23764413595199585, "learning_rate": 5.5084434492142125e-05, "loss": 0.0091, "step": 148260 }, { "epoch": 94.49968132568515, "grad_norm": 0.28203296661376953, "learning_rate": 5.507895146567622e-05, "loss": 0.0117, "step": 148270 }, { "epoch": 94.50605481198215, "grad_norm": 0.2948940396308899, "learning_rate": 5.5073468377496265e-05, "loss": 0.0133, "step": 148280 }, { "epoch": 94.51242829827916, "grad_norm": 0.5270401835441589, "learning_rate": 5.506798522766887e-05, "loss": 0.0114, "step": 148290 }, { "epoch": 94.51880178457617, "grad_norm": 0.31761911511421204, "learning_rate": 5.506250201626068e-05, "loss": 0.0119, "step": 148300 }, { "epoch": 94.52517527087316, "grad_norm": 0.4103859066963196, "learning_rate": 5.505701874333831e-05, "loss": 0.0102, "step": 148310 }, { "epoch": 94.53154875717017, "grad_norm": 0.2289949208498001, "learning_rate": 5.5051535408968366e-05, "loss": 0.0186, "step": 148320 }, { "epoch": 94.53792224346718, "grad_norm": 0.28073060512542725, "learning_rate": 5.504605201321752e-05, "loss": 0.0158, "step": 148330 }, { "epoch": 94.54429572976419, "grad_norm": 0.3429391384124756, "learning_rate": 5.504056855615237e-05, "loss": 0.019, "step": 148340 }, { "epoch": 94.55066921606118, "grad_norm": 0.20877911150455475, "learning_rate": 5.503508503783956e-05, "loss": 0.0173, "step": 148350 }, { "epoch": 94.55704270235819, "grad_norm": 0.29286444187164307, "learning_rate": 5.5029601458345705e-05, "loss": 0.0172, "step": 148360 }, { "epoch": 94.5634161886552, "grad_norm": 0.31284311413764954, "learning_rate": 5.502411781773745e-05, "loss": 0.0107, "step": 148370 }, { "epoch": 94.5697896749522, "grad_norm": 0.25922414660453796, "learning_rate": 5.5018634116081415e-05, "loss": 0.0125, "step": 148380 }, { "epoch": 94.5761631612492, "grad_norm": 0.452631413936615, "learning_rate": 5.501315035344423e-05, "loss": 0.0214, "step": 148390 }, { "epoch": 94.5825366475462, "grad_norm": 0.19491618871688843, "learning_rate": 5.500766652989254e-05, "loss": 0.0097, "step": 148400 }, { "epoch": 94.58891013384321, "grad_norm": 0.3739275336265564, "learning_rate": 5.500218264549298e-05, "loss": 0.0088, "step": 148410 }, { "epoch": 94.59528362014022, "grad_norm": 0.6323074102401733, "learning_rate": 5.499669870031218e-05, "loss": 0.0219, "step": 148420 }, { "epoch": 94.60165710643722, "grad_norm": 0.6122003197669983, "learning_rate": 5.499121469441677e-05, "loss": 0.0129, "step": 148430 }, { "epoch": 94.60803059273422, "grad_norm": 0.46616896986961365, "learning_rate": 5.498573062787339e-05, "loss": 0.0151, "step": 148440 }, { "epoch": 94.61440407903123, "grad_norm": 0.44036102294921875, "learning_rate": 5.4980246500748665e-05, "loss": 0.0138, "step": 148450 }, { "epoch": 94.62077756532824, "grad_norm": 0.5817452073097229, "learning_rate": 5.497476231310926e-05, "loss": 0.0167, "step": 148460 }, { "epoch": 94.62715105162523, "grad_norm": 0.6669417023658752, "learning_rate": 5.496927806502178e-05, "loss": 0.0173, "step": 148470 }, { "epoch": 94.63352453792224, "grad_norm": 0.2698137164115906, "learning_rate": 5.496379375655289e-05, "loss": 0.0121, "step": 148480 }, { "epoch": 94.63989802421925, "grad_norm": 0.22615528106689453, "learning_rate": 5.495830938776921e-05, "loss": 0.0124, "step": 148490 }, { "epoch": 94.64627151051626, "grad_norm": 0.18765072524547577, "learning_rate": 5.49528249587374e-05, "loss": 0.0178, "step": 148500 }, { "epoch": 94.65264499681325, "grad_norm": 0.22492346167564392, "learning_rate": 5.494734046952409e-05, "loss": 0.0142, "step": 148510 }, { "epoch": 94.65901848311026, "grad_norm": 0.44900962710380554, "learning_rate": 5.494185592019592e-05, "loss": 0.0138, "step": 148520 }, { "epoch": 94.66539196940727, "grad_norm": 0.4493436813354492, "learning_rate": 5.493637131081953e-05, "loss": 0.0142, "step": 148530 }, { "epoch": 94.67176545570427, "grad_norm": 0.37618401646614075, "learning_rate": 5.493088664146156e-05, "loss": 0.0149, "step": 148540 }, { "epoch": 94.67813894200127, "grad_norm": 0.45867666602134705, "learning_rate": 5.492540191218869e-05, "loss": 0.0182, "step": 148550 }, { "epoch": 94.68451242829828, "grad_norm": 0.4445740580558777, "learning_rate": 5.491991712306751e-05, "loss": 0.0127, "step": 148560 }, { "epoch": 94.69088591459528, "grad_norm": 0.28907260298728943, "learning_rate": 5.491443227416471e-05, "loss": 0.0174, "step": 148570 }, { "epoch": 94.69725940089229, "grad_norm": 0.35643860697746277, "learning_rate": 5.4908947365546895e-05, "loss": 0.0249, "step": 148580 }, { "epoch": 94.70363288718929, "grad_norm": 0.19916582107543945, "learning_rate": 5.4903462397280756e-05, "loss": 0.0204, "step": 148590 }, { "epoch": 94.7100063734863, "grad_norm": 0.3528454303741455, "learning_rate": 5.489797736943292e-05, "loss": 0.0191, "step": 148600 }, { "epoch": 94.7163798597833, "grad_norm": 0.5787343978881836, "learning_rate": 5.4892492282070016e-05, "loss": 0.0104, "step": 148610 }, { "epoch": 94.72275334608031, "grad_norm": 0.36331266164779663, "learning_rate": 5.4887007135258724e-05, "loss": 0.015, "step": 148620 }, { "epoch": 94.7291268323773, "grad_norm": 0.34050503373146057, "learning_rate": 5.488152192906567e-05, "loss": 0.0139, "step": 148630 }, { "epoch": 94.73550031867431, "grad_norm": 0.3950363099575043, "learning_rate": 5.4876036663557526e-05, "loss": 0.0124, "step": 148640 }, { "epoch": 94.74187380497132, "grad_norm": 0.22955359518527985, "learning_rate": 5.4870551338800936e-05, "loss": 0.0115, "step": 148650 }, { "epoch": 94.74824729126833, "grad_norm": 0.2201840728521347, "learning_rate": 5.486506595486254e-05, "loss": 0.0107, "step": 148660 }, { "epoch": 94.75462077756532, "grad_norm": 0.26805540919303894, "learning_rate": 5.485958051180902e-05, "loss": 0.0131, "step": 148670 }, { "epoch": 94.76099426386233, "grad_norm": 0.4261666238307953, "learning_rate": 5.485409500970697e-05, "loss": 0.0091, "step": 148680 }, { "epoch": 94.76736775015934, "grad_norm": 0.3640148341655731, "learning_rate": 5.484860944862311e-05, "loss": 0.0175, "step": 148690 }, { "epoch": 94.77374123645635, "grad_norm": 0.46711912751197815, "learning_rate": 5.484312382862407e-05, "loss": 0.0173, "step": 148700 }, { "epoch": 94.78011472275334, "grad_norm": 0.22272107005119324, "learning_rate": 5.4837638149776495e-05, "loss": 0.0198, "step": 148710 }, { "epoch": 94.78648820905035, "grad_norm": 0.19235128164291382, "learning_rate": 5.4832152412147044e-05, "loss": 0.0137, "step": 148720 }, { "epoch": 94.79286169534736, "grad_norm": 0.4468098282814026, "learning_rate": 5.482666661580238e-05, "loss": 0.0175, "step": 148730 }, { "epoch": 94.79923518164436, "grad_norm": 0.341447114944458, "learning_rate": 5.4821180760809155e-05, "loss": 0.0157, "step": 148740 }, { "epoch": 94.80560866794136, "grad_norm": 0.3378106951713562, "learning_rate": 5.481569484723405e-05, "loss": 0.0167, "step": 148750 }, { "epoch": 94.81198215423836, "grad_norm": 0.47642233967781067, "learning_rate": 5.4810208875143695e-05, "loss": 0.0195, "step": 148760 }, { "epoch": 94.81835564053537, "grad_norm": 0.20794612169265747, "learning_rate": 5.480472284460476e-05, "loss": 0.0101, "step": 148770 }, { "epoch": 94.82472912683238, "grad_norm": 0.279787540435791, "learning_rate": 5.4799236755683914e-05, "loss": 0.0198, "step": 148780 }, { "epoch": 94.83110261312939, "grad_norm": 0.2321513146162033, "learning_rate": 5.479375060844779e-05, "loss": 0.0124, "step": 148790 }, { "epoch": 94.83747609942638, "grad_norm": 0.3568318486213684, "learning_rate": 5.47882644029631e-05, "loss": 0.0133, "step": 148800 }, { "epoch": 94.84384958572339, "grad_norm": 0.346884161233902, "learning_rate": 5.4782778139296445e-05, "loss": 0.0139, "step": 148810 }, { "epoch": 94.8502230720204, "grad_norm": 0.2635842561721802, "learning_rate": 5.4777291817514544e-05, "loss": 0.0175, "step": 148820 }, { "epoch": 94.8565965583174, "grad_norm": 0.5264608263969421, "learning_rate": 5.477180543768403e-05, "loss": 0.0146, "step": 148830 }, { "epoch": 94.8629700446144, "grad_norm": 0.692413866519928, "learning_rate": 5.4766318999871566e-05, "loss": 0.0151, "step": 148840 }, { "epoch": 94.86934353091141, "grad_norm": 0.612583339214325, "learning_rate": 5.476083250414383e-05, "loss": 0.0288, "step": 148850 }, { "epoch": 94.87571701720842, "grad_norm": 0.3315301835536957, "learning_rate": 5.475534595056748e-05, "loss": 0.01, "step": 148860 }, { "epoch": 94.88209050350542, "grad_norm": 0.2904511094093323, "learning_rate": 5.474985933920921e-05, "loss": 0.0124, "step": 148870 }, { "epoch": 94.88846398980242, "grad_norm": 0.6543716192245483, "learning_rate": 5.474437267013563e-05, "loss": 0.0086, "step": 148880 }, { "epoch": 94.89483747609943, "grad_norm": 0.3598393499851227, "learning_rate": 5.4738885943413476e-05, "loss": 0.0152, "step": 148890 }, { "epoch": 94.90121096239643, "grad_norm": 0.5844879150390625, "learning_rate": 5.4733399159109356e-05, "loss": 0.0193, "step": 148900 }, { "epoch": 94.90758444869344, "grad_norm": 0.4949098527431488, "learning_rate": 5.4727912317289985e-05, "loss": 0.0127, "step": 148910 }, { "epoch": 94.91395793499044, "grad_norm": 0.26134559512138367, "learning_rate": 5.472242541802202e-05, "loss": 0.0072, "step": 148920 }, { "epoch": 94.92033142128744, "grad_norm": 0.21850579977035522, "learning_rate": 5.4716938461372106e-05, "loss": 0.0192, "step": 148930 }, { "epoch": 94.92670490758445, "grad_norm": 0.6526776552200317, "learning_rate": 5.471145144740695e-05, "loss": 0.0087, "step": 148940 }, { "epoch": 94.93307839388146, "grad_norm": 0.3279213011264801, "learning_rate": 5.47059643761932e-05, "loss": 0.0094, "step": 148950 }, { "epoch": 94.93945188017845, "grad_norm": 0.27727341651916504, "learning_rate": 5.470047724779756e-05, "loss": 0.0114, "step": 148960 }, { "epoch": 94.94582536647546, "grad_norm": 0.2846534848213196, "learning_rate": 5.469499006228665e-05, "loss": 0.0178, "step": 148970 }, { "epoch": 94.95219885277247, "grad_norm": 0.23814576864242554, "learning_rate": 5.468950281972721e-05, "loss": 0.013, "step": 148980 }, { "epoch": 94.95857233906948, "grad_norm": 0.27495288848876953, "learning_rate": 5.468401552018587e-05, "loss": 0.0101, "step": 148990 }, { "epoch": 94.96494582536647, "grad_norm": 0.6820880770683289, "learning_rate": 5.4678528163729326e-05, "loss": 0.0185, "step": 149000 }, { "epoch": 94.97131931166348, "grad_norm": 0.1700764298439026, "learning_rate": 5.467304075042423e-05, "loss": 0.0108, "step": 149010 }, { "epoch": 94.97769279796049, "grad_norm": 0.2720614969730377, "learning_rate": 5.46675532803373e-05, "loss": 0.0161, "step": 149020 }, { "epoch": 94.9840662842575, "grad_norm": 0.23125040531158447, "learning_rate": 5.466206575353518e-05, "loss": 0.0106, "step": 149030 }, { "epoch": 94.99043977055449, "grad_norm": 0.46922576427459717, "learning_rate": 5.465657817008456e-05, "loss": 0.008, "step": 149040 }, { "epoch": 94.9968132568515, "grad_norm": 0.18791848421096802, "learning_rate": 5.4651090530052105e-05, "loss": 0.0124, "step": 149050 }, { "epoch": 95.0031867431485, "grad_norm": 0.376442015171051, "learning_rate": 5.4645602833504526e-05, "loss": 0.014, "step": 149060 }, { "epoch": 95.00956022944551, "grad_norm": 0.27130743861198425, "learning_rate": 5.464011508050848e-05, "loss": 0.0147, "step": 149070 }, { "epoch": 95.0159337157425, "grad_norm": 0.1544101983308792, "learning_rate": 5.4634627271130667e-05, "loss": 0.0102, "step": 149080 }, { "epoch": 95.02230720203951, "grad_norm": 0.2704155147075653, "learning_rate": 5.4629139405437736e-05, "loss": 0.0125, "step": 149090 }, { "epoch": 95.02868068833652, "grad_norm": 0.25516629219055176, "learning_rate": 5.4623651483496404e-05, "loss": 0.0119, "step": 149100 }, { "epoch": 95.03505417463353, "grad_norm": 0.2663826644420624, "learning_rate": 5.4618163505373335e-05, "loss": 0.0093, "step": 149110 }, { "epoch": 95.04142766093052, "grad_norm": 0.30743229389190674, "learning_rate": 5.461267547113522e-05, "loss": 0.0199, "step": 149120 }, { "epoch": 95.04780114722753, "grad_norm": 0.29381394386291504, "learning_rate": 5.4607187380848746e-05, "loss": 0.0166, "step": 149130 }, { "epoch": 95.05417463352454, "grad_norm": 0.28823691606521606, "learning_rate": 5.4601699234580604e-05, "loss": 0.0108, "step": 149140 }, { "epoch": 95.06054811982155, "grad_norm": 0.2896229922771454, "learning_rate": 5.459621103239746e-05, "loss": 0.0086, "step": 149150 }, { "epoch": 95.06692160611854, "grad_norm": 0.2861909568309784, "learning_rate": 5.459072277436601e-05, "loss": 0.0138, "step": 149160 }, { "epoch": 95.07329509241555, "grad_norm": 0.5005630850791931, "learning_rate": 5.4585234460552955e-05, "loss": 0.0093, "step": 149170 }, { "epoch": 95.07966857871256, "grad_norm": 0.36937305331230164, "learning_rate": 5.457974609102497e-05, "loss": 0.007, "step": 149180 }, { "epoch": 95.08604206500956, "grad_norm": 0.30674371123313904, "learning_rate": 5.457425766584874e-05, "loss": 0.0146, "step": 149190 }, { "epoch": 95.09241555130656, "grad_norm": 0.22086775302886963, "learning_rate": 5.456876918509095e-05, "loss": 0.0068, "step": 149200 }, { "epoch": 95.09878903760357, "grad_norm": 0.39696353673934937, "learning_rate": 5.4563280648818325e-05, "loss": 0.011, "step": 149210 }, { "epoch": 95.10516252390057, "grad_norm": 0.35711950063705444, "learning_rate": 5.4557792057097515e-05, "loss": 0.0089, "step": 149220 }, { "epoch": 95.11153601019758, "grad_norm": 0.14046573638916016, "learning_rate": 5.455230340999523e-05, "loss": 0.0187, "step": 149230 }, { "epoch": 95.11790949649458, "grad_norm": 0.36534738540649414, "learning_rate": 5.454681470757818e-05, "loss": 0.0089, "step": 149240 }, { "epoch": 95.12428298279158, "grad_norm": 0.32343366742134094, "learning_rate": 5.4541325949913e-05, "loss": 0.0104, "step": 149250 }, { "epoch": 95.13065646908859, "grad_norm": 0.5546572804450989, "learning_rate": 5.453583713706646e-05, "loss": 0.0161, "step": 149260 }, { "epoch": 95.1370299553856, "grad_norm": 0.32366976141929626, "learning_rate": 5.453034826910519e-05, "loss": 0.0151, "step": 149270 }, { "epoch": 95.1434034416826, "grad_norm": 0.30678093433380127, "learning_rate": 5.4524859346095926e-05, "loss": 0.0145, "step": 149280 }, { "epoch": 95.1497769279796, "grad_norm": 0.3341193199157715, "learning_rate": 5.4519370368105336e-05, "loss": 0.0153, "step": 149290 }, { "epoch": 95.15615041427661, "grad_norm": 0.3049311637878418, "learning_rate": 5.4513881335200144e-05, "loss": 0.0102, "step": 149300 }, { "epoch": 95.16252390057362, "grad_norm": 0.29205048084259033, "learning_rate": 5.4508392247447014e-05, "loss": 0.0157, "step": 149310 }, { "epoch": 95.16889738687061, "grad_norm": 0.23826740682125092, "learning_rate": 5.4502903104912685e-05, "loss": 0.0126, "step": 149320 }, { "epoch": 95.17527087316762, "grad_norm": 0.25184834003448486, "learning_rate": 5.4497413907663816e-05, "loss": 0.0121, "step": 149330 }, { "epoch": 95.18164435946463, "grad_norm": 0.2282467931509018, "learning_rate": 5.4491924655767126e-05, "loss": 0.0102, "step": 149340 }, { "epoch": 95.18801784576164, "grad_norm": 0.3355931341648102, "learning_rate": 5.448643534928931e-05, "loss": 0.0092, "step": 149350 }, { "epoch": 95.19439133205863, "grad_norm": 0.1604497879743576, "learning_rate": 5.448094598829707e-05, "loss": 0.014, "step": 149360 }, { "epoch": 95.20076481835564, "grad_norm": 0.6062396764755249, "learning_rate": 5.447545657285711e-05, "loss": 0.0165, "step": 149370 }, { "epoch": 95.20713830465264, "grad_norm": 0.1915375143289566, "learning_rate": 5.446996710303611e-05, "loss": 0.0066, "step": 149380 }, { "epoch": 95.21351179094965, "grad_norm": 0.29346534609794617, "learning_rate": 5.446447757890081e-05, "loss": 0.0071, "step": 149390 }, { "epoch": 95.21988527724666, "grad_norm": 0.4728187620639801, "learning_rate": 5.4458988000517885e-05, "loss": 0.0151, "step": 149400 }, { "epoch": 95.22625876354365, "grad_norm": 0.1910751610994339, "learning_rate": 5.445349836795404e-05, "loss": 0.0137, "step": 149410 }, { "epoch": 95.23263224984066, "grad_norm": 0.512949526309967, "learning_rate": 5.4448008681275985e-05, "loss": 0.0111, "step": 149420 }, { "epoch": 95.23900573613767, "grad_norm": 0.3026995062828064, "learning_rate": 5.444251894055042e-05, "loss": 0.0118, "step": 149430 }, { "epoch": 95.24537922243468, "grad_norm": 0.3116975426673889, "learning_rate": 5.443702914584408e-05, "loss": 0.0159, "step": 149440 }, { "epoch": 95.25175270873167, "grad_norm": 0.20224547386169434, "learning_rate": 5.443153929722362e-05, "loss": 0.019, "step": 149450 }, { "epoch": 95.25812619502868, "grad_norm": 0.3210320472717285, "learning_rate": 5.4426049394755786e-05, "loss": 0.0143, "step": 149460 }, { "epoch": 95.26449968132569, "grad_norm": 0.36309534311294556, "learning_rate": 5.442055943850727e-05, "loss": 0.0163, "step": 149470 }, { "epoch": 95.2708731676227, "grad_norm": 0.461465060710907, "learning_rate": 5.441506942854478e-05, "loss": 0.0113, "step": 149480 }, { "epoch": 95.27724665391969, "grad_norm": 0.3632928133010864, "learning_rate": 5.4409579364935035e-05, "loss": 0.0127, "step": 149490 }, { "epoch": 95.2836201402167, "grad_norm": 0.31910446286201477, "learning_rate": 5.4404089247744736e-05, "loss": 0.0108, "step": 149500 }, { "epoch": 95.2899936265137, "grad_norm": 0.25607970356941223, "learning_rate": 5.439859907704059e-05, "loss": 0.0158, "step": 149510 }, { "epoch": 95.29636711281071, "grad_norm": 0.20795413851737976, "learning_rate": 5.439310885288932e-05, "loss": 0.0096, "step": 149520 }, { "epoch": 95.30274059910771, "grad_norm": 0.492428719997406, "learning_rate": 5.438761857535762e-05, "loss": 0.0213, "step": 149530 }, { "epoch": 95.30911408540472, "grad_norm": 0.25877267122268677, "learning_rate": 5.438212824451221e-05, "loss": 0.0249, "step": 149540 }, { "epoch": 95.31548757170172, "grad_norm": 0.38214582204818726, "learning_rate": 5.437663786041982e-05, "loss": 0.0176, "step": 149550 }, { "epoch": 95.32186105799873, "grad_norm": 0.34114670753479004, "learning_rate": 5.437114742314714e-05, "loss": 0.0096, "step": 149560 }, { "epoch": 95.32823454429573, "grad_norm": 0.4007178843021393, "learning_rate": 5.436565693276089e-05, "loss": 0.0102, "step": 149570 }, { "epoch": 95.33460803059273, "grad_norm": 0.23746730387210846, "learning_rate": 5.436016638932779e-05, "loss": 0.0101, "step": 149580 }, { "epoch": 95.34098151688974, "grad_norm": 0.40467655658721924, "learning_rate": 5.435467579291454e-05, "loss": 0.0139, "step": 149590 }, { "epoch": 95.34735500318675, "grad_norm": 0.15319301187992096, "learning_rate": 5.434918514358788e-05, "loss": 0.0127, "step": 149600 }, { "epoch": 95.35372848948374, "grad_norm": 0.2824612259864807, "learning_rate": 5.4343694441414507e-05, "loss": 0.024, "step": 149610 }, { "epoch": 95.36010197578075, "grad_norm": 0.3946649432182312, "learning_rate": 5.433820368646115e-05, "loss": 0.0195, "step": 149620 }, { "epoch": 95.36647546207776, "grad_norm": 0.3371158838272095, "learning_rate": 5.433271287879451e-05, "loss": 0.0142, "step": 149630 }, { "epoch": 95.37284894837477, "grad_norm": 0.29002615809440613, "learning_rate": 5.432722201848134e-05, "loss": 0.016, "step": 149640 }, { "epoch": 95.37922243467176, "grad_norm": 0.19360649585723877, "learning_rate": 5.432173110558834e-05, "loss": 0.01, "step": 149650 }, { "epoch": 95.38559592096877, "grad_norm": 0.30188387632369995, "learning_rate": 5.4316240140182194e-05, "loss": 0.0098, "step": 149660 }, { "epoch": 95.39196940726578, "grad_norm": 0.43326354026794434, "learning_rate": 5.4310749122329686e-05, "loss": 0.0129, "step": 149670 }, { "epoch": 95.39834289356278, "grad_norm": 0.3087943494319916, "learning_rate": 5.430525805209749e-05, "loss": 0.0149, "step": 149680 }, { "epoch": 95.40471637985978, "grad_norm": 0.2337004691362381, "learning_rate": 5.4299766929552345e-05, "loss": 0.0146, "step": 149690 }, { "epoch": 95.41108986615679, "grad_norm": 0.5700981616973877, "learning_rate": 5.429427575476097e-05, "loss": 0.0194, "step": 149700 }, { "epoch": 95.4174633524538, "grad_norm": 0.1887144297361374, "learning_rate": 5.4288784527790096e-05, "loss": 0.0175, "step": 149710 }, { "epoch": 95.4238368387508, "grad_norm": 0.28061339259147644, "learning_rate": 5.4283293248706434e-05, "loss": 0.0093, "step": 149720 }, { "epoch": 95.4302103250478, "grad_norm": 0.1816241592168808, "learning_rate": 5.427780191757672e-05, "loss": 0.0126, "step": 149730 }, { "epoch": 95.4365838113448, "grad_norm": 0.55616295337677, "learning_rate": 5.427231053446767e-05, "loss": 0.0176, "step": 149740 }, { "epoch": 95.44295729764181, "grad_norm": 0.37318408489227295, "learning_rate": 5.426681909944602e-05, "loss": 0.0137, "step": 149750 }, { "epoch": 95.44933078393882, "grad_norm": 0.3433413505554199, "learning_rate": 5.4261327612578485e-05, "loss": 0.0135, "step": 149760 }, { "epoch": 95.45570427023581, "grad_norm": 0.29974570870399475, "learning_rate": 5.425583607393181e-05, "loss": 0.0107, "step": 149770 }, { "epoch": 95.46207775653282, "grad_norm": 0.3011167645454407, "learning_rate": 5.425034448357268e-05, "loss": 0.0111, "step": 149780 }, { "epoch": 95.46845124282983, "grad_norm": 0.5191665887832642, "learning_rate": 5.424485284156787e-05, "loss": 0.0256, "step": 149790 }, { "epoch": 95.47482472912684, "grad_norm": 0.30862298607826233, "learning_rate": 5.423936114798409e-05, "loss": 0.0116, "step": 149800 }, { "epoch": 95.48119821542383, "grad_norm": 0.36959734559059143, "learning_rate": 5.423386940288807e-05, "loss": 0.0127, "step": 149810 }, { "epoch": 95.48757170172084, "grad_norm": 0.4745253324508667, "learning_rate": 5.422837760634654e-05, "loss": 0.0175, "step": 149820 }, { "epoch": 95.49394518801785, "grad_norm": 0.2764197289943695, "learning_rate": 5.4222885758426224e-05, "loss": 0.0089, "step": 149830 }, { "epoch": 95.50031867431485, "grad_norm": 0.27762097120285034, "learning_rate": 5.4217393859193855e-05, "loss": 0.0119, "step": 149840 }, { "epoch": 95.50669216061185, "grad_norm": 0.16980919241905212, "learning_rate": 5.421190190871618e-05, "loss": 0.0117, "step": 149850 }, { "epoch": 95.51306564690886, "grad_norm": 0.3427116572856903, "learning_rate": 5.420640990705991e-05, "loss": 0.0115, "step": 149860 }, { "epoch": 95.51943913320586, "grad_norm": 0.20824280381202698, "learning_rate": 5.42009178542918e-05, "loss": 0.0061, "step": 149870 }, { "epoch": 95.52581261950287, "grad_norm": 0.6053286790847778, "learning_rate": 5.4195425750478565e-05, "loss": 0.0152, "step": 149880 }, { "epoch": 95.53218610579987, "grad_norm": 0.48166704177856445, "learning_rate": 5.4189933595686936e-05, "loss": 0.0198, "step": 149890 }, { "epoch": 95.53855959209687, "grad_norm": 0.2955203056335449, "learning_rate": 5.4184441389983674e-05, "loss": 0.012, "step": 149900 }, { "epoch": 95.54493307839388, "grad_norm": 0.2665337026119232, "learning_rate": 5.417894913343548e-05, "loss": 0.0145, "step": 149910 }, { "epoch": 95.55130656469089, "grad_norm": 0.31912973523139954, "learning_rate": 5.417345682610914e-05, "loss": 0.0257, "step": 149920 }, { "epoch": 95.55768005098788, "grad_norm": 0.5165213346481323, "learning_rate": 5.416796446807133e-05, "loss": 0.0215, "step": 149930 }, { "epoch": 95.56405353728489, "grad_norm": 0.3781658709049225, "learning_rate": 5.416247205938883e-05, "loss": 0.0147, "step": 149940 }, { "epoch": 95.5704270235819, "grad_norm": 0.4234388470649719, "learning_rate": 5.415697960012835e-05, "loss": 0.0134, "step": 149950 }, { "epoch": 95.57680050987891, "grad_norm": 0.227054625749588, "learning_rate": 5.4151487090356666e-05, "loss": 0.0095, "step": 149960 }, { "epoch": 95.5831739961759, "grad_norm": 0.2017863243818283, "learning_rate": 5.414599453014049e-05, "loss": 0.0191, "step": 149970 }, { "epoch": 95.58954748247291, "grad_norm": 0.18486297130584717, "learning_rate": 5.414050191954655e-05, "loss": 0.0168, "step": 149980 }, { "epoch": 95.59592096876992, "grad_norm": 0.3358275890350342, "learning_rate": 5.413500925864161e-05, "loss": 0.012, "step": 149990 }, { "epoch": 95.60229445506693, "grad_norm": 0.24106521904468536, "learning_rate": 5.41295165474924e-05, "loss": 0.0128, "step": 150000 } ], "logging_steps": 10, "max_steps": 300000, "num_input_tokens_seen": 0, "num_train_epochs": 192, "save_steps": 150000, "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": 8, "trial_name": null, "trial_params": null }