{ "best_metric": null, "best_model_checkpoint": null, "epoch": 229.00763358778627, "eval_steps": 500, "global_step": 30000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.07633587786259542, "grad_norm": 168.92755126953125, "learning_rate": 0.0001, "loss": 21.1213, "step": 10 }, { "epoch": 0.15267175572519084, "grad_norm": 87.27042388916016, "learning_rate": 0.0001, "loss": 5.2154, "step": 20 }, { "epoch": 0.22900763358778625, "grad_norm": 59.843685150146484, "learning_rate": 0.0001, "loss": 3.8252, "step": 30 }, { "epoch": 0.3053435114503817, "grad_norm": 61.74694061279297, "learning_rate": 0.0001, "loss": 3.3132, "step": 40 }, { "epoch": 0.3816793893129771, "grad_norm": 61.5179328918457, "learning_rate": 0.0001, "loss": 3.0223, "step": 50 }, { "epoch": 0.4580152671755725, "grad_norm": 47.61620330810547, "learning_rate": 0.0001, "loss": 2.6841, "step": 60 }, { "epoch": 0.5343511450381679, "grad_norm": 45.58167266845703, "learning_rate": 0.0001, "loss": 2.5108, "step": 70 }, { "epoch": 0.6106870229007634, "grad_norm": 43.55134582519531, "learning_rate": 0.0001, "loss": 2.2675, "step": 80 }, { "epoch": 0.6870229007633588, "grad_norm": 45.57367706298828, "learning_rate": 0.0001, "loss": 2.1627, "step": 90 }, { "epoch": 0.7633587786259542, "grad_norm": 51.9700927734375, "learning_rate": 0.0001, "loss": 2.0652, "step": 100 }, { "epoch": 0.8396946564885496, "grad_norm": 47.24523162841797, "learning_rate": 0.0001, "loss": 1.8909, "step": 110 }, { "epoch": 0.916030534351145, "grad_norm": 44.4114990234375, "learning_rate": 0.0001, "loss": 1.8201, "step": 120 }, { "epoch": 0.9923664122137404, "grad_norm": 43.364898681640625, "learning_rate": 0.0001, "loss": 1.6964, "step": 130 }, { "epoch": 1.0687022900763359, "grad_norm": 38.24567413330078, "learning_rate": 0.0001, "loss": 1.5982, "step": 140 }, { "epoch": 1.1450381679389312, "grad_norm": 43.68265914916992, "learning_rate": 0.0001, "loss": 1.5308, "step": 150 }, { "epoch": 1.2213740458015268, "grad_norm": 46.67293930053711, "learning_rate": 0.0001, "loss": 1.4872, "step": 160 }, { "epoch": 1.297709923664122, "grad_norm": 39.62367248535156, "learning_rate": 0.0001, "loss": 1.3923, "step": 170 }, { "epoch": 1.3740458015267176, "grad_norm": 41.08576583862305, "learning_rate": 0.0001, "loss": 1.3167, "step": 180 }, { "epoch": 1.450381679389313, "grad_norm": 36.37272644042969, "learning_rate": 0.0001, "loss": 1.2402, "step": 190 }, { "epoch": 1.5267175572519083, "grad_norm": 35.016685485839844, "learning_rate": 0.0001, "loss": 1.1674, "step": 200 }, { "epoch": 1.6030534351145038, "grad_norm": 35.531612396240234, "learning_rate": 0.0001, "loss": 1.1045, "step": 210 }, { "epoch": 1.6793893129770994, "grad_norm": 31.628305435180664, "learning_rate": 0.0001, "loss": 1.0617, "step": 220 }, { "epoch": 1.7557251908396947, "grad_norm": 36.71664810180664, "learning_rate": 0.0001, "loss": 1.0107, "step": 230 }, { "epoch": 1.83206106870229, "grad_norm": 30.35818099975586, "learning_rate": 0.0001, "loss": 0.953, "step": 240 }, { "epoch": 1.9083969465648853, "grad_norm": 32.68187713623047, "learning_rate": 0.0001, "loss": 0.8861, "step": 250 }, { "epoch": 1.984732824427481, "grad_norm": 28.810409545898438, "learning_rate": 0.0001, "loss": 0.8351, "step": 260 }, { "epoch": 2.0610687022900764, "grad_norm": 29.773344039916992, "learning_rate": 0.0001, "loss": 0.8068, "step": 270 }, { "epoch": 2.1374045801526718, "grad_norm": 29.348405838012695, "learning_rate": 0.0001, "loss": 0.7612, "step": 280 }, { "epoch": 2.213740458015267, "grad_norm": 26.97745704650879, "learning_rate": 0.0001, "loss": 0.7234, "step": 290 }, { "epoch": 2.2900763358778624, "grad_norm": 28.27231216430664, "learning_rate": 0.0001, "loss": 0.6904, "step": 300 }, { "epoch": 2.366412213740458, "grad_norm": 30.91962242126465, "learning_rate": 0.0001, "loss": 0.6724, "step": 310 }, { "epoch": 2.4427480916030535, "grad_norm": 24.390432357788086, "learning_rate": 0.0001, "loss": 0.6307, "step": 320 }, { "epoch": 2.519083969465649, "grad_norm": 27.383405685424805, "learning_rate": 0.0001, "loss": 0.5961, "step": 330 }, { "epoch": 2.595419847328244, "grad_norm": 25.17072868347168, "learning_rate": 0.0001, "loss": 0.5621, "step": 340 }, { "epoch": 2.67175572519084, "grad_norm": 23.66916847229004, "learning_rate": 0.0001, "loss": 0.5445, "step": 350 }, { "epoch": 2.7480916030534353, "grad_norm": 28.739665985107422, "learning_rate": 0.0001, "loss": 0.5276, "step": 360 }, { "epoch": 2.8244274809160306, "grad_norm": 23.73203468322754, "learning_rate": 0.0001, "loss": 0.502, "step": 370 }, { "epoch": 2.900763358778626, "grad_norm": 21.6328125, "learning_rate": 0.0001, "loss": 0.4923, "step": 380 }, { "epoch": 2.9770992366412212, "grad_norm": 25.200016021728516, "learning_rate": 0.0001, "loss": 0.4607, "step": 390 }, { "epoch": 3.053435114503817, "grad_norm": 20.37843894958496, "learning_rate": 0.0001, "loss": 0.4427, "step": 400 }, { "epoch": 3.1297709923664123, "grad_norm": 25.30202293395996, "learning_rate": 0.0001, "loss": 0.4391, "step": 410 }, { "epoch": 3.2061068702290076, "grad_norm": 22.56325340270996, "learning_rate": 0.0001, "loss": 0.4222, "step": 420 }, { "epoch": 3.282442748091603, "grad_norm": 21.053443908691406, "learning_rate": 0.0001, "loss": 0.4169, "step": 430 }, { "epoch": 3.3587786259541983, "grad_norm": 17.73294448852539, "learning_rate": 0.0001, "loss": 0.3881, "step": 440 }, { "epoch": 3.435114503816794, "grad_norm": 20.485143661499023, "learning_rate": 0.0001, "loss": 0.3738, "step": 450 }, { "epoch": 3.5114503816793894, "grad_norm": 19.157276153564453, "learning_rate": 0.0001, "loss": 0.3611, "step": 460 }, { "epoch": 3.5877862595419847, "grad_norm": 18.875654220581055, "learning_rate": 0.0001, "loss": 0.3609, "step": 470 }, { "epoch": 3.66412213740458, "grad_norm": 19.00319480895996, "learning_rate": 0.0001, "loss": 0.3387, "step": 480 }, { "epoch": 3.7404580152671754, "grad_norm": 19.173879623413086, "learning_rate": 0.0001, "loss": 0.3352, "step": 490 }, { "epoch": 3.816793893129771, "grad_norm": 19.501708984375, "learning_rate": 0.0001, "loss": 0.3217, "step": 500 }, { "epoch": 3.8931297709923665, "grad_norm": 16.747928619384766, "learning_rate": 0.0001, "loss": 0.3199, "step": 510 }, { "epoch": 3.969465648854962, "grad_norm": 18.285846710205078, "learning_rate": 0.0001, "loss": 0.2979, "step": 520 }, { "epoch": 4.0458015267175576, "grad_norm": 15.963340759277344, "learning_rate": 0.0001, "loss": 0.2975, "step": 530 }, { "epoch": 4.122137404580153, "grad_norm": 13.798179626464844, "learning_rate": 0.0001, "loss": 0.2866, "step": 540 }, { "epoch": 4.198473282442748, "grad_norm": 15.340399742126465, "learning_rate": 0.0001, "loss": 0.2841, "step": 550 }, { "epoch": 4.2748091603053435, "grad_norm": 14.734049797058105, "learning_rate": 0.0001, "loss": 0.2759, "step": 560 }, { "epoch": 4.351145038167939, "grad_norm": 15.308343887329102, "learning_rate": 0.0001, "loss": 0.2736, "step": 570 }, { "epoch": 4.427480916030534, "grad_norm": 17.409940719604492, "learning_rate": 0.0001, "loss": 0.2669, "step": 580 }, { "epoch": 4.5038167938931295, "grad_norm": 12.640336036682129, "learning_rate": 0.0001, "loss": 0.257, "step": 590 }, { "epoch": 4.580152671755725, "grad_norm": 15.863229751586914, "learning_rate": 0.0001, "loss": 0.253, "step": 600 }, { "epoch": 4.65648854961832, "grad_norm": 16.420909881591797, "learning_rate": 0.0001, "loss": 0.2488, "step": 610 }, { "epoch": 4.732824427480916, "grad_norm": 13.079082489013672, "learning_rate": 0.0001, "loss": 0.2415, "step": 620 }, { "epoch": 4.809160305343512, "grad_norm": 14.312274932861328, "learning_rate": 0.0001, "loss": 0.2401, "step": 630 }, { "epoch": 4.885496183206107, "grad_norm": 13.296807289123535, "learning_rate": 0.0001, "loss": 0.2338, "step": 640 }, { "epoch": 4.961832061068702, "grad_norm": 13.039434432983398, "learning_rate": 0.0001, "loss": 0.2307, "step": 650 }, { "epoch": 5.038167938931298, "grad_norm": 12.49682903289795, "learning_rate": 0.0001, "loss": 0.225, "step": 660 }, { "epoch": 5.114503816793893, "grad_norm": 11.922616004943848, "learning_rate": 0.0001, "loss": 0.2214, "step": 670 }, { "epoch": 5.190839694656488, "grad_norm": 16.84244728088379, "learning_rate": 0.0001, "loss": 0.2157, "step": 680 }, { "epoch": 5.267175572519084, "grad_norm": 14.172879219055176, "learning_rate": 0.0001, "loss": 0.2171, "step": 690 }, { "epoch": 5.34351145038168, "grad_norm": 13.99837875366211, "learning_rate": 0.0001, "loss": 0.2121, "step": 700 }, { "epoch": 5.419847328244275, "grad_norm": 13.507291793823242, "learning_rate": 0.0001, "loss": 0.211, "step": 710 }, { "epoch": 5.4961832061068705, "grad_norm": 11.915947914123535, "learning_rate": 0.0001, "loss": 0.2025, "step": 720 }, { "epoch": 5.572519083969466, "grad_norm": 11.74112319946289, "learning_rate": 0.0001, "loss": 0.2016, "step": 730 }, { "epoch": 5.648854961832061, "grad_norm": 11.204144477844238, "learning_rate": 0.0001, "loss": 0.1995, "step": 740 }, { "epoch": 5.7251908396946565, "grad_norm": 12.914752006530762, "learning_rate": 0.0001, "loss": 0.2017, "step": 750 }, { "epoch": 5.801526717557252, "grad_norm": 11.5668306350708, "learning_rate": 0.0001, "loss": 0.1997, "step": 760 }, { "epoch": 5.877862595419847, "grad_norm": 12.363494873046875, "learning_rate": 0.0001, "loss": 0.1921, "step": 770 }, { "epoch": 5.9541984732824424, "grad_norm": 9.735773086547852, "learning_rate": 0.0001, "loss": 0.1893, "step": 780 }, { "epoch": 6.030534351145038, "grad_norm": 15.016815185546875, "learning_rate": 0.0001, "loss": 0.1992, "step": 790 }, { "epoch": 6.106870229007634, "grad_norm": 13.192752838134766, "learning_rate": 0.0001, "loss": 0.1878, "step": 800 }, { "epoch": 6.183206106870229, "grad_norm": 11.377209663391113, "learning_rate": 0.0001, "loss": 0.1848, "step": 810 }, { "epoch": 6.259541984732825, "grad_norm": 12.316781997680664, "learning_rate": 0.0001, "loss": 0.1834, "step": 820 }, { "epoch": 6.33587786259542, "grad_norm": 10.229947090148926, "learning_rate": 0.0001, "loss": 0.1791, "step": 830 }, { "epoch": 6.412213740458015, "grad_norm": 11.881203651428223, "learning_rate": 0.0001, "loss": 0.1765, "step": 840 }, { "epoch": 6.488549618320611, "grad_norm": 13.067709922790527, "learning_rate": 0.0001, "loss": 0.1767, "step": 850 }, { "epoch": 6.564885496183206, "grad_norm": 12.92533016204834, "learning_rate": 0.0001, "loss": 0.1789, "step": 860 }, { "epoch": 6.641221374045801, "grad_norm": 13.507234573364258, "learning_rate": 0.0001, "loss": 0.1701, "step": 870 }, { "epoch": 6.717557251908397, "grad_norm": 10.344167709350586, "learning_rate": 0.0001, "loss": 0.1695, "step": 880 }, { "epoch": 6.793893129770993, "grad_norm": 11.722400665283203, "learning_rate": 0.0001, "loss": 0.1664, "step": 890 }, { "epoch": 6.870229007633588, "grad_norm": 11.22995662689209, "learning_rate": 0.0001, "loss": 0.1634, "step": 900 }, { "epoch": 6.9465648854961835, "grad_norm": 10.70693588256836, "learning_rate": 0.0001, "loss": 0.1563, "step": 910 }, { "epoch": 7.022900763358779, "grad_norm": 11.981927871704102, "learning_rate": 0.0001, "loss": 0.1617, "step": 920 }, { "epoch": 7.099236641221374, "grad_norm": 12.469937324523926, "learning_rate": 0.0001, "loss": 0.1571, "step": 930 }, { "epoch": 7.175572519083969, "grad_norm": 12.589916229248047, "learning_rate": 0.0001, "loss": 0.155, "step": 940 }, { "epoch": 7.251908396946565, "grad_norm": 11.459300994873047, "learning_rate": 0.0001, "loss": 0.1554, "step": 950 }, { "epoch": 7.32824427480916, "grad_norm": 10.55831241607666, "learning_rate": 0.0001, "loss": 0.149, "step": 960 }, { "epoch": 7.404580152671755, "grad_norm": 10.322016716003418, "learning_rate": 0.0001, "loss": 0.1496, "step": 970 }, { "epoch": 7.480916030534351, "grad_norm": 10.562490463256836, "learning_rate": 0.0001, "loss": 0.1507, "step": 980 }, { "epoch": 7.557251908396947, "grad_norm": 11.646677017211914, "learning_rate": 0.0001, "loss": 0.1471, "step": 990 }, { "epoch": 7.633587786259542, "grad_norm": 10.703932762145996, "learning_rate": 0.0001, "loss": 0.1561, "step": 1000 }, { "epoch": 7.709923664122138, "grad_norm": 12.677861213684082, "learning_rate": 0.0001, "loss": 0.1521, "step": 1010 }, { "epoch": 7.786259541984733, "grad_norm": 10.73009967803955, "learning_rate": 0.0001, "loss": 0.1465, "step": 1020 }, { "epoch": 7.862595419847328, "grad_norm": 8.363862991333008, "learning_rate": 0.0001, "loss": 0.1483, "step": 1030 }, { "epoch": 7.938931297709924, "grad_norm": 11.016866683959961, "learning_rate": 0.0001, "loss": 0.1513, "step": 1040 }, { "epoch": 8.01526717557252, "grad_norm": 8.85919189453125, "learning_rate": 0.0001, "loss": 0.1484, "step": 1050 }, { "epoch": 8.091603053435115, "grad_norm": 11.612984657287598, "learning_rate": 0.0001, "loss": 0.1508, "step": 1060 }, { "epoch": 8.16793893129771, "grad_norm": 10.121712684631348, "learning_rate": 0.0001, "loss": 0.1461, "step": 1070 }, { "epoch": 8.244274809160306, "grad_norm": 9.81918716430664, "learning_rate": 0.0001, "loss": 0.1411, "step": 1080 }, { "epoch": 8.320610687022901, "grad_norm": 9.005017280578613, "learning_rate": 0.0001, "loss": 0.1388, "step": 1090 }, { "epoch": 8.396946564885496, "grad_norm": 9.334559440612793, "learning_rate": 0.0001, "loss": 0.1371, "step": 1100 }, { "epoch": 8.473282442748092, "grad_norm": 9.8480863571167, "learning_rate": 0.0001, "loss": 0.1362, "step": 1110 }, { "epoch": 8.549618320610687, "grad_norm": 8.730084419250488, "learning_rate": 0.0001, "loss": 0.1402, "step": 1120 }, { "epoch": 8.625954198473282, "grad_norm": 10.462005615234375, "learning_rate": 0.0001, "loss": 0.1315, "step": 1130 }, { "epoch": 8.702290076335878, "grad_norm": 8.740320205688477, "learning_rate": 0.0001, "loss": 0.1333, "step": 1140 }, { "epoch": 8.778625954198473, "grad_norm": 9.550580978393555, "learning_rate": 0.0001, "loss": 0.1349, "step": 1150 }, { "epoch": 8.854961832061068, "grad_norm": 10.238468170166016, "learning_rate": 0.0001, "loss": 0.1289, "step": 1160 }, { "epoch": 8.931297709923664, "grad_norm": 9.37375259399414, "learning_rate": 0.0001, "loss": 0.1315, "step": 1170 }, { "epoch": 9.007633587786259, "grad_norm": 7.704553127288818, "learning_rate": 0.0001, "loss": 0.1276, "step": 1180 }, { "epoch": 9.083969465648854, "grad_norm": 8.280585289001465, "learning_rate": 0.0001, "loss": 0.1257, "step": 1190 }, { "epoch": 9.16030534351145, "grad_norm": 9.050798416137695, "learning_rate": 0.0001, "loss": 0.1244, "step": 1200 }, { "epoch": 9.236641221374045, "grad_norm": 7.744107246398926, "learning_rate": 0.0001, "loss": 0.1231, "step": 1210 }, { "epoch": 9.312977099236642, "grad_norm": 8.575624465942383, "learning_rate": 0.0001, "loss": 0.1246, "step": 1220 }, { "epoch": 9.389312977099237, "grad_norm": 10.146769523620605, "learning_rate": 0.0001, "loss": 0.1256, "step": 1230 }, { "epoch": 9.465648854961833, "grad_norm": 9.420551300048828, "learning_rate": 0.0001, "loss": 0.124, "step": 1240 }, { "epoch": 9.541984732824428, "grad_norm": 7.888007640838623, "learning_rate": 0.0001, "loss": 0.1188, "step": 1250 }, { "epoch": 9.618320610687023, "grad_norm": 9.856849670410156, "learning_rate": 0.0001, "loss": 0.1213, "step": 1260 }, { "epoch": 9.694656488549619, "grad_norm": 8.719855308532715, "learning_rate": 0.0001, "loss": 0.1187, "step": 1270 }, { "epoch": 9.770992366412214, "grad_norm": 8.88354206085205, "learning_rate": 0.0001, "loss": 0.1204, "step": 1280 }, { "epoch": 9.84732824427481, "grad_norm": 9.407536506652832, "learning_rate": 0.0001, "loss": 0.1189, "step": 1290 }, { "epoch": 9.923664122137405, "grad_norm": 8.780763626098633, "learning_rate": 0.0001, "loss": 0.1143, "step": 1300 }, { "epoch": 10.0, "grad_norm": 9.952812194824219, "learning_rate": 0.0001, "loss": 0.1124, "step": 1310 }, { "epoch": 10.076335877862595, "grad_norm": 9.866719245910645, "learning_rate": 0.0001, "loss": 0.1138, "step": 1320 }, { "epoch": 10.15267175572519, "grad_norm": 9.185962677001953, "learning_rate": 0.0001, "loss": 0.1157, "step": 1330 }, { "epoch": 10.229007633587786, "grad_norm": 9.978385925292969, "learning_rate": 0.0001, "loss": 0.1124, "step": 1340 }, { "epoch": 10.305343511450381, "grad_norm": 10.240349769592285, "learning_rate": 0.0001, "loss": 0.1116, "step": 1350 }, { "epoch": 10.381679389312977, "grad_norm": 8.836553573608398, "learning_rate": 0.0001, "loss": 0.1125, "step": 1360 }, { "epoch": 10.458015267175572, "grad_norm": 9.968942642211914, "learning_rate": 0.0001, "loss": 0.1068, "step": 1370 }, { "epoch": 10.534351145038167, "grad_norm": 9.093345642089844, "learning_rate": 0.0001, "loss": 0.1058, "step": 1380 }, { "epoch": 10.610687022900763, "grad_norm": 8.468681335449219, "learning_rate": 0.0001, "loss": 0.1072, "step": 1390 }, { "epoch": 10.68702290076336, "grad_norm": 7.485656261444092, "learning_rate": 0.0001, "loss": 0.1045, "step": 1400 }, { "epoch": 10.763358778625955, "grad_norm": 7.843764305114746, "learning_rate": 0.0001, "loss": 0.1028, "step": 1410 }, { "epoch": 10.83969465648855, "grad_norm": 7.74562406539917, "learning_rate": 0.0001, "loss": 0.1046, "step": 1420 }, { "epoch": 10.916030534351146, "grad_norm": 7.549198150634766, "learning_rate": 0.0001, "loss": 0.1066, "step": 1430 }, { "epoch": 10.992366412213741, "grad_norm": 7.393519401550293, "learning_rate": 0.0001, "loss": 0.1053, "step": 1440 }, { "epoch": 11.068702290076336, "grad_norm": 8.308021545410156, "learning_rate": 0.0001, "loss": 0.1036, "step": 1450 }, { "epoch": 11.145038167938932, "grad_norm": 7.227427005767822, "learning_rate": 0.0001, "loss": 0.1035, "step": 1460 }, { "epoch": 11.221374045801527, "grad_norm": 8.64355182647705, "learning_rate": 0.0001, "loss": 0.1049, "step": 1470 }, { "epoch": 11.297709923664122, "grad_norm": 8.155220031738281, "learning_rate": 0.0001, "loss": 0.1003, "step": 1480 }, { "epoch": 11.374045801526718, "grad_norm": 6.244787693023682, "learning_rate": 0.0001, "loss": 0.0984, "step": 1490 }, { "epoch": 11.450381679389313, "grad_norm": 7.557957172393799, "learning_rate": 0.0001, "loss": 0.0995, "step": 1500 }, { "epoch": 11.526717557251908, "grad_norm": 9.104961395263672, "learning_rate": 0.0001, "loss": 0.1034, "step": 1510 }, { "epoch": 11.603053435114504, "grad_norm": 7.173948287963867, "learning_rate": 0.0001, "loss": 0.0995, "step": 1520 }, { "epoch": 11.679389312977099, "grad_norm": 5.6769256591796875, "learning_rate": 0.0001, "loss": 0.0957, "step": 1530 }, { "epoch": 11.755725190839694, "grad_norm": 7.955283164978027, "learning_rate": 0.0001, "loss": 0.0974, "step": 1540 }, { "epoch": 11.83206106870229, "grad_norm": 8.542166709899902, "learning_rate": 0.0001, "loss": 0.1019, "step": 1550 }, { "epoch": 11.908396946564885, "grad_norm": 6.941015243530273, "learning_rate": 0.0001, "loss": 0.1027, "step": 1560 }, { "epoch": 11.98473282442748, "grad_norm": 9.677477836608887, "learning_rate": 0.0001, "loss": 0.1002, "step": 1570 }, { "epoch": 12.061068702290076, "grad_norm": 7.5489182472229, "learning_rate": 0.0001, "loss": 0.0977, "step": 1580 }, { "epoch": 12.137404580152673, "grad_norm": 7.113029956817627, "learning_rate": 0.0001, "loss": 0.0955, "step": 1590 }, { "epoch": 12.213740458015268, "grad_norm": 6.386077404022217, "learning_rate": 0.0001, "loss": 0.0937, "step": 1600 }, { "epoch": 12.290076335877863, "grad_norm": 7.5228681564331055, "learning_rate": 0.0001, "loss": 0.0936, "step": 1610 }, { "epoch": 12.366412213740459, "grad_norm": 7.112820148468018, "learning_rate": 0.0001, "loss": 0.0987, "step": 1620 }, { "epoch": 12.442748091603054, "grad_norm": 7.284379482269287, "learning_rate": 0.0001, "loss": 0.0974, "step": 1630 }, { "epoch": 12.51908396946565, "grad_norm": 7.478989601135254, "learning_rate": 0.0001, "loss": 0.0983, "step": 1640 }, { "epoch": 12.595419847328245, "grad_norm": 7.1693620681762695, "learning_rate": 0.0001, "loss": 0.0962, "step": 1650 }, { "epoch": 12.67175572519084, "grad_norm": 7.737825393676758, "learning_rate": 0.0001, "loss": 0.092, "step": 1660 }, { "epoch": 12.748091603053435, "grad_norm": 8.236441612243652, "learning_rate": 0.0001, "loss": 0.0919, "step": 1670 }, { "epoch": 12.82442748091603, "grad_norm": 7.869682788848877, "learning_rate": 0.0001, "loss": 0.0927, "step": 1680 }, { "epoch": 12.900763358778626, "grad_norm": 7.423285961151123, "learning_rate": 0.0001, "loss": 0.0912, "step": 1690 }, { "epoch": 12.977099236641221, "grad_norm": 6.838749885559082, "learning_rate": 0.0001, "loss": 0.0896, "step": 1700 }, { "epoch": 13.053435114503817, "grad_norm": 7.5142388343811035, "learning_rate": 0.0001, "loss": 0.0898, "step": 1710 }, { "epoch": 13.129770992366412, "grad_norm": 8.024205207824707, "learning_rate": 0.0001, "loss": 0.0897, "step": 1720 }, { "epoch": 13.206106870229007, "grad_norm": 7.345959663391113, "learning_rate": 0.0001, "loss": 0.0888, "step": 1730 }, { "epoch": 13.282442748091603, "grad_norm": 7.675141334533691, "learning_rate": 0.0001, "loss": 0.0877, "step": 1740 }, { "epoch": 13.358778625954198, "grad_norm": 8.565728187561035, "learning_rate": 0.0001, "loss": 0.0901, "step": 1750 }, { "epoch": 13.435114503816793, "grad_norm": 6.637859344482422, "learning_rate": 0.0001, "loss": 0.087, "step": 1760 }, { "epoch": 13.511450381679388, "grad_norm": 6.416836261749268, "learning_rate": 0.0001, "loss": 0.0878, "step": 1770 }, { "epoch": 13.587786259541986, "grad_norm": 6.845940589904785, "learning_rate": 0.0001, "loss": 0.0858, "step": 1780 }, { "epoch": 13.664122137404581, "grad_norm": 7.624638080596924, "learning_rate": 0.0001, "loss": 0.0851, "step": 1790 }, { "epoch": 13.740458015267176, "grad_norm": 6.991026401519775, "learning_rate": 0.0001, "loss": 0.0858, "step": 1800 }, { "epoch": 13.816793893129772, "grad_norm": 7.503624439239502, "learning_rate": 0.0001, "loss": 0.0852, "step": 1810 }, { "epoch": 13.893129770992367, "grad_norm": 7.202059745788574, "learning_rate": 0.0001, "loss": 0.0847, "step": 1820 }, { "epoch": 13.969465648854962, "grad_norm": 7.312540054321289, "learning_rate": 0.0001, "loss": 0.084, "step": 1830 }, { "epoch": 14.045801526717558, "grad_norm": 8.166820526123047, "learning_rate": 0.0001, "loss": 0.0867, "step": 1840 }, { "epoch": 14.122137404580153, "grad_norm": 6.808027744293213, "learning_rate": 0.0001, "loss": 0.0881, "step": 1850 }, { "epoch": 14.198473282442748, "grad_norm": 7.274450302124023, "learning_rate": 0.0001, "loss": 0.0851, "step": 1860 }, { "epoch": 14.274809160305344, "grad_norm": 6.350966453552246, "learning_rate": 0.0001, "loss": 0.0839, "step": 1870 }, { "epoch": 14.351145038167939, "grad_norm": 7.1431193351745605, "learning_rate": 0.0001, "loss": 0.0826, "step": 1880 }, { "epoch": 14.427480916030534, "grad_norm": 5.488149642944336, "learning_rate": 0.0001, "loss": 0.0824, "step": 1890 }, { "epoch": 14.50381679389313, "grad_norm": 6.318151950836182, "learning_rate": 0.0001, "loss": 0.081, "step": 1900 }, { "epoch": 14.580152671755725, "grad_norm": 5.92701530456543, "learning_rate": 0.0001, "loss": 0.0787, "step": 1910 }, { "epoch": 14.65648854961832, "grad_norm": 5.832387924194336, "learning_rate": 0.0001, "loss": 0.0775, "step": 1920 }, { "epoch": 14.732824427480915, "grad_norm": 6.097518444061279, "learning_rate": 0.0001, "loss": 0.0809, "step": 1930 }, { "epoch": 14.80916030534351, "grad_norm": 6.0153398513793945, "learning_rate": 0.0001, "loss": 0.0814, "step": 1940 }, { "epoch": 14.885496183206106, "grad_norm": 6.778148174285889, "learning_rate": 0.0001, "loss": 0.0808, "step": 1950 }, { "epoch": 14.961832061068701, "grad_norm": 6.84468412399292, "learning_rate": 0.0001, "loss": 0.0818, "step": 1960 }, { "epoch": 15.038167938931299, "grad_norm": 5.8659491539001465, "learning_rate": 0.0001, "loss": 0.0814, "step": 1970 }, { "epoch": 15.114503816793894, "grad_norm": 6.39572286605835, "learning_rate": 0.0001, "loss": 0.0818, "step": 1980 }, { "epoch": 15.19083969465649, "grad_norm": 6.94228982925415, "learning_rate": 0.0001, "loss": 0.0807, "step": 1990 }, { "epoch": 15.267175572519085, "grad_norm": 6.229966163635254, "learning_rate": 0.0001, "loss": 0.0785, "step": 2000 }, { "epoch": 15.34351145038168, "grad_norm": 6.096153259277344, "learning_rate": 0.0001, "loss": 0.0795, "step": 2010 }, { "epoch": 15.419847328244275, "grad_norm": 6.71791934967041, "learning_rate": 0.0001, "loss": 0.0802, "step": 2020 }, { "epoch": 15.49618320610687, "grad_norm": 6.773488521575928, "learning_rate": 0.0001, "loss": 0.0771, "step": 2030 }, { "epoch": 15.572519083969466, "grad_norm": 5.644485950469971, "learning_rate": 0.0001, "loss": 0.0751, "step": 2040 }, { "epoch": 15.648854961832061, "grad_norm": 6.287095546722412, "learning_rate": 0.0001, "loss": 0.0762, "step": 2050 }, { "epoch": 15.725190839694656, "grad_norm": 6.729592800140381, "learning_rate": 0.0001, "loss": 0.0765, "step": 2060 }, { "epoch": 15.801526717557252, "grad_norm": 5.234157085418701, "learning_rate": 0.0001, "loss": 0.0738, "step": 2070 }, { "epoch": 15.877862595419847, "grad_norm": 5.021719932556152, "learning_rate": 0.0001, "loss": 0.0726, "step": 2080 }, { "epoch": 15.954198473282442, "grad_norm": 5.897186279296875, "learning_rate": 0.0001, "loss": 0.0729, "step": 2090 }, { "epoch": 16.03053435114504, "grad_norm": 6.141025543212891, "learning_rate": 0.0001, "loss": 0.0724, "step": 2100 }, { "epoch": 16.106870229007633, "grad_norm": 5.613735675811768, "learning_rate": 0.0001, "loss": 0.0718, "step": 2110 }, { "epoch": 16.18320610687023, "grad_norm": 5.455859661102295, "learning_rate": 0.0001, "loss": 0.0719, "step": 2120 }, { "epoch": 16.259541984732824, "grad_norm": 5.105410575866699, "learning_rate": 0.0001, "loss": 0.0745, "step": 2130 }, { "epoch": 16.33587786259542, "grad_norm": 6.43524169921875, "learning_rate": 0.0001, "loss": 0.0733, "step": 2140 }, { "epoch": 16.412213740458014, "grad_norm": 6.013232707977295, "learning_rate": 0.0001, "loss": 0.072, "step": 2150 }, { "epoch": 16.48854961832061, "grad_norm": 5.530680179595947, "learning_rate": 0.0001, "loss": 0.0726, "step": 2160 }, { "epoch": 16.564885496183205, "grad_norm": 5.1906914710998535, "learning_rate": 0.0001, "loss": 0.0724, "step": 2170 }, { "epoch": 16.641221374045802, "grad_norm": 6.035505294799805, "learning_rate": 0.0001, "loss": 0.0734, "step": 2180 }, { "epoch": 16.717557251908396, "grad_norm": 6.849437236785889, "learning_rate": 0.0001, "loss": 0.0742, "step": 2190 }, { "epoch": 16.793893129770993, "grad_norm": 5.905850887298584, "learning_rate": 0.0001, "loss": 0.0743, "step": 2200 }, { "epoch": 16.870229007633586, "grad_norm": 5.868034839630127, "learning_rate": 0.0001, "loss": 0.071, "step": 2210 }, { "epoch": 16.946564885496183, "grad_norm": 5.603492259979248, "learning_rate": 0.0001, "loss": 0.0706, "step": 2220 }, { "epoch": 17.022900763358777, "grad_norm": 6.782287120819092, "learning_rate": 0.0001, "loss": 0.0739, "step": 2230 }, { "epoch": 17.099236641221374, "grad_norm": 6.456106185913086, "learning_rate": 0.0001, "loss": 0.0731, "step": 2240 }, { "epoch": 17.17557251908397, "grad_norm": 5.722738742828369, "learning_rate": 0.0001, "loss": 0.0713, "step": 2250 }, { "epoch": 17.251908396946565, "grad_norm": 7.123944282531738, "learning_rate": 0.0001, "loss": 0.0701, "step": 2260 }, { "epoch": 17.328244274809162, "grad_norm": 6.290699481964111, "learning_rate": 0.0001, "loss": 0.0704, "step": 2270 }, { "epoch": 17.404580152671755, "grad_norm": 5.507736682891846, "learning_rate": 0.0001, "loss": 0.0689, "step": 2280 }, { "epoch": 17.480916030534353, "grad_norm": 5.154501914978027, "learning_rate": 0.0001, "loss": 0.0683, "step": 2290 }, { "epoch": 17.557251908396946, "grad_norm": 5.525034427642822, "learning_rate": 0.0001, "loss": 0.0678, "step": 2300 }, { "epoch": 17.633587786259543, "grad_norm": 5.395140171051025, "learning_rate": 0.0001, "loss": 0.0704, "step": 2310 }, { "epoch": 17.709923664122137, "grad_norm": 5.8200154304504395, "learning_rate": 0.0001, "loss": 0.0694, "step": 2320 }, { "epoch": 17.786259541984734, "grad_norm": 6.472496509552002, "learning_rate": 0.0001, "loss": 0.0692, "step": 2330 }, { "epoch": 17.862595419847327, "grad_norm": 5.067510604858398, "learning_rate": 0.0001, "loss": 0.0694, "step": 2340 }, { "epoch": 17.938931297709924, "grad_norm": 5.070642948150635, "learning_rate": 0.0001, "loss": 0.0684, "step": 2350 }, { "epoch": 18.015267175572518, "grad_norm": 5.140161991119385, "learning_rate": 0.0001, "loss": 0.0703, "step": 2360 }, { "epoch": 18.091603053435115, "grad_norm": 5.678916931152344, "learning_rate": 0.0001, "loss": 0.0681, "step": 2370 }, { "epoch": 18.16793893129771, "grad_norm": 5.2955002784729, "learning_rate": 0.0001, "loss": 0.0679, "step": 2380 }, { "epoch": 18.244274809160306, "grad_norm": 5.867574214935303, "learning_rate": 0.0001, "loss": 0.0681, "step": 2390 }, { "epoch": 18.3206106870229, "grad_norm": 4.9059553146362305, "learning_rate": 0.0001, "loss": 0.0679, "step": 2400 }, { "epoch": 18.396946564885496, "grad_norm": 6.183122634887695, "learning_rate": 0.0001, "loss": 0.0663, "step": 2410 }, { "epoch": 18.47328244274809, "grad_norm": 5.018169403076172, "learning_rate": 0.0001, "loss": 0.0638, "step": 2420 }, { "epoch": 18.549618320610687, "grad_norm": 5.290459632873535, "learning_rate": 0.0001, "loss": 0.0656, "step": 2430 }, { "epoch": 18.625954198473284, "grad_norm": 5.464596748352051, "learning_rate": 0.0001, "loss": 0.0643, "step": 2440 }, { "epoch": 18.702290076335878, "grad_norm": 5.608114719390869, "learning_rate": 0.0001, "loss": 0.0663, "step": 2450 }, { "epoch": 18.778625954198475, "grad_norm": 5.931028842926025, "learning_rate": 0.0001, "loss": 0.0644, "step": 2460 }, { "epoch": 18.85496183206107, "grad_norm": 5.58274507522583, "learning_rate": 0.0001, "loss": 0.0654, "step": 2470 }, { "epoch": 18.931297709923665, "grad_norm": 4.923034191131592, "learning_rate": 0.0001, "loss": 0.0638, "step": 2480 }, { "epoch": 19.00763358778626, "grad_norm": 5.379124164581299, "learning_rate": 0.0001, "loss": 0.065, "step": 2490 }, { "epoch": 19.083969465648856, "grad_norm": 5.7871599197387695, "learning_rate": 0.0001, "loss": 0.0635, "step": 2500 }, { "epoch": 19.16030534351145, "grad_norm": 4.753558158874512, "learning_rate": 0.0001, "loss": 0.0653, "step": 2510 }, { "epoch": 19.236641221374047, "grad_norm": 4.794043064117432, "learning_rate": 0.0001, "loss": 0.0643, "step": 2520 }, { "epoch": 19.31297709923664, "grad_norm": 5.980827808380127, "learning_rate": 0.0001, "loss": 0.0643, "step": 2530 }, { "epoch": 19.389312977099237, "grad_norm": 5.7473602294921875, "learning_rate": 0.0001, "loss": 0.0668, "step": 2540 }, { "epoch": 19.46564885496183, "grad_norm": 5.7471137046813965, "learning_rate": 0.0001, "loss": 0.0667, "step": 2550 }, { "epoch": 19.541984732824428, "grad_norm": 4.844244003295898, "learning_rate": 0.0001, "loss": 0.0656, "step": 2560 }, { "epoch": 19.61832061068702, "grad_norm": 4.9728007316589355, "learning_rate": 0.0001, "loss": 0.0619, "step": 2570 }, { "epoch": 19.69465648854962, "grad_norm": 5.148043632507324, "learning_rate": 0.0001, "loss": 0.0623, "step": 2580 }, { "epoch": 19.770992366412212, "grad_norm": 4.875810146331787, "learning_rate": 0.0001, "loss": 0.0612, "step": 2590 }, { "epoch": 19.84732824427481, "grad_norm": 4.321455955505371, "learning_rate": 0.0001, "loss": 0.0614, "step": 2600 }, { "epoch": 19.923664122137403, "grad_norm": 4.757236957550049, "learning_rate": 0.0001, "loss": 0.0611, "step": 2610 }, { "epoch": 20.0, "grad_norm": 4.378152370452881, "learning_rate": 0.0001, "loss": 0.0628, "step": 2620 }, { "epoch": 20.076335877862597, "grad_norm": 4.362908840179443, "learning_rate": 0.0001, "loss": 0.0617, "step": 2630 }, { "epoch": 20.15267175572519, "grad_norm": 4.613173961639404, "learning_rate": 0.0001, "loss": 0.0627, "step": 2640 }, { "epoch": 20.229007633587788, "grad_norm": 5.0382795333862305, "learning_rate": 0.0001, "loss": 0.0639, "step": 2650 }, { "epoch": 20.30534351145038, "grad_norm": 6.1410675048828125, "learning_rate": 0.0001, "loss": 0.0639, "step": 2660 }, { "epoch": 20.38167938931298, "grad_norm": 5.2981181144714355, "learning_rate": 0.0001, "loss": 0.061, "step": 2670 }, { "epoch": 20.458015267175572, "grad_norm": 4.971876621246338, "learning_rate": 0.0001, "loss": 0.0644, "step": 2680 }, { "epoch": 20.53435114503817, "grad_norm": 4.710940837860107, "learning_rate": 0.0001, "loss": 0.0625, "step": 2690 }, { "epoch": 20.610687022900763, "grad_norm": 5.167057991027832, "learning_rate": 0.0001, "loss": 0.0616, "step": 2700 }, { "epoch": 20.68702290076336, "grad_norm": 4.96181058883667, "learning_rate": 0.0001, "loss": 0.0609, "step": 2710 }, { "epoch": 20.763358778625953, "grad_norm": 4.925744533538818, "learning_rate": 0.0001, "loss": 0.0603, "step": 2720 }, { "epoch": 20.83969465648855, "grad_norm": 5.478791236877441, "learning_rate": 0.0001, "loss": 0.0617, "step": 2730 }, { "epoch": 20.916030534351144, "grad_norm": 5.078333377838135, "learning_rate": 0.0001, "loss": 0.0629, "step": 2740 }, { "epoch": 20.99236641221374, "grad_norm": 5.133959770202637, "learning_rate": 0.0001, "loss": 0.0618, "step": 2750 }, { "epoch": 21.068702290076335, "grad_norm": 5.244518280029297, "learning_rate": 0.0001, "loss": 0.0597, "step": 2760 }, { "epoch": 21.14503816793893, "grad_norm": 5.390507221221924, "learning_rate": 0.0001, "loss": 0.0601, "step": 2770 }, { "epoch": 21.221374045801525, "grad_norm": 4.441818714141846, "learning_rate": 0.0001, "loss": 0.058, "step": 2780 }, { "epoch": 21.297709923664122, "grad_norm": 5.100788116455078, "learning_rate": 0.0001, "loss": 0.06, "step": 2790 }, { "epoch": 21.374045801526716, "grad_norm": 4.860738277435303, "learning_rate": 0.0001, "loss": 0.0616, "step": 2800 }, { "epoch": 21.450381679389313, "grad_norm": 4.385000228881836, "learning_rate": 0.0001, "loss": 0.0581, "step": 2810 }, { "epoch": 21.52671755725191, "grad_norm": 4.946840763092041, "learning_rate": 0.0001, "loss": 0.061, "step": 2820 }, { "epoch": 21.603053435114504, "grad_norm": 4.521257400512695, "learning_rate": 0.0001, "loss": 0.0591, "step": 2830 }, { "epoch": 21.6793893129771, "grad_norm": 4.535577774047852, "learning_rate": 0.0001, "loss": 0.0591, "step": 2840 }, { "epoch": 21.755725190839694, "grad_norm": 4.7203755378723145, "learning_rate": 0.0001, "loss": 0.0588, "step": 2850 }, { "epoch": 21.83206106870229, "grad_norm": 4.223835468292236, "learning_rate": 0.0001, "loss": 0.0573, "step": 2860 }, { "epoch": 21.908396946564885, "grad_norm": 4.7818708419799805, "learning_rate": 0.0001, "loss": 0.0586, "step": 2870 }, { "epoch": 21.984732824427482, "grad_norm": 5.4875078201293945, "learning_rate": 0.0001, "loss": 0.0573, "step": 2880 }, { "epoch": 22.061068702290076, "grad_norm": 4.094913005828857, "learning_rate": 0.0001, "loss": 0.0563, "step": 2890 }, { "epoch": 22.137404580152673, "grad_norm": 4.53065824508667, "learning_rate": 0.0001, "loss": 0.0562, "step": 2900 }, { "epoch": 22.213740458015266, "grad_norm": 4.862125396728516, "learning_rate": 0.0001, "loss": 0.058, "step": 2910 }, { "epoch": 22.290076335877863, "grad_norm": 4.263631343841553, "learning_rate": 0.0001, "loss": 0.0578, "step": 2920 }, { "epoch": 22.366412213740457, "grad_norm": 5.581209182739258, "learning_rate": 0.0001, "loss": 0.0571, "step": 2930 }, { "epoch": 22.442748091603054, "grad_norm": 5.414520740509033, "learning_rate": 0.0001, "loss": 0.0579, "step": 2940 }, { "epoch": 22.519083969465647, "grad_norm": 4.728989601135254, "learning_rate": 0.0001, "loss": 0.0564, "step": 2950 }, { "epoch": 22.595419847328245, "grad_norm": 4.530935287475586, "learning_rate": 0.0001, "loss": 0.0576, "step": 2960 }, { "epoch": 22.671755725190838, "grad_norm": 5.538379192352295, "learning_rate": 0.0001, "loss": 0.0566, "step": 2970 }, { "epoch": 22.748091603053435, "grad_norm": 4.991380214691162, "learning_rate": 0.0001, "loss": 0.0593, "step": 2980 }, { "epoch": 22.82442748091603, "grad_norm": 4.083966255187988, "learning_rate": 0.0001, "loss": 0.0569, "step": 2990 }, { "epoch": 22.900763358778626, "grad_norm": 5.470835208892822, "learning_rate": 0.0001, "loss": 0.0574, "step": 3000 }, { "epoch": 22.977099236641223, "grad_norm": 3.868380308151245, "learning_rate": 0.0001, "loss": 0.0573, "step": 3010 }, { "epoch": 23.053435114503817, "grad_norm": 3.8256640434265137, "learning_rate": 0.0001, "loss": 0.0563, "step": 3020 }, { "epoch": 23.129770992366414, "grad_norm": 4.314810752868652, "learning_rate": 0.0001, "loss": 0.0542, "step": 3030 }, { "epoch": 23.206106870229007, "grad_norm": 4.500524044036865, "learning_rate": 0.0001, "loss": 0.0548, "step": 3040 }, { "epoch": 23.282442748091604, "grad_norm": 3.629347801208496, "learning_rate": 0.0001, "loss": 0.0545, "step": 3050 }, { "epoch": 23.358778625954198, "grad_norm": 4.847570896148682, "learning_rate": 0.0001, "loss": 0.0558, "step": 3060 }, { "epoch": 23.435114503816795, "grad_norm": 4.475305557250977, "learning_rate": 0.0001, "loss": 0.0571, "step": 3070 }, { "epoch": 23.51145038167939, "grad_norm": 4.393753528594971, "learning_rate": 0.0001, "loss": 0.0566, "step": 3080 }, { "epoch": 23.587786259541986, "grad_norm": 4.348673343658447, "learning_rate": 0.0001, "loss": 0.0569, "step": 3090 }, { "epoch": 23.66412213740458, "grad_norm": 3.8200294971466064, "learning_rate": 0.0001, "loss": 0.0553, "step": 3100 }, { "epoch": 23.740458015267176, "grad_norm": 3.551671266555786, "learning_rate": 0.0001, "loss": 0.0563, "step": 3110 }, { "epoch": 23.81679389312977, "grad_norm": 4.088159084320068, "learning_rate": 0.0001, "loss": 0.0545, "step": 3120 }, { "epoch": 23.893129770992367, "grad_norm": 3.8160500526428223, "learning_rate": 0.0001, "loss": 0.0561, "step": 3130 }, { "epoch": 23.96946564885496, "grad_norm": 4.097822666168213, "learning_rate": 0.0001, "loss": 0.0556, "step": 3140 }, { "epoch": 24.045801526717558, "grad_norm": 4.243167400360107, "learning_rate": 0.0001, "loss": 0.0585, "step": 3150 }, { "epoch": 24.12213740458015, "grad_norm": 4.92620325088501, "learning_rate": 0.0001, "loss": 0.0566, "step": 3160 }, { "epoch": 24.198473282442748, "grad_norm": 5.0826520919799805, "learning_rate": 0.0001, "loss": 0.0568, "step": 3170 }, { "epoch": 24.274809160305345, "grad_norm": 4.148538589477539, "learning_rate": 0.0001, "loss": 0.0547, "step": 3180 }, { "epoch": 24.35114503816794, "grad_norm": 3.6605608463287354, "learning_rate": 0.0001, "loss": 0.0545, "step": 3190 }, { "epoch": 24.427480916030536, "grad_norm": 3.980734348297119, "learning_rate": 0.0001, "loss": 0.0546, "step": 3200 }, { "epoch": 24.50381679389313, "grad_norm": 3.8442630767822266, "learning_rate": 0.0001, "loss": 0.0542, "step": 3210 }, { "epoch": 24.580152671755727, "grad_norm": 3.624903917312622, "learning_rate": 0.0001, "loss": 0.0546, "step": 3220 }, { "epoch": 24.65648854961832, "grad_norm": 3.5705676078796387, "learning_rate": 0.0001, "loss": 0.0541, "step": 3230 }, { "epoch": 24.732824427480917, "grad_norm": 4.1891255378723145, "learning_rate": 0.0001, "loss": 0.0551, "step": 3240 }, { "epoch": 24.80916030534351, "grad_norm": 3.8709261417388916, "learning_rate": 0.0001, "loss": 0.0524, "step": 3250 }, { "epoch": 24.885496183206108, "grad_norm": 4.208442211151123, "learning_rate": 0.0001, "loss": 0.0531, "step": 3260 }, { "epoch": 24.9618320610687, "grad_norm": 4.348855495452881, "learning_rate": 0.0001, "loss": 0.0537, "step": 3270 }, { "epoch": 25.0381679389313, "grad_norm": 4.784348011016846, "learning_rate": 0.0001, "loss": 0.0532, "step": 3280 }, { "epoch": 25.114503816793892, "grad_norm": 3.909681558609009, "learning_rate": 0.0001, "loss": 0.0529, "step": 3290 }, { "epoch": 25.19083969465649, "grad_norm": 4.281042575836182, "learning_rate": 0.0001, "loss": 0.0528, "step": 3300 }, { "epoch": 25.267175572519083, "grad_norm": 4.067004203796387, "learning_rate": 0.0001, "loss": 0.0524, "step": 3310 }, { "epoch": 25.34351145038168, "grad_norm": 4.297754287719727, "learning_rate": 0.0001, "loss": 0.0545, "step": 3320 }, { "epoch": 25.419847328244273, "grad_norm": 4.891724109649658, "learning_rate": 0.0001, "loss": 0.0526, "step": 3330 }, { "epoch": 25.49618320610687, "grad_norm": 4.336523056030273, "learning_rate": 0.0001, "loss": 0.0514, "step": 3340 }, { "epoch": 25.572519083969464, "grad_norm": 3.574252128601074, "learning_rate": 0.0001, "loss": 0.0514, "step": 3350 }, { "epoch": 25.64885496183206, "grad_norm": 3.998688220977783, "learning_rate": 0.0001, "loss": 0.0525, "step": 3360 }, { "epoch": 25.725190839694655, "grad_norm": 4.02937126159668, "learning_rate": 0.0001, "loss": 0.0508, "step": 3370 }, { "epoch": 25.801526717557252, "grad_norm": 3.6698431968688965, "learning_rate": 0.0001, "loss": 0.0522, "step": 3380 }, { "epoch": 25.87786259541985, "grad_norm": 3.8178515434265137, "learning_rate": 0.0001, "loss": 0.0502, "step": 3390 }, { "epoch": 25.954198473282442, "grad_norm": 3.739509344100952, "learning_rate": 0.0001, "loss": 0.0515, "step": 3400 }, { "epoch": 26.03053435114504, "grad_norm": 4.049863338470459, "learning_rate": 0.0001, "loss": 0.053, "step": 3410 }, { "epoch": 26.106870229007633, "grad_norm": 3.3929524421691895, "learning_rate": 0.0001, "loss": 0.0521, "step": 3420 }, { "epoch": 26.18320610687023, "grad_norm": 3.193859815597534, "learning_rate": 0.0001, "loss": 0.0512, "step": 3430 }, { "epoch": 26.259541984732824, "grad_norm": 4.224793434143066, "learning_rate": 0.0001, "loss": 0.0503, "step": 3440 }, { "epoch": 26.33587786259542, "grad_norm": 3.577955961227417, "learning_rate": 0.0001, "loss": 0.0515, "step": 3450 }, { "epoch": 26.412213740458014, "grad_norm": 3.4221303462982178, "learning_rate": 0.0001, "loss": 0.0499, "step": 3460 }, { "epoch": 26.48854961832061, "grad_norm": 3.89853572845459, "learning_rate": 0.0001, "loss": 0.0498, "step": 3470 }, { "epoch": 26.564885496183205, "grad_norm": 3.5571982860565186, "learning_rate": 0.0001, "loss": 0.0494, "step": 3480 }, { "epoch": 26.641221374045802, "grad_norm": 3.567416191101074, "learning_rate": 0.0001, "loss": 0.0514, "step": 3490 }, { "epoch": 26.717557251908396, "grad_norm": 4.274509429931641, "learning_rate": 0.0001, "loss": 0.0533, "step": 3500 }, { "epoch": 26.793893129770993, "grad_norm": 5.235034465789795, "learning_rate": 0.0001, "loss": 0.0544, "step": 3510 }, { "epoch": 26.870229007633586, "grad_norm": 3.2366955280303955, "learning_rate": 0.0001, "loss": 0.0521, "step": 3520 }, { "epoch": 26.946564885496183, "grad_norm": 2.9122447967529297, "learning_rate": 0.0001, "loss": 0.0499, "step": 3530 }, { "epoch": 27.022900763358777, "grad_norm": 3.433436155319214, "learning_rate": 0.0001, "loss": 0.0489, "step": 3540 }, { "epoch": 27.099236641221374, "grad_norm": 3.9485082626342773, "learning_rate": 0.0001, "loss": 0.048, "step": 3550 }, { "epoch": 27.17557251908397, "grad_norm": 3.3704323768615723, "learning_rate": 0.0001, "loss": 0.0487, "step": 3560 }, { "epoch": 27.251908396946565, "grad_norm": 3.1673593521118164, "learning_rate": 0.0001, "loss": 0.0489, "step": 3570 }, { "epoch": 27.328244274809162, "grad_norm": 4.008748531341553, "learning_rate": 0.0001, "loss": 0.051, "step": 3580 }, { "epoch": 27.404580152671755, "grad_norm": 4.217282295227051, "learning_rate": 0.0001, "loss": 0.0531, "step": 3590 }, { "epoch": 27.480916030534353, "grad_norm": 3.9805526733398438, "learning_rate": 0.0001, "loss": 0.0517, "step": 3600 }, { "epoch": 27.557251908396946, "grad_norm": 3.290966272354126, "learning_rate": 0.0001, "loss": 0.0507, "step": 3610 }, { "epoch": 27.633587786259543, "grad_norm": 4.409199237823486, "learning_rate": 0.0001, "loss": 0.0507, "step": 3620 }, { "epoch": 27.709923664122137, "grad_norm": 3.840510368347168, "learning_rate": 0.0001, "loss": 0.0495, "step": 3630 }, { "epoch": 27.786259541984734, "grad_norm": 4.1309733390808105, "learning_rate": 0.0001, "loss": 0.051, "step": 3640 }, { "epoch": 27.862595419847327, "grad_norm": 3.6765079498291016, "learning_rate": 0.0001, "loss": 0.05, "step": 3650 }, { "epoch": 27.938931297709924, "grad_norm": 3.800614356994629, "learning_rate": 0.0001, "loss": 0.0497, "step": 3660 }, { "epoch": 28.015267175572518, "grad_norm": 4.894189834594727, "learning_rate": 0.0001, "loss": 0.0485, "step": 3670 }, { "epoch": 28.091603053435115, "grad_norm": 4.065109729766846, "learning_rate": 0.0001, "loss": 0.0488, "step": 3680 }, { "epoch": 28.16793893129771, "grad_norm": 3.8664438724517822, "learning_rate": 0.0001, "loss": 0.047, "step": 3690 }, { "epoch": 28.244274809160306, "grad_norm": 3.733966827392578, "learning_rate": 0.0001, "loss": 0.0462, "step": 3700 }, { "epoch": 28.3206106870229, "grad_norm": 3.9565484523773193, "learning_rate": 0.0001, "loss": 0.0484, "step": 3710 }, { "epoch": 28.396946564885496, "grad_norm": 4.484684944152832, "learning_rate": 0.0001, "loss": 0.0504, "step": 3720 }, { "epoch": 28.47328244274809, "grad_norm": 4.26271390914917, "learning_rate": 0.0001, "loss": 0.0495, "step": 3730 }, { "epoch": 28.549618320610687, "grad_norm": 3.4989476203918457, "learning_rate": 0.0001, "loss": 0.048, "step": 3740 }, { "epoch": 28.625954198473284, "grad_norm": 3.423330545425415, "learning_rate": 0.0001, "loss": 0.0463, "step": 3750 }, { "epoch": 28.702290076335878, "grad_norm": 3.5422756671905518, "learning_rate": 0.0001, "loss": 0.0478, "step": 3760 }, { "epoch": 28.778625954198475, "grad_norm": 3.5913641452789307, "learning_rate": 0.0001, "loss": 0.0467, "step": 3770 }, { "epoch": 28.85496183206107, "grad_norm": 3.647221326828003, "learning_rate": 0.0001, "loss": 0.0492, "step": 3780 }, { "epoch": 28.931297709923665, "grad_norm": 2.952324390411377, "learning_rate": 0.0001, "loss": 0.0484, "step": 3790 }, { "epoch": 29.00763358778626, "grad_norm": 3.4018542766571045, "learning_rate": 0.0001, "loss": 0.0502, "step": 3800 }, { "epoch": 29.083969465648856, "grad_norm": 3.516319990158081, "learning_rate": 0.0001, "loss": 0.048, "step": 3810 }, { "epoch": 29.16030534351145, "grad_norm": 3.3765485286712646, "learning_rate": 0.0001, "loss": 0.0472, "step": 3820 }, { "epoch": 29.236641221374047, "grad_norm": 3.4809155464172363, "learning_rate": 0.0001, "loss": 0.0478, "step": 3830 }, { "epoch": 29.31297709923664, "grad_norm": 3.65321683883667, "learning_rate": 0.0001, "loss": 0.0492, "step": 3840 }, { "epoch": 29.389312977099237, "grad_norm": 4.674622058868408, "learning_rate": 0.0001, "loss": 0.0513, "step": 3850 }, { "epoch": 29.46564885496183, "grad_norm": 3.851452350616455, "learning_rate": 0.0001, "loss": 0.0491, "step": 3860 }, { "epoch": 29.541984732824428, "grad_norm": 3.864873170852661, "learning_rate": 0.0001, "loss": 0.0488, "step": 3870 }, { "epoch": 29.61832061068702, "grad_norm": 3.288259506225586, "learning_rate": 0.0001, "loss": 0.0509, "step": 3880 }, { "epoch": 29.69465648854962, "grad_norm": 2.7707817554473877, "learning_rate": 0.0001, "loss": 0.0482, "step": 3890 }, { "epoch": 29.770992366412212, "grad_norm": 3.1165928840637207, "learning_rate": 0.0001, "loss": 0.0482, "step": 3900 }, { "epoch": 29.84732824427481, "grad_norm": 3.1843955516815186, "learning_rate": 0.0001, "loss": 0.0485, "step": 3910 }, { "epoch": 29.923664122137403, "grad_norm": 3.3647572994232178, "learning_rate": 0.0001, "loss": 0.0501, "step": 3920 }, { "epoch": 30.0, "grad_norm": 3.520442008972168, "learning_rate": 0.0001, "loss": 0.048, "step": 3930 }, { "epoch": 30.076335877862597, "grad_norm": 2.9003231525421143, "learning_rate": 0.0001, "loss": 0.0478, "step": 3940 }, { "epoch": 30.15267175572519, "grad_norm": 3.0314769744873047, "learning_rate": 0.0001, "loss": 0.0449, "step": 3950 }, { "epoch": 30.229007633587788, "grad_norm": 3.2121691703796387, "learning_rate": 0.0001, "loss": 0.0446, "step": 3960 }, { "epoch": 30.30534351145038, "grad_norm": 3.1008832454681396, "learning_rate": 0.0001, "loss": 0.0444, "step": 3970 }, { "epoch": 30.38167938931298, "grad_norm": 2.917156219482422, "learning_rate": 0.0001, "loss": 0.0466, "step": 3980 }, { "epoch": 30.458015267175572, "grad_norm": 3.308389186859131, "learning_rate": 0.0001, "loss": 0.0466, "step": 3990 }, { "epoch": 30.53435114503817, "grad_norm": 3.5539300441741943, "learning_rate": 0.0001, "loss": 0.0477, "step": 4000 }, { "epoch": 30.610687022900763, "grad_norm": 3.7347187995910645, "learning_rate": 0.0001, "loss": 0.0474, "step": 4010 }, { "epoch": 30.68702290076336, "grad_norm": 3.404816150665283, "learning_rate": 0.0001, "loss": 0.047, "step": 4020 }, { "epoch": 30.763358778625953, "grad_norm": 3.5113792419433594, "learning_rate": 0.0001, "loss": 0.0477, "step": 4030 }, { "epoch": 30.83969465648855, "grad_norm": 3.765782356262207, "learning_rate": 0.0001, "loss": 0.0471, "step": 4040 }, { "epoch": 30.916030534351144, "grad_norm": 3.2591331005096436, "learning_rate": 0.0001, "loss": 0.0463, "step": 4050 }, { "epoch": 30.99236641221374, "grad_norm": 3.5785152912139893, "learning_rate": 0.0001, "loss": 0.0457, "step": 4060 }, { "epoch": 31.068702290076335, "grad_norm": 3.3190696239471436, "learning_rate": 0.0001, "loss": 0.0463, "step": 4070 }, { "epoch": 31.14503816793893, "grad_norm": 3.2588956356048584, "learning_rate": 0.0001, "loss": 0.0458, "step": 4080 }, { "epoch": 31.221374045801525, "grad_norm": 3.1497199535369873, "learning_rate": 0.0001, "loss": 0.0456, "step": 4090 }, { "epoch": 31.297709923664122, "grad_norm": 3.077242136001587, "learning_rate": 0.0001, "loss": 0.0432, "step": 4100 }, { "epoch": 31.374045801526716, "grad_norm": 2.980234146118164, "learning_rate": 0.0001, "loss": 0.0442, "step": 4110 }, { "epoch": 31.450381679389313, "grad_norm": 3.100454330444336, "learning_rate": 0.0001, "loss": 0.0441, "step": 4120 }, { "epoch": 31.52671755725191, "grad_norm": 3.129984140396118, "learning_rate": 0.0001, "loss": 0.0446, "step": 4130 }, { "epoch": 31.603053435114504, "grad_norm": 3.687347888946533, "learning_rate": 0.0001, "loss": 0.046, "step": 4140 }, { "epoch": 31.6793893129771, "grad_norm": 3.2343502044677734, "learning_rate": 0.0001, "loss": 0.0469, "step": 4150 }, { "epoch": 31.755725190839694, "grad_norm": 3.523177146911621, "learning_rate": 0.0001, "loss": 0.0476, "step": 4160 }, { "epoch": 31.83206106870229, "grad_norm": 3.032170057296753, "learning_rate": 0.0001, "loss": 0.0453, "step": 4170 }, { "epoch": 31.908396946564885, "grad_norm": 3.380631923675537, "learning_rate": 0.0001, "loss": 0.0459, "step": 4180 }, { "epoch": 31.984732824427482, "grad_norm": 3.342435598373413, "learning_rate": 0.0001, "loss": 0.0439, "step": 4190 }, { "epoch": 32.06106870229008, "grad_norm": 3.850493907928467, "learning_rate": 0.0001, "loss": 0.0455, "step": 4200 }, { "epoch": 32.13740458015267, "grad_norm": 3.762910842895508, "learning_rate": 0.0001, "loss": 0.0469, "step": 4210 }, { "epoch": 32.213740458015266, "grad_norm": 3.8790574073791504, "learning_rate": 0.0001, "loss": 0.0453, "step": 4220 }, { "epoch": 32.29007633587786, "grad_norm": 3.1606740951538086, "learning_rate": 0.0001, "loss": 0.0447, "step": 4230 }, { "epoch": 32.36641221374046, "grad_norm": 3.0923469066619873, "learning_rate": 0.0001, "loss": 0.0435, "step": 4240 }, { "epoch": 32.44274809160305, "grad_norm": 2.8528671264648438, "learning_rate": 0.0001, "loss": 0.0441, "step": 4250 }, { "epoch": 32.51908396946565, "grad_norm": 2.963791608810425, "learning_rate": 0.0001, "loss": 0.0448, "step": 4260 }, { "epoch": 32.595419847328245, "grad_norm": 2.594233751296997, "learning_rate": 0.0001, "loss": 0.0443, "step": 4270 }, { "epoch": 32.67175572519084, "grad_norm": 3.1827516555786133, "learning_rate": 0.0001, "loss": 0.0452, "step": 4280 }, { "epoch": 32.74809160305343, "grad_norm": 2.4301981925964355, "learning_rate": 0.0001, "loss": 0.0458, "step": 4290 }, { "epoch": 32.82442748091603, "grad_norm": 2.866853952407837, "learning_rate": 0.0001, "loss": 0.0454, "step": 4300 }, { "epoch": 32.900763358778626, "grad_norm": 3.389662981033325, "learning_rate": 0.0001, "loss": 0.0449, "step": 4310 }, { "epoch": 32.97709923664122, "grad_norm": 3.413158893585205, "learning_rate": 0.0001, "loss": 0.0456, "step": 4320 }, { "epoch": 33.05343511450382, "grad_norm": 3.0129153728485107, "learning_rate": 0.0001, "loss": 0.0442, "step": 4330 }, { "epoch": 33.12977099236641, "grad_norm": 2.75960373878479, "learning_rate": 0.0001, "loss": 0.0456, "step": 4340 }, { "epoch": 33.20610687022901, "grad_norm": 3.1550495624542236, "learning_rate": 0.0001, "loss": 0.0442, "step": 4350 }, { "epoch": 33.282442748091604, "grad_norm": 2.968376636505127, "learning_rate": 0.0001, "loss": 0.0438, "step": 4360 }, { "epoch": 33.3587786259542, "grad_norm": 3.2726564407348633, "learning_rate": 0.0001, "loss": 0.0437, "step": 4370 }, { "epoch": 33.43511450381679, "grad_norm": 3.1569831371307373, "learning_rate": 0.0001, "loss": 0.0433, "step": 4380 }, { "epoch": 33.51145038167939, "grad_norm": 3.159256935119629, "learning_rate": 0.0001, "loss": 0.0431, "step": 4390 }, { "epoch": 33.587786259541986, "grad_norm": 2.9669647216796875, "learning_rate": 0.0001, "loss": 0.0437, "step": 4400 }, { "epoch": 33.66412213740458, "grad_norm": 3.1114189624786377, "learning_rate": 0.0001, "loss": 0.0431, "step": 4410 }, { "epoch": 33.74045801526717, "grad_norm": 3.374591588973999, "learning_rate": 0.0001, "loss": 0.0429, "step": 4420 }, { "epoch": 33.81679389312977, "grad_norm": 2.7850701808929443, "learning_rate": 0.0001, "loss": 0.0453, "step": 4430 }, { "epoch": 33.89312977099237, "grad_norm": 3.1140894889831543, "learning_rate": 0.0001, "loss": 0.0435, "step": 4440 }, { "epoch": 33.969465648854964, "grad_norm": 3.252207040786743, "learning_rate": 0.0001, "loss": 0.0432, "step": 4450 }, { "epoch": 34.045801526717554, "grad_norm": 2.6837997436523438, "learning_rate": 0.0001, "loss": 0.0426, "step": 4460 }, { "epoch": 34.12213740458015, "grad_norm": 2.6828255653381348, "learning_rate": 0.0001, "loss": 0.0421, "step": 4470 }, { "epoch": 34.19847328244275, "grad_norm": 2.5278425216674805, "learning_rate": 0.0001, "loss": 0.0431, "step": 4480 }, { "epoch": 34.274809160305345, "grad_norm": 2.5507678985595703, "learning_rate": 0.0001, "loss": 0.0421, "step": 4490 }, { "epoch": 34.35114503816794, "grad_norm": 3.5572123527526855, "learning_rate": 0.0001, "loss": 0.0425, "step": 4500 }, { "epoch": 34.42748091603053, "grad_norm": 2.7339582443237305, "learning_rate": 0.0001, "loss": 0.0429, "step": 4510 }, { "epoch": 34.50381679389313, "grad_norm": 3.3866958618164062, "learning_rate": 0.0001, "loss": 0.0441, "step": 4520 }, { "epoch": 34.58015267175573, "grad_norm": 3.5509848594665527, "learning_rate": 0.0001, "loss": 0.0434, "step": 4530 }, { "epoch": 34.656488549618324, "grad_norm": 3.0683412551879883, "learning_rate": 0.0001, "loss": 0.043, "step": 4540 }, { "epoch": 34.732824427480914, "grad_norm": 2.501969814300537, "learning_rate": 0.0001, "loss": 0.042, "step": 4550 }, { "epoch": 34.80916030534351, "grad_norm": 2.817521333694458, "learning_rate": 0.0001, "loss": 0.0413, "step": 4560 }, { "epoch": 34.88549618320611, "grad_norm": 2.5012943744659424, "learning_rate": 0.0001, "loss": 0.0434, "step": 4570 }, { "epoch": 34.961832061068705, "grad_norm": 2.5811054706573486, "learning_rate": 0.0001, "loss": 0.0433, "step": 4580 }, { "epoch": 35.038167938931295, "grad_norm": 3.1649036407470703, "learning_rate": 0.0001, "loss": 0.0426, "step": 4590 }, { "epoch": 35.11450381679389, "grad_norm": 3.225064992904663, "learning_rate": 0.0001, "loss": 0.0412, "step": 4600 }, { "epoch": 35.19083969465649, "grad_norm": 3.1672935485839844, "learning_rate": 0.0001, "loss": 0.0413, "step": 4610 }, { "epoch": 35.267175572519086, "grad_norm": 2.804443836212158, "learning_rate": 0.0001, "loss": 0.0401, "step": 4620 }, { "epoch": 35.343511450381676, "grad_norm": 3.11222243309021, "learning_rate": 0.0001, "loss": 0.04, "step": 4630 }, { "epoch": 35.41984732824427, "grad_norm": 3.1439590454101562, "learning_rate": 0.0001, "loss": 0.0412, "step": 4640 }, { "epoch": 35.49618320610687, "grad_norm": 3.019904136657715, "learning_rate": 0.0001, "loss": 0.0406, "step": 4650 }, { "epoch": 35.57251908396947, "grad_norm": 2.9445254802703857, "learning_rate": 0.0001, "loss": 0.0394, "step": 4660 }, { "epoch": 35.64885496183206, "grad_norm": 3.5982344150543213, "learning_rate": 0.0001, "loss": 0.0397, "step": 4670 }, { "epoch": 35.725190839694655, "grad_norm": 3.4939627647399902, "learning_rate": 0.0001, "loss": 0.0422, "step": 4680 }, { "epoch": 35.80152671755725, "grad_norm": 3.161851644515991, "learning_rate": 0.0001, "loss": 0.0429, "step": 4690 }, { "epoch": 35.87786259541985, "grad_norm": 2.664153814315796, "learning_rate": 0.0001, "loss": 0.0418, "step": 4700 }, { "epoch": 35.954198473282446, "grad_norm": 2.573061227798462, "learning_rate": 0.0001, "loss": 0.042, "step": 4710 }, { "epoch": 36.030534351145036, "grad_norm": 2.3937745094299316, "learning_rate": 0.0001, "loss": 0.0403, "step": 4720 }, { "epoch": 36.10687022900763, "grad_norm": 2.597156524658203, "learning_rate": 0.0001, "loss": 0.04, "step": 4730 }, { "epoch": 36.18320610687023, "grad_norm": 2.66945743560791, "learning_rate": 0.0001, "loss": 0.0421, "step": 4740 }, { "epoch": 36.25954198473283, "grad_norm": 3.4073524475097656, "learning_rate": 0.0001, "loss": 0.0431, "step": 4750 }, { "epoch": 36.33587786259542, "grad_norm": 2.706364154815674, "learning_rate": 0.0001, "loss": 0.0402, "step": 4760 }, { "epoch": 36.412213740458014, "grad_norm": 2.8333280086517334, "learning_rate": 0.0001, "loss": 0.0404, "step": 4770 }, { "epoch": 36.48854961832061, "grad_norm": 2.706805944442749, "learning_rate": 0.0001, "loss": 0.0408, "step": 4780 }, { "epoch": 36.56488549618321, "grad_norm": 2.7377679347991943, "learning_rate": 0.0001, "loss": 0.0405, "step": 4790 }, { "epoch": 36.6412213740458, "grad_norm": 2.769543409347534, "learning_rate": 0.0001, "loss": 0.0406, "step": 4800 }, { "epoch": 36.717557251908396, "grad_norm": 2.555346727371216, "learning_rate": 0.0001, "loss": 0.0419, "step": 4810 }, { "epoch": 36.79389312977099, "grad_norm": 2.3741202354431152, "learning_rate": 0.0001, "loss": 0.0424, "step": 4820 }, { "epoch": 36.87022900763359, "grad_norm": 2.5709145069122314, "learning_rate": 0.0001, "loss": 0.0426, "step": 4830 }, { "epoch": 36.94656488549618, "grad_norm": 3.14483642578125, "learning_rate": 0.0001, "loss": 0.0418, "step": 4840 }, { "epoch": 37.02290076335878, "grad_norm": 2.7833096981048584, "learning_rate": 0.0001, "loss": 0.0431, "step": 4850 }, { "epoch": 37.099236641221374, "grad_norm": 2.817727565765381, "learning_rate": 0.0001, "loss": 0.0421, "step": 4860 }, { "epoch": 37.17557251908397, "grad_norm": 2.7858357429504395, "learning_rate": 0.0001, "loss": 0.0427, "step": 4870 }, { "epoch": 37.25190839694657, "grad_norm": 2.6241023540496826, "learning_rate": 0.0001, "loss": 0.0421, "step": 4880 }, { "epoch": 37.32824427480916, "grad_norm": 2.654731512069702, "learning_rate": 0.0001, "loss": 0.0402, "step": 4890 }, { "epoch": 37.404580152671755, "grad_norm": 2.5916359424591064, "learning_rate": 0.0001, "loss": 0.0404, "step": 4900 }, { "epoch": 37.48091603053435, "grad_norm": 3.016453504562378, "learning_rate": 0.0001, "loss": 0.042, "step": 4910 }, { "epoch": 37.55725190839695, "grad_norm": 2.312607526779175, "learning_rate": 0.0001, "loss": 0.0429, "step": 4920 }, { "epoch": 37.63358778625954, "grad_norm": 3.5747108459472656, "learning_rate": 0.0001, "loss": 0.0435, "step": 4930 }, { "epoch": 37.70992366412214, "grad_norm": 2.547776222229004, "learning_rate": 0.0001, "loss": 0.0421, "step": 4940 }, { "epoch": 37.786259541984734, "grad_norm": 3.1391854286193848, "learning_rate": 0.0001, "loss": 0.0414, "step": 4950 }, { "epoch": 37.86259541984733, "grad_norm": 3.249826669692993, "learning_rate": 0.0001, "loss": 0.0413, "step": 4960 }, { "epoch": 37.93893129770992, "grad_norm": 2.9902076721191406, "learning_rate": 0.0001, "loss": 0.0417, "step": 4970 }, { "epoch": 38.01526717557252, "grad_norm": 3.0540292263031006, "learning_rate": 0.0001, "loss": 0.0417, "step": 4980 }, { "epoch": 38.091603053435115, "grad_norm": 2.906388998031616, "learning_rate": 0.0001, "loss": 0.0415, "step": 4990 }, { "epoch": 38.16793893129771, "grad_norm": 2.5135858058929443, "learning_rate": 0.0001, "loss": 0.0416, "step": 5000 }, { "epoch": 38.2442748091603, "grad_norm": 3.1996307373046875, "learning_rate": 0.0001, "loss": 0.0412, "step": 5010 }, { "epoch": 38.3206106870229, "grad_norm": 2.4462337493896484, "learning_rate": 0.0001, "loss": 0.0417, "step": 5020 }, { "epoch": 38.396946564885496, "grad_norm": 2.772047281265259, "learning_rate": 0.0001, "loss": 0.0414, "step": 5030 }, { "epoch": 38.47328244274809, "grad_norm": 2.959993600845337, "learning_rate": 0.0001, "loss": 0.0405, "step": 5040 }, { "epoch": 38.54961832061069, "grad_norm": 2.269501209259033, "learning_rate": 0.0001, "loss": 0.043, "step": 5050 }, { "epoch": 38.62595419847328, "grad_norm": 2.965453624725342, "learning_rate": 0.0001, "loss": 0.0428, "step": 5060 }, { "epoch": 38.70229007633588, "grad_norm": 2.3477022647857666, "learning_rate": 0.0001, "loss": 0.0429, "step": 5070 }, { "epoch": 38.778625954198475, "grad_norm": 2.213515043258667, "learning_rate": 0.0001, "loss": 0.0414, "step": 5080 }, { "epoch": 38.85496183206107, "grad_norm": 2.86283016204834, "learning_rate": 0.0001, "loss": 0.0428, "step": 5090 }, { "epoch": 38.93129770992366, "grad_norm": 2.793640375137329, "learning_rate": 0.0001, "loss": 0.0424, "step": 5100 }, { "epoch": 39.00763358778626, "grad_norm": 2.6359457969665527, "learning_rate": 0.0001, "loss": 0.0424, "step": 5110 }, { "epoch": 39.083969465648856, "grad_norm": 3.023926258087158, "learning_rate": 0.0001, "loss": 0.0431, "step": 5120 }, { "epoch": 39.16030534351145, "grad_norm": 3.0659825801849365, "learning_rate": 0.0001, "loss": 0.0413, "step": 5130 }, { "epoch": 39.23664122137404, "grad_norm": 2.9539597034454346, "learning_rate": 0.0001, "loss": 0.04, "step": 5140 }, { "epoch": 39.31297709923664, "grad_norm": 3.309925079345703, "learning_rate": 0.0001, "loss": 0.0401, "step": 5150 }, { "epoch": 39.38931297709924, "grad_norm": 3.2164924144744873, "learning_rate": 0.0001, "loss": 0.039, "step": 5160 }, { "epoch": 39.465648854961835, "grad_norm": 2.320535182952881, "learning_rate": 0.0001, "loss": 0.0383, "step": 5170 }, { "epoch": 39.541984732824424, "grad_norm": 2.5562503337860107, "learning_rate": 0.0001, "loss": 0.0386, "step": 5180 }, { "epoch": 39.61832061068702, "grad_norm": 2.869114875793457, "learning_rate": 0.0001, "loss": 0.0371, "step": 5190 }, { "epoch": 39.69465648854962, "grad_norm": 3.0522701740264893, "learning_rate": 0.0001, "loss": 0.0374, "step": 5200 }, { "epoch": 39.770992366412216, "grad_norm": 2.6509947776794434, "learning_rate": 0.0001, "loss": 0.0398, "step": 5210 }, { "epoch": 39.847328244274806, "grad_norm": 2.6635944843292236, "learning_rate": 0.0001, "loss": 0.0379, "step": 5220 }, { "epoch": 39.9236641221374, "grad_norm": 2.1523704528808594, "learning_rate": 0.0001, "loss": 0.0387, "step": 5230 }, { "epoch": 40.0, "grad_norm": 2.5757007598876953, "learning_rate": 0.0001, "loss": 0.039, "step": 5240 }, { "epoch": 40.0763358778626, "grad_norm": 2.3721554279327393, "learning_rate": 0.0001, "loss": 0.0378, "step": 5250 }, { "epoch": 40.152671755725194, "grad_norm": 2.766690492630005, "learning_rate": 0.0001, "loss": 0.0387, "step": 5260 }, { "epoch": 40.229007633587784, "grad_norm": 3.0696401596069336, "learning_rate": 0.0001, "loss": 0.0395, "step": 5270 }, { "epoch": 40.30534351145038, "grad_norm": 2.8735852241516113, "learning_rate": 0.0001, "loss": 0.0405, "step": 5280 }, { "epoch": 40.38167938931298, "grad_norm": 3.6144680976867676, "learning_rate": 0.0001, "loss": 0.0392, "step": 5290 }, { "epoch": 40.458015267175576, "grad_norm": 2.605578660964966, "learning_rate": 0.0001, "loss": 0.0396, "step": 5300 }, { "epoch": 40.534351145038165, "grad_norm": 2.1938908100128174, "learning_rate": 0.0001, "loss": 0.0407, "step": 5310 }, { "epoch": 40.61068702290076, "grad_norm": 3.2314584255218506, "learning_rate": 0.0001, "loss": 0.0424, "step": 5320 }, { "epoch": 40.68702290076336, "grad_norm": 2.520944356918335, "learning_rate": 0.0001, "loss": 0.0399, "step": 5330 }, { "epoch": 40.76335877862596, "grad_norm": 2.6674437522888184, "learning_rate": 0.0001, "loss": 0.0398, "step": 5340 }, { "epoch": 40.83969465648855, "grad_norm": 2.5482187271118164, "learning_rate": 0.0001, "loss": 0.0381, "step": 5350 }, { "epoch": 40.916030534351144, "grad_norm": 3.999814510345459, "learning_rate": 0.0001, "loss": 0.042, "step": 5360 }, { "epoch": 40.99236641221374, "grad_norm": 2.9117038249969482, "learning_rate": 0.0001, "loss": 0.0403, "step": 5370 }, { "epoch": 41.06870229007634, "grad_norm": 2.936765193939209, "learning_rate": 0.0001, "loss": 0.0388, "step": 5380 }, { "epoch": 41.14503816793893, "grad_norm": 2.87949538230896, "learning_rate": 0.0001, "loss": 0.0395, "step": 5390 }, { "epoch": 41.221374045801525, "grad_norm": 2.8703343868255615, "learning_rate": 0.0001, "loss": 0.0387, "step": 5400 }, { "epoch": 41.29770992366412, "grad_norm": 3.175351858139038, "learning_rate": 0.0001, "loss": 0.0401, "step": 5410 }, { "epoch": 41.37404580152672, "grad_norm": 2.6117746829986572, "learning_rate": 0.0001, "loss": 0.0388, "step": 5420 }, { "epoch": 41.45038167938931, "grad_norm": 2.757021427154541, "learning_rate": 0.0001, "loss": 0.0375, "step": 5430 }, { "epoch": 41.52671755725191, "grad_norm": 2.846672534942627, "learning_rate": 0.0001, "loss": 0.0373, "step": 5440 }, { "epoch": 41.603053435114504, "grad_norm": 3.156965494155884, "learning_rate": 0.0001, "loss": 0.0394, "step": 5450 }, { "epoch": 41.6793893129771, "grad_norm": 2.277841329574585, "learning_rate": 0.0001, "loss": 0.0386, "step": 5460 }, { "epoch": 41.7557251908397, "grad_norm": 2.198606252670288, "learning_rate": 0.0001, "loss": 0.0389, "step": 5470 }, { "epoch": 41.83206106870229, "grad_norm": 2.734288215637207, "learning_rate": 0.0001, "loss": 0.0375, "step": 5480 }, { "epoch": 41.908396946564885, "grad_norm": 2.5212509632110596, "learning_rate": 0.0001, "loss": 0.0375, "step": 5490 }, { "epoch": 41.98473282442748, "grad_norm": 2.5822556018829346, "learning_rate": 0.0001, "loss": 0.0377, "step": 5500 }, { "epoch": 42.06106870229008, "grad_norm": 2.8420071601867676, "learning_rate": 0.0001, "loss": 0.0373, "step": 5510 }, { "epoch": 42.13740458015267, "grad_norm": 2.6207613945007324, "learning_rate": 0.0001, "loss": 0.038, "step": 5520 }, { "epoch": 42.213740458015266, "grad_norm": 2.6019492149353027, "learning_rate": 0.0001, "loss": 0.0377, "step": 5530 }, { "epoch": 42.29007633587786, "grad_norm": 2.3905889987945557, "learning_rate": 0.0001, "loss": 0.0369, "step": 5540 }, { "epoch": 42.36641221374046, "grad_norm": 2.601881742477417, "learning_rate": 0.0001, "loss": 0.0377, "step": 5550 }, { "epoch": 42.44274809160305, "grad_norm": 2.4978208541870117, "learning_rate": 0.0001, "loss": 0.0375, "step": 5560 }, { "epoch": 42.51908396946565, "grad_norm": 2.7590718269348145, "learning_rate": 0.0001, "loss": 0.039, "step": 5570 }, { "epoch": 42.595419847328245, "grad_norm": 2.519303798675537, "learning_rate": 0.0001, "loss": 0.0386, "step": 5580 }, { "epoch": 42.67175572519084, "grad_norm": 2.4339895248413086, "learning_rate": 0.0001, "loss": 0.0386, "step": 5590 }, { "epoch": 42.74809160305343, "grad_norm": 2.0476951599121094, "learning_rate": 0.0001, "loss": 0.0385, "step": 5600 }, { "epoch": 42.82442748091603, "grad_norm": 2.4573683738708496, "learning_rate": 0.0001, "loss": 0.0384, "step": 5610 }, { "epoch": 42.900763358778626, "grad_norm": 3.260197639465332, "learning_rate": 0.0001, "loss": 0.0391, "step": 5620 }, { "epoch": 42.97709923664122, "grad_norm": 2.3142921924591064, "learning_rate": 0.0001, "loss": 0.0385, "step": 5630 }, { "epoch": 43.05343511450382, "grad_norm": 2.687488317489624, "learning_rate": 0.0001, "loss": 0.0379, "step": 5640 }, { "epoch": 43.12977099236641, "grad_norm": 2.3007423877716064, "learning_rate": 0.0001, "loss": 0.0369, "step": 5650 }, { "epoch": 43.20610687022901, "grad_norm": 2.827714204788208, "learning_rate": 0.0001, "loss": 0.0388, "step": 5660 }, { "epoch": 43.282442748091604, "grad_norm": 2.5639729499816895, "learning_rate": 0.0001, "loss": 0.0381, "step": 5670 }, { "epoch": 43.3587786259542, "grad_norm": 2.606452226638794, "learning_rate": 0.0001, "loss": 0.0385, "step": 5680 }, { "epoch": 43.43511450381679, "grad_norm": 2.1021220684051514, "learning_rate": 0.0001, "loss": 0.0387, "step": 5690 }, { "epoch": 43.51145038167939, "grad_norm": 3.2288801670074463, "learning_rate": 0.0001, "loss": 0.0398, "step": 5700 }, { "epoch": 43.587786259541986, "grad_norm": 2.6732211112976074, "learning_rate": 0.0001, "loss": 0.0397, "step": 5710 }, { "epoch": 43.66412213740458, "grad_norm": 3.0404343605041504, "learning_rate": 0.0001, "loss": 0.0405, "step": 5720 }, { "epoch": 43.74045801526717, "grad_norm": 3.020749568939209, "learning_rate": 0.0001, "loss": 0.0385, "step": 5730 }, { "epoch": 43.81679389312977, "grad_norm": 2.623088836669922, "learning_rate": 0.0001, "loss": 0.0372, "step": 5740 }, { "epoch": 43.89312977099237, "grad_norm": 2.6659088134765625, "learning_rate": 0.0001, "loss": 0.0367, "step": 5750 }, { "epoch": 43.969465648854964, "grad_norm": 2.6762301921844482, "learning_rate": 0.0001, "loss": 0.0361, "step": 5760 }, { "epoch": 44.045801526717554, "grad_norm": 2.7373850345611572, "learning_rate": 0.0001, "loss": 0.0369, "step": 5770 }, { "epoch": 44.12213740458015, "grad_norm": 2.8411030769348145, "learning_rate": 0.0001, "loss": 0.0358, "step": 5780 }, { "epoch": 44.19847328244275, "grad_norm": 2.2531847953796387, "learning_rate": 0.0001, "loss": 0.0364, "step": 5790 }, { "epoch": 44.274809160305345, "grad_norm": 2.5787270069122314, "learning_rate": 0.0001, "loss": 0.0351, "step": 5800 }, { "epoch": 44.35114503816794, "grad_norm": 2.7405145168304443, "learning_rate": 0.0001, "loss": 0.0366, "step": 5810 }, { "epoch": 44.42748091603053, "grad_norm": 2.2316906452178955, "learning_rate": 0.0001, "loss": 0.0364, "step": 5820 }, { "epoch": 44.50381679389313, "grad_norm": 2.316192865371704, "learning_rate": 0.0001, "loss": 0.0359, "step": 5830 }, { "epoch": 44.58015267175573, "grad_norm": 2.497990846633911, "learning_rate": 0.0001, "loss": 0.036, "step": 5840 }, { "epoch": 44.656488549618324, "grad_norm": 2.7521727085113525, "learning_rate": 0.0001, "loss": 0.0371, "step": 5850 }, { "epoch": 44.732824427480914, "grad_norm": 3.0083391666412354, "learning_rate": 0.0001, "loss": 0.0383, "step": 5860 }, { "epoch": 44.80916030534351, "grad_norm": 2.2850751876831055, "learning_rate": 0.0001, "loss": 0.0377, "step": 5870 }, { "epoch": 44.88549618320611, "grad_norm": 2.4022581577301025, "learning_rate": 0.0001, "loss": 0.0363, "step": 5880 }, { "epoch": 44.961832061068705, "grad_norm": 2.6845862865448, "learning_rate": 0.0001, "loss": 0.0365, "step": 5890 }, { "epoch": 45.038167938931295, "grad_norm": 2.6063520908355713, "learning_rate": 0.0001, "loss": 0.0366, "step": 5900 }, { "epoch": 45.11450381679389, "grad_norm": 2.1554946899414062, "learning_rate": 0.0001, "loss": 0.0358, "step": 5910 }, { "epoch": 45.19083969465649, "grad_norm": 2.1477222442626953, "learning_rate": 0.0001, "loss": 0.0361, "step": 5920 }, { "epoch": 45.267175572519086, "grad_norm": 2.355727434158325, "learning_rate": 0.0001, "loss": 0.037, "step": 5930 }, { "epoch": 45.343511450381676, "grad_norm": 2.6387155055999756, "learning_rate": 0.0001, "loss": 0.036, "step": 5940 }, { "epoch": 45.41984732824427, "grad_norm": 1.66975998878479, "learning_rate": 0.0001, "loss": 0.0358, "step": 5950 }, { "epoch": 45.49618320610687, "grad_norm": 2.0654666423797607, "learning_rate": 0.0001, "loss": 0.036, "step": 5960 }, { "epoch": 45.57251908396947, "grad_norm": 2.176196813583374, "learning_rate": 0.0001, "loss": 0.0361, "step": 5970 }, { "epoch": 45.64885496183206, "grad_norm": 2.388503074645996, "learning_rate": 0.0001, "loss": 0.0375, "step": 5980 }, { "epoch": 45.725190839694655, "grad_norm": 2.734039545059204, "learning_rate": 0.0001, "loss": 0.0408, "step": 5990 }, { "epoch": 45.80152671755725, "grad_norm": 2.451106309890747, "learning_rate": 0.0001, "loss": 0.0407, "step": 6000 }, { "epoch": 45.87786259541985, "grad_norm": 2.5229203701019287, "learning_rate": 0.0001, "loss": 0.0407, "step": 6010 }, { "epoch": 45.954198473282446, "grad_norm": 2.5338292121887207, "learning_rate": 0.0001, "loss": 0.0391, "step": 6020 }, { "epoch": 46.030534351145036, "grad_norm": 2.627972364425659, "learning_rate": 0.0001, "loss": 0.0387, "step": 6030 }, { "epoch": 46.10687022900763, "grad_norm": 2.2555620670318604, "learning_rate": 0.0001, "loss": 0.0387, "step": 6040 }, { "epoch": 46.18320610687023, "grad_norm": 1.9861828088760376, "learning_rate": 0.0001, "loss": 0.0384, "step": 6050 }, { "epoch": 46.25954198473283, "grad_norm": 1.7634623050689697, "learning_rate": 0.0001, "loss": 0.0378, "step": 6060 }, { "epoch": 46.33587786259542, "grad_norm": 2.021008014678955, "learning_rate": 0.0001, "loss": 0.0382, "step": 6070 }, { "epoch": 46.412213740458014, "grad_norm": 2.508978843688965, "learning_rate": 0.0001, "loss": 0.0371, "step": 6080 }, { "epoch": 46.48854961832061, "grad_norm": 2.227961301803589, "learning_rate": 0.0001, "loss": 0.0369, "step": 6090 }, { "epoch": 46.56488549618321, "grad_norm": 1.9804142713546753, "learning_rate": 0.0001, "loss": 0.0363, "step": 6100 }, { "epoch": 46.6412213740458, "grad_norm": 2.3933331966400146, "learning_rate": 0.0001, "loss": 0.0346, "step": 6110 }, { "epoch": 46.717557251908396, "grad_norm": 2.5680012702941895, "learning_rate": 0.0001, "loss": 0.0368, "step": 6120 }, { "epoch": 46.79389312977099, "grad_norm": 2.6341497898101807, "learning_rate": 0.0001, "loss": 0.0387, "step": 6130 }, { "epoch": 46.87022900763359, "grad_norm": 2.839428424835205, "learning_rate": 0.0001, "loss": 0.0368, "step": 6140 }, { "epoch": 46.94656488549618, "grad_norm": 2.2205631732940674, "learning_rate": 0.0001, "loss": 0.0385, "step": 6150 }, { "epoch": 47.02290076335878, "grad_norm": 2.480875015258789, "learning_rate": 0.0001, "loss": 0.0384, "step": 6160 }, { "epoch": 47.099236641221374, "grad_norm": 2.3821444511413574, "learning_rate": 0.0001, "loss": 0.0378, "step": 6170 }, { "epoch": 47.17557251908397, "grad_norm": 2.958724021911621, "learning_rate": 0.0001, "loss": 0.0367, "step": 6180 }, { "epoch": 47.25190839694657, "grad_norm": 2.0941898822784424, "learning_rate": 0.0001, "loss": 0.0357, "step": 6190 }, { "epoch": 47.32824427480916, "grad_norm": 2.3248085975646973, "learning_rate": 0.0001, "loss": 0.0358, "step": 6200 }, { "epoch": 47.404580152671755, "grad_norm": 2.7085769176483154, "learning_rate": 0.0001, "loss": 0.035, "step": 6210 }, { "epoch": 47.48091603053435, "grad_norm": 2.28778338432312, "learning_rate": 0.0001, "loss": 0.0374, "step": 6220 }, { "epoch": 47.55725190839695, "grad_norm": 2.7257072925567627, "learning_rate": 0.0001, "loss": 0.0385, "step": 6230 }, { "epoch": 47.63358778625954, "grad_norm": 3.0156049728393555, "learning_rate": 0.0001, "loss": 0.0378, "step": 6240 }, { "epoch": 47.70992366412214, "grad_norm": 2.686002254486084, "learning_rate": 0.0001, "loss": 0.0384, "step": 6250 }, { "epoch": 47.786259541984734, "grad_norm": 2.6361515522003174, "learning_rate": 0.0001, "loss": 0.0367, "step": 6260 }, { "epoch": 47.86259541984733, "grad_norm": 2.016510009765625, "learning_rate": 0.0001, "loss": 0.0371, "step": 6270 }, { "epoch": 47.93893129770992, "grad_norm": 2.050410270690918, "learning_rate": 0.0001, "loss": 0.0371, "step": 6280 }, { "epoch": 48.01526717557252, "grad_norm": 2.5446548461914062, "learning_rate": 0.0001, "loss": 0.0361, "step": 6290 }, { "epoch": 48.091603053435115, "grad_norm": 2.8214776515960693, "learning_rate": 0.0001, "loss": 0.0359, "step": 6300 }, { "epoch": 48.16793893129771, "grad_norm": 2.3898675441741943, "learning_rate": 0.0001, "loss": 0.0372, "step": 6310 }, { "epoch": 48.2442748091603, "grad_norm": 2.7222471237182617, "learning_rate": 0.0001, "loss": 0.0372, "step": 6320 }, { "epoch": 48.3206106870229, "grad_norm": 2.2025489807128906, "learning_rate": 0.0001, "loss": 0.0376, "step": 6330 }, { "epoch": 48.396946564885496, "grad_norm": 2.16463041305542, "learning_rate": 0.0001, "loss": 0.0357, "step": 6340 }, { "epoch": 48.47328244274809, "grad_norm": 2.5995137691497803, "learning_rate": 0.0001, "loss": 0.0363, "step": 6350 }, { "epoch": 48.54961832061069, "grad_norm": 2.5997602939605713, "learning_rate": 0.0001, "loss": 0.0359, "step": 6360 }, { "epoch": 48.62595419847328, "grad_norm": 2.7237067222595215, "learning_rate": 0.0001, "loss": 0.0358, "step": 6370 }, { "epoch": 48.70229007633588, "grad_norm": 2.3695788383483887, "learning_rate": 0.0001, "loss": 0.0351, "step": 6380 }, { "epoch": 48.778625954198475, "grad_norm": 2.2954654693603516, "learning_rate": 0.0001, "loss": 0.0333, "step": 6390 }, { "epoch": 48.85496183206107, "grad_norm": 2.5193090438842773, "learning_rate": 0.0001, "loss": 0.0349, "step": 6400 }, { "epoch": 48.93129770992366, "grad_norm": 1.9484264850616455, "learning_rate": 0.0001, "loss": 0.035, "step": 6410 }, { "epoch": 49.00763358778626, "grad_norm": 2.5910024642944336, "learning_rate": 0.0001, "loss": 0.0357, "step": 6420 }, { "epoch": 49.083969465648856, "grad_norm": 2.1642863750457764, "learning_rate": 0.0001, "loss": 0.0356, "step": 6430 }, { "epoch": 49.16030534351145, "grad_norm": 2.317484140396118, "learning_rate": 0.0001, "loss": 0.035, "step": 6440 }, { "epoch": 49.23664122137404, "grad_norm": 2.260068416595459, "learning_rate": 0.0001, "loss": 0.0365, "step": 6450 }, { "epoch": 49.31297709923664, "grad_norm": 2.4356212615966797, "learning_rate": 0.0001, "loss": 0.0348, "step": 6460 }, { "epoch": 49.38931297709924, "grad_norm": 2.4276459217071533, "learning_rate": 0.0001, "loss": 0.0345, "step": 6470 }, { "epoch": 49.465648854961835, "grad_norm": 2.5710182189941406, "learning_rate": 0.0001, "loss": 0.0362, "step": 6480 }, { "epoch": 49.541984732824424, "grad_norm": 3.2164297103881836, "learning_rate": 0.0001, "loss": 0.0348, "step": 6490 }, { "epoch": 49.61832061068702, "grad_norm": 2.473411798477173, "learning_rate": 0.0001, "loss": 0.0355, "step": 6500 }, { "epoch": 49.69465648854962, "grad_norm": 2.725154399871826, "learning_rate": 0.0001, "loss": 0.035, "step": 6510 }, { "epoch": 49.770992366412216, "grad_norm": 2.3573341369628906, "learning_rate": 0.0001, "loss": 0.0344, "step": 6520 }, { "epoch": 49.847328244274806, "grad_norm": 1.8914446830749512, "learning_rate": 0.0001, "loss": 0.0346, "step": 6530 }, { "epoch": 49.9236641221374, "grad_norm": 1.9303611516952515, "learning_rate": 0.0001, "loss": 0.0336, "step": 6540 }, { "epoch": 50.0, "grad_norm": 1.8684591054916382, "learning_rate": 0.0001, "loss": 0.033, "step": 6550 }, { "epoch": 50.0763358778626, "grad_norm": 1.9339630603790283, "learning_rate": 0.0001, "loss": 0.0349, "step": 6560 }, { "epoch": 50.152671755725194, "grad_norm": 2.3936593532562256, "learning_rate": 0.0001, "loss": 0.0348, "step": 6570 }, { "epoch": 50.229007633587784, "grad_norm": 1.9896608591079712, "learning_rate": 0.0001, "loss": 0.0345, "step": 6580 }, { "epoch": 50.30534351145038, "grad_norm": 2.535001754760742, "learning_rate": 0.0001, "loss": 0.0365, "step": 6590 }, { "epoch": 50.38167938931298, "grad_norm": 2.274082899093628, "learning_rate": 0.0001, "loss": 0.0361, "step": 6600 }, { "epoch": 50.458015267175576, "grad_norm": 2.206608533859253, "learning_rate": 0.0001, "loss": 0.0348, "step": 6610 }, { "epoch": 50.534351145038165, "grad_norm": 2.018345832824707, "learning_rate": 0.0001, "loss": 0.0341, "step": 6620 }, { "epoch": 50.61068702290076, "grad_norm": 2.2754480838775635, "learning_rate": 0.0001, "loss": 0.0343, "step": 6630 }, { "epoch": 50.68702290076336, "grad_norm": 2.2988085746765137, "learning_rate": 0.0001, "loss": 0.0342, "step": 6640 }, { "epoch": 50.76335877862596, "grad_norm": 2.1542584896087646, "learning_rate": 0.0001, "loss": 0.0344, "step": 6650 }, { "epoch": 50.83969465648855, "grad_norm": 1.9080113172531128, "learning_rate": 0.0001, "loss": 0.0351, "step": 6660 }, { "epoch": 50.916030534351144, "grad_norm": 2.6312742233276367, "learning_rate": 0.0001, "loss": 0.0359, "step": 6670 }, { "epoch": 50.99236641221374, "grad_norm": 2.002800703048706, "learning_rate": 0.0001, "loss": 0.0345, "step": 6680 }, { "epoch": 51.06870229007634, "grad_norm": 2.1637744903564453, "learning_rate": 0.0001, "loss": 0.0344, "step": 6690 }, { "epoch": 51.14503816793893, "grad_norm": 1.958133339881897, "learning_rate": 0.0001, "loss": 0.0356, "step": 6700 }, { "epoch": 51.221374045801525, "grad_norm": 2.380970001220703, "learning_rate": 0.0001, "loss": 0.0357, "step": 6710 }, { "epoch": 51.29770992366412, "grad_norm": 2.008175849914551, "learning_rate": 0.0001, "loss": 0.0353, "step": 6720 }, { "epoch": 51.37404580152672, "grad_norm": 2.097261667251587, "learning_rate": 0.0001, "loss": 0.0368, "step": 6730 }, { "epoch": 51.45038167938931, "grad_norm": 2.2324330806732178, "learning_rate": 0.0001, "loss": 0.0359, "step": 6740 }, { "epoch": 51.52671755725191, "grad_norm": 1.9775235652923584, "learning_rate": 0.0001, "loss": 0.0367, "step": 6750 }, { "epoch": 51.603053435114504, "grad_norm": 2.281738519668579, "learning_rate": 0.0001, "loss": 0.0356, "step": 6760 }, { "epoch": 51.6793893129771, "grad_norm": 2.6676409244537354, "learning_rate": 0.0001, "loss": 0.0368, "step": 6770 }, { "epoch": 51.7557251908397, "grad_norm": 2.555079460144043, "learning_rate": 0.0001, "loss": 0.0368, "step": 6780 }, { "epoch": 51.83206106870229, "grad_norm": 2.4438016414642334, "learning_rate": 0.0001, "loss": 0.0362, "step": 6790 }, { "epoch": 51.908396946564885, "grad_norm": 2.0046701431274414, "learning_rate": 0.0001, "loss": 0.0356, "step": 6800 }, { "epoch": 51.98473282442748, "grad_norm": 1.6842252016067505, "learning_rate": 0.0001, "loss": 0.0365, "step": 6810 }, { "epoch": 52.06106870229008, "grad_norm": 2.6614902019500732, "learning_rate": 0.0001, "loss": 0.0362, "step": 6820 }, { "epoch": 52.13740458015267, "grad_norm": 2.728041887283325, "learning_rate": 0.0001, "loss": 0.0364, "step": 6830 }, { "epoch": 52.213740458015266, "grad_norm": 2.527210235595703, "learning_rate": 0.0001, "loss": 0.0355, "step": 6840 }, { "epoch": 52.29007633587786, "grad_norm": 2.3304595947265625, "learning_rate": 0.0001, "loss": 0.0353, "step": 6850 }, { "epoch": 52.36641221374046, "grad_norm": 2.0097153186798096, "learning_rate": 0.0001, "loss": 0.0348, "step": 6860 }, { "epoch": 52.44274809160305, "grad_norm": 1.765758752822876, "learning_rate": 0.0001, "loss": 0.035, "step": 6870 }, { "epoch": 52.51908396946565, "grad_norm": 2.134899616241455, "learning_rate": 0.0001, "loss": 0.0345, "step": 6880 }, { "epoch": 52.595419847328245, "grad_norm": 2.6896471977233887, "learning_rate": 0.0001, "loss": 0.0357, "step": 6890 }, { "epoch": 52.67175572519084, "grad_norm": 2.36889910697937, "learning_rate": 0.0001, "loss": 0.035, "step": 6900 }, { "epoch": 52.74809160305343, "grad_norm": 2.735297203063965, "learning_rate": 0.0001, "loss": 0.0366, "step": 6910 }, { "epoch": 52.82442748091603, "grad_norm": 2.2026584148406982, "learning_rate": 0.0001, "loss": 0.0346, "step": 6920 }, { "epoch": 52.900763358778626, "grad_norm": 2.9628148078918457, "learning_rate": 0.0001, "loss": 0.0358, "step": 6930 }, { "epoch": 52.97709923664122, "grad_norm": 2.1205177307128906, "learning_rate": 0.0001, "loss": 0.0364, "step": 6940 }, { "epoch": 53.05343511450382, "grad_norm": 2.1675751209259033, "learning_rate": 0.0001, "loss": 0.0374, "step": 6950 }, { "epoch": 53.12977099236641, "grad_norm": 3.127397060394287, "learning_rate": 0.0001, "loss": 0.0361, "step": 6960 }, { "epoch": 53.20610687022901, "grad_norm": 2.1476891040802, "learning_rate": 0.0001, "loss": 0.0344, "step": 6970 }, { "epoch": 53.282442748091604, "grad_norm": 1.985656499862671, "learning_rate": 0.0001, "loss": 0.0327, "step": 6980 }, { "epoch": 53.3587786259542, "grad_norm": 2.0476458072662354, "learning_rate": 0.0001, "loss": 0.0337, "step": 6990 }, { "epoch": 53.43511450381679, "grad_norm": 2.593846082687378, "learning_rate": 0.0001, "loss": 0.0322, "step": 7000 }, { "epoch": 53.51145038167939, "grad_norm": 2.6332850456237793, "learning_rate": 0.0001, "loss": 0.0327, "step": 7010 }, { "epoch": 53.587786259541986, "grad_norm": 2.657205104827881, "learning_rate": 0.0001, "loss": 0.0332, "step": 7020 }, { "epoch": 53.66412213740458, "grad_norm": 2.7496702671051025, "learning_rate": 0.0001, "loss": 0.0338, "step": 7030 }, { "epoch": 53.74045801526717, "grad_norm": 2.704888343811035, "learning_rate": 0.0001, "loss": 0.033, "step": 7040 }, { "epoch": 53.81679389312977, "grad_norm": 2.0050015449523926, "learning_rate": 0.0001, "loss": 0.0323, "step": 7050 }, { "epoch": 53.89312977099237, "grad_norm": 2.837059736251831, "learning_rate": 0.0001, "loss": 0.0333, "step": 7060 }, { "epoch": 53.969465648854964, "grad_norm": 2.4059176445007324, "learning_rate": 0.0001, "loss": 0.0329, "step": 7070 }, { "epoch": 54.045801526717554, "grad_norm": 2.162334442138672, "learning_rate": 0.0001, "loss": 0.0328, "step": 7080 }, { "epoch": 54.12213740458015, "grad_norm": 2.8191678524017334, "learning_rate": 0.0001, "loss": 0.0329, "step": 7090 }, { "epoch": 54.19847328244275, "grad_norm": 1.891901969909668, "learning_rate": 0.0001, "loss": 0.0335, "step": 7100 }, { "epoch": 54.274809160305345, "grad_norm": 2.989685297012329, "learning_rate": 0.0001, "loss": 0.0329, "step": 7110 }, { "epoch": 54.35114503816794, "grad_norm": 1.939099907875061, "learning_rate": 0.0001, "loss": 0.0323, "step": 7120 }, { "epoch": 54.42748091603053, "grad_norm": 2.4207868576049805, "learning_rate": 0.0001, "loss": 0.035, "step": 7130 }, { "epoch": 54.50381679389313, "grad_norm": 2.2917399406433105, "learning_rate": 0.0001, "loss": 0.0339, "step": 7140 }, { "epoch": 54.58015267175573, "grad_norm": 2.973111867904663, "learning_rate": 0.0001, "loss": 0.0338, "step": 7150 }, { "epoch": 54.656488549618324, "grad_norm": 2.516721725463867, "learning_rate": 0.0001, "loss": 0.034, "step": 7160 }, { "epoch": 54.732824427480914, "grad_norm": 2.585022449493408, "learning_rate": 0.0001, "loss": 0.0351, "step": 7170 }, { "epoch": 54.80916030534351, "grad_norm": 2.2500391006469727, "learning_rate": 0.0001, "loss": 0.034, "step": 7180 }, { "epoch": 54.88549618320611, "grad_norm": 2.537076234817505, "learning_rate": 0.0001, "loss": 0.0341, "step": 7190 }, { "epoch": 54.961832061068705, "grad_norm": 2.7653725147247314, "learning_rate": 0.0001, "loss": 0.0334, "step": 7200 }, { "epoch": 55.038167938931295, "grad_norm": 2.5402090549468994, "learning_rate": 0.0001, "loss": 0.0344, "step": 7210 }, { "epoch": 55.11450381679389, "grad_norm": 2.3640823364257812, "learning_rate": 0.0001, "loss": 0.0322, "step": 7220 }, { "epoch": 55.19083969465649, "grad_norm": 2.2265453338623047, "learning_rate": 0.0001, "loss": 0.0345, "step": 7230 }, { "epoch": 55.267175572519086, "grad_norm": 2.858018398284912, "learning_rate": 0.0001, "loss": 0.0331, "step": 7240 }, { "epoch": 55.343511450381676, "grad_norm": 2.3819127082824707, "learning_rate": 0.0001, "loss": 0.0339, "step": 7250 }, { "epoch": 55.41984732824427, "grad_norm": 2.491863250732422, "learning_rate": 0.0001, "loss": 0.0324, "step": 7260 }, { "epoch": 55.49618320610687, "grad_norm": 2.265744209289551, "learning_rate": 0.0001, "loss": 0.0326, "step": 7270 }, { "epoch": 55.57251908396947, "grad_norm": 1.7130805253982544, "learning_rate": 0.0001, "loss": 0.0325, "step": 7280 }, { "epoch": 55.64885496183206, "grad_norm": 1.94850754737854, "learning_rate": 0.0001, "loss": 0.0332, "step": 7290 }, { "epoch": 55.725190839694655, "grad_norm": 1.782880425453186, "learning_rate": 0.0001, "loss": 0.0324, "step": 7300 }, { "epoch": 55.80152671755725, "grad_norm": 2.1322708129882812, "learning_rate": 0.0001, "loss": 0.0338, "step": 7310 }, { "epoch": 55.87786259541985, "grad_norm": 1.7269765138626099, "learning_rate": 0.0001, "loss": 0.0335, "step": 7320 }, { "epoch": 55.954198473282446, "grad_norm": 2.0144431591033936, "learning_rate": 0.0001, "loss": 0.0327, "step": 7330 }, { "epoch": 56.030534351145036, "grad_norm": 2.0713956356048584, "learning_rate": 0.0001, "loss": 0.0344, "step": 7340 }, { "epoch": 56.10687022900763, "grad_norm": 2.053267002105713, "learning_rate": 0.0001, "loss": 0.0347, "step": 7350 }, { "epoch": 56.18320610687023, "grad_norm": 2.925492286682129, "learning_rate": 0.0001, "loss": 0.0352, "step": 7360 }, { "epoch": 56.25954198473283, "grad_norm": 2.0910351276397705, "learning_rate": 0.0001, "loss": 0.0354, "step": 7370 }, { "epoch": 56.33587786259542, "grad_norm": 1.8954061269760132, "learning_rate": 0.0001, "loss": 0.0349, "step": 7380 }, { "epoch": 56.412213740458014, "grad_norm": 2.2161002159118652, "learning_rate": 0.0001, "loss": 0.0354, "step": 7390 }, { "epoch": 56.48854961832061, "grad_norm": 2.919419288635254, "learning_rate": 0.0001, "loss": 0.0348, "step": 7400 }, { "epoch": 56.56488549618321, "grad_norm": 1.7207791805267334, "learning_rate": 0.0001, "loss": 0.0348, "step": 7410 }, { "epoch": 56.6412213740458, "grad_norm": 1.977713704109192, "learning_rate": 0.0001, "loss": 0.0332, "step": 7420 }, { "epoch": 56.717557251908396, "grad_norm": 2.1116206645965576, "learning_rate": 0.0001, "loss": 0.0322, "step": 7430 }, { "epoch": 56.79389312977099, "grad_norm": 2.2648444175720215, "learning_rate": 0.0001, "loss": 0.032, "step": 7440 }, { "epoch": 56.87022900763359, "grad_norm": 2.1030471324920654, "learning_rate": 0.0001, "loss": 0.0317, "step": 7450 }, { "epoch": 56.94656488549618, "grad_norm": 1.970046043395996, "learning_rate": 0.0001, "loss": 0.0327, "step": 7460 }, { "epoch": 57.02290076335878, "grad_norm": 2.1334269046783447, "learning_rate": 0.0001, "loss": 0.0333, "step": 7470 }, { "epoch": 57.099236641221374, "grad_norm": 1.4281752109527588, "learning_rate": 0.0001, "loss": 0.0328, "step": 7480 }, { "epoch": 57.17557251908397, "grad_norm": 2.596101999282837, "learning_rate": 0.0001, "loss": 0.0351, "step": 7490 }, { "epoch": 57.25190839694657, "grad_norm": 1.6012048721313477, "learning_rate": 0.0001, "loss": 0.0334, "step": 7500 }, { "epoch": 57.32824427480916, "grad_norm": 1.8773671388626099, "learning_rate": 0.0001, "loss": 0.0337, "step": 7510 }, { "epoch": 57.404580152671755, "grad_norm": 1.9184192419052124, "learning_rate": 0.0001, "loss": 0.0324, "step": 7520 }, { "epoch": 57.48091603053435, "grad_norm": 1.9644181728363037, "learning_rate": 0.0001, "loss": 0.0326, "step": 7530 }, { "epoch": 57.55725190839695, "grad_norm": 1.7865904569625854, "learning_rate": 0.0001, "loss": 0.0323, "step": 7540 }, { "epoch": 57.63358778625954, "grad_norm": 1.8247275352478027, "learning_rate": 0.0001, "loss": 0.0329, "step": 7550 }, { "epoch": 57.70992366412214, "grad_norm": 2.0242197513580322, "learning_rate": 0.0001, "loss": 0.0309, "step": 7560 }, { "epoch": 57.786259541984734, "grad_norm": 2.0151021480560303, "learning_rate": 0.0001, "loss": 0.0334, "step": 7570 }, { "epoch": 57.86259541984733, "grad_norm": 2.4653589725494385, "learning_rate": 0.0001, "loss": 0.0328, "step": 7580 }, { "epoch": 57.93893129770992, "grad_norm": 2.6118927001953125, "learning_rate": 0.0001, "loss": 0.0328, "step": 7590 }, { "epoch": 58.01526717557252, "grad_norm": 2.2515602111816406, "learning_rate": 0.0001, "loss": 0.0321, "step": 7600 }, { "epoch": 58.091603053435115, "grad_norm": 2.073371410369873, "learning_rate": 0.0001, "loss": 0.0318, "step": 7610 }, { "epoch": 58.16793893129771, "grad_norm": 2.035658359527588, "learning_rate": 0.0001, "loss": 0.0315, "step": 7620 }, { "epoch": 58.2442748091603, "grad_norm": 1.8616833686828613, "learning_rate": 0.0001, "loss": 0.0311, "step": 7630 }, { "epoch": 58.3206106870229, "grad_norm": 2.1964519023895264, "learning_rate": 0.0001, "loss": 0.033, "step": 7640 }, { "epoch": 58.396946564885496, "grad_norm": 1.7429636716842651, "learning_rate": 0.0001, "loss": 0.0328, "step": 7650 }, { "epoch": 58.47328244274809, "grad_norm": 2.429835796356201, "learning_rate": 0.0001, "loss": 0.0327, "step": 7660 }, { "epoch": 58.54961832061069, "grad_norm": 2.497576951980591, "learning_rate": 0.0001, "loss": 0.0318, "step": 7670 }, { "epoch": 58.62595419847328, "grad_norm": 1.5992636680603027, "learning_rate": 0.0001, "loss": 0.0321, "step": 7680 }, { "epoch": 58.70229007633588, "grad_norm": 2.013201951980591, "learning_rate": 0.0001, "loss": 0.031, "step": 7690 }, { "epoch": 58.778625954198475, "grad_norm": 2.041584014892578, "learning_rate": 0.0001, "loss": 0.0322, "step": 7700 }, { "epoch": 58.85496183206107, "grad_norm": 1.6401121616363525, "learning_rate": 0.0001, "loss": 0.032, "step": 7710 }, { "epoch": 58.93129770992366, "grad_norm": 1.9825340509414673, "learning_rate": 0.0001, "loss": 0.0326, "step": 7720 }, { "epoch": 59.00763358778626, "grad_norm": 1.916630744934082, "learning_rate": 0.0001, "loss": 0.0323, "step": 7730 }, { "epoch": 59.083969465648856, "grad_norm": 1.6304489374160767, "learning_rate": 0.0001, "loss": 0.0326, "step": 7740 }, { "epoch": 59.16030534351145, "grad_norm": 1.7412917613983154, "learning_rate": 0.0001, "loss": 0.0327, "step": 7750 }, { "epoch": 59.23664122137404, "grad_norm": 1.8241571187973022, "learning_rate": 0.0001, "loss": 0.0334, "step": 7760 }, { "epoch": 59.31297709923664, "grad_norm": 2.0048587322235107, "learning_rate": 0.0001, "loss": 0.032, "step": 7770 }, { "epoch": 59.38931297709924, "grad_norm": 1.583746314048767, "learning_rate": 0.0001, "loss": 0.0332, "step": 7780 }, { "epoch": 59.465648854961835, "grad_norm": 2.321847677230835, "learning_rate": 0.0001, "loss": 0.0327, "step": 7790 }, { "epoch": 59.541984732824424, "grad_norm": 2.098334789276123, "learning_rate": 0.0001, "loss": 0.0336, "step": 7800 }, { "epoch": 59.61832061068702, "grad_norm": 2.038996934890747, "learning_rate": 0.0001, "loss": 0.0327, "step": 7810 }, { "epoch": 59.69465648854962, "grad_norm": 1.6981151103973389, "learning_rate": 0.0001, "loss": 0.0309, "step": 7820 }, { "epoch": 59.770992366412216, "grad_norm": 1.7717534303665161, "learning_rate": 0.0001, "loss": 0.0318, "step": 7830 }, { "epoch": 59.847328244274806, "grad_norm": 1.8323487043380737, "learning_rate": 0.0001, "loss": 0.0317, "step": 7840 }, { "epoch": 59.9236641221374, "grad_norm": 1.6514848470687866, "learning_rate": 0.0001, "loss": 0.0316, "step": 7850 }, { "epoch": 60.0, "grad_norm": 2.716301918029785, "learning_rate": 0.0001, "loss": 0.034, "step": 7860 }, { "epoch": 60.0763358778626, "grad_norm": 2.144780397415161, "learning_rate": 0.0001, "loss": 0.0331, "step": 7870 }, { "epoch": 60.152671755725194, "grad_norm": 2.1545653343200684, "learning_rate": 0.0001, "loss": 0.0335, "step": 7880 }, { "epoch": 60.229007633587784, "grad_norm": 1.8637224435806274, "learning_rate": 0.0001, "loss": 0.0328, "step": 7890 }, { "epoch": 60.30534351145038, "grad_norm": 1.6419209241867065, "learning_rate": 0.0001, "loss": 0.0332, "step": 7900 }, { "epoch": 60.38167938931298, "grad_norm": 3.0104448795318604, "learning_rate": 0.0001, "loss": 0.0324, "step": 7910 }, { "epoch": 60.458015267175576, "grad_norm": 2.6341285705566406, "learning_rate": 0.0001, "loss": 0.0336, "step": 7920 }, { "epoch": 60.534351145038165, "grad_norm": 2.2004048824310303, "learning_rate": 0.0001, "loss": 0.0335, "step": 7930 }, { "epoch": 60.61068702290076, "grad_norm": 2.3205771446228027, "learning_rate": 0.0001, "loss": 0.0319, "step": 7940 }, { "epoch": 60.68702290076336, "grad_norm": 1.5460617542266846, "learning_rate": 0.0001, "loss": 0.0324, "step": 7950 }, { "epoch": 60.76335877862596, "grad_norm": 2.311814308166504, "learning_rate": 0.0001, "loss": 0.0329, "step": 7960 }, { "epoch": 60.83969465648855, "grad_norm": 2.3538529872894287, "learning_rate": 0.0001, "loss": 0.0326, "step": 7970 }, { "epoch": 60.916030534351144, "grad_norm": 2.143474578857422, "learning_rate": 0.0001, "loss": 0.0332, "step": 7980 }, { "epoch": 60.99236641221374, "grad_norm": 3.050978183746338, "learning_rate": 0.0001, "loss": 0.0334, "step": 7990 }, { "epoch": 61.06870229007634, "grad_norm": 2.331542730331421, "learning_rate": 0.0001, "loss": 0.0343, "step": 8000 }, { "epoch": 61.14503816793893, "grad_norm": 1.5952768325805664, "learning_rate": 0.0001, "loss": 0.0325, "step": 8010 }, { "epoch": 61.221374045801525, "grad_norm": 2.111820936203003, "learning_rate": 0.0001, "loss": 0.0313, "step": 8020 }, { "epoch": 61.29770992366412, "grad_norm": 1.6829665899276733, "learning_rate": 0.0001, "loss": 0.0314, "step": 8030 }, { "epoch": 61.37404580152672, "grad_norm": 2.3493752479553223, "learning_rate": 0.0001, "loss": 0.031, "step": 8040 }, { "epoch": 61.45038167938931, "grad_norm": 1.84051513671875, "learning_rate": 0.0001, "loss": 0.0308, "step": 8050 }, { "epoch": 61.52671755725191, "grad_norm": 1.9063640832901, "learning_rate": 0.0001, "loss": 0.0315, "step": 8060 }, { "epoch": 61.603053435114504, "grad_norm": 2.8346660137176514, "learning_rate": 0.0001, "loss": 0.0321, "step": 8070 }, { "epoch": 61.6793893129771, "grad_norm": 1.514216423034668, "learning_rate": 0.0001, "loss": 0.0313, "step": 8080 }, { "epoch": 61.7557251908397, "grad_norm": 1.5355480909347534, "learning_rate": 0.0001, "loss": 0.0315, "step": 8090 }, { "epoch": 61.83206106870229, "grad_norm": 2.2017767429351807, "learning_rate": 0.0001, "loss": 0.0319, "step": 8100 }, { "epoch": 61.908396946564885, "grad_norm": 1.8373870849609375, "learning_rate": 0.0001, "loss": 0.0319, "step": 8110 }, { "epoch": 61.98473282442748, "grad_norm": 2.342411518096924, "learning_rate": 0.0001, "loss": 0.0322, "step": 8120 }, { "epoch": 62.06106870229008, "grad_norm": 1.7069913148880005, "learning_rate": 0.0001, "loss": 0.0313, "step": 8130 }, { "epoch": 62.13740458015267, "grad_norm": 2.950815200805664, "learning_rate": 0.0001, "loss": 0.0319, "step": 8140 }, { "epoch": 62.213740458015266, "grad_norm": 1.7024372816085815, "learning_rate": 0.0001, "loss": 0.0321, "step": 8150 }, { "epoch": 62.29007633587786, "grad_norm": 1.990932583808899, "learning_rate": 0.0001, "loss": 0.0328, "step": 8160 }, { "epoch": 62.36641221374046, "grad_norm": 2.318087339401245, "learning_rate": 0.0001, "loss": 0.0335, "step": 8170 }, { "epoch": 62.44274809160305, "grad_norm": 2.125663995742798, "learning_rate": 0.0001, "loss": 0.0319, "step": 8180 }, { "epoch": 62.51908396946565, "grad_norm": 2.327949285507202, "learning_rate": 0.0001, "loss": 0.0326, "step": 8190 }, { "epoch": 62.595419847328245, "grad_norm": 2.0915346145629883, "learning_rate": 0.0001, "loss": 0.031, "step": 8200 }, { "epoch": 62.67175572519084, "grad_norm": 1.49560546875, "learning_rate": 0.0001, "loss": 0.0312, "step": 8210 }, { "epoch": 62.74809160305343, "grad_norm": 1.881609559059143, "learning_rate": 0.0001, "loss": 0.0328, "step": 8220 }, { "epoch": 62.82442748091603, "grad_norm": 2.365471363067627, "learning_rate": 0.0001, "loss": 0.0341, "step": 8230 }, { "epoch": 62.900763358778626, "grad_norm": 2.1752216815948486, "learning_rate": 0.0001, "loss": 0.035, "step": 8240 }, { "epoch": 62.97709923664122, "grad_norm": 1.811848521232605, "learning_rate": 0.0001, "loss": 0.0352, "step": 8250 }, { "epoch": 63.05343511450382, "grad_norm": 1.6423739194869995, "learning_rate": 0.0001, "loss": 0.0321, "step": 8260 }, { "epoch": 63.12977099236641, "grad_norm": 1.621764898300171, "learning_rate": 0.0001, "loss": 0.0324, "step": 8270 }, { "epoch": 63.20610687022901, "grad_norm": 2.1209471225738525, "learning_rate": 0.0001, "loss": 0.032, "step": 8280 }, { "epoch": 63.282442748091604, "grad_norm": 2.2083353996276855, "learning_rate": 0.0001, "loss": 0.0334, "step": 8290 }, { "epoch": 63.3587786259542, "grad_norm": 1.9260131120681763, "learning_rate": 0.0001, "loss": 0.0326, "step": 8300 }, { "epoch": 63.43511450381679, "grad_norm": 1.876798152923584, "learning_rate": 0.0001, "loss": 0.0323, "step": 8310 }, { "epoch": 63.51145038167939, "grad_norm": 2.4608616828918457, "learning_rate": 0.0001, "loss": 0.0326, "step": 8320 }, { "epoch": 63.587786259541986, "grad_norm": 2.0825273990631104, "learning_rate": 0.0001, "loss": 0.0319, "step": 8330 }, { "epoch": 63.66412213740458, "grad_norm": 1.3804361820220947, "learning_rate": 0.0001, "loss": 0.0313, "step": 8340 }, { "epoch": 63.74045801526717, "grad_norm": 1.5615490674972534, "learning_rate": 0.0001, "loss": 0.0304, "step": 8350 }, { "epoch": 63.81679389312977, "grad_norm": 1.8977432250976562, "learning_rate": 0.0001, "loss": 0.0312, "step": 8360 }, { "epoch": 63.89312977099237, "grad_norm": 1.9342325925827026, "learning_rate": 0.0001, "loss": 0.0307, "step": 8370 }, { "epoch": 63.969465648854964, "grad_norm": 1.5919638872146606, "learning_rate": 0.0001, "loss": 0.0297, "step": 8380 }, { "epoch": 64.04580152671755, "grad_norm": 2.1904683113098145, "learning_rate": 0.0001, "loss": 0.0317, "step": 8390 }, { "epoch": 64.12213740458016, "grad_norm": 2.1755967140197754, "learning_rate": 0.0001, "loss": 0.0316, "step": 8400 }, { "epoch": 64.19847328244275, "grad_norm": 1.6037392616271973, "learning_rate": 0.0001, "loss": 0.0317, "step": 8410 }, { "epoch": 64.27480916030534, "grad_norm": 1.6413397789001465, "learning_rate": 0.0001, "loss": 0.0312, "step": 8420 }, { "epoch": 64.35114503816794, "grad_norm": 2.2234275341033936, "learning_rate": 0.0001, "loss": 0.0306, "step": 8430 }, { "epoch": 64.42748091603053, "grad_norm": 1.5503214597702026, "learning_rate": 0.0001, "loss": 0.0305, "step": 8440 }, { "epoch": 64.50381679389314, "grad_norm": 2.0822324752807617, "learning_rate": 0.0001, "loss": 0.0312, "step": 8450 }, { "epoch": 64.58015267175573, "grad_norm": 1.8392794132232666, "learning_rate": 0.0001, "loss": 0.031, "step": 8460 }, { "epoch": 64.65648854961832, "grad_norm": 1.9951685667037964, "learning_rate": 0.0001, "loss": 0.0317, "step": 8470 }, { "epoch": 64.73282442748092, "grad_norm": 1.8509739637374878, "learning_rate": 0.0001, "loss": 0.0314, "step": 8480 }, { "epoch": 64.80916030534351, "grad_norm": 2.0644590854644775, "learning_rate": 0.0001, "loss": 0.0309, "step": 8490 }, { "epoch": 64.8854961832061, "grad_norm": 1.9337153434753418, "learning_rate": 0.0001, "loss": 0.0311, "step": 8500 }, { "epoch": 64.9618320610687, "grad_norm": 1.9227486848831177, "learning_rate": 0.0001, "loss": 0.0303, "step": 8510 }, { "epoch": 65.0381679389313, "grad_norm": 2.275709867477417, "learning_rate": 0.0001, "loss": 0.0297, "step": 8520 }, { "epoch": 65.1145038167939, "grad_norm": 1.6467446088790894, "learning_rate": 0.0001, "loss": 0.0294, "step": 8530 }, { "epoch": 65.19083969465649, "grad_norm": 2.466478109359741, "learning_rate": 0.0001, "loss": 0.0299, "step": 8540 }, { "epoch": 65.26717557251908, "grad_norm": 1.909367322921753, "learning_rate": 0.0001, "loss": 0.0302, "step": 8550 }, { "epoch": 65.34351145038168, "grad_norm": 2.462709665298462, "learning_rate": 0.0001, "loss": 0.0317, "step": 8560 }, { "epoch": 65.41984732824427, "grad_norm": 2.014777183532715, "learning_rate": 0.0001, "loss": 0.0307, "step": 8570 }, { "epoch": 65.49618320610686, "grad_norm": 2.4112908840179443, "learning_rate": 0.0001, "loss": 0.0305, "step": 8580 }, { "epoch": 65.57251908396947, "grad_norm": 1.6771920919418335, "learning_rate": 0.0001, "loss": 0.0305, "step": 8590 }, { "epoch": 65.64885496183206, "grad_norm": 2.231489896774292, "learning_rate": 0.0001, "loss": 0.0306, "step": 8600 }, { "epoch": 65.72519083969466, "grad_norm": 2.6609926223754883, "learning_rate": 0.0001, "loss": 0.0308, "step": 8610 }, { "epoch": 65.80152671755725, "grad_norm": 2.828763008117676, "learning_rate": 0.0001, "loss": 0.0318, "step": 8620 }, { "epoch": 65.87786259541984, "grad_norm": 2.236938714981079, "learning_rate": 0.0001, "loss": 0.0301, "step": 8630 }, { "epoch": 65.95419847328245, "grad_norm": 2.286282539367676, "learning_rate": 0.0001, "loss": 0.0295, "step": 8640 }, { "epoch": 66.03053435114504, "grad_norm": 1.5397930145263672, "learning_rate": 0.0001, "loss": 0.0296, "step": 8650 }, { "epoch": 66.10687022900764, "grad_norm": 1.7071449756622314, "learning_rate": 0.0001, "loss": 0.0301, "step": 8660 }, { "epoch": 66.18320610687023, "grad_norm": 2.1714563369750977, "learning_rate": 0.0001, "loss": 0.0311, "step": 8670 }, { "epoch": 66.25954198473282, "grad_norm": 1.9439994096755981, "learning_rate": 0.0001, "loss": 0.0294, "step": 8680 }, { "epoch": 66.33587786259542, "grad_norm": 2.1424508094787598, "learning_rate": 0.0001, "loss": 0.0308, "step": 8690 }, { "epoch": 66.41221374045801, "grad_norm": 2.277506113052368, "learning_rate": 0.0001, "loss": 0.0302, "step": 8700 }, { "epoch": 66.4885496183206, "grad_norm": 1.9151673316955566, "learning_rate": 0.0001, "loss": 0.0308, "step": 8710 }, { "epoch": 66.56488549618321, "grad_norm": 1.4847289323806763, "learning_rate": 0.0001, "loss": 0.0294, "step": 8720 }, { "epoch": 66.6412213740458, "grad_norm": 2.068592071533203, "learning_rate": 0.0001, "loss": 0.0282, "step": 8730 }, { "epoch": 66.7175572519084, "grad_norm": 1.727120280265808, "learning_rate": 0.0001, "loss": 0.0297, "step": 8740 }, { "epoch": 66.79389312977099, "grad_norm": 2.142831563949585, "learning_rate": 0.0001, "loss": 0.0307, "step": 8750 }, { "epoch": 66.87022900763358, "grad_norm": 2.236034393310547, "learning_rate": 0.0001, "loss": 0.0313, "step": 8760 }, { "epoch": 66.94656488549619, "grad_norm": 2.2216806411743164, "learning_rate": 0.0001, "loss": 0.0298, "step": 8770 }, { "epoch": 67.02290076335878, "grad_norm": 1.7911618947982788, "learning_rate": 0.0001, "loss": 0.0304, "step": 8780 }, { "epoch": 67.09923664122137, "grad_norm": 1.9789682626724243, "learning_rate": 0.0001, "loss": 0.0297, "step": 8790 }, { "epoch": 67.17557251908397, "grad_norm": 2.192772150039673, "learning_rate": 0.0001, "loss": 0.0301, "step": 8800 }, { "epoch": 67.25190839694656, "grad_norm": 1.6653828620910645, "learning_rate": 0.0001, "loss": 0.0298, "step": 8810 }, { "epoch": 67.32824427480917, "grad_norm": 1.9383299350738525, "learning_rate": 0.0001, "loss": 0.029, "step": 8820 }, { "epoch": 67.40458015267176, "grad_norm": 1.8837453126907349, "learning_rate": 0.0001, "loss": 0.0304, "step": 8830 }, { "epoch": 67.48091603053435, "grad_norm": 1.9213292598724365, "learning_rate": 0.0001, "loss": 0.0287, "step": 8840 }, { "epoch": 67.55725190839695, "grad_norm": 2.7693567276000977, "learning_rate": 0.0001, "loss": 0.0299, "step": 8850 }, { "epoch": 67.63358778625954, "grad_norm": 2.2570409774780273, "learning_rate": 0.0001, "loss": 0.0314, "step": 8860 }, { "epoch": 67.70992366412214, "grad_norm": 1.8556944131851196, "learning_rate": 0.0001, "loss": 0.03, "step": 8870 }, { "epoch": 67.78625954198473, "grad_norm": 2.182081460952759, "learning_rate": 0.0001, "loss": 0.0294, "step": 8880 }, { "epoch": 67.86259541984732, "grad_norm": 2.377044200897217, "learning_rate": 0.0001, "loss": 0.0302, "step": 8890 }, { "epoch": 67.93893129770993, "grad_norm": 2.4464709758758545, "learning_rate": 0.0001, "loss": 0.0311, "step": 8900 }, { "epoch": 68.01526717557252, "grad_norm": 1.8607606887817383, "learning_rate": 0.0001, "loss": 0.0305, "step": 8910 }, { "epoch": 68.09160305343511, "grad_norm": 1.8662669658660889, "learning_rate": 0.0001, "loss": 0.0303, "step": 8920 }, { "epoch": 68.16793893129771, "grad_norm": 1.6468815803527832, "learning_rate": 0.0001, "loss": 0.03, "step": 8930 }, { "epoch": 68.2442748091603, "grad_norm": 2.4743740558624268, "learning_rate": 0.0001, "loss": 0.03, "step": 8940 }, { "epoch": 68.3206106870229, "grad_norm": 1.975601077079773, "learning_rate": 0.0001, "loss": 0.03, "step": 8950 }, { "epoch": 68.3969465648855, "grad_norm": 2.6657161712646484, "learning_rate": 0.0001, "loss": 0.0304, "step": 8960 }, { "epoch": 68.47328244274809, "grad_norm": 2.122692823410034, "learning_rate": 0.0001, "loss": 0.0296, "step": 8970 }, { "epoch": 68.54961832061069, "grad_norm": 1.6443560123443604, "learning_rate": 0.0001, "loss": 0.0307, "step": 8980 }, { "epoch": 68.62595419847328, "grad_norm": 2.11207914352417, "learning_rate": 0.0001, "loss": 0.0299, "step": 8990 }, { "epoch": 68.70229007633588, "grad_norm": 2.5480427742004395, "learning_rate": 0.0001, "loss": 0.0296, "step": 9000 }, { "epoch": 68.77862595419847, "grad_norm": 2.2168571949005127, "learning_rate": 0.0001, "loss": 0.029, "step": 9010 }, { "epoch": 68.85496183206106, "grad_norm": 1.7049243450164795, "learning_rate": 0.0001, "loss": 0.0292, "step": 9020 }, { "epoch": 68.93129770992367, "grad_norm": 2.519770860671997, "learning_rate": 0.0001, "loss": 0.03, "step": 9030 }, { "epoch": 69.00763358778626, "grad_norm": 1.943392038345337, "learning_rate": 0.0001, "loss": 0.0305, "step": 9040 }, { "epoch": 69.08396946564885, "grad_norm": 2.431184768676758, "learning_rate": 0.0001, "loss": 0.0305, "step": 9050 }, { "epoch": 69.16030534351145, "grad_norm": 2.2565202713012695, "learning_rate": 0.0001, "loss": 0.0295, "step": 9060 }, { "epoch": 69.23664122137404, "grad_norm": 2.6503212451934814, "learning_rate": 0.0001, "loss": 0.029, "step": 9070 }, { "epoch": 69.31297709923665, "grad_norm": 2.1392672061920166, "learning_rate": 0.0001, "loss": 0.03, "step": 9080 }, { "epoch": 69.38931297709924, "grad_norm": 1.9566749334335327, "learning_rate": 0.0001, "loss": 0.0292, "step": 9090 }, { "epoch": 69.46564885496183, "grad_norm": 2.2549238204956055, "learning_rate": 0.0001, "loss": 0.0301, "step": 9100 }, { "epoch": 69.54198473282443, "grad_norm": 2.414083480834961, "learning_rate": 0.0001, "loss": 0.0298, "step": 9110 }, { "epoch": 69.61832061068702, "grad_norm": 2.1689066886901855, "learning_rate": 0.0001, "loss": 0.0305, "step": 9120 }, { "epoch": 69.69465648854961, "grad_norm": 1.7097729444503784, "learning_rate": 0.0001, "loss": 0.0302, "step": 9130 }, { "epoch": 69.77099236641222, "grad_norm": 1.8869540691375732, "learning_rate": 0.0001, "loss": 0.0295, "step": 9140 }, { "epoch": 69.8473282442748, "grad_norm": 1.8318934440612793, "learning_rate": 0.0001, "loss": 0.0304, "step": 9150 }, { "epoch": 69.92366412213741, "grad_norm": 3.150242567062378, "learning_rate": 0.0001, "loss": 0.0305, "step": 9160 }, { "epoch": 70.0, "grad_norm": 2.5037059783935547, "learning_rate": 0.0001, "loss": 0.0299, "step": 9170 }, { "epoch": 70.07633587786259, "grad_norm": 2.7803537845611572, "learning_rate": 0.0001, "loss": 0.0307, "step": 9180 }, { "epoch": 70.1526717557252, "grad_norm": 2.747732639312744, "learning_rate": 0.0001, "loss": 0.031, "step": 9190 }, { "epoch": 70.22900763358778, "grad_norm": 2.3899874687194824, "learning_rate": 0.0001, "loss": 0.0295, "step": 9200 }, { "epoch": 70.30534351145039, "grad_norm": 2.3623979091644287, "learning_rate": 0.0001, "loss": 0.0299, "step": 9210 }, { "epoch": 70.38167938931298, "grad_norm": 2.176391839981079, "learning_rate": 0.0001, "loss": 0.0294, "step": 9220 }, { "epoch": 70.45801526717557, "grad_norm": 2.0698840618133545, "learning_rate": 0.0001, "loss": 0.031, "step": 9230 }, { "epoch": 70.53435114503817, "grad_norm": 2.47269344329834, "learning_rate": 0.0001, "loss": 0.0298, "step": 9240 }, { "epoch": 70.61068702290076, "grad_norm": 2.088634967803955, "learning_rate": 0.0001, "loss": 0.0295, "step": 9250 }, { "epoch": 70.68702290076335, "grad_norm": 1.8751108646392822, "learning_rate": 0.0001, "loss": 0.0301, "step": 9260 }, { "epoch": 70.76335877862596, "grad_norm": 1.9406391382217407, "learning_rate": 0.0001, "loss": 0.0308, "step": 9270 }, { "epoch": 70.83969465648855, "grad_norm": 1.9143279790878296, "learning_rate": 0.0001, "loss": 0.0303, "step": 9280 }, { "epoch": 70.91603053435115, "grad_norm": 2.2935869693756104, "learning_rate": 0.0001, "loss": 0.0305, "step": 9290 }, { "epoch": 70.99236641221374, "grad_norm": 1.6144917011260986, "learning_rate": 0.0001, "loss": 0.029, "step": 9300 }, { "epoch": 71.06870229007633, "grad_norm": 2.1397719383239746, "learning_rate": 0.0001, "loss": 0.0288, "step": 9310 }, { "epoch": 71.14503816793894, "grad_norm": 1.9069713354110718, "learning_rate": 0.0001, "loss": 0.0307, "step": 9320 }, { "epoch": 71.22137404580153, "grad_norm": 3.019467830657959, "learning_rate": 0.0001, "loss": 0.0308, "step": 9330 }, { "epoch": 71.29770992366412, "grad_norm": 2.0691001415252686, "learning_rate": 0.0001, "loss": 0.0315, "step": 9340 }, { "epoch": 71.37404580152672, "grad_norm": 1.8356878757476807, "learning_rate": 0.0001, "loss": 0.0297, "step": 9350 }, { "epoch": 71.45038167938931, "grad_norm": 1.8802990913391113, "learning_rate": 0.0001, "loss": 0.0312, "step": 9360 }, { "epoch": 71.52671755725191, "grad_norm": 2.2259182929992676, "learning_rate": 0.0001, "loss": 0.0303, "step": 9370 }, { "epoch": 71.6030534351145, "grad_norm": 1.5587525367736816, "learning_rate": 0.0001, "loss": 0.0302, "step": 9380 }, { "epoch": 71.6793893129771, "grad_norm": 2.035710096359253, "learning_rate": 0.0001, "loss": 0.0298, "step": 9390 }, { "epoch": 71.7557251908397, "grad_norm": 1.9290261268615723, "learning_rate": 0.0001, "loss": 0.0314, "step": 9400 }, { "epoch": 71.83206106870229, "grad_norm": 2.1120641231536865, "learning_rate": 0.0001, "loss": 0.0313, "step": 9410 }, { "epoch": 71.90839694656489, "grad_norm": 2.2737133502960205, "learning_rate": 0.0001, "loss": 0.0305, "step": 9420 }, { "epoch": 71.98473282442748, "grad_norm": 1.554010033607483, "learning_rate": 0.0001, "loss": 0.0301, "step": 9430 }, { "epoch": 72.06106870229007, "grad_norm": 2.8816843032836914, "learning_rate": 0.0001, "loss": 0.0295, "step": 9440 }, { "epoch": 72.13740458015268, "grad_norm": 2.125767469406128, "learning_rate": 0.0001, "loss": 0.0308, "step": 9450 }, { "epoch": 72.21374045801527, "grad_norm": 1.8103172779083252, "learning_rate": 0.0001, "loss": 0.0307, "step": 9460 }, { "epoch": 72.29007633587786, "grad_norm": 1.8527859449386597, "learning_rate": 0.0001, "loss": 0.0301, "step": 9470 }, { "epoch": 72.36641221374046, "grad_norm": 1.8353875875473022, "learning_rate": 0.0001, "loss": 0.0316, "step": 9480 }, { "epoch": 72.44274809160305, "grad_norm": 2.836209774017334, "learning_rate": 0.0001, "loss": 0.0303, "step": 9490 }, { "epoch": 72.51908396946565, "grad_norm": 2.621262550354004, "learning_rate": 0.0001, "loss": 0.031, "step": 9500 }, { "epoch": 72.59541984732824, "grad_norm": 2.0114669799804688, "learning_rate": 0.0001, "loss": 0.0314, "step": 9510 }, { "epoch": 72.67175572519083, "grad_norm": 2.806640625, "learning_rate": 0.0001, "loss": 0.03, "step": 9520 }, { "epoch": 72.74809160305344, "grad_norm": 2.534691095352173, "learning_rate": 0.0001, "loss": 0.0295, "step": 9530 }, { "epoch": 72.82442748091603, "grad_norm": 1.637236475944519, "learning_rate": 0.0001, "loss": 0.03, "step": 9540 }, { "epoch": 72.90076335877862, "grad_norm": 2.1319808959960938, "learning_rate": 0.0001, "loss": 0.0281, "step": 9550 }, { "epoch": 72.97709923664122, "grad_norm": 2.411827564239502, "learning_rate": 0.0001, "loss": 0.0287, "step": 9560 }, { "epoch": 73.05343511450381, "grad_norm": 2.218604326248169, "learning_rate": 0.0001, "loss": 0.0285, "step": 9570 }, { "epoch": 73.12977099236642, "grad_norm": 2.4074063301086426, "learning_rate": 0.0001, "loss": 0.0273, "step": 9580 }, { "epoch": 73.20610687022901, "grad_norm": 2.165762424468994, "learning_rate": 0.0001, "loss": 0.0278, "step": 9590 }, { "epoch": 73.2824427480916, "grad_norm": 1.3627911806106567, "learning_rate": 0.0001, "loss": 0.0275, "step": 9600 }, { "epoch": 73.3587786259542, "grad_norm": 2.5073766708374023, "learning_rate": 0.0001, "loss": 0.0286, "step": 9610 }, { "epoch": 73.43511450381679, "grad_norm": 2.648125648498535, "learning_rate": 0.0001, "loss": 0.0291, "step": 9620 }, { "epoch": 73.5114503816794, "grad_norm": 2.7140467166900635, "learning_rate": 0.0001, "loss": 0.0296, "step": 9630 }, { "epoch": 73.58778625954199, "grad_norm": 3.3176913261413574, "learning_rate": 0.0001, "loss": 0.0289, "step": 9640 }, { "epoch": 73.66412213740458, "grad_norm": 2.556940793991089, "learning_rate": 0.0001, "loss": 0.0286, "step": 9650 }, { "epoch": 73.74045801526718, "grad_norm": 2.041752338409424, "learning_rate": 0.0001, "loss": 0.0284, "step": 9660 }, { "epoch": 73.81679389312977, "grad_norm": 2.5002834796905518, "learning_rate": 0.0001, "loss": 0.0287, "step": 9670 }, { "epoch": 73.89312977099236, "grad_norm": 1.8303453922271729, "learning_rate": 0.0001, "loss": 0.0286, "step": 9680 }, { "epoch": 73.96946564885496, "grad_norm": 2.2524070739746094, "learning_rate": 0.0001, "loss": 0.0296, "step": 9690 }, { "epoch": 74.04580152671755, "grad_norm": 1.784694790840149, "learning_rate": 0.0001, "loss": 0.0283, "step": 9700 }, { "epoch": 74.12213740458016, "grad_norm": 2.318023443222046, "learning_rate": 0.0001, "loss": 0.0285, "step": 9710 }, { "epoch": 74.19847328244275, "grad_norm": 2.2711362838745117, "learning_rate": 0.0001, "loss": 0.0286, "step": 9720 }, { "epoch": 74.27480916030534, "grad_norm": 2.1202738285064697, "learning_rate": 0.0001, "loss": 0.0283, "step": 9730 }, { "epoch": 74.35114503816794, "grad_norm": 1.8524848222732544, "learning_rate": 0.0001, "loss": 0.0283, "step": 9740 }, { "epoch": 74.42748091603053, "grad_norm": 2.2905337810516357, "learning_rate": 0.0001, "loss": 0.0292, "step": 9750 }, { "epoch": 74.50381679389314, "grad_norm": 1.506243109703064, "learning_rate": 0.0001, "loss": 0.0286, "step": 9760 }, { "epoch": 74.58015267175573, "grad_norm": 2.2872180938720703, "learning_rate": 0.0001, "loss": 0.0287, "step": 9770 }, { "epoch": 74.65648854961832, "grad_norm": 2.450125217437744, "learning_rate": 0.0001, "loss": 0.0286, "step": 9780 }, { "epoch": 74.73282442748092, "grad_norm": 2.1647050380706787, "learning_rate": 0.0001, "loss": 0.0288, "step": 9790 }, { "epoch": 74.80916030534351, "grad_norm": 2.5376462936401367, "learning_rate": 0.0001, "loss": 0.0293, "step": 9800 }, { "epoch": 74.8854961832061, "grad_norm": 2.1900815963745117, "learning_rate": 0.0001, "loss": 0.0313, "step": 9810 }, { "epoch": 74.9618320610687, "grad_norm": 2.4247231483459473, "learning_rate": 0.0001, "loss": 0.0297, "step": 9820 }, { "epoch": 75.0381679389313, "grad_norm": 2.3287341594696045, "learning_rate": 0.0001, "loss": 0.0303, "step": 9830 }, { "epoch": 75.1145038167939, "grad_norm": 2.5684101581573486, "learning_rate": 0.0001, "loss": 0.0308, "step": 9840 }, { "epoch": 75.19083969465649, "grad_norm": 1.663659930229187, "learning_rate": 0.0001, "loss": 0.032, "step": 9850 }, { "epoch": 75.26717557251908, "grad_norm": 2.4680287837982178, "learning_rate": 0.0001, "loss": 0.0304, "step": 9860 }, { "epoch": 75.34351145038168, "grad_norm": 2.381572723388672, "learning_rate": 0.0001, "loss": 0.0293, "step": 9870 }, { "epoch": 75.41984732824427, "grad_norm": 1.9898970127105713, "learning_rate": 0.0001, "loss": 0.0308, "step": 9880 }, { "epoch": 75.49618320610686, "grad_norm": 2.1753761768341064, "learning_rate": 0.0001, "loss": 0.0297, "step": 9890 }, { "epoch": 75.57251908396947, "grad_norm": 2.3156304359436035, "learning_rate": 0.0001, "loss": 0.0301, "step": 9900 }, { "epoch": 75.64885496183206, "grad_norm": 1.7426077127456665, "learning_rate": 0.0001, "loss": 0.0287, "step": 9910 }, { "epoch": 75.72519083969466, "grad_norm": 2.5900280475616455, "learning_rate": 0.0001, "loss": 0.0291, "step": 9920 }, { "epoch": 75.80152671755725, "grad_norm": 1.5610605478286743, "learning_rate": 0.0001, "loss": 0.0285, "step": 9930 }, { "epoch": 75.87786259541984, "grad_norm": 1.944419026374817, "learning_rate": 0.0001, "loss": 0.03, "step": 9940 }, { "epoch": 75.95419847328245, "grad_norm": 2.1751949787139893, "learning_rate": 0.0001, "loss": 0.0296, "step": 9950 }, { "epoch": 76.03053435114504, "grad_norm": 1.5059646368026733, "learning_rate": 0.0001, "loss": 0.0283, "step": 9960 }, { "epoch": 76.10687022900764, "grad_norm": 2.0945818424224854, "learning_rate": 0.0001, "loss": 0.0288, "step": 9970 }, { "epoch": 76.18320610687023, "grad_norm": 1.8100814819335938, "learning_rate": 0.0001, "loss": 0.0284, "step": 9980 }, { "epoch": 76.25954198473282, "grad_norm": 1.6310064792633057, "learning_rate": 0.0001, "loss": 0.0287, "step": 9990 }, { "epoch": 76.33587786259542, "grad_norm": 2.748358726501465, "learning_rate": 0.0001, "loss": 0.03, "step": 10000 }, { "epoch": 76.41221374045801, "grad_norm": 2.7476203441619873, "learning_rate": 0.0001, "loss": 0.0304, "step": 10010 }, { "epoch": 76.4885496183206, "grad_norm": 2.604482889175415, "learning_rate": 0.0001, "loss": 0.0297, "step": 10020 }, { "epoch": 76.56488549618321, "grad_norm": 1.827165961265564, "learning_rate": 0.0001, "loss": 0.029, "step": 10030 }, { "epoch": 76.6412213740458, "grad_norm": 2.595947265625, "learning_rate": 0.0001, "loss": 0.0291, "step": 10040 }, { "epoch": 76.7175572519084, "grad_norm": 2.1675007343292236, "learning_rate": 0.0001, "loss": 0.0293, "step": 10050 }, { "epoch": 76.79389312977099, "grad_norm": 2.1655330657958984, "learning_rate": 0.0001, "loss": 0.0291, "step": 10060 }, { "epoch": 76.87022900763358, "grad_norm": 2.1297690868377686, "learning_rate": 0.0001, "loss": 0.0286, "step": 10070 }, { "epoch": 76.94656488549619, "grad_norm": 1.943449854850769, "learning_rate": 0.0001, "loss": 0.0288, "step": 10080 }, { "epoch": 77.02290076335878, "grad_norm": 2.023320198059082, "learning_rate": 0.0001, "loss": 0.0284, "step": 10090 }, { "epoch": 77.09923664122137, "grad_norm": 2.7108616828918457, "learning_rate": 0.0001, "loss": 0.028, "step": 10100 }, { "epoch": 77.17557251908397, "grad_norm": 1.997301697731018, "learning_rate": 0.0001, "loss": 0.0276, "step": 10110 }, { "epoch": 77.25190839694656, "grad_norm": 1.879205346107483, "learning_rate": 0.0001, "loss": 0.0284, "step": 10120 }, { "epoch": 77.32824427480917, "grad_norm": 2.0820703506469727, "learning_rate": 0.0001, "loss": 0.0285, "step": 10130 }, { "epoch": 77.40458015267176, "grad_norm": 2.2258620262145996, "learning_rate": 0.0001, "loss": 0.0283, "step": 10140 }, { "epoch": 77.48091603053435, "grad_norm": 2.322016477584839, "learning_rate": 0.0001, "loss": 0.0288, "step": 10150 }, { "epoch": 77.55725190839695, "grad_norm": 2.0846176147460938, "learning_rate": 0.0001, "loss": 0.0289, "step": 10160 }, { "epoch": 77.63358778625954, "grad_norm": 2.1014726161956787, "learning_rate": 0.0001, "loss": 0.0284, "step": 10170 }, { "epoch": 77.70992366412214, "grad_norm": 1.6727752685546875, "learning_rate": 0.0001, "loss": 0.0285, "step": 10180 }, { "epoch": 77.78625954198473, "grad_norm": 1.4181956052780151, "learning_rate": 0.0001, "loss": 0.0275, "step": 10190 }, { "epoch": 77.86259541984732, "grad_norm": 2.3122684955596924, "learning_rate": 0.0001, "loss": 0.0276, "step": 10200 }, { "epoch": 77.93893129770993, "grad_norm": 2.0261049270629883, "learning_rate": 0.0001, "loss": 0.0282, "step": 10210 }, { "epoch": 78.01526717557252, "grad_norm": 2.0695736408233643, "learning_rate": 0.0001, "loss": 0.0274, "step": 10220 }, { "epoch": 78.09160305343511, "grad_norm": 2.347654342651367, "learning_rate": 0.0001, "loss": 0.0275, "step": 10230 }, { "epoch": 78.16793893129771, "grad_norm": 2.5935475826263428, "learning_rate": 0.0001, "loss": 0.0277, "step": 10240 }, { "epoch": 78.2442748091603, "grad_norm": 1.8814716339111328, "learning_rate": 0.0001, "loss": 0.028, "step": 10250 }, { "epoch": 78.3206106870229, "grad_norm": 2.15913462638855, "learning_rate": 0.0001, "loss": 0.028, "step": 10260 }, { "epoch": 78.3969465648855, "grad_norm": 2.820437431335449, "learning_rate": 0.0001, "loss": 0.0289, "step": 10270 }, { "epoch": 78.47328244274809, "grad_norm": 2.35988712310791, "learning_rate": 0.0001, "loss": 0.0292, "step": 10280 }, { "epoch": 78.54961832061069, "grad_norm": 2.7713005542755127, "learning_rate": 0.0001, "loss": 0.0288, "step": 10290 }, { "epoch": 78.62595419847328, "grad_norm": 2.0153582096099854, "learning_rate": 0.0001, "loss": 0.0286, "step": 10300 }, { "epoch": 78.70229007633588, "grad_norm": 2.7163286209106445, "learning_rate": 0.0001, "loss": 0.0284, "step": 10310 }, { "epoch": 78.77862595419847, "grad_norm": 2.9722514152526855, "learning_rate": 0.0001, "loss": 0.0279, "step": 10320 }, { "epoch": 78.85496183206106, "grad_norm": 1.6790387630462646, "learning_rate": 0.0001, "loss": 0.0285, "step": 10330 }, { "epoch": 78.93129770992367, "grad_norm": 2.349343776702881, "learning_rate": 0.0001, "loss": 0.0272, "step": 10340 }, { "epoch": 79.00763358778626, "grad_norm": 2.0683915615081787, "learning_rate": 0.0001, "loss": 0.0283, "step": 10350 }, { "epoch": 79.08396946564885, "grad_norm": 1.6854232549667358, "learning_rate": 0.0001, "loss": 0.028, "step": 10360 }, { "epoch": 79.16030534351145, "grad_norm": 2.7603068351745605, "learning_rate": 0.0001, "loss": 0.028, "step": 10370 }, { "epoch": 79.23664122137404, "grad_norm": 2.642169952392578, "learning_rate": 0.0001, "loss": 0.029, "step": 10380 }, { "epoch": 79.31297709923665, "grad_norm": 2.6648600101470947, "learning_rate": 0.0001, "loss": 0.0295, "step": 10390 }, { "epoch": 79.38931297709924, "grad_norm": 1.959434151649475, "learning_rate": 0.0001, "loss": 0.028, "step": 10400 }, { "epoch": 79.46564885496183, "grad_norm": 1.9033215045928955, "learning_rate": 0.0001, "loss": 0.0282, "step": 10410 }, { "epoch": 79.54198473282443, "grad_norm": 1.7636406421661377, "learning_rate": 0.0001, "loss": 0.0301, "step": 10420 }, { "epoch": 79.61832061068702, "grad_norm": 2.5184988975524902, "learning_rate": 0.0001, "loss": 0.0292, "step": 10430 }, { "epoch": 79.69465648854961, "grad_norm": 2.366982936859131, "learning_rate": 0.0001, "loss": 0.03, "step": 10440 }, { "epoch": 79.77099236641222, "grad_norm": 2.3025264739990234, "learning_rate": 0.0001, "loss": 0.0296, "step": 10450 }, { "epoch": 79.8473282442748, "grad_norm": 2.626521348953247, "learning_rate": 0.0001, "loss": 0.0292, "step": 10460 }, { "epoch": 79.92366412213741, "grad_norm": 1.679948329925537, "learning_rate": 0.0001, "loss": 0.0287, "step": 10470 }, { "epoch": 80.0, "grad_norm": 1.9731758832931519, "learning_rate": 0.0001, "loss": 0.0294, "step": 10480 }, { "epoch": 80.07633587786259, "grad_norm": 1.9730417728424072, "learning_rate": 0.0001, "loss": 0.0283, "step": 10490 }, { "epoch": 80.1526717557252, "grad_norm": 2.9638490676879883, "learning_rate": 0.0001, "loss": 0.0291, "step": 10500 }, { "epoch": 80.22900763358778, "grad_norm": 2.3904917240142822, "learning_rate": 0.0001, "loss": 0.029, "step": 10510 }, { "epoch": 80.30534351145039, "grad_norm": 2.5046310424804688, "learning_rate": 0.0001, "loss": 0.0289, "step": 10520 }, { "epoch": 80.38167938931298, "grad_norm": 2.801680088043213, "learning_rate": 0.0001, "loss": 0.0284, "step": 10530 }, { "epoch": 80.45801526717557, "grad_norm": 2.300738573074341, "learning_rate": 0.0001, "loss": 0.028, "step": 10540 }, { "epoch": 80.53435114503817, "grad_norm": 2.722491502761841, "learning_rate": 0.0001, "loss": 0.027, "step": 10550 }, { "epoch": 80.61068702290076, "grad_norm": 2.40254282951355, "learning_rate": 0.0001, "loss": 0.0264, "step": 10560 }, { "epoch": 80.68702290076335, "grad_norm": 1.7349532842636108, "learning_rate": 0.0001, "loss": 0.0274, "step": 10570 }, { "epoch": 80.76335877862596, "grad_norm": 2.5299434661865234, "learning_rate": 0.0001, "loss": 0.0273, "step": 10580 }, { "epoch": 80.83969465648855, "grad_norm": 3.0167794227600098, "learning_rate": 0.0001, "loss": 0.0275, "step": 10590 }, { "epoch": 80.91603053435115, "grad_norm": 2.757657289505005, "learning_rate": 0.0001, "loss": 0.0263, "step": 10600 }, { "epoch": 80.99236641221374, "grad_norm": 2.82710862159729, "learning_rate": 0.0001, "loss": 0.0272, "step": 10610 }, { "epoch": 81.06870229007633, "grad_norm": 2.7465686798095703, "learning_rate": 0.0001, "loss": 0.0273, "step": 10620 }, { "epoch": 81.14503816793894, "grad_norm": 2.4258642196655273, "learning_rate": 0.0001, "loss": 0.0261, "step": 10630 }, { "epoch": 81.22137404580153, "grad_norm": 2.520108222961426, "learning_rate": 0.0001, "loss": 0.0275, "step": 10640 }, { "epoch": 81.29770992366412, "grad_norm": 1.6708160638809204, "learning_rate": 0.0001, "loss": 0.0265, "step": 10650 }, { "epoch": 81.37404580152672, "grad_norm": 2.5221779346466064, "learning_rate": 0.0001, "loss": 0.0274, "step": 10660 }, { "epoch": 81.45038167938931, "grad_norm": 1.7794641256332397, "learning_rate": 0.0001, "loss": 0.027, "step": 10670 }, { "epoch": 81.52671755725191, "grad_norm": 1.539909839630127, "learning_rate": 0.0001, "loss": 0.0265, "step": 10680 }, { "epoch": 81.6030534351145, "grad_norm": 2.518326997756958, "learning_rate": 0.0001, "loss": 0.0277, "step": 10690 }, { "epoch": 81.6793893129771, "grad_norm": 1.9832795858383179, "learning_rate": 0.0001, "loss": 0.0275, "step": 10700 }, { "epoch": 81.7557251908397, "grad_norm": 2.2501864433288574, "learning_rate": 0.0001, "loss": 0.0276, "step": 10710 }, { "epoch": 81.83206106870229, "grad_norm": 2.2928073406219482, "learning_rate": 0.0001, "loss": 0.0275, "step": 10720 }, { "epoch": 81.90839694656489, "grad_norm": 1.766321063041687, "learning_rate": 0.0001, "loss": 0.0282, "step": 10730 }, { "epoch": 81.98473282442748, "grad_norm": 1.7931703329086304, "learning_rate": 0.0001, "loss": 0.0284, "step": 10740 }, { "epoch": 82.06106870229007, "grad_norm": 2.121523380279541, "learning_rate": 0.0001, "loss": 0.0281, "step": 10750 }, { "epoch": 82.13740458015268, "grad_norm": 1.7455849647521973, "learning_rate": 0.0001, "loss": 0.0289, "step": 10760 }, { "epoch": 82.21374045801527, "grad_norm": 2.0523483753204346, "learning_rate": 0.0001, "loss": 0.0273, "step": 10770 }, { "epoch": 82.29007633587786, "grad_norm": 2.081961154937744, "learning_rate": 0.0001, "loss": 0.029, "step": 10780 }, { "epoch": 82.36641221374046, "grad_norm": 3.3333113193511963, "learning_rate": 0.0001, "loss": 0.0293, "step": 10790 }, { "epoch": 82.44274809160305, "grad_norm": 2.593715190887451, "learning_rate": 0.0001, "loss": 0.0298, "step": 10800 }, { "epoch": 82.51908396946565, "grad_norm": 2.091723918914795, "learning_rate": 0.0001, "loss": 0.0303, "step": 10810 }, { "epoch": 82.59541984732824, "grad_norm": 2.1461970806121826, "learning_rate": 0.0001, "loss": 0.0288, "step": 10820 }, { "epoch": 82.67175572519083, "grad_norm": 2.9187562465667725, "learning_rate": 0.0001, "loss": 0.0305, "step": 10830 }, { "epoch": 82.74809160305344, "grad_norm": 2.254857301712036, "learning_rate": 0.0001, "loss": 0.03, "step": 10840 }, { "epoch": 82.82442748091603, "grad_norm": 2.656959295272827, "learning_rate": 0.0001, "loss": 0.0279, "step": 10850 }, { "epoch": 82.90076335877862, "grad_norm": 2.8794617652893066, "learning_rate": 0.0001, "loss": 0.0277, "step": 10860 }, { "epoch": 82.97709923664122, "grad_norm": 2.2824857234954834, "learning_rate": 0.0001, "loss": 0.0278, "step": 10870 }, { "epoch": 83.05343511450381, "grad_norm": 2.040933847427368, "learning_rate": 0.0001, "loss": 0.0272, "step": 10880 }, { "epoch": 83.12977099236642, "grad_norm": 1.7349995374679565, "learning_rate": 0.0001, "loss": 0.0273, "step": 10890 }, { "epoch": 83.20610687022901, "grad_norm": 2.5776727199554443, "learning_rate": 0.0001, "loss": 0.0278, "step": 10900 }, { "epoch": 83.2824427480916, "grad_norm": 2.4218461513519287, "learning_rate": 0.0001, "loss": 0.0271, "step": 10910 }, { "epoch": 83.3587786259542, "grad_norm": 1.8032950162887573, "learning_rate": 0.0001, "loss": 0.0274, "step": 10920 }, { "epoch": 83.43511450381679, "grad_norm": 1.939735770225525, "learning_rate": 0.0001, "loss": 0.0272, "step": 10930 }, { "epoch": 83.5114503816794, "grad_norm": 2.151937961578369, "learning_rate": 0.0001, "loss": 0.0272, "step": 10940 }, { "epoch": 83.58778625954199, "grad_norm": 1.9332836866378784, "learning_rate": 0.0001, "loss": 0.028, "step": 10950 }, { "epoch": 83.66412213740458, "grad_norm": 2.5597593784332275, "learning_rate": 0.0001, "loss": 0.0276, "step": 10960 }, { "epoch": 83.74045801526718, "grad_norm": 1.9307562112808228, "learning_rate": 0.0001, "loss": 0.0277, "step": 10970 }, { "epoch": 83.81679389312977, "grad_norm": 3.188930034637451, "learning_rate": 0.0001, "loss": 0.0273, "step": 10980 }, { "epoch": 83.89312977099236, "grad_norm": 2.5842020511627197, "learning_rate": 0.0001, "loss": 0.0282, "step": 10990 }, { "epoch": 83.96946564885496, "grad_norm": 2.562777042388916, "learning_rate": 0.0001, "loss": 0.0279, "step": 11000 }, { "epoch": 84.04580152671755, "grad_norm": 2.7715718746185303, "learning_rate": 0.0001, "loss": 0.028, "step": 11010 }, { "epoch": 84.12213740458016, "grad_norm": 2.366283893585205, "learning_rate": 0.0001, "loss": 0.0277, "step": 11020 }, { "epoch": 84.19847328244275, "grad_norm": 2.171762466430664, "learning_rate": 0.0001, "loss": 0.0267, "step": 11030 }, { "epoch": 84.27480916030534, "grad_norm": 1.8973623514175415, "learning_rate": 0.0001, "loss": 0.0272, "step": 11040 }, { "epoch": 84.35114503816794, "grad_norm": 1.6622511148452759, "learning_rate": 0.0001, "loss": 0.0269, "step": 11050 }, { "epoch": 84.42748091603053, "grad_norm": 2.2300472259521484, "learning_rate": 0.0001, "loss": 0.0271, "step": 11060 }, { "epoch": 84.50381679389314, "grad_norm": 2.7135040760040283, "learning_rate": 0.0001, "loss": 0.0279, "step": 11070 }, { "epoch": 84.58015267175573, "grad_norm": 2.898020029067993, "learning_rate": 0.0001, "loss": 0.0282, "step": 11080 }, { "epoch": 84.65648854961832, "grad_norm": 1.697792410850525, "learning_rate": 0.0001, "loss": 0.0284, "step": 11090 }, { "epoch": 84.73282442748092, "grad_norm": 2.4063503742218018, "learning_rate": 0.0001, "loss": 0.0279, "step": 11100 }, { "epoch": 84.80916030534351, "grad_norm": 2.911066770553589, "learning_rate": 0.0001, "loss": 0.0277, "step": 11110 }, { "epoch": 84.8854961832061, "grad_norm": 2.2331111431121826, "learning_rate": 0.0001, "loss": 0.028, "step": 11120 }, { "epoch": 84.9618320610687, "grad_norm": 2.07033634185791, "learning_rate": 0.0001, "loss": 0.0277, "step": 11130 }, { "epoch": 85.0381679389313, "grad_norm": 1.8405708074569702, "learning_rate": 0.0001, "loss": 0.0265, "step": 11140 }, { "epoch": 85.1145038167939, "grad_norm": 2.921661615371704, "learning_rate": 0.0001, "loss": 0.0274, "step": 11150 }, { "epoch": 85.19083969465649, "grad_norm": 2.8261125087738037, "learning_rate": 0.0001, "loss": 0.0294, "step": 11160 }, { "epoch": 85.26717557251908, "grad_norm": 2.247525453567505, "learning_rate": 0.0001, "loss": 0.0288, "step": 11170 }, { "epoch": 85.34351145038168, "grad_norm": 2.924842596054077, "learning_rate": 0.0001, "loss": 0.0286, "step": 11180 }, { "epoch": 85.41984732824427, "grad_norm": 1.7985937595367432, "learning_rate": 0.0001, "loss": 0.0282, "step": 11190 }, { "epoch": 85.49618320610686, "grad_norm": 2.387685537338257, "learning_rate": 0.0001, "loss": 0.0278, "step": 11200 }, { "epoch": 85.57251908396947, "grad_norm": 2.0068447589874268, "learning_rate": 0.0001, "loss": 0.0277, "step": 11210 }, { "epoch": 85.64885496183206, "grad_norm": 2.6553947925567627, "learning_rate": 0.0001, "loss": 0.0275, "step": 11220 }, { "epoch": 85.72519083969466, "grad_norm": 1.978235125541687, "learning_rate": 0.0001, "loss": 0.028, "step": 11230 }, { "epoch": 85.80152671755725, "grad_norm": 2.1442441940307617, "learning_rate": 0.0001, "loss": 0.028, "step": 11240 }, { "epoch": 85.87786259541984, "grad_norm": 2.2226665019989014, "learning_rate": 0.0001, "loss": 0.0294, "step": 11250 }, { "epoch": 85.95419847328245, "grad_norm": 2.2620153427124023, "learning_rate": 0.0001, "loss": 0.0289, "step": 11260 }, { "epoch": 86.03053435114504, "grad_norm": 1.8819456100463867, "learning_rate": 0.0001, "loss": 0.0271, "step": 11270 }, { "epoch": 86.10687022900764, "grad_norm": 2.2033238410949707, "learning_rate": 0.0001, "loss": 0.0269, "step": 11280 }, { "epoch": 86.18320610687023, "grad_norm": 2.174774169921875, "learning_rate": 0.0001, "loss": 0.0266, "step": 11290 }, { "epoch": 86.25954198473282, "grad_norm": 2.260216474533081, "learning_rate": 0.0001, "loss": 0.0275, "step": 11300 }, { "epoch": 86.33587786259542, "grad_norm": 2.1385982036590576, "learning_rate": 0.0001, "loss": 0.0274, "step": 11310 }, { "epoch": 86.41221374045801, "grad_norm": 2.2329347133636475, "learning_rate": 0.0001, "loss": 0.0271, "step": 11320 }, { "epoch": 86.4885496183206, "grad_norm": 2.1605770587921143, "learning_rate": 0.0001, "loss": 0.0276, "step": 11330 }, { "epoch": 86.56488549618321, "grad_norm": 3.1473443508148193, "learning_rate": 0.0001, "loss": 0.0296, "step": 11340 }, { "epoch": 86.6412213740458, "grad_norm": 3.6369874477386475, "learning_rate": 0.0001, "loss": 0.0296, "step": 11350 }, { "epoch": 86.7175572519084, "grad_norm": 2.714406728744507, "learning_rate": 0.0001, "loss": 0.0275, "step": 11360 }, { "epoch": 86.79389312977099, "grad_norm": 2.714616298675537, "learning_rate": 0.0001, "loss": 0.0274, "step": 11370 }, { "epoch": 86.87022900763358, "grad_norm": 2.96529483795166, "learning_rate": 0.0001, "loss": 0.026, "step": 11380 }, { "epoch": 86.94656488549619, "grad_norm": 2.035886764526367, "learning_rate": 0.0001, "loss": 0.0253, "step": 11390 }, { "epoch": 87.02290076335878, "grad_norm": 1.8550589084625244, "learning_rate": 0.0001, "loss": 0.0248, "step": 11400 }, { "epoch": 87.09923664122137, "grad_norm": 1.7593374252319336, "learning_rate": 0.0001, "loss": 0.0248, "step": 11410 }, { "epoch": 87.17557251908397, "grad_norm": 2.0461885929107666, "learning_rate": 0.0001, "loss": 0.026, "step": 11420 }, { "epoch": 87.25190839694656, "grad_norm": 2.147975444793701, "learning_rate": 0.0001, "loss": 0.0257, "step": 11430 }, { "epoch": 87.32824427480917, "grad_norm": 2.6490743160247803, "learning_rate": 0.0001, "loss": 0.0258, "step": 11440 }, { "epoch": 87.40458015267176, "grad_norm": 1.9579747915267944, "learning_rate": 0.0001, "loss": 0.026, "step": 11450 }, { "epoch": 87.48091603053435, "grad_norm": 2.7576167583465576, "learning_rate": 0.0001, "loss": 0.0269, "step": 11460 }, { "epoch": 87.55725190839695, "grad_norm": 2.137805461883545, "learning_rate": 0.0001, "loss": 0.0267, "step": 11470 }, { "epoch": 87.63358778625954, "grad_norm": 2.038662910461426, "learning_rate": 0.0001, "loss": 0.026, "step": 11480 }, { "epoch": 87.70992366412214, "grad_norm": 2.3412673473358154, "learning_rate": 0.0001, "loss": 0.0262, "step": 11490 }, { "epoch": 87.78625954198473, "grad_norm": 1.759331464767456, "learning_rate": 0.0001, "loss": 0.0259, "step": 11500 }, { "epoch": 87.86259541984732, "grad_norm": 1.6385581493377686, "learning_rate": 0.0001, "loss": 0.0262, "step": 11510 }, { "epoch": 87.93893129770993, "grad_norm": 1.9937957525253296, "learning_rate": 0.0001, "loss": 0.0282, "step": 11520 }, { "epoch": 88.01526717557252, "grad_norm": 2.134979486465454, "learning_rate": 0.0001, "loss": 0.027, "step": 11530 }, { "epoch": 88.09160305343511, "grad_norm": 3.0720973014831543, "learning_rate": 0.0001, "loss": 0.0267, "step": 11540 }, { "epoch": 88.16793893129771, "grad_norm": 3.3801724910736084, "learning_rate": 0.0001, "loss": 0.0278, "step": 11550 }, { "epoch": 88.2442748091603, "grad_norm": 2.244098663330078, "learning_rate": 0.0001, "loss": 0.0267, "step": 11560 }, { "epoch": 88.3206106870229, "grad_norm": 2.288437843322754, "learning_rate": 0.0001, "loss": 0.0272, "step": 11570 }, { "epoch": 88.3969465648855, "grad_norm": 1.9747563600540161, "learning_rate": 0.0001, "loss": 0.0274, "step": 11580 }, { "epoch": 88.47328244274809, "grad_norm": 2.6894655227661133, "learning_rate": 0.0001, "loss": 0.0263, "step": 11590 }, { "epoch": 88.54961832061069, "grad_norm": 2.0527844429016113, "learning_rate": 0.0001, "loss": 0.0266, "step": 11600 }, { "epoch": 88.62595419847328, "grad_norm": 2.1973631381988525, "learning_rate": 0.0001, "loss": 0.0266, "step": 11610 }, { "epoch": 88.70229007633588, "grad_norm": 1.8795274496078491, "learning_rate": 0.0001, "loss": 0.0272, "step": 11620 }, { "epoch": 88.77862595419847, "grad_norm": 1.9769906997680664, "learning_rate": 0.0001, "loss": 0.0287, "step": 11630 }, { "epoch": 88.85496183206106, "grad_norm": 1.4388185739517212, "learning_rate": 0.0001, "loss": 0.0293, "step": 11640 }, { "epoch": 88.93129770992367, "grad_norm": 2.935965061187744, "learning_rate": 0.0001, "loss": 0.0279, "step": 11650 }, { "epoch": 89.00763358778626, "grad_norm": 2.46012806892395, "learning_rate": 0.0001, "loss": 0.0273, "step": 11660 }, { "epoch": 89.08396946564885, "grad_norm": 1.5942413806915283, "learning_rate": 0.0001, "loss": 0.0273, "step": 11670 }, { "epoch": 89.16030534351145, "grad_norm": 2.5535054206848145, "learning_rate": 0.0001, "loss": 0.0271, "step": 11680 }, { "epoch": 89.23664122137404, "grad_norm": 1.9476726055145264, "learning_rate": 0.0001, "loss": 0.0272, "step": 11690 }, { "epoch": 89.31297709923665, "grad_norm": 1.99789559841156, "learning_rate": 0.0001, "loss": 0.0268, "step": 11700 }, { "epoch": 89.38931297709924, "grad_norm": 1.3621854782104492, "learning_rate": 0.0001, "loss": 0.0264, "step": 11710 }, { "epoch": 89.46564885496183, "grad_norm": 1.9725546836853027, "learning_rate": 0.0001, "loss": 0.0268, "step": 11720 }, { "epoch": 89.54198473282443, "grad_norm": 2.200479745864868, "learning_rate": 0.0001, "loss": 0.0279, "step": 11730 }, { "epoch": 89.61832061068702, "grad_norm": 2.4446043968200684, "learning_rate": 0.0001, "loss": 0.0281, "step": 11740 }, { "epoch": 89.69465648854961, "grad_norm": 1.9628406763076782, "learning_rate": 0.0001, "loss": 0.0274, "step": 11750 }, { "epoch": 89.77099236641222, "grad_norm": 3.4169468879699707, "learning_rate": 0.0001, "loss": 0.0268, "step": 11760 }, { "epoch": 89.8473282442748, "grad_norm": 2.542243003845215, "learning_rate": 0.0001, "loss": 0.0264, "step": 11770 }, { "epoch": 89.92366412213741, "grad_norm": 2.8820714950561523, "learning_rate": 0.0001, "loss": 0.0264, "step": 11780 }, { "epoch": 90.0, "grad_norm": 2.2055888175964355, "learning_rate": 0.0001, "loss": 0.0263, "step": 11790 }, { "epoch": 90.07633587786259, "grad_norm": 2.675469398498535, "learning_rate": 0.0001, "loss": 0.0269, "step": 11800 }, { "epoch": 90.1526717557252, "grad_norm": 2.095036745071411, "learning_rate": 0.0001, "loss": 0.0275, "step": 11810 }, { "epoch": 90.22900763358778, "grad_norm": 2.6725072860717773, "learning_rate": 0.0001, "loss": 0.0274, "step": 11820 }, { "epoch": 90.30534351145039, "grad_norm": 3.2174298763275146, "learning_rate": 0.0001, "loss": 0.0273, "step": 11830 }, { "epoch": 90.38167938931298, "grad_norm": 2.6546199321746826, "learning_rate": 0.0001, "loss": 0.028, "step": 11840 }, { "epoch": 90.45801526717557, "grad_norm": 2.589028835296631, "learning_rate": 0.0001, "loss": 0.0264, "step": 11850 }, { "epoch": 90.53435114503817, "grad_norm": 2.3547821044921875, "learning_rate": 0.0001, "loss": 0.0267, "step": 11860 }, { "epoch": 90.61068702290076, "grad_norm": 2.279137372970581, "learning_rate": 0.0001, "loss": 0.0261, "step": 11870 }, { "epoch": 90.68702290076335, "grad_norm": 2.4125990867614746, "learning_rate": 0.0001, "loss": 0.0263, "step": 11880 }, { "epoch": 90.76335877862596, "grad_norm": 2.2775156497955322, "learning_rate": 0.0001, "loss": 0.0262, "step": 11890 }, { "epoch": 90.83969465648855, "grad_norm": 2.160188674926758, "learning_rate": 0.0001, "loss": 0.0259, "step": 11900 }, { "epoch": 90.91603053435115, "grad_norm": 2.4646904468536377, "learning_rate": 0.0001, "loss": 0.0274, "step": 11910 }, { "epoch": 90.99236641221374, "grad_norm": 2.0003888607025146, "learning_rate": 0.0001, "loss": 0.0265, "step": 11920 }, { "epoch": 91.06870229007633, "grad_norm": 1.9841735363006592, "learning_rate": 0.0001, "loss": 0.026, "step": 11930 }, { "epoch": 91.14503816793894, "grad_norm": 1.876184105873108, "learning_rate": 0.0001, "loss": 0.0261, "step": 11940 }, { "epoch": 91.22137404580153, "grad_norm": 1.6449109315872192, "learning_rate": 0.0001, "loss": 0.0259, "step": 11950 }, { "epoch": 91.29770992366412, "grad_norm": 2.1331706047058105, "learning_rate": 0.0001, "loss": 0.0267, "step": 11960 }, { "epoch": 91.37404580152672, "grad_norm": 1.804682731628418, "learning_rate": 0.0001, "loss": 0.0269, "step": 11970 }, { "epoch": 91.45038167938931, "grad_norm": 2.599132537841797, "learning_rate": 0.0001, "loss": 0.0277, "step": 11980 }, { "epoch": 91.52671755725191, "grad_norm": 2.1297760009765625, "learning_rate": 0.0001, "loss": 0.0288, "step": 11990 }, { "epoch": 91.6030534351145, "grad_norm": 2.3124494552612305, "learning_rate": 0.0001, "loss": 0.0278, "step": 12000 }, { "epoch": 91.6793893129771, "grad_norm": 1.6630727052688599, "learning_rate": 0.0001, "loss": 0.0269, "step": 12010 }, { "epoch": 91.7557251908397, "grad_norm": 2.5629305839538574, "learning_rate": 0.0001, "loss": 0.0288, "step": 12020 }, { "epoch": 91.83206106870229, "grad_norm": 2.616098165512085, "learning_rate": 0.0001, "loss": 0.0295, "step": 12030 }, { "epoch": 91.90839694656489, "grad_norm": 2.0494155883789062, "learning_rate": 0.0001, "loss": 0.0274, "step": 12040 }, { "epoch": 91.98473282442748, "grad_norm": 1.736792802810669, "learning_rate": 0.0001, "loss": 0.0268, "step": 12050 }, { "epoch": 92.06106870229007, "grad_norm": 1.6075241565704346, "learning_rate": 0.0001, "loss": 0.0267, "step": 12060 }, { "epoch": 92.13740458015268, "grad_norm": 1.940314769744873, "learning_rate": 0.0001, "loss": 0.0284, "step": 12070 }, { "epoch": 92.21374045801527, "grad_norm": 2.0320138931274414, "learning_rate": 0.0001, "loss": 0.0284, "step": 12080 }, { "epoch": 92.29007633587786, "grad_norm": 1.979528546333313, "learning_rate": 0.0001, "loss": 0.0291, "step": 12090 }, { "epoch": 92.36641221374046, "grad_norm": 2.8313446044921875, "learning_rate": 0.0001, "loss": 0.0296, "step": 12100 }, { "epoch": 92.44274809160305, "grad_norm": 3.553704261779785, "learning_rate": 0.0001, "loss": 0.0315, "step": 12110 }, { "epoch": 92.51908396946565, "grad_norm": 2.941283941268921, "learning_rate": 0.0001, "loss": 0.0284, "step": 12120 }, { "epoch": 92.59541984732824, "grad_norm": 2.710987091064453, "learning_rate": 0.0001, "loss": 0.0279, "step": 12130 }, { "epoch": 92.67175572519083, "grad_norm": 1.8682197332382202, "learning_rate": 0.0001, "loss": 0.0264, "step": 12140 }, { "epoch": 92.74809160305344, "grad_norm": 1.8129642009735107, "learning_rate": 0.0001, "loss": 0.0263, "step": 12150 }, { "epoch": 92.82442748091603, "grad_norm": 2.1207287311553955, "learning_rate": 0.0001, "loss": 0.0269, "step": 12160 }, { "epoch": 92.90076335877862, "grad_norm": 1.6483654975891113, "learning_rate": 0.0001, "loss": 0.0268, "step": 12170 }, { "epoch": 92.97709923664122, "grad_norm": 2.1568973064422607, "learning_rate": 0.0001, "loss": 0.0262, "step": 12180 }, { "epoch": 93.05343511450381, "grad_norm": 1.9511538743972778, "learning_rate": 0.0001, "loss": 0.0265, "step": 12190 }, { "epoch": 93.12977099236642, "grad_norm": 1.701849102973938, "learning_rate": 0.0001, "loss": 0.0262, "step": 12200 }, { "epoch": 93.20610687022901, "grad_norm": 2.1576130390167236, "learning_rate": 0.0001, "loss": 0.0271, "step": 12210 }, { "epoch": 93.2824427480916, "grad_norm": 1.8654911518096924, "learning_rate": 0.0001, "loss": 0.0266, "step": 12220 }, { "epoch": 93.3587786259542, "grad_norm": 2.542178153991699, "learning_rate": 0.0001, "loss": 0.0271, "step": 12230 }, { "epoch": 93.43511450381679, "grad_norm": 2.7562649250030518, "learning_rate": 0.0001, "loss": 0.0322, "step": 12240 }, { "epoch": 93.5114503816794, "grad_norm": 3.43674635887146, "learning_rate": 0.0001, "loss": 0.028, "step": 12250 }, { "epoch": 93.58778625954199, "grad_norm": 2.3735146522521973, "learning_rate": 0.0001, "loss": 0.028, "step": 12260 }, { "epoch": 93.66412213740458, "grad_norm": 2.7813379764556885, "learning_rate": 0.0001, "loss": 0.0274, "step": 12270 }, { "epoch": 93.74045801526718, "grad_norm": 2.0569212436676025, "learning_rate": 0.0001, "loss": 0.0276, "step": 12280 }, { "epoch": 93.81679389312977, "grad_norm": 2.2714474201202393, "learning_rate": 0.0001, "loss": 0.0261, "step": 12290 }, { "epoch": 93.89312977099236, "grad_norm": 2.0890860557556152, "learning_rate": 0.0001, "loss": 0.0265, "step": 12300 }, { "epoch": 93.96946564885496, "grad_norm": 1.767375111579895, "learning_rate": 0.0001, "loss": 0.0258, "step": 12310 }, { "epoch": 94.04580152671755, "grad_norm": 2.6445882320404053, "learning_rate": 0.0001, "loss": 0.0268, "step": 12320 }, { "epoch": 94.12213740458016, "grad_norm": 2.270380973815918, "learning_rate": 0.0001, "loss": 0.0275, "step": 12330 }, { "epoch": 94.19847328244275, "grad_norm": 1.983239769935608, "learning_rate": 0.0001, "loss": 0.0276, "step": 12340 }, { "epoch": 94.27480916030534, "grad_norm": 1.4643745422363281, "learning_rate": 0.0001, "loss": 0.0271, "step": 12350 }, { "epoch": 94.35114503816794, "grad_norm": 1.9631189107894897, "learning_rate": 0.0001, "loss": 0.0262, "step": 12360 }, { "epoch": 94.42748091603053, "grad_norm": 2.1273958683013916, "learning_rate": 0.0001, "loss": 0.0262, "step": 12370 }, { "epoch": 94.50381679389314, "grad_norm": 2.6540958881378174, "learning_rate": 0.0001, "loss": 0.0267, "step": 12380 }, { "epoch": 94.58015267175573, "grad_norm": 2.3812499046325684, "learning_rate": 0.0001, "loss": 0.0267, "step": 12390 }, { "epoch": 94.65648854961832, "grad_norm": 2.554792881011963, "learning_rate": 0.0001, "loss": 0.0274, "step": 12400 }, { "epoch": 94.73282442748092, "grad_norm": 1.4970659017562866, "learning_rate": 0.0001, "loss": 0.0277, "step": 12410 }, { "epoch": 94.80916030534351, "grad_norm": 3.5073297023773193, "learning_rate": 0.0001, "loss": 0.0295, "step": 12420 }, { "epoch": 94.8854961832061, "grad_norm": 1.9948437213897705, "learning_rate": 0.0001, "loss": 0.0286, "step": 12430 }, { "epoch": 94.9618320610687, "grad_norm": 1.8702677488327026, "learning_rate": 0.0001, "loss": 0.0281, "step": 12440 }, { "epoch": 95.0381679389313, "grad_norm": 2.4937241077423096, "learning_rate": 0.0001, "loss": 0.0283, "step": 12450 }, { "epoch": 95.1145038167939, "grad_norm": 2.118450164794922, "learning_rate": 0.0001, "loss": 0.0289, "step": 12460 }, { "epoch": 95.19083969465649, "grad_norm": 1.8362172842025757, "learning_rate": 0.0001, "loss": 0.0298, "step": 12470 }, { "epoch": 95.26717557251908, "grad_norm": 2.0029568672180176, "learning_rate": 0.0001, "loss": 0.0291, "step": 12480 }, { "epoch": 95.34351145038168, "grad_norm": 2.0437755584716797, "learning_rate": 0.0001, "loss": 0.0298, "step": 12490 }, { "epoch": 95.41984732824427, "grad_norm": 2.391571521759033, "learning_rate": 0.0001, "loss": 0.0293, "step": 12500 }, { "epoch": 95.49618320610686, "grad_norm": 2.47033953666687, "learning_rate": 0.0001, "loss": 0.0282, "step": 12510 }, { "epoch": 95.57251908396947, "grad_norm": 2.147152900695801, "learning_rate": 0.0001, "loss": 0.0268, "step": 12520 }, { "epoch": 95.64885496183206, "grad_norm": 1.8828831911087036, "learning_rate": 0.0001, "loss": 0.0267, "step": 12530 }, { "epoch": 95.72519083969466, "grad_norm": 2.222689628601074, "learning_rate": 0.0001, "loss": 0.0286, "step": 12540 }, { "epoch": 95.80152671755725, "grad_norm": 1.936876893043518, "learning_rate": 0.0001, "loss": 0.0287, "step": 12550 }, { "epoch": 95.87786259541984, "grad_norm": 2.520296096801758, "learning_rate": 0.0001, "loss": 0.0277, "step": 12560 }, { "epoch": 95.95419847328245, "grad_norm": 2.202451229095459, "learning_rate": 0.0001, "loss": 0.0272, "step": 12570 }, { "epoch": 96.03053435114504, "grad_norm": 2.4258410930633545, "learning_rate": 0.0001, "loss": 0.027, "step": 12580 }, { "epoch": 96.10687022900764, "grad_norm": 1.5114727020263672, "learning_rate": 0.0001, "loss": 0.0267, "step": 12590 }, { "epoch": 96.18320610687023, "grad_norm": 2.437631845474243, "learning_rate": 0.0001, "loss": 0.0275, "step": 12600 }, { "epoch": 96.25954198473282, "grad_norm": 2.919373035430908, "learning_rate": 0.0001, "loss": 0.0269, "step": 12610 }, { "epoch": 96.33587786259542, "grad_norm": 2.049731731414795, "learning_rate": 0.0001, "loss": 0.0274, "step": 12620 }, { "epoch": 96.41221374045801, "grad_norm": 2.110135316848755, "learning_rate": 0.0001, "loss": 0.026, "step": 12630 }, { "epoch": 96.4885496183206, "grad_norm": 1.9516268968582153, "learning_rate": 0.0001, "loss": 0.026, "step": 12640 }, { "epoch": 96.56488549618321, "grad_norm": 2.519014835357666, "learning_rate": 0.0001, "loss": 0.0263, "step": 12650 }, { "epoch": 96.6412213740458, "grad_norm": 1.9871870279312134, "learning_rate": 0.0001, "loss": 0.027, "step": 12660 }, { "epoch": 96.7175572519084, "grad_norm": 1.9281364679336548, "learning_rate": 0.0001, "loss": 0.027, "step": 12670 }, { "epoch": 96.79389312977099, "grad_norm": 1.8452962636947632, "learning_rate": 0.0001, "loss": 0.0264, "step": 12680 }, { "epoch": 96.87022900763358, "grad_norm": 1.5770654678344727, "learning_rate": 0.0001, "loss": 0.0266, "step": 12690 }, { "epoch": 96.94656488549619, "grad_norm": 1.9786511659622192, "learning_rate": 0.0001, "loss": 0.0272, "step": 12700 }, { "epoch": 97.02290076335878, "grad_norm": 1.9076733589172363, "learning_rate": 0.0001, "loss": 0.0265, "step": 12710 }, { "epoch": 97.09923664122137, "grad_norm": 1.8837732076644897, "learning_rate": 0.0001, "loss": 0.0262, "step": 12720 }, { "epoch": 97.17557251908397, "grad_norm": 2.1225030422210693, "learning_rate": 0.0001, "loss": 0.0267, "step": 12730 }, { "epoch": 97.25190839694656, "grad_norm": 2.5011379718780518, "learning_rate": 0.0001, "loss": 0.0269, "step": 12740 }, { "epoch": 97.32824427480917, "grad_norm": 1.5919737815856934, "learning_rate": 0.0001, "loss": 0.0263, "step": 12750 }, { "epoch": 97.40458015267176, "grad_norm": 1.636897325515747, "learning_rate": 0.0001, "loss": 0.0262, "step": 12760 }, { "epoch": 97.48091603053435, "grad_norm": 1.0833041667938232, "learning_rate": 0.0001, "loss": 0.0263, "step": 12770 }, { "epoch": 97.55725190839695, "grad_norm": 1.9773684740066528, "learning_rate": 0.0001, "loss": 0.0287, "step": 12780 }, { "epoch": 97.63358778625954, "grad_norm": 1.9876443147659302, "learning_rate": 0.0001, "loss": 0.0291, "step": 12790 }, { "epoch": 97.70992366412214, "grad_norm": 2.0529608726501465, "learning_rate": 0.0001, "loss": 0.0275, "step": 12800 }, { "epoch": 97.78625954198473, "grad_norm": 1.7003059387207031, "learning_rate": 0.0001, "loss": 0.0282, "step": 12810 }, { "epoch": 97.86259541984732, "grad_norm": 1.9323927164077759, "learning_rate": 0.0001, "loss": 0.0272, "step": 12820 }, { "epoch": 97.93893129770993, "grad_norm": 1.6131038665771484, "learning_rate": 0.0001, "loss": 0.0269, "step": 12830 }, { "epoch": 98.01526717557252, "grad_norm": 1.63638436794281, "learning_rate": 0.0001, "loss": 0.0265, "step": 12840 }, { "epoch": 98.09160305343511, "grad_norm": 1.3910961151123047, "learning_rate": 0.0001, "loss": 0.0254, "step": 12850 }, { "epoch": 98.16793893129771, "grad_norm": 2.1709914207458496, "learning_rate": 0.0001, "loss": 0.0262, "step": 12860 }, { "epoch": 98.2442748091603, "grad_norm": 1.7440389394760132, "learning_rate": 0.0001, "loss": 0.0266, "step": 12870 }, { "epoch": 98.3206106870229, "grad_norm": 2.699617862701416, "learning_rate": 0.0001, "loss": 0.0281, "step": 12880 }, { "epoch": 98.3969465648855, "grad_norm": 2.7482492923736572, "learning_rate": 0.0001, "loss": 0.0284, "step": 12890 }, { "epoch": 98.47328244274809, "grad_norm": 2.827089786529541, "learning_rate": 0.0001, "loss": 0.0277, "step": 12900 }, { "epoch": 98.54961832061069, "grad_norm": 2.6283609867095947, "learning_rate": 0.0001, "loss": 0.0264, "step": 12910 }, { "epoch": 98.62595419847328, "grad_norm": 2.8441927433013916, "learning_rate": 0.0001, "loss": 0.0264, "step": 12920 }, { "epoch": 98.70229007633588, "grad_norm": 2.3401236534118652, "learning_rate": 0.0001, "loss": 0.0252, "step": 12930 }, { "epoch": 98.77862595419847, "grad_norm": 2.2555291652679443, "learning_rate": 0.0001, "loss": 0.0254, "step": 12940 }, { "epoch": 98.85496183206106, "grad_norm": 1.9515525102615356, "learning_rate": 0.0001, "loss": 0.0247, "step": 12950 }, { "epoch": 98.93129770992367, "grad_norm": 2.2192654609680176, "learning_rate": 0.0001, "loss": 0.0262, "step": 12960 }, { "epoch": 99.00763358778626, "grad_norm": 1.8601863384246826, "learning_rate": 0.0001, "loss": 0.0257, "step": 12970 }, { "epoch": 99.08396946564885, "grad_norm": 1.718033790588379, "learning_rate": 0.0001, "loss": 0.0258, "step": 12980 }, { "epoch": 99.16030534351145, "grad_norm": 3.2218809127807617, "learning_rate": 0.0001, "loss": 0.026, "step": 12990 }, { "epoch": 99.23664122137404, "grad_norm": 2.6817145347595215, "learning_rate": 0.0001, "loss": 0.027, "step": 13000 }, { "epoch": 99.31297709923665, "grad_norm": 2.4839651584625244, "learning_rate": 0.0001, "loss": 0.025, "step": 13010 }, { "epoch": 99.38931297709924, "grad_norm": 2.777679204940796, "learning_rate": 0.0001, "loss": 0.0263, "step": 13020 }, { "epoch": 99.46564885496183, "grad_norm": 1.944774866104126, "learning_rate": 0.0001, "loss": 0.0256, "step": 13030 }, { "epoch": 99.54198473282443, "grad_norm": 1.5498093366622925, "learning_rate": 0.0001, "loss": 0.0247, "step": 13040 }, { "epoch": 99.61832061068702, "grad_norm": 1.9581953287124634, "learning_rate": 0.0001, "loss": 0.0257, "step": 13050 }, { "epoch": 99.69465648854961, "grad_norm": 2.437225580215454, "learning_rate": 0.0001, "loss": 0.0257, "step": 13060 }, { "epoch": 99.77099236641222, "grad_norm": 2.0014047622680664, "learning_rate": 0.0001, "loss": 0.0267, "step": 13070 }, { "epoch": 99.8473282442748, "grad_norm": 1.6492891311645508, "learning_rate": 0.0001, "loss": 0.0265, "step": 13080 }, { "epoch": 99.92366412213741, "grad_norm": 1.7730087041854858, "learning_rate": 0.0001, "loss": 0.0264, "step": 13090 }, { "epoch": 100.0, "grad_norm": 1.9110543727874756, "learning_rate": 0.0001, "loss": 0.027, "step": 13100 }, { "epoch": 100.07633587786259, "grad_norm": 2.593916654586792, "learning_rate": 0.0001, "loss": 0.027, "step": 13110 }, { "epoch": 100.1526717557252, "grad_norm": 1.9830400943756104, "learning_rate": 0.0001, "loss": 0.0266, "step": 13120 }, { "epoch": 100.22900763358778, "grad_norm": 1.9248664379119873, "learning_rate": 0.0001, "loss": 0.0261, "step": 13130 }, { "epoch": 100.30534351145039, "grad_norm": 1.3814231157302856, "learning_rate": 0.0001, "loss": 0.0262, "step": 13140 }, { "epoch": 100.38167938931298, "grad_norm": 2.674006223678589, "learning_rate": 0.0001, "loss": 0.0268, "step": 13150 }, { "epoch": 100.45801526717557, "grad_norm": 2.1744117736816406, "learning_rate": 0.0001, "loss": 0.0267, "step": 13160 }, { "epoch": 100.53435114503817, "grad_norm": 1.8403598070144653, "learning_rate": 0.0001, "loss": 0.0267, "step": 13170 }, { "epoch": 100.61068702290076, "grad_norm": 1.9773123264312744, "learning_rate": 0.0001, "loss": 0.027, "step": 13180 }, { "epoch": 100.68702290076335, "grad_norm": 1.864845871925354, "learning_rate": 0.0001, "loss": 0.0266, "step": 13190 }, { "epoch": 100.76335877862596, "grad_norm": 1.8461503982543945, "learning_rate": 0.0001, "loss": 0.0265, "step": 13200 }, { "epoch": 100.83969465648855, "grad_norm": 1.7676866054534912, "learning_rate": 0.0001, "loss": 0.0257, "step": 13210 }, { "epoch": 100.91603053435115, "grad_norm": 1.608507752418518, "learning_rate": 0.0001, "loss": 0.0265, "step": 13220 }, { "epoch": 100.99236641221374, "grad_norm": 2.0789263248443604, "learning_rate": 0.0001, "loss": 0.027, "step": 13230 }, { "epoch": 101.06870229007633, "grad_norm": 1.9388922452926636, "learning_rate": 0.0001, "loss": 0.0263, "step": 13240 }, { "epoch": 101.14503816793894, "grad_norm": 2.257690906524658, "learning_rate": 0.0001, "loss": 0.0265, "step": 13250 }, { "epoch": 101.22137404580153, "grad_norm": 2.1055948734283447, "learning_rate": 0.0001, "loss": 0.0257, "step": 13260 }, { "epoch": 101.29770992366412, "grad_norm": 1.7157933712005615, "learning_rate": 0.0001, "loss": 0.0263, "step": 13270 }, { "epoch": 101.37404580152672, "grad_norm": 1.5311808586120605, "learning_rate": 0.0001, "loss": 0.0262, "step": 13280 }, { "epoch": 101.45038167938931, "grad_norm": 2.2825655937194824, "learning_rate": 0.0001, "loss": 0.0269, "step": 13290 }, { "epoch": 101.52671755725191, "grad_norm": 2.837780714035034, "learning_rate": 0.0001, "loss": 0.0279, "step": 13300 }, { "epoch": 101.6030534351145, "grad_norm": 1.9594930410385132, "learning_rate": 0.0001, "loss": 0.0273, "step": 13310 }, { "epoch": 101.6793893129771, "grad_norm": 1.9205213785171509, "learning_rate": 0.0001, "loss": 0.0271, "step": 13320 }, { "epoch": 101.7557251908397, "grad_norm": 2.6183323860168457, "learning_rate": 0.0001, "loss": 0.0273, "step": 13330 }, { "epoch": 101.83206106870229, "grad_norm": 2.1924021244049072, "learning_rate": 0.0001, "loss": 0.0265, "step": 13340 }, { "epoch": 101.90839694656489, "grad_norm": 1.9418336153030396, "learning_rate": 0.0001, "loss": 0.0254, "step": 13350 }, { "epoch": 101.98473282442748, "grad_norm": 2.281630277633667, "learning_rate": 0.0001, "loss": 0.0253, "step": 13360 }, { "epoch": 102.06106870229007, "grad_norm": 1.817372441291809, "learning_rate": 0.0001, "loss": 0.0254, "step": 13370 }, { "epoch": 102.13740458015268, "grad_norm": 1.7329566478729248, "learning_rate": 0.0001, "loss": 0.0244, "step": 13380 }, { "epoch": 102.21374045801527, "grad_norm": 2.245183229446411, "learning_rate": 0.0001, "loss": 0.0267, "step": 13390 }, { "epoch": 102.29007633587786, "grad_norm": 2.141514539718628, "learning_rate": 0.0001, "loss": 0.0264, "step": 13400 }, { "epoch": 102.36641221374046, "grad_norm": 1.868302583694458, "learning_rate": 0.0001, "loss": 0.0271, "step": 13410 }, { "epoch": 102.44274809160305, "grad_norm": 1.9047666788101196, "learning_rate": 0.0001, "loss": 0.0264, "step": 13420 }, { "epoch": 102.51908396946565, "grad_norm": 1.4527908563613892, "learning_rate": 0.0001, "loss": 0.0259, "step": 13430 }, { "epoch": 102.59541984732824, "grad_norm": 1.6587125062942505, "learning_rate": 0.0001, "loss": 0.0262, "step": 13440 }, { "epoch": 102.67175572519083, "grad_norm": 1.9685027599334717, "learning_rate": 0.0001, "loss": 0.0264, "step": 13450 }, { "epoch": 102.74809160305344, "grad_norm": 2.385047197341919, "learning_rate": 0.0001, "loss": 0.0315, "step": 13460 }, { "epoch": 102.82442748091603, "grad_norm": 2.1906399726867676, "learning_rate": 0.0001, "loss": 0.0275, "step": 13470 }, { "epoch": 102.90076335877862, "grad_norm": 2.975116014480591, "learning_rate": 0.0001, "loss": 0.0275, "step": 13480 }, { "epoch": 102.97709923664122, "grad_norm": 2.452561855316162, "learning_rate": 0.0001, "loss": 0.0268, "step": 13490 }, { "epoch": 103.05343511450381, "grad_norm": 2.3257410526275635, "learning_rate": 0.0001, "loss": 0.0264, "step": 13500 }, { "epoch": 103.12977099236642, "grad_norm": 2.1003623008728027, "learning_rate": 0.0001, "loss": 0.0271, "step": 13510 }, { "epoch": 103.20610687022901, "grad_norm": 1.8771532773971558, "learning_rate": 0.0001, "loss": 0.0265, "step": 13520 }, { "epoch": 103.2824427480916, "grad_norm": 3.3052337169647217, "learning_rate": 0.0001, "loss": 0.0317, "step": 13530 }, { "epoch": 103.3587786259542, "grad_norm": 2.5287954807281494, "learning_rate": 0.0001, "loss": 0.0264, "step": 13540 }, { "epoch": 103.43511450381679, "grad_norm": 2.1215076446533203, "learning_rate": 0.0001, "loss": 0.0256, "step": 13550 }, { "epoch": 103.5114503816794, "grad_norm": 2.151548147201538, "learning_rate": 0.0001, "loss": 0.0253, "step": 13560 }, { "epoch": 103.58778625954199, "grad_norm": 2.0747971534729004, "learning_rate": 0.0001, "loss": 0.0248, "step": 13570 }, { "epoch": 103.66412213740458, "grad_norm": 1.8518166542053223, "learning_rate": 0.0001, "loss": 0.0246, "step": 13580 }, { "epoch": 103.74045801526718, "grad_norm": 1.7765434980392456, "learning_rate": 0.0001, "loss": 0.0258, "step": 13590 }, { "epoch": 103.81679389312977, "grad_norm": 2.391880989074707, "learning_rate": 0.0001, "loss": 0.0257, "step": 13600 }, { "epoch": 103.89312977099236, "grad_norm": 1.7466745376586914, "learning_rate": 0.0001, "loss": 0.0248, "step": 13610 }, { "epoch": 103.96946564885496, "grad_norm": 2.612107515335083, "learning_rate": 0.0001, "loss": 0.0264, "step": 13620 }, { "epoch": 104.04580152671755, "grad_norm": 2.5531797409057617, "learning_rate": 0.0001, "loss": 0.025, "step": 13630 }, { "epoch": 104.12213740458016, "grad_norm": 2.0130014419555664, "learning_rate": 0.0001, "loss": 0.026, "step": 13640 }, { "epoch": 104.19847328244275, "grad_norm": 3.0866012573242188, "learning_rate": 0.0001, "loss": 0.0258, "step": 13650 }, { "epoch": 104.27480916030534, "grad_norm": 1.7225936651229858, "learning_rate": 0.0001, "loss": 0.025, "step": 13660 }, { "epoch": 104.35114503816794, "grad_norm": 1.9726008176803589, "learning_rate": 0.0001, "loss": 0.0262, "step": 13670 }, { "epoch": 104.42748091603053, "grad_norm": 2.141629457473755, "learning_rate": 0.0001, "loss": 0.0254, "step": 13680 }, { "epoch": 104.50381679389314, "grad_norm": 1.9255377054214478, "learning_rate": 0.0001, "loss": 0.0251, "step": 13690 }, { "epoch": 104.58015267175573, "grad_norm": 2.3301713466644287, "learning_rate": 0.0001, "loss": 0.0254, "step": 13700 }, { "epoch": 104.65648854961832, "grad_norm": 1.502343773841858, "learning_rate": 0.0001, "loss": 0.0247, "step": 13710 }, { "epoch": 104.73282442748092, "grad_norm": 2.382239818572998, "learning_rate": 0.0001, "loss": 0.026, "step": 13720 }, { "epoch": 104.80916030534351, "grad_norm": 1.9556808471679688, "learning_rate": 0.0001, "loss": 0.0257, "step": 13730 }, { "epoch": 104.8854961832061, "grad_norm": 1.9662972688674927, "learning_rate": 0.0001, "loss": 0.0253, "step": 13740 }, { "epoch": 104.9618320610687, "grad_norm": 1.9106040000915527, "learning_rate": 0.0001, "loss": 0.0254, "step": 13750 }, { "epoch": 105.0381679389313, "grad_norm": 1.9143030643463135, "learning_rate": 0.0001, "loss": 0.0255, "step": 13760 }, { "epoch": 105.1145038167939, "grad_norm": 1.670398235321045, "learning_rate": 0.0001, "loss": 0.0255, "step": 13770 }, { "epoch": 105.19083969465649, "grad_norm": 2.3935201168060303, "learning_rate": 0.0001, "loss": 0.0255, "step": 13780 }, { "epoch": 105.26717557251908, "grad_norm": 2.1972644329071045, "learning_rate": 0.0001, "loss": 0.0252, "step": 13790 }, { "epoch": 105.34351145038168, "grad_norm": 2.434650421142578, "learning_rate": 0.0001, "loss": 0.026, "step": 13800 }, { "epoch": 105.41984732824427, "grad_norm": 2.4319562911987305, "learning_rate": 0.0001, "loss": 0.0266, "step": 13810 }, { "epoch": 105.49618320610686, "grad_norm": 2.9726366996765137, "learning_rate": 0.0001, "loss": 0.027, "step": 13820 }, { "epoch": 105.57251908396947, "grad_norm": 1.8264877796173096, "learning_rate": 0.0001, "loss": 0.0258, "step": 13830 }, { "epoch": 105.64885496183206, "grad_norm": 2.0912246704101562, "learning_rate": 0.0001, "loss": 0.0255, "step": 13840 }, { "epoch": 105.72519083969466, "grad_norm": 1.6885946989059448, "learning_rate": 0.0001, "loss": 0.0253, "step": 13850 }, { "epoch": 105.80152671755725, "grad_norm": 2.030979871749878, "learning_rate": 0.0001, "loss": 0.0259, "step": 13860 }, { "epoch": 105.87786259541984, "grad_norm": 1.9114108085632324, "learning_rate": 0.0001, "loss": 0.0261, "step": 13870 }, { "epoch": 105.95419847328245, "grad_norm": 1.7938579320907593, "learning_rate": 0.0001, "loss": 0.0267, "step": 13880 }, { "epoch": 106.03053435114504, "grad_norm": 2.1011531352996826, "learning_rate": 0.0001, "loss": 0.0266, "step": 13890 }, { "epoch": 106.10687022900764, "grad_norm": 1.5344115495681763, "learning_rate": 0.0001, "loss": 0.0261, "step": 13900 }, { "epoch": 106.18320610687023, "grad_norm": 1.7902506589889526, "learning_rate": 0.0001, "loss": 0.0261, "step": 13910 }, { "epoch": 106.25954198473282, "grad_norm": 1.2303615808486938, "learning_rate": 0.0001, "loss": 0.0265, "step": 13920 }, { "epoch": 106.33587786259542, "grad_norm": 2.060385227203369, "learning_rate": 0.0001, "loss": 0.0267, "step": 13930 }, { "epoch": 106.41221374045801, "grad_norm": 2.898751735687256, "learning_rate": 0.0001, "loss": 0.0272, "step": 13940 }, { "epoch": 106.4885496183206, "grad_norm": 2.0475878715515137, "learning_rate": 0.0001, "loss": 0.0277, "step": 13950 }, { "epoch": 106.56488549618321, "grad_norm": 2.111868381500244, "learning_rate": 0.0001, "loss": 0.0264, "step": 13960 }, { "epoch": 106.6412213740458, "grad_norm": 3.0741000175476074, "learning_rate": 0.0001, "loss": 0.026, "step": 13970 }, { "epoch": 106.7175572519084, "grad_norm": 3.1445024013519287, "learning_rate": 0.0001, "loss": 0.0262, "step": 13980 }, { "epoch": 106.79389312977099, "grad_norm": 1.571978211402893, "learning_rate": 0.0001, "loss": 0.0267, "step": 13990 }, { "epoch": 106.87022900763358, "grad_norm": 2.149178981781006, "learning_rate": 0.0001, "loss": 0.0272, "step": 14000 }, { "epoch": 106.94656488549619, "grad_norm": 2.199023485183716, "learning_rate": 0.0001, "loss": 0.0258, "step": 14010 }, { "epoch": 107.02290076335878, "grad_norm": 1.6317484378814697, "learning_rate": 0.0001, "loss": 0.0251, "step": 14020 }, { "epoch": 107.09923664122137, "grad_norm": 2.061647891998291, "learning_rate": 0.0001, "loss": 0.0263, "step": 14030 }, { "epoch": 107.17557251908397, "grad_norm": 1.744196891784668, "learning_rate": 0.0001, "loss": 0.0257, "step": 14040 }, { "epoch": 107.25190839694656, "grad_norm": 1.7950681447982788, "learning_rate": 0.0001, "loss": 0.026, "step": 14050 }, { "epoch": 107.32824427480917, "grad_norm": 1.8815538883209229, "learning_rate": 0.0001, "loss": 0.0252, "step": 14060 }, { "epoch": 107.40458015267176, "grad_norm": 1.4833840131759644, "learning_rate": 0.0001, "loss": 0.0255, "step": 14070 }, { "epoch": 107.48091603053435, "grad_norm": 2.272127628326416, "learning_rate": 0.0001, "loss": 0.0259, "step": 14080 }, { "epoch": 107.55725190839695, "grad_norm": 1.6246694326400757, "learning_rate": 0.0001, "loss": 0.0254, "step": 14090 }, { "epoch": 107.63358778625954, "grad_norm": 1.773882269859314, "learning_rate": 0.0001, "loss": 0.0262, "step": 14100 }, { "epoch": 107.70992366412214, "grad_norm": 2.1254212856292725, "learning_rate": 0.0001, "loss": 0.0286, "step": 14110 }, { "epoch": 107.78625954198473, "grad_norm": 2.6394572257995605, "learning_rate": 0.0001, "loss": 0.0275, "step": 14120 }, { "epoch": 107.86259541984732, "grad_norm": 1.7560313940048218, "learning_rate": 0.0001, "loss": 0.0273, "step": 14130 }, { "epoch": 107.93893129770993, "grad_norm": 2.160723924636841, "learning_rate": 0.0001, "loss": 0.0269, "step": 14140 }, { "epoch": 108.01526717557252, "grad_norm": 1.7521841526031494, "learning_rate": 0.0001, "loss": 0.0264, "step": 14150 }, { "epoch": 108.09160305343511, "grad_norm": 2.285616159439087, "learning_rate": 0.0001, "loss": 0.0269, "step": 14160 }, { "epoch": 108.16793893129771, "grad_norm": 2.025925636291504, "learning_rate": 0.0001, "loss": 0.0272, "step": 14170 }, { "epoch": 108.2442748091603, "grad_norm": 2.6377503871917725, "learning_rate": 0.0001, "loss": 0.0264, "step": 14180 }, { "epoch": 108.3206106870229, "grad_norm": 2.6374945640563965, "learning_rate": 0.0001, "loss": 0.0288, "step": 14190 }, { "epoch": 108.3969465648855, "grad_norm": 2.27520751953125, "learning_rate": 0.0001, "loss": 0.028, "step": 14200 }, { "epoch": 108.47328244274809, "grad_norm": 2.311246871948242, "learning_rate": 0.0001, "loss": 0.0274, "step": 14210 }, { "epoch": 108.54961832061069, "grad_norm": 2.076172351837158, "learning_rate": 0.0001, "loss": 0.0283, "step": 14220 }, { "epoch": 108.62595419847328, "grad_norm": 2.5163216590881348, "learning_rate": 0.0001, "loss": 0.0268, "step": 14230 }, { "epoch": 108.70229007633588, "grad_norm": 1.7224295139312744, "learning_rate": 0.0001, "loss": 0.0279, "step": 14240 }, { "epoch": 108.77862595419847, "grad_norm": 1.9800406694412231, "learning_rate": 0.0001, "loss": 0.0275, "step": 14250 }, { "epoch": 108.85496183206106, "grad_norm": 2.4573442935943604, "learning_rate": 0.0001, "loss": 0.0274, "step": 14260 }, { "epoch": 108.93129770992367, "grad_norm": 2.324249744415283, "learning_rate": 0.0001, "loss": 0.029, "step": 14270 }, { "epoch": 109.00763358778626, "grad_norm": 2.0618770122528076, "learning_rate": 0.0001, "loss": 0.0268, "step": 14280 }, { "epoch": 109.08396946564885, "grad_norm": 2.4860265254974365, "learning_rate": 0.0001, "loss": 0.0272, "step": 14290 }, { "epoch": 109.16030534351145, "grad_norm": 1.7677160501480103, "learning_rate": 0.0001, "loss": 0.0279, "step": 14300 }, { "epoch": 109.23664122137404, "grad_norm": 3.158130407333374, "learning_rate": 0.0001, "loss": 0.0268, "step": 14310 }, { "epoch": 109.31297709923665, "grad_norm": 2.150164842605591, "learning_rate": 0.0001, "loss": 0.0278, "step": 14320 }, { "epoch": 109.38931297709924, "grad_norm": 2.3639869689941406, "learning_rate": 0.0001, "loss": 0.0268, "step": 14330 }, { "epoch": 109.46564885496183, "grad_norm": 2.216660499572754, "learning_rate": 0.0001, "loss": 0.0265, "step": 14340 }, { "epoch": 109.54198473282443, "grad_norm": 1.7041683197021484, "learning_rate": 0.0001, "loss": 0.0257, "step": 14350 }, { "epoch": 109.61832061068702, "grad_norm": 1.875625491142273, "learning_rate": 0.0001, "loss": 0.0258, "step": 14360 }, { "epoch": 109.69465648854961, "grad_norm": 3.2453742027282715, "learning_rate": 0.0001, "loss": 0.0263, "step": 14370 }, { "epoch": 109.77099236641222, "grad_norm": 1.4797676801681519, "learning_rate": 0.0001, "loss": 0.0269, "step": 14380 }, { "epoch": 109.8473282442748, "grad_norm": 2.0921294689178467, "learning_rate": 0.0001, "loss": 0.0262, "step": 14390 }, { "epoch": 109.92366412213741, "grad_norm": 1.428452491760254, "learning_rate": 0.0001, "loss": 0.0274, "step": 14400 }, { "epoch": 110.0, "grad_norm": 1.6458938121795654, "learning_rate": 0.0001, "loss": 0.0271, "step": 14410 }, { "epoch": 110.07633587786259, "grad_norm": 1.9492511749267578, "learning_rate": 0.0001, "loss": 0.0272, "step": 14420 }, { "epoch": 110.1526717557252, "grad_norm": 1.6527072191238403, "learning_rate": 0.0001, "loss": 0.0276, "step": 14430 }, { "epoch": 110.22900763358778, "grad_norm": 2.148014783859253, "learning_rate": 0.0001, "loss": 0.0272, "step": 14440 }, { "epoch": 110.30534351145039, "grad_norm": 1.9488916397094727, "learning_rate": 0.0001, "loss": 0.027, "step": 14450 }, { "epoch": 110.38167938931298, "grad_norm": 2.081972599029541, "learning_rate": 0.0001, "loss": 0.0259, "step": 14460 }, { "epoch": 110.45801526717557, "grad_norm": 1.8138233423233032, "learning_rate": 0.0001, "loss": 0.0264, "step": 14470 }, { "epoch": 110.53435114503817, "grad_norm": 1.699871301651001, "learning_rate": 0.0001, "loss": 0.0272, "step": 14480 }, { "epoch": 110.61068702290076, "grad_norm": 1.660079836845398, "learning_rate": 0.0001, "loss": 0.0268, "step": 14490 }, { "epoch": 110.68702290076335, "grad_norm": 1.831562876701355, "learning_rate": 0.0001, "loss": 0.0265, "step": 14500 }, { "epoch": 110.76335877862596, "grad_norm": 2.212834596633911, "learning_rate": 0.0001, "loss": 0.026, "step": 14510 }, { "epoch": 110.83969465648855, "grad_norm": 2.119967460632324, "learning_rate": 0.0001, "loss": 0.0262, "step": 14520 }, { "epoch": 110.91603053435115, "grad_norm": 1.9459441900253296, "learning_rate": 0.0001, "loss": 0.0262, "step": 14530 }, { "epoch": 110.99236641221374, "grad_norm": 1.869023084640503, "learning_rate": 0.0001, "loss": 0.0266, "step": 14540 }, { "epoch": 111.06870229007633, "grad_norm": 2.0053560733795166, "learning_rate": 0.0001, "loss": 0.0263, "step": 14550 }, { "epoch": 111.14503816793894, "grad_norm": 1.5239641666412354, "learning_rate": 0.0001, "loss": 0.0257, "step": 14560 }, { "epoch": 111.22137404580153, "grad_norm": 2.209461212158203, "learning_rate": 0.0001, "loss": 0.0263, "step": 14570 }, { "epoch": 111.29770992366412, "grad_norm": 2.075442314147949, "learning_rate": 0.0001, "loss": 0.0267, "step": 14580 }, { "epoch": 111.37404580152672, "grad_norm": 2.1874117851257324, "learning_rate": 0.0001, "loss": 0.0259, "step": 14590 }, { "epoch": 111.45038167938931, "grad_norm": 1.9970927238464355, "learning_rate": 0.0001, "loss": 0.0258, "step": 14600 }, { "epoch": 111.52671755725191, "grad_norm": 2.628873109817505, "learning_rate": 0.0001, "loss": 0.0266, "step": 14610 }, { "epoch": 111.6030534351145, "grad_norm": 1.3176394701004028, "learning_rate": 0.0001, "loss": 0.0256, "step": 14620 }, { "epoch": 111.6793893129771, "grad_norm": 2.2550082206726074, "learning_rate": 0.0001, "loss": 0.0264, "step": 14630 }, { "epoch": 111.7557251908397, "grad_norm": 1.7920382022857666, "learning_rate": 0.0001, "loss": 0.026, "step": 14640 }, { "epoch": 111.83206106870229, "grad_norm": 2.475863218307495, "learning_rate": 0.0001, "loss": 0.0261, "step": 14650 }, { "epoch": 111.90839694656489, "grad_norm": 1.4396247863769531, "learning_rate": 0.0001, "loss": 0.0262, "step": 14660 }, { "epoch": 111.98473282442748, "grad_norm": 2.1318399906158447, "learning_rate": 0.0001, "loss": 0.0273, "step": 14670 }, { "epoch": 112.06106870229007, "grad_norm": 2.320446252822876, "learning_rate": 0.0001, "loss": 0.0255, "step": 14680 }, { "epoch": 112.13740458015268, "grad_norm": 2.496549367904663, "learning_rate": 0.0001, "loss": 0.0265, "step": 14690 }, { "epoch": 112.21374045801527, "grad_norm": 1.9661532640457153, "learning_rate": 0.0001, "loss": 0.0271, "step": 14700 }, { "epoch": 112.29007633587786, "grad_norm": 1.9215774536132812, "learning_rate": 0.0001, "loss": 0.027, "step": 14710 }, { "epoch": 112.36641221374046, "grad_norm": 1.4611661434173584, "learning_rate": 0.0001, "loss": 0.0272, "step": 14720 }, { "epoch": 112.44274809160305, "grad_norm": 2.085303783416748, "learning_rate": 0.0001, "loss": 0.0276, "step": 14730 }, { "epoch": 112.51908396946565, "grad_norm": 1.0519261360168457, "learning_rate": 0.0001, "loss": 0.0265, "step": 14740 }, { "epoch": 112.59541984732824, "grad_norm": 1.988202452659607, "learning_rate": 0.0001, "loss": 0.0276, "step": 14750 }, { "epoch": 112.67175572519083, "grad_norm": 2.1803975105285645, "learning_rate": 0.0001, "loss": 0.0277, "step": 14760 }, { "epoch": 112.74809160305344, "grad_norm": 1.841810941696167, "learning_rate": 0.0001, "loss": 0.0273, "step": 14770 }, { "epoch": 112.82442748091603, "grad_norm": 2.1181576251983643, "learning_rate": 0.0001, "loss": 0.0283, "step": 14780 }, { "epoch": 112.90076335877862, "grad_norm": 1.7345706224441528, "learning_rate": 0.0001, "loss": 0.0281, "step": 14790 }, { "epoch": 112.97709923664122, "grad_norm": 1.7214415073394775, "learning_rate": 0.0001, "loss": 0.0277, "step": 14800 }, { "epoch": 113.05343511450381, "grad_norm": 1.9596589803695679, "learning_rate": 0.0001, "loss": 0.0272, "step": 14810 }, { "epoch": 113.12977099236642, "grad_norm": 1.7209182977676392, "learning_rate": 0.0001, "loss": 0.0287, "step": 14820 }, { "epoch": 113.20610687022901, "grad_norm": 1.487409234046936, "learning_rate": 0.0001, "loss": 0.0268, "step": 14830 }, { "epoch": 113.2824427480916, "grad_norm": 1.814766526222229, "learning_rate": 0.0001, "loss": 0.0274, "step": 14840 }, { "epoch": 113.3587786259542, "grad_norm": 1.5842312574386597, "learning_rate": 0.0001, "loss": 0.0259, "step": 14850 }, { "epoch": 113.43511450381679, "grad_norm": 1.2899078130722046, "learning_rate": 0.0001, "loss": 0.0271, "step": 14860 }, { "epoch": 113.5114503816794, "grad_norm": 1.9995806217193604, "learning_rate": 0.0001, "loss": 0.0273, "step": 14870 }, { "epoch": 113.58778625954199, "grad_norm": 2.1332333087921143, "learning_rate": 0.0001, "loss": 0.0278, "step": 14880 }, { "epoch": 113.66412213740458, "grad_norm": 1.4960154294967651, "learning_rate": 0.0001, "loss": 0.0261, "step": 14890 }, { "epoch": 113.74045801526718, "grad_norm": 2.2078776359558105, "learning_rate": 0.0001, "loss": 0.0263, "step": 14900 }, { "epoch": 113.81679389312977, "grad_norm": 1.7981834411621094, "learning_rate": 0.0001, "loss": 0.0267, "step": 14910 }, { "epoch": 113.89312977099236, "grad_norm": 1.8380239009857178, "learning_rate": 0.0001, "loss": 0.0262, "step": 14920 }, { "epoch": 113.96946564885496, "grad_norm": 2.5005035400390625, "learning_rate": 0.0001, "loss": 0.0256, "step": 14930 }, { "epoch": 114.04580152671755, "grad_norm": 2.5531742572784424, "learning_rate": 0.0001, "loss": 0.0257, "step": 14940 }, { "epoch": 114.12213740458016, "grad_norm": 2.342524528503418, "learning_rate": 0.0001, "loss": 0.0253, "step": 14950 }, { "epoch": 114.19847328244275, "grad_norm": 1.6001231670379639, "learning_rate": 0.0001, "loss": 0.0252, "step": 14960 }, { "epoch": 114.27480916030534, "grad_norm": 1.5401047468185425, "learning_rate": 0.0001, "loss": 0.0248, "step": 14970 }, { "epoch": 114.35114503816794, "grad_norm": 1.4266992807388306, "learning_rate": 0.0001, "loss": 0.0241, "step": 14980 }, { "epoch": 114.42748091603053, "grad_norm": 2.5369367599487305, "learning_rate": 0.0001, "loss": 0.025, "step": 14990 }, { "epoch": 114.50381679389314, "grad_norm": 1.9132202863693237, "learning_rate": 0.0001, "loss": 0.0253, "step": 15000 }, { "epoch": 114.58015267175573, "grad_norm": 1.8563328981399536, "learning_rate": 0.0001, "loss": 0.0251, "step": 15010 }, { "epoch": 114.65648854961832, "grad_norm": 2.314810037612915, "learning_rate": 0.0001, "loss": 0.0256, "step": 15020 }, { "epoch": 114.73282442748092, "grad_norm": 1.5925146341323853, "learning_rate": 0.0001, "loss": 0.0251, "step": 15030 }, { "epoch": 114.80916030534351, "grad_norm": 1.6348196268081665, "learning_rate": 0.0001, "loss": 0.0248, "step": 15040 }, { "epoch": 114.8854961832061, "grad_norm": 3.0218279361724854, "learning_rate": 0.0001, "loss": 0.0255, "step": 15050 }, { "epoch": 114.9618320610687, "grad_norm": 2.471376657485962, "learning_rate": 0.0001, "loss": 0.0264, "step": 15060 }, { "epoch": 115.0381679389313, "grad_norm": 2.5577564239501953, "learning_rate": 0.0001, "loss": 0.0271, "step": 15070 }, { "epoch": 115.1145038167939, "grad_norm": 2.4267125129699707, "learning_rate": 0.0001, "loss": 0.0267, "step": 15080 }, { "epoch": 115.19083969465649, "grad_norm": 2.6858878135681152, "learning_rate": 0.0001, "loss": 0.0261, "step": 15090 }, { "epoch": 115.26717557251908, "grad_norm": 2.1606760025024414, "learning_rate": 0.0001, "loss": 0.0255, "step": 15100 }, { "epoch": 115.34351145038168, "grad_norm": 2.5755081176757812, "learning_rate": 0.0001, "loss": 0.0247, "step": 15110 }, { "epoch": 115.41984732824427, "grad_norm": 2.9287302494049072, "learning_rate": 0.0001, "loss": 0.0251, "step": 15120 }, { "epoch": 115.49618320610686, "grad_norm": 2.459038734436035, "learning_rate": 0.0001, "loss": 0.0251, "step": 15130 }, { "epoch": 115.57251908396947, "grad_norm": 2.318486213684082, "learning_rate": 0.0001, "loss": 0.0245, "step": 15140 }, { "epoch": 115.64885496183206, "grad_norm": 2.0854971408843994, "learning_rate": 0.0001, "loss": 0.0239, "step": 15150 }, { "epoch": 115.72519083969466, "grad_norm": 1.7474807500839233, "learning_rate": 0.0001, "loss": 0.0245, "step": 15160 }, { "epoch": 115.80152671755725, "grad_norm": 1.97526216506958, "learning_rate": 0.0001, "loss": 0.0235, "step": 15170 }, { "epoch": 115.87786259541984, "grad_norm": 1.4873026609420776, "learning_rate": 0.0001, "loss": 0.0245, "step": 15180 }, { "epoch": 115.95419847328245, "grad_norm": 2.5225839614868164, "learning_rate": 0.0001, "loss": 0.0247, "step": 15190 }, { "epoch": 116.03053435114504, "grad_norm": 2.6586527824401855, "learning_rate": 0.0001, "loss": 0.0239, "step": 15200 }, { "epoch": 116.10687022900764, "grad_norm": 2.249241590499878, "learning_rate": 0.0001, "loss": 0.0237, "step": 15210 }, { "epoch": 116.18320610687023, "grad_norm": 2.8993890285491943, "learning_rate": 0.0001, "loss": 0.0239, "step": 15220 }, { "epoch": 116.25954198473282, "grad_norm": 2.9991607666015625, "learning_rate": 0.0001, "loss": 0.0245, "step": 15230 }, { "epoch": 116.33587786259542, "grad_norm": 2.3404102325439453, "learning_rate": 0.0001, "loss": 0.0238, "step": 15240 }, { "epoch": 116.41221374045801, "grad_norm": 1.1817855834960938, "learning_rate": 0.0001, "loss": 0.0234, "step": 15250 }, { "epoch": 116.4885496183206, "grad_norm": 2.328691005706787, "learning_rate": 0.0001, "loss": 0.0234, "step": 15260 }, { "epoch": 116.56488549618321, "grad_norm": 1.7467302083969116, "learning_rate": 0.0001, "loss": 0.0241, "step": 15270 }, { "epoch": 116.6412213740458, "grad_norm": 1.8969968557357788, "learning_rate": 0.0001, "loss": 0.0246, "step": 15280 }, { "epoch": 116.7175572519084, "grad_norm": 1.2914165258407593, "learning_rate": 0.0001, "loss": 0.0246, "step": 15290 }, { "epoch": 116.79389312977099, "grad_norm": 2.4162673950195312, "learning_rate": 0.0001, "loss": 0.0255, "step": 15300 }, { "epoch": 116.87022900763358, "grad_norm": 2.5609350204467773, "learning_rate": 0.0001, "loss": 0.0249, "step": 15310 }, { "epoch": 116.94656488549619, "grad_norm": 2.5717668533325195, "learning_rate": 0.0001, "loss": 0.025, "step": 15320 }, { "epoch": 117.02290076335878, "grad_norm": 2.017791509628296, "learning_rate": 0.0001, "loss": 0.0242, "step": 15330 }, { "epoch": 117.09923664122137, "grad_norm": 1.892228126525879, "learning_rate": 0.0001, "loss": 0.025, "step": 15340 }, { "epoch": 117.17557251908397, "grad_norm": 1.5655754804611206, "learning_rate": 0.0001, "loss": 0.0244, "step": 15350 }, { "epoch": 117.25190839694656, "grad_norm": 1.4662784337997437, "learning_rate": 0.0001, "loss": 0.0248, "step": 15360 }, { "epoch": 117.32824427480917, "grad_norm": 1.4378795623779297, "learning_rate": 0.0001, "loss": 0.024, "step": 15370 }, { "epoch": 117.40458015267176, "grad_norm": 1.8588016033172607, "learning_rate": 0.0001, "loss": 0.0246, "step": 15380 }, { "epoch": 117.48091603053435, "grad_norm": 2.0430283546447754, "learning_rate": 0.0001, "loss": 0.0252, "step": 15390 }, { "epoch": 117.55725190839695, "grad_norm": 2.9123165607452393, "learning_rate": 0.0001, "loss": 0.0248, "step": 15400 }, { "epoch": 117.63358778625954, "grad_norm": 1.9609043598175049, "learning_rate": 0.0001, "loss": 0.024, "step": 15410 }, { "epoch": 117.70992366412214, "grad_norm": 1.840674877166748, "learning_rate": 0.0001, "loss": 0.0242, "step": 15420 }, { "epoch": 117.78625954198473, "grad_norm": 1.7681403160095215, "learning_rate": 0.0001, "loss": 0.0242, "step": 15430 }, { "epoch": 117.86259541984732, "grad_norm": 1.702487826347351, "learning_rate": 0.0001, "loss": 0.0248, "step": 15440 }, { "epoch": 117.93893129770993, "grad_norm": 1.639277458190918, "learning_rate": 0.0001, "loss": 0.0242, "step": 15450 }, { "epoch": 118.01526717557252, "grad_norm": 2.175025224685669, "learning_rate": 0.0001, "loss": 0.024, "step": 15460 }, { "epoch": 118.09160305343511, "grad_norm": 2.0539023876190186, "learning_rate": 0.0001, "loss": 0.0241, "step": 15470 }, { "epoch": 118.16793893129771, "grad_norm": 2.061069965362549, "learning_rate": 0.0001, "loss": 0.0245, "step": 15480 }, { "epoch": 118.2442748091603, "grad_norm": 2.0635929107666016, "learning_rate": 0.0001, "loss": 0.0239, "step": 15490 }, { "epoch": 118.3206106870229, "grad_norm": 2.075137138366699, "learning_rate": 0.0001, "loss": 0.0249, "step": 15500 }, { "epoch": 118.3969465648855, "grad_norm": 2.2840263843536377, "learning_rate": 0.0001, "loss": 0.0256, "step": 15510 }, { "epoch": 118.47328244274809, "grad_norm": 2.3434224128723145, "learning_rate": 0.0001, "loss": 0.0237, "step": 15520 }, { "epoch": 118.54961832061069, "grad_norm": 2.0291318893432617, "learning_rate": 0.0001, "loss": 0.0244, "step": 15530 }, { "epoch": 118.62595419847328, "grad_norm": 2.1068997383117676, "learning_rate": 0.0001, "loss": 0.0256, "step": 15540 }, { "epoch": 118.70229007633588, "grad_norm": 1.7506309747695923, "learning_rate": 0.0001, "loss": 0.0257, "step": 15550 }, { "epoch": 118.77862595419847, "grad_norm": 1.6364175081253052, "learning_rate": 0.0001, "loss": 0.0255, "step": 15560 }, { "epoch": 118.85496183206106, "grad_norm": 1.4477061033248901, "learning_rate": 0.0001, "loss": 0.0255, "step": 15570 }, { "epoch": 118.93129770992367, "grad_norm": 2.6273930072784424, "learning_rate": 0.0001, "loss": 0.0257, "step": 15580 }, { "epoch": 119.00763358778626, "grad_norm": 1.8002591133117676, "learning_rate": 0.0001, "loss": 0.0261, "step": 15590 }, { "epoch": 119.08396946564885, "grad_norm": 2.0550897121429443, "learning_rate": 0.0001, "loss": 0.0264, "step": 15600 }, { "epoch": 119.16030534351145, "grad_norm": 2.2471096515655518, "learning_rate": 0.0001, "loss": 0.026, "step": 15610 }, { "epoch": 119.23664122137404, "grad_norm": 1.2224024534225464, "learning_rate": 0.0001, "loss": 0.0257, "step": 15620 }, { "epoch": 119.31297709923665, "grad_norm": 1.9191336631774902, "learning_rate": 0.0001, "loss": 0.0243, "step": 15630 }, { "epoch": 119.38931297709924, "grad_norm": 1.6931322813034058, "learning_rate": 0.0001, "loss": 0.0239, "step": 15640 }, { "epoch": 119.46564885496183, "grad_norm": 2.370849370956421, "learning_rate": 0.0001, "loss": 0.0239, "step": 15650 }, { "epoch": 119.54198473282443, "grad_norm": 2.9323511123657227, "learning_rate": 0.0001, "loss": 0.0238, "step": 15660 }, { "epoch": 119.61832061068702, "grad_norm": 1.4664177894592285, "learning_rate": 0.0001, "loss": 0.0237, "step": 15670 }, { "epoch": 119.69465648854961, "grad_norm": 1.7453854084014893, "learning_rate": 0.0001, "loss": 0.0242, "step": 15680 }, { "epoch": 119.77099236641222, "grad_norm": 1.619956374168396, "learning_rate": 0.0001, "loss": 0.0238, "step": 15690 }, { "epoch": 119.8473282442748, "grad_norm": 1.8349858522415161, "learning_rate": 0.0001, "loss": 0.025, "step": 15700 }, { "epoch": 119.92366412213741, "grad_norm": 2.0322132110595703, "learning_rate": 0.0001, "loss": 0.0255, "step": 15710 }, { "epoch": 120.0, "grad_norm": 1.6165908575057983, "learning_rate": 0.0001, "loss": 0.0245, "step": 15720 }, { "epoch": 120.07633587786259, "grad_norm": 1.8520020246505737, "learning_rate": 0.0001, "loss": 0.0244, "step": 15730 }, { "epoch": 120.1526717557252, "grad_norm": 1.761335015296936, "learning_rate": 0.0001, "loss": 0.025, "step": 15740 }, { "epoch": 120.22900763358778, "grad_norm": 1.8247337341308594, "learning_rate": 0.0001, "loss": 0.0245, "step": 15750 }, { "epoch": 120.30534351145039, "grad_norm": 2.2505698204040527, "learning_rate": 0.0001, "loss": 0.0244, "step": 15760 }, { "epoch": 120.38167938931298, "grad_norm": 1.6356956958770752, "learning_rate": 0.0001, "loss": 0.0244, "step": 15770 }, { "epoch": 120.45801526717557, "grad_norm": 1.8992877006530762, "learning_rate": 0.0001, "loss": 0.0238, "step": 15780 }, { "epoch": 120.53435114503817, "grad_norm": 2.519810438156128, "learning_rate": 0.0001, "loss": 0.0242, "step": 15790 }, { "epoch": 120.61068702290076, "grad_norm": 1.8599034547805786, "learning_rate": 0.0001, "loss": 0.0245, "step": 15800 }, { "epoch": 120.68702290076335, "grad_norm": 2.08980655670166, "learning_rate": 0.0001, "loss": 0.0254, "step": 15810 }, { "epoch": 120.76335877862596, "grad_norm": 1.8496781587600708, "learning_rate": 0.0001, "loss": 0.0246, "step": 15820 }, { "epoch": 120.83969465648855, "grad_norm": 2.3666319847106934, "learning_rate": 0.0001, "loss": 0.0249, "step": 15830 }, { "epoch": 120.91603053435115, "grad_norm": 1.9057577848434448, "learning_rate": 0.0001, "loss": 0.0248, "step": 15840 }, { "epoch": 120.99236641221374, "grad_norm": 2.53497576713562, "learning_rate": 0.0001, "loss": 0.0245, "step": 15850 }, { "epoch": 121.06870229007633, "grad_norm": 2.1285314559936523, "learning_rate": 0.0001, "loss": 0.0242, "step": 15860 }, { "epoch": 121.14503816793894, "grad_norm": 1.7215737104415894, "learning_rate": 0.0001, "loss": 0.0247, "step": 15870 }, { "epoch": 121.22137404580153, "grad_norm": 2.6900715827941895, "learning_rate": 0.0001, "loss": 0.0235, "step": 15880 }, { "epoch": 121.29770992366412, "grad_norm": 2.2493953704833984, "learning_rate": 0.0001, "loss": 0.0239, "step": 15890 }, { "epoch": 121.37404580152672, "grad_norm": 2.6272735595703125, "learning_rate": 0.0001, "loss": 0.0234, "step": 15900 }, { "epoch": 121.45038167938931, "grad_norm": 2.4724934101104736, "learning_rate": 0.0001, "loss": 0.0241, "step": 15910 }, { "epoch": 121.52671755725191, "grad_norm": 1.789008378982544, "learning_rate": 0.0001, "loss": 0.0242, "step": 15920 }, { "epoch": 121.6030534351145, "grad_norm": 1.454261302947998, "learning_rate": 0.0001, "loss": 0.023, "step": 15930 }, { "epoch": 121.6793893129771, "grad_norm": 1.9153112173080444, "learning_rate": 0.0001, "loss": 0.0229, "step": 15940 }, { "epoch": 121.7557251908397, "grad_norm": 1.9114553928375244, "learning_rate": 0.0001, "loss": 0.0242, "step": 15950 }, { "epoch": 121.83206106870229, "grad_norm": 1.8078442811965942, "learning_rate": 0.0001, "loss": 0.0246, "step": 15960 }, { "epoch": 121.90839694656489, "grad_norm": 1.569886565208435, "learning_rate": 0.0001, "loss": 0.0244, "step": 15970 }, { "epoch": 121.98473282442748, "grad_norm": 1.994455099105835, "learning_rate": 0.0001, "loss": 0.0255, "step": 15980 }, { "epoch": 122.06106870229007, "grad_norm": 1.606343150138855, "learning_rate": 0.0001, "loss": 0.0263, "step": 15990 }, { "epoch": 122.13740458015268, "grad_norm": 1.417449951171875, "learning_rate": 0.0001, "loss": 0.0257, "step": 16000 }, { "epoch": 122.21374045801527, "grad_norm": 2.146721363067627, "learning_rate": 0.0001, "loss": 0.025, "step": 16010 }, { "epoch": 122.29007633587786, "grad_norm": 1.578862190246582, "learning_rate": 0.0001, "loss": 0.0252, "step": 16020 }, { "epoch": 122.36641221374046, "grad_norm": 1.2085710763931274, "learning_rate": 0.0001, "loss": 0.0245, "step": 16030 }, { "epoch": 122.44274809160305, "grad_norm": 1.5111885070800781, "learning_rate": 0.0001, "loss": 0.024, "step": 16040 }, { "epoch": 122.51908396946565, "grad_norm": 1.8656097650527954, "learning_rate": 0.0001, "loss": 0.0249, "step": 16050 }, { "epoch": 122.59541984732824, "grad_norm": 1.5108895301818848, "learning_rate": 0.0001, "loss": 0.024, "step": 16060 }, { "epoch": 122.67175572519083, "grad_norm": 1.4296106100082397, "learning_rate": 0.0001, "loss": 0.0245, "step": 16070 }, { "epoch": 122.74809160305344, "grad_norm": 1.7553739547729492, "learning_rate": 0.0001, "loss": 0.0243, "step": 16080 }, { "epoch": 122.82442748091603, "grad_norm": 2.311823606491089, "learning_rate": 0.0001, "loss": 0.0249, "step": 16090 }, { "epoch": 122.90076335877862, "grad_norm": 2.3516499996185303, "learning_rate": 0.0001, "loss": 0.025, "step": 16100 }, { "epoch": 122.97709923664122, "grad_norm": 2.228538751602173, "learning_rate": 0.0001, "loss": 0.0246, "step": 16110 }, { "epoch": 123.05343511450381, "grad_norm": 1.9106272459030151, "learning_rate": 0.0001, "loss": 0.0251, "step": 16120 }, { "epoch": 123.12977099236642, "grad_norm": 1.0804166793823242, "learning_rate": 0.0001, "loss": 0.0249, "step": 16130 }, { "epoch": 123.20610687022901, "grad_norm": 1.8840841054916382, "learning_rate": 0.0001, "loss": 0.0252, "step": 16140 }, { "epoch": 123.2824427480916, "grad_norm": 1.9093974828720093, "learning_rate": 0.0001, "loss": 0.025, "step": 16150 }, { "epoch": 123.3587786259542, "grad_norm": 1.716206431388855, "learning_rate": 0.0001, "loss": 0.0258, "step": 16160 }, { "epoch": 123.43511450381679, "grad_norm": 2.1801977157592773, "learning_rate": 0.0001, "loss": 0.0263, "step": 16170 }, { "epoch": 123.5114503816794, "grad_norm": 2.2246596813201904, "learning_rate": 0.0001, "loss": 0.0264, "step": 16180 }, { "epoch": 123.58778625954199, "grad_norm": 2.61885929107666, "learning_rate": 0.0001, "loss": 0.0274, "step": 16190 }, { "epoch": 123.66412213740458, "grad_norm": 1.5009523630142212, "learning_rate": 0.0001, "loss": 0.026, "step": 16200 }, { "epoch": 123.74045801526718, "grad_norm": 1.685074806213379, "learning_rate": 0.0001, "loss": 0.0248, "step": 16210 }, { "epoch": 123.81679389312977, "grad_norm": 1.4073107242584229, "learning_rate": 0.0001, "loss": 0.0248, "step": 16220 }, { "epoch": 123.89312977099236, "grad_norm": 1.7385926246643066, "learning_rate": 0.0001, "loss": 0.0248, "step": 16230 }, { "epoch": 123.96946564885496, "grad_norm": 2.715198278427124, "learning_rate": 0.0001, "loss": 0.0249, "step": 16240 }, { "epoch": 124.04580152671755, "grad_norm": 1.4922488927841187, "learning_rate": 0.0001, "loss": 0.0239, "step": 16250 }, { "epoch": 124.12213740458016, "grad_norm": 2.1269845962524414, "learning_rate": 0.0001, "loss": 0.0237, "step": 16260 }, { "epoch": 124.19847328244275, "grad_norm": 2.33278226852417, "learning_rate": 0.0001, "loss": 0.024, "step": 16270 }, { "epoch": 124.27480916030534, "grad_norm": 1.8886468410491943, "learning_rate": 0.0001, "loss": 0.024, "step": 16280 }, { "epoch": 124.35114503816794, "grad_norm": 1.5079668760299683, "learning_rate": 0.0001, "loss": 0.0244, "step": 16290 }, { "epoch": 124.42748091603053, "grad_norm": 1.828126072883606, "learning_rate": 0.0001, "loss": 0.0244, "step": 16300 }, { "epoch": 124.50381679389314, "grad_norm": 2.2024781703948975, "learning_rate": 0.0001, "loss": 0.0252, "step": 16310 }, { "epoch": 124.58015267175573, "grad_norm": 1.8092633485794067, "learning_rate": 0.0001, "loss": 0.0251, "step": 16320 }, { "epoch": 124.65648854961832, "grad_norm": 1.783573031425476, "learning_rate": 0.0001, "loss": 0.0244, "step": 16330 }, { "epoch": 124.73282442748092, "grad_norm": 1.5551701784133911, "learning_rate": 0.0001, "loss": 0.0254, "step": 16340 }, { "epoch": 124.80916030534351, "grad_norm": 1.403311014175415, "learning_rate": 0.0001, "loss": 0.0246, "step": 16350 }, { "epoch": 124.8854961832061, "grad_norm": 2.099445104598999, "learning_rate": 0.0001, "loss": 0.0248, "step": 16360 }, { "epoch": 124.9618320610687, "grad_norm": 2.640312433242798, "learning_rate": 0.0001, "loss": 0.0253, "step": 16370 }, { "epoch": 125.0381679389313, "grad_norm": 2.173469066619873, "learning_rate": 0.0001, "loss": 0.0257, "step": 16380 }, { "epoch": 125.1145038167939, "grad_norm": 2.045154333114624, "learning_rate": 0.0001, "loss": 0.0258, "step": 16390 }, { "epoch": 125.19083969465649, "grad_norm": 1.5636930465698242, "learning_rate": 0.0001, "loss": 0.025, "step": 16400 }, { "epoch": 125.26717557251908, "grad_norm": 1.406623363494873, "learning_rate": 0.0001, "loss": 0.0247, "step": 16410 }, { "epoch": 125.34351145038168, "grad_norm": 1.4970333576202393, "learning_rate": 0.0001, "loss": 0.0246, "step": 16420 }, { "epoch": 125.41984732824427, "grad_norm": 1.933078646659851, "learning_rate": 0.0001, "loss": 0.0251, "step": 16430 }, { "epoch": 125.49618320610686, "grad_norm": 2.6897997856140137, "learning_rate": 0.0001, "loss": 0.0257, "step": 16440 }, { "epoch": 125.57251908396947, "grad_norm": 1.6014680862426758, "learning_rate": 0.0001, "loss": 0.025, "step": 16450 }, { "epoch": 125.64885496183206, "grad_norm": 1.978671669960022, "learning_rate": 0.0001, "loss": 0.0249, "step": 16460 }, { "epoch": 125.72519083969466, "grad_norm": 3.407970666885376, "learning_rate": 0.0001, "loss": 0.0267, "step": 16470 }, { "epoch": 125.80152671755725, "grad_norm": 1.9870110750198364, "learning_rate": 0.0001, "loss": 0.0263, "step": 16480 }, { "epoch": 125.87786259541984, "grad_norm": 2.014089822769165, "learning_rate": 0.0001, "loss": 0.0257, "step": 16490 }, { "epoch": 125.95419847328245, "grad_norm": 2.6592674255371094, "learning_rate": 0.0001, "loss": 0.0239, "step": 16500 }, { "epoch": 126.03053435114504, "grad_norm": 1.9115325212478638, "learning_rate": 0.0001, "loss": 0.0234, "step": 16510 }, { "epoch": 126.10687022900764, "grad_norm": 2.484633207321167, "learning_rate": 0.0001, "loss": 0.0228, "step": 16520 }, { "epoch": 126.18320610687023, "grad_norm": 2.5079545974731445, "learning_rate": 0.0001, "loss": 0.0233, "step": 16530 }, { "epoch": 126.25954198473282, "grad_norm": 2.4172589778900146, "learning_rate": 0.0001, "loss": 0.0228, "step": 16540 }, { "epoch": 126.33587786259542, "grad_norm": 2.044318914413452, "learning_rate": 0.0001, "loss": 0.0231, "step": 16550 }, { "epoch": 126.41221374045801, "grad_norm": 1.971238136291504, "learning_rate": 0.0001, "loss": 0.0236, "step": 16560 }, { "epoch": 126.4885496183206, "grad_norm": 1.5323655605316162, "learning_rate": 0.0001, "loss": 0.0241, "step": 16570 }, { "epoch": 126.56488549618321, "grad_norm": 2.4429073333740234, "learning_rate": 0.0001, "loss": 0.0238, "step": 16580 }, { "epoch": 126.6412213740458, "grad_norm": 2.4460911750793457, "learning_rate": 0.0001, "loss": 0.0232, "step": 16590 }, { "epoch": 126.7175572519084, "grad_norm": 2.7748730182647705, "learning_rate": 0.0001, "loss": 0.024, "step": 16600 }, { "epoch": 126.79389312977099, "grad_norm": 2.668083906173706, "learning_rate": 0.0001, "loss": 0.0231, "step": 16610 }, { "epoch": 126.87022900763358, "grad_norm": 1.8329863548278809, "learning_rate": 0.0001, "loss": 0.024, "step": 16620 }, { "epoch": 126.94656488549619, "grad_norm": 2.0759153366088867, "learning_rate": 0.0001, "loss": 0.0237, "step": 16630 }, { "epoch": 127.02290076335878, "grad_norm": 1.6104083061218262, "learning_rate": 0.0001, "loss": 0.0236, "step": 16640 }, { "epoch": 127.09923664122137, "grad_norm": 1.470834732055664, "learning_rate": 0.0001, "loss": 0.0219, "step": 16650 }, { "epoch": 127.17557251908397, "grad_norm": 2.223426342010498, "learning_rate": 0.0001, "loss": 0.0224, "step": 16660 }, { "epoch": 127.25190839694656, "grad_norm": 1.6001783609390259, "learning_rate": 0.0001, "loss": 0.023, "step": 16670 }, { "epoch": 127.32824427480917, "grad_norm": 2.39385724067688, "learning_rate": 0.0001, "loss": 0.0236, "step": 16680 }, { "epoch": 127.40458015267176, "grad_norm": 2.6426987648010254, "learning_rate": 0.0001, "loss": 0.0233, "step": 16690 }, { "epoch": 127.48091603053435, "grad_norm": 2.4115514755249023, "learning_rate": 0.0001, "loss": 0.0235, "step": 16700 }, { "epoch": 127.55725190839695, "grad_norm": 1.9211236238479614, "learning_rate": 0.0001, "loss": 0.0237, "step": 16710 }, { "epoch": 127.63358778625954, "grad_norm": 1.9920649528503418, "learning_rate": 0.0001, "loss": 0.0232, "step": 16720 }, { "epoch": 127.70992366412214, "grad_norm": 1.3294391632080078, "learning_rate": 0.0001, "loss": 0.0225, "step": 16730 }, { "epoch": 127.78625954198473, "grad_norm": 2.018923282623291, "learning_rate": 0.0001, "loss": 0.0235, "step": 16740 }, { "epoch": 127.86259541984732, "grad_norm": 1.4288134574890137, "learning_rate": 0.0001, "loss": 0.0226, "step": 16750 }, { "epoch": 127.93893129770993, "grad_norm": 1.699035882949829, "learning_rate": 0.0001, "loss": 0.0232, "step": 16760 }, { "epoch": 128.01526717557252, "grad_norm": 2.265871524810791, "learning_rate": 0.0001, "loss": 0.0235, "step": 16770 }, { "epoch": 128.0916030534351, "grad_norm": 2.2055561542510986, "learning_rate": 0.0001, "loss": 0.0243, "step": 16780 }, { "epoch": 128.1679389312977, "grad_norm": 2.4187886714935303, "learning_rate": 0.0001, "loss": 0.0242, "step": 16790 }, { "epoch": 128.24427480916032, "grad_norm": 1.7839254140853882, "learning_rate": 0.0001, "loss": 0.0239, "step": 16800 }, { "epoch": 128.3206106870229, "grad_norm": 2.0504486560821533, "learning_rate": 0.0001, "loss": 0.0236, "step": 16810 }, { "epoch": 128.3969465648855, "grad_norm": 1.9998629093170166, "learning_rate": 0.0001, "loss": 0.024, "step": 16820 }, { "epoch": 128.4732824427481, "grad_norm": 2.167717695236206, "learning_rate": 0.0001, "loss": 0.0232, "step": 16830 }, { "epoch": 128.54961832061068, "grad_norm": 2.2405080795288086, "learning_rate": 0.0001, "loss": 0.0233, "step": 16840 }, { "epoch": 128.6259541984733, "grad_norm": 1.903322696685791, "learning_rate": 0.0001, "loss": 0.0234, "step": 16850 }, { "epoch": 128.70229007633588, "grad_norm": 2.6256179809570312, "learning_rate": 0.0001, "loss": 0.0228, "step": 16860 }, { "epoch": 128.77862595419847, "grad_norm": 1.5426400899887085, "learning_rate": 0.0001, "loss": 0.0226, "step": 16870 }, { "epoch": 128.85496183206106, "grad_norm": 1.7430808544158936, "learning_rate": 0.0001, "loss": 0.0232, "step": 16880 }, { "epoch": 128.93129770992365, "grad_norm": 1.567038655281067, "learning_rate": 0.0001, "loss": 0.0228, "step": 16890 }, { "epoch": 129.00763358778627, "grad_norm": 1.9716322422027588, "learning_rate": 0.0001, "loss": 0.0244, "step": 16900 }, { "epoch": 129.08396946564886, "grad_norm": 2.7702853679656982, "learning_rate": 0.0001, "loss": 0.0244, "step": 16910 }, { "epoch": 129.16030534351145, "grad_norm": 2.676506519317627, "learning_rate": 0.0001, "loss": 0.0233, "step": 16920 }, { "epoch": 129.23664122137404, "grad_norm": 2.5509068965911865, "learning_rate": 0.0001, "loss": 0.0231, "step": 16930 }, { "epoch": 129.31297709923663, "grad_norm": 2.454807996749878, "learning_rate": 0.0001, "loss": 0.0239, "step": 16940 }, { "epoch": 129.38931297709922, "grad_norm": 2.748016119003296, "learning_rate": 0.0001, "loss": 0.0231, "step": 16950 }, { "epoch": 129.46564885496184, "grad_norm": 2.288818359375, "learning_rate": 0.0001, "loss": 0.0237, "step": 16960 }, { "epoch": 129.54198473282443, "grad_norm": 2.6273117065429688, "learning_rate": 0.0001, "loss": 0.0243, "step": 16970 }, { "epoch": 129.61832061068702, "grad_norm": 1.739262580871582, "learning_rate": 0.0001, "loss": 0.0243, "step": 16980 }, { "epoch": 129.6946564885496, "grad_norm": 2.302551746368408, "learning_rate": 0.0001, "loss": 0.0232, "step": 16990 }, { "epoch": 129.7709923664122, "grad_norm": 1.4240367412567139, "learning_rate": 0.0001, "loss": 0.0224, "step": 17000 }, { "epoch": 129.84732824427482, "grad_norm": 2.234084129333496, "learning_rate": 0.0001, "loss": 0.0231, "step": 17010 }, { "epoch": 129.9236641221374, "grad_norm": 2.1059212684631348, "learning_rate": 0.0001, "loss": 0.0245, "step": 17020 }, { "epoch": 130.0, "grad_norm": 2.2144315242767334, "learning_rate": 0.0001, "loss": 0.0251, "step": 17030 }, { "epoch": 130.0763358778626, "grad_norm": 2.4574151039123535, "learning_rate": 0.0001, "loss": 0.0251, "step": 17040 }, { "epoch": 130.15267175572518, "grad_norm": 1.8324997425079346, "learning_rate": 0.0001, "loss": 0.0244, "step": 17050 }, { "epoch": 130.2290076335878, "grad_norm": 2.768584728240967, "learning_rate": 0.0001, "loss": 0.025, "step": 17060 }, { "epoch": 130.3053435114504, "grad_norm": 1.6315059661865234, "learning_rate": 0.0001, "loss": 0.0237, "step": 17070 }, { "epoch": 130.38167938931298, "grad_norm": 1.8941564559936523, "learning_rate": 0.0001, "loss": 0.0241, "step": 17080 }, { "epoch": 130.45801526717557, "grad_norm": 2.0110528469085693, "learning_rate": 0.0001, "loss": 0.0244, "step": 17090 }, { "epoch": 130.53435114503816, "grad_norm": 2.6245663166046143, "learning_rate": 0.0001, "loss": 0.0246, "step": 17100 }, { "epoch": 130.61068702290078, "grad_norm": 2.602757215499878, "learning_rate": 0.0001, "loss": 0.0251, "step": 17110 }, { "epoch": 130.68702290076337, "grad_norm": 2.5330708026885986, "learning_rate": 0.0001, "loss": 0.0238, "step": 17120 }, { "epoch": 130.76335877862596, "grad_norm": 1.6286498308181763, "learning_rate": 0.0001, "loss": 0.0237, "step": 17130 }, { "epoch": 130.83969465648855, "grad_norm": 1.2735488414764404, "learning_rate": 0.0001, "loss": 0.0246, "step": 17140 }, { "epoch": 130.91603053435114, "grad_norm": 2.287062644958496, "learning_rate": 0.0001, "loss": 0.0242, "step": 17150 }, { "epoch": 130.99236641221373, "grad_norm": 3.1358642578125, "learning_rate": 0.0001, "loss": 0.0246, "step": 17160 }, { "epoch": 131.06870229007635, "grad_norm": 3.1548497676849365, "learning_rate": 0.0001, "loss": 0.024, "step": 17170 }, { "epoch": 131.14503816793894, "grad_norm": 2.182913303375244, "learning_rate": 0.0001, "loss": 0.0235, "step": 17180 }, { "epoch": 131.22137404580153, "grad_norm": 1.8393830060958862, "learning_rate": 0.0001, "loss": 0.0236, "step": 17190 }, { "epoch": 131.29770992366412, "grad_norm": 1.4923710823059082, "learning_rate": 0.0001, "loss": 0.0249, "step": 17200 }, { "epoch": 131.3740458015267, "grad_norm": 1.3316115140914917, "learning_rate": 0.0001, "loss": 0.0242, "step": 17210 }, { "epoch": 131.45038167938932, "grad_norm": 2.199976682662964, "learning_rate": 0.0001, "loss": 0.0237, "step": 17220 }, { "epoch": 131.5267175572519, "grad_norm": 1.5019402503967285, "learning_rate": 0.0001, "loss": 0.0234, "step": 17230 }, { "epoch": 131.6030534351145, "grad_norm": 1.7773449420928955, "learning_rate": 0.0001, "loss": 0.0238, "step": 17240 }, { "epoch": 131.6793893129771, "grad_norm": 2.2185046672821045, "learning_rate": 0.0001, "loss": 0.0248, "step": 17250 }, { "epoch": 131.75572519083968, "grad_norm": 1.5093896389007568, "learning_rate": 0.0001, "loss": 0.0241, "step": 17260 }, { "epoch": 131.8320610687023, "grad_norm": 1.8566250801086426, "learning_rate": 0.0001, "loss": 0.0246, "step": 17270 }, { "epoch": 131.9083969465649, "grad_norm": 2.5687170028686523, "learning_rate": 0.0001, "loss": 0.0256, "step": 17280 }, { "epoch": 131.98473282442748, "grad_norm": 2.1301796436309814, "learning_rate": 0.0001, "loss": 0.0253, "step": 17290 }, { "epoch": 132.06106870229007, "grad_norm": 1.9783841371536255, "learning_rate": 0.0001, "loss": 0.0241, "step": 17300 }, { "epoch": 132.13740458015266, "grad_norm": 2.194040536880493, "learning_rate": 0.0001, "loss": 0.0251, "step": 17310 }, { "epoch": 132.21374045801528, "grad_norm": 1.9301542043685913, "learning_rate": 0.0001, "loss": 0.0246, "step": 17320 }, { "epoch": 132.29007633587787, "grad_norm": 1.7430975437164307, "learning_rate": 0.0001, "loss": 0.0242, "step": 17330 }, { "epoch": 132.36641221374046, "grad_norm": 2.263097047805786, "learning_rate": 0.0001, "loss": 0.0246, "step": 17340 }, { "epoch": 132.44274809160305, "grad_norm": 1.9472745656967163, "learning_rate": 0.0001, "loss": 0.0234, "step": 17350 }, { "epoch": 132.51908396946564, "grad_norm": 2.459352970123291, "learning_rate": 0.0001, "loss": 0.0235, "step": 17360 }, { "epoch": 132.59541984732823, "grad_norm": 1.2990909814834595, "learning_rate": 0.0001, "loss": 0.0244, "step": 17370 }, { "epoch": 132.67175572519085, "grad_norm": 2.0318782329559326, "learning_rate": 0.0001, "loss": 0.0243, "step": 17380 }, { "epoch": 132.74809160305344, "grad_norm": 1.5344359874725342, "learning_rate": 0.0001, "loss": 0.0245, "step": 17390 }, { "epoch": 132.82442748091603, "grad_norm": 1.3245092630386353, "learning_rate": 0.0001, "loss": 0.0243, "step": 17400 }, { "epoch": 132.90076335877862, "grad_norm": 1.6624939441680908, "learning_rate": 0.0001, "loss": 0.0244, "step": 17410 }, { "epoch": 132.9770992366412, "grad_norm": 1.247812032699585, "learning_rate": 0.0001, "loss": 0.0239, "step": 17420 }, { "epoch": 133.05343511450383, "grad_norm": 2.4932608604431152, "learning_rate": 0.0001, "loss": 0.0242, "step": 17430 }, { "epoch": 133.12977099236642, "grad_norm": 1.5695245265960693, "learning_rate": 0.0001, "loss": 0.0241, "step": 17440 }, { "epoch": 133.206106870229, "grad_norm": 2.2067346572875977, "learning_rate": 0.0001, "loss": 0.024, "step": 17450 }, { "epoch": 133.2824427480916, "grad_norm": 2.6182620525360107, "learning_rate": 0.0001, "loss": 0.0232, "step": 17460 }, { "epoch": 133.3587786259542, "grad_norm": 2.2251272201538086, "learning_rate": 0.0001, "loss": 0.0229, "step": 17470 }, { "epoch": 133.4351145038168, "grad_norm": 1.7760385274887085, "learning_rate": 0.0001, "loss": 0.0226, "step": 17480 }, { "epoch": 133.5114503816794, "grad_norm": 2.3665664196014404, "learning_rate": 0.0001, "loss": 0.0228, "step": 17490 }, { "epoch": 133.58778625954199, "grad_norm": 1.820529818534851, "learning_rate": 0.0001, "loss": 0.0233, "step": 17500 }, { "epoch": 133.66412213740458, "grad_norm": 2.3367795944213867, "learning_rate": 0.0001, "loss": 0.0239, "step": 17510 }, { "epoch": 133.74045801526717, "grad_norm": 2.4237756729125977, "learning_rate": 0.0001, "loss": 0.0233, "step": 17520 }, { "epoch": 133.81679389312978, "grad_norm": 1.8744934797286987, "learning_rate": 0.0001, "loss": 0.0232, "step": 17530 }, { "epoch": 133.89312977099237, "grad_norm": 2.206275463104248, "learning_rate": 0.0001, "loss": 0.0226, "step": 17540 }, { "epoch": 133.96946564885496, "grad_norm": 2.367539644241333, "learning_rate": 0.0001, "loss": 0.0235, "step": 17550 }, { "epoch": 134.04580152671755, "grad_norm": 2.706902503967285, "learning_rate": 0.0001, "loss": 0.0233, "step": 17560 }, { "epoch": 134.12213740458014, "grad_norm": 1.8736778497695923, "learning_rate": 0.0001, "loss": 0.0231, "step": 17570 }, { "epoch": 134.19847328244273, "grad_norm": 2.6433279514312744, "learning_rate": 0.0001, "loss": 0.0233, "step": 17580 }, { "epoch": 134.27480916030535, "grad_norm": 2.4147403240203857, "learning_rate": 0.0001, "loss": 0.0235, "step": 17590 }, { "epoch": 134.35114503816794, "grad_norm": 1.8503918647766113, "learning_rate": 0.0001, "loss": 0.023, "step": 17600 }, { "epoch": 134.42748091603053, "grad_norm": 2.960439443588257, "learning_rate": 0.0001, "loss": 0.0228, "step": 17610 }, { "epoch": 134.50381679389312, "grad_norm": 1.3852510452270508, "learning_rate": 0.0001, "loss": 0.0227, "step": 17620 }, { "epoch": 134.5801526717557, "grad_norm": 1.7641721963882446, "learning_rate": 0.0001, "loss": 0.0222, "step": 17630 }, { "epoch": 134.65648854961833, "grad_norm": 1.5121361017227173, "learning_rate": 0.0001, "loss": 0.0238, "step": 17640 }, { "epoch": 134.73282442748092, "grad_norm": 2.3068606853485107, "learning_rate": 0.0001, "loss": 0.0238, "step": 17650 }, { "epoch": 134.8091603053435, "grad_norm": 1.716588020324707, "learning_rate": 0.0001, "loss": 0.0252, "step": 17660 }, { "epoch": 134.8854961832061, "grad_norm": 2.560833692550659, "learning_rate": 0.0001, "loss": 0.0243, "step": 17670 }, { "epoch": 134.9618320610687, "grad_norm": 2.049750328063965, "learning_rate": 0.0001, "loss": 0.0247, "step": 17680 }, { "epoch": 135.0381679389313, "grad_norm": 1.6938486099243164, "learning_rate": 0.0001, "loss": 0.0238, "step": 17690 }, { "epoch": 135.1145038167939, "grad_norm": 2.614863395690918, "learning_rate": 0.0001, "loss": 0.0254, "step": 17700 }, { "epoch": 135.1908396946565, "grad_norm": 2.718453884124756, "learning_rate": 0.0001, "loss": 0.0255, "step": 17710 }, { "epoch": 135.26717557251908, "grad_norm": 2.672191858291626, "learning_rate": 0.0001, "loss": 0.0253, "step": 17720 }, { "epoch": 135.34351145038167, "grad_norm": 1.8606363534927368, "learning_rate": 0.0001, "loss": 0.0245, "step": 17730 }, { "epoch": 135.4198473282443, "grad_norm": 1.7322572469711304, "learning_rate": 0.0001, "loss": 0.0239, "step": 17740 }, { "epoch": 135.49618320610688, "grad_norm": 1.3159083127975464, "learning_rate": 0.0001, "loss": 0.0238, "step": 17750 }, { "epoch": 135.57251908396947, "grad_norm": 1.6337357759475708, "learning_rate": 0.0001, "loss": 0.024, "step": 17760 }, { "epoch": 135.64885496183206, "grad_norm": 2.69419527053833, "learning_rate": 0.0001, "loss": 0.0241, "step": 17770 }, { "epoch": 135.72519083969465, "grad_norm": 2.2264959812164307, "learning_rate": 0.0001, "loss": 0.0251, "step": 17780 }, { "epoch": 135.80152671755727, "grad_norm": 1.8453428745269775, "learning_rate": 0.0001, "loss": 0.0244, "step": 17790 }, { "epoch": 135.87786259541986, "grad_norm": 2.204101085662842, "learning_rate": 0.0001, "loss": 0.0238, "step": 17800 }, { "epoch": 135.95419847328245, "grad_norm": 1.814440369606018, "learning_rate": 0.0001, "loss": 0.0234, "step": 17810 }, { "epoch": 136.03053435114504, "grad_norm": 2.256760358810425, "learning_rate": 0.0001, "loss": 0.0242, "step": 17820 }, { "epoch": 136.10687022900763, "grad_norm": 2.301678419113159, "learning_rate": 0.0001, "loss": 0.0245, "step": 17830 }, { "epoch": 136.18320610687022, "grad_norm": 2.0630905628204346, "learning_rate": 0.0001, "loss": 0.0236, "step": 17840 }, { "epoch": 136.25954198473283, "grad_norm": 2.215071678161621, "learning_rate": 0.0001, "loss": 0.024, "step": 17850 }, { "epoch": 136.33587786259542, "grad_norm": 2.200763702392578, "learning_rate": 0.0001, "loss": 0.0244, "step": 17860 }, { "epoch": 136.41221374045801, "grad_norm": 1.8786540031433105, "learning_rate": 0.0001, "loss": 0.0237, "step": 17870 }, { "epoch": 136.4885496183206, "grad_norm": 2.1192283630371094, "learning_rate": 0.0001, "loss": 0.0241, "step": 17880 }, { "epoch": 136.5648854961832, "grad_norm": 1.7447792291641235, "learning_rate": 0.0001, "loss": 0.024, "step": 17890 }, { "epoch": 136.6412213740458, "grad_norm": 2.151350259780884, "learning_rate": 0.0001, "loss": 0.0238, "step": 17900 }, { "epoch": 136.7175572519084, "grad_norm": 2.182978868484497, "learning_rate": 0.0001, "loss": 0.024, "step": 17910 }, { "epoch": 136.793893129771, "grad_norm": 2.5491943359375, "learning_rate": 0.0001, "loss": 0.0226, "step": 17920 }, { "epoch": 136.87022900763358, "grad_norm": 1.852717638015747, "learning_rate": 0.0001, "loss": 0.024, "step": 17930 }, { "epoch": 136.94656488549617, "grad_norm": 2.8841519355773926, "learning_rate": 0.0001, "loss": 0.0236, "step": 17940 }, { "epoch": 137.0229007633588, "grad_norm": 1.4681836366653442, "learning_rate": 0.0001, "loss": 0.0237, "step": 17950 }, { "epoch": 137.09923664122138, "grad_norm": 2.398796319961548, "learning_rate": 0.0001, "loss": 0.0238, "step": 17960 }, { "epoch": 137.17557251908397, "grad_norm": 2.713977336883545, "learning_rate": 0.0001, "loss": 0.024, "step": 17970 }, { "epoch": 137.25190839694656, "grad_norm": 1.9905749559402466, "learning_rate": 0.0001, "loss": 0.0242, "step": 17980 }, { "epoch": 137.32824427480915, "grad_norm": 1.2388859987258911, "learning_rate": 0.0001, "loss": 0.0244, "step": 17990 }, { "epoch": 137.40458015267177, "grad_norm": 1.5954746007919312, "learning_rate": 0.0001, "loss": 0.0238, "step": 18000 }, { "epoch": 137.48091603053436, "grad_norm": 2.5332436561584473, "learning_rate": 0.0001, "loss": 0.0237, "step": 18010 }, { "epoch": 137.55725190839695, "grad_norm": 1.9164764881134033, "learning_rate": 0.0001, "loss": 0.025, "step": 18020 }, { "epoch": 137.63358778625954, "grad_norm": 1.9297348260879517, "learning_rate": 0.0001, "loss": 0.025, "step": 18030 }, { "epoch": 137.70992366412213, "grad_norm": 1.6118085384368896, "learning_rate": 0.0001, "loss": 0.0245, "step": 18040 }, { "epoch": 137.78625954198472, "grad_norm": 2.1878342628479004, "learning_rate": 0.0001, "loss": 0.025, "step": 18050 }, { "epoch": 137.86259541984734, "grad_norm": 3.0654454231262207, "learning_rate": 0.0001, "loss": 0.0277, "step": 18060 }, { "epoch": 137.93893129770993, "grad_norm": 1.7301170825958252, "learning_rate": 0.0001, "loss": 0.0259, "step": 18070 }, { "epoch": 138.01526717557252, "grad_norm": 2.535501718521118, "learning_rate": 0.0001, "loss": 0.0243, "step": 18080 }, { "epoch": 138.0916030534351, "grad_norm": 2.1926183700561523, "learning_rate": 0.0001, "loss": 0.0243, "step": 18090 }, { "epoch": 138.1679389312977, "grad_norm": 1.818994164466858, "learning_rate": 0.0001, "loss": 0.0232, "step": 18100 }, { "epoch": 138.24427480916032, "grad_norm": 1.4814082384109497, "learning_rate": 0.0001, "loss": 0.0228, "step": 18110 }, { "epoch": 138.3206106870229, "grad_norm": 1.6262168884277344, "learning_rate": 0.0001, "loss": 0.0232, "step": 18120 }, { "epoch": 138.3969465648855, "grad_norm": 1.6219680309295654, "learning_rate": 0.0001, "loss": 0.0243, "step": 18130 }, { "epoch": 138.4732824427481, "grad_norm": 1.7726970911026, "learning_rate": 0.0001, "loss": 0.0242, "step": 18140 }, { "epoch": 138.54961832061068, "grad_norm": 1.5728394985198975, "learning_rate": 0.0001, "loss": 0.0237, "step": 18150 }, { "epoch": 138.6259541984733, "grad_norm": 2.7487504482269287, "learning_rate": 0.0001, "loss": 0.0257, "step": 18160 }, { "epoch": 138.70229007633588, "grad_norm": 2.474066972732544, "learning_rate": 0.0001, "loss": 0.0244, "step": 18170 }, { "epoch": 138.77862595419847, "grad_norm": 1.9651178121566772, "learning_rate": 0.0001, "loss": 0.0231, "step": 18180 }, { "epoch": 138.85496183206106, "grad_norm": 2.0430376529693604, "learning_rate": 0.0001, "loss": 0.0225, "step": 18190 }, { "epoch": 138.93129770992365, "grad_norm": 1.4653342962265015, "learning_rate": 0.0001, "loss": 0.0227, "step": 18200 }, { "epoch": 139.00763358778627, "grad_norm": 1.6509342193603516, "learning_rate": 0.0001, "loss": 0.0227, "step": 18210 }, { "epoch": 139.08396946564886, "grad_norm": 1.786604881286621, "learning_rate": 0.0001, "loss": 0.0232, "step": 18220 }, { "epoch": 139.16030534351145, "grad_norm": 2.336493730545044, "learning_rate": 0.0001, "loss": 0.0228, "step": 18230 }, { "epoch": 139.23664122137404, "grad_norm": 2.0069239139556885, "learning_rate": 0.0001, "loss": 0.023, "step": 18240 }, { "epoch": 139.31297709923663, "grad_norm": 1.6837098598480225, "learning_rate": 0.0001, "loss": 0.0238, "step": 18250 }, { "epoch": 139.38931297709922, "grad_norm": 1.7687920331954956, "learning_rate": 0.0001, "loss": 0.0237, "step": 18260 }, { "epoch": 139.46564885496184, "grad_norm": 1.802767276763916, "learning_rate": 0.0001, "loss": 0.0235, "step": 18270 }, { "epoch": 139.54198473282443, "grad_norm": 2.0011942386627197, "learning_rate": 0.0001, "loss": 0.0246, "step": 18280 }, { "epoch": 139.61832061068702, "grad_norm": 2.678366184234619, "learning_rate": 0.0001, "loss": 0.0247, "step": 18290 }, { "epoch": 139.6946564885496, "grad_norm": 1.9679418802261353, "learning_rate": 0.0001, "loss": 0.024, "step": 18300 }, { "epoch": 139.7709923664122, "grad_norm": 1.9045889377593994, "learning_rate": 0.0001, "loss": 0.0234, "step": 18310 }, { "epoch": 139.84732824427482, "grad_norm": 2.1383426189422607, "learning_rate": 0.0001, "loss": 0.024, "step": 18320 }, { "epoch": 139.9236641221374, "grad_norm": 2.010624885559082, "learning_rate": 0.0001, "loss": 0.0243, "step": 18330 }, { "epoch": 140.0, "grad_norm": 2.1159169673919678, "learning_rate": 0.0001, "loss": 0.0239, "step": 18340 }, { "epoch": 140.0763358778626, "grad_norm": 2.0660452842712402, "learning_rate": 0.0001, "loss": 0.0235, "step": 18350 }, { "epoch": 140.15267175572518, "grad_norm": 1.5873801708221436, "learning_rate": 0.0001, "loss": 0.0239, "step": 18360 }, { "epoch": 140.2290076335878, "grad_norm": 1.531684398651123, "learning_rate": 0.0001, "loss": 0.0235, "step": 18370 }, { "epoch": 140.3053435114504, "grad_norm": 1.8974181413650513, "learning_rate": 0.0001, "loss": 0.0233, "step": 18380 }, { "epoch": 140.38167938931298, "grad_norm": 1.2750846147537231, "learning_rate": 0.0001, "loss": 0.0239, "step": 18390 }, { "epoch": 140.45801526717557, "grad_norm": 1.6573597192764282, "learning_rate": 0.0001, "loss": 0.0242, "step": 18400 }, { "epoch": 140.53435114503816, "grad_norm": 1.568307638168335, "learning_rate": 0.0001, "loss": 0.0237, "step": 18410 }, { "epoch": 140.61068702290078, "grad_norm": 1.7128914594650269, "learning_rate": 0.0001, "loss": 0.0245, "step": 18420 }, { "epoch": 140.68702290076337, "grad_norm": 1.6476033926010132, "learning_rate": 0.0001, "loss": 0.0237, "step": 18430 }, { "epoch": 140.76335877862596, "grad_norm": 2.3594841957092285, "learning_rate": 0.0001, "loss": 0.0232, "step": 18440 }, { "epoch": 140.83969465648855, "grad_norm": 1.8596796989440918, "learning_rate": 0.0001, "loss": 0.0235, "step": 18450 }, { "epoch": 140.91603053435114, "grad_norm": 1.6548274755477905, "learning_rate": 0.0001, "loss": 0.0241, "step": 18460 }, { "epoch": 140.99236641221373, "grad_norm": 1.747327446937561, "learning_rate": 0.0001, "loss": 0.0236, "step": 18470 }, { "epoch": 141.06870229007635, "grad_norm": 1.7708501815795898, "learning_rate": 0.0001, "loss": 0.0237, "step": 18480 }, { "epoch": 141.14503816793894, "grad_norm": 2.177659273147583, "learning_rate": 0.0001, "loss": 0.0236, "step": 18490 }, { "epoch": 141.22137404580153, "grad_norm": 1.5947973728179932, "learning_rate": 0.0001, "loss": 0.0237, "step": 18500 }, { "epoch": 141.29770992366412, "grad_norm": 1.5913785696029663, "learning_rate": 0.0001, "loss": 0.0241, "step": 18510 }, { "epoch": 141.3740458015267, "grad_norm": 1.8323575258255005, "learning_rate": 0.0001, "loss": 0.0241, "step": 18520 }, { "epoch": 141.45038167938932, "grad_norm": 2.03694748878479, "learning_rate": 0.0001, "loss": 0.0234, "step": 18530 }, { "epoch": 141.5267175572519, "grad_norm": 1.7177472114562988, "learning_rate": 0.0001, "loss": 0.0236, "step": 18540 }, { "epoch": 141.6030534351145, "grad_norm": 2.5359511375427246, "learning_rate": 0.0001, "loss": 0.0232, "step": 18550 }, { "epoch": 141.6793893129771, "grad_norm": 1.6932958364486694, "learning_rate": 0.0001, "loss": 0.0235, "step": 18560 }, { "epoch": 141.75572519083968, "grad_norm": 1.111154317855835, "learning_rate": 0.0001, "loss": 0.0228, "step": 18570 }, { "epoch": 141.8320610687023, "grad_norm": 1.9316298961639404, "learning_rate": 0.0001, "loss": 0.0228, "step": 18580 }, { "epoch": 141.9083969465649, "grad_norm": 1.3265175819396973, "learning_rate": 0.0001, "loss": 0.0229, "step": 18590 }, { "epoch": 141.98473282442748, "grad_norm": 1.7748003005981445, "learning_rate": 0.0001, "loss": 0.0229, "step": 18600 }, { "epoch": 142.06106870229007, "grad_norm": 2.6775388717651367, "learning_rate": 0.0001, "loss": 0.0239, "step": 18610 }, { "epoch": 142.13740458015266, "grad_norm": 2.3784589767456055, "learning_rate": 0.0001, "loss": 0.0234, "step": 18620 }, { "epoch": 142.21374045801528, "grad_norm": 1.7620410919189453, "learning_rate": 0.0001, "loss": 0.023, "step": 18630 }, { "epoch": 142.29007633587787, "grad_norm": 1.6391719579696655, "learning_rate": 0.0001, "loss": 0.0238, "step": 18640 }, { "epoch": 142.36641221374046, "grad_norm": 2.2096164226531982, "learning_rate": 0.0001, "loss": 0.0233, "step": 18650 }, { "epoch": 142.44274809160305, "grad_norm": 2.438231945037842, "learning_rate": 0.0001, "loss": 0.0233, "step": 18660 }, { "epoch": 142.51908396946564, "grad_norm": 1.441009521484375, "learning_rate": 0.0001, "loss": 0.0237, "step": 18670 }, { "epoch": 142.59541984732823, "grad_norm": 1.7045871019363403, "learning_rate": 0.0001, "loss": 0.0224, "step": 18680 }, { "epoch": 142.67175572519085, "grad_norm": 1.710010051727295, "learning_rate": 0.0001, "loss": 0.0232, "step": 18690 }, { "epoch": 142.74809160305344, "grad_norm": 2.75773024559021, "learning_rate": 0.0001, "loss": 0.0227, "step": 18700 }, { "epoch": 142.82442748091603, "grad_norm": 2.2741661071777344, "learning_rate": 0.0001, "loss": 0.0235, "step": 18710 }, { "epoch": 142.90076335877862, "grad_norm": 1.9336555004119873, "learning_rate": 0.0001, "loss": 0.0238, "step": 18720 }, { "epoch": 142.9770992366412, "grad_norm": 1.9988118410110474, "learning_rate": 0.0001, "loss": 0.0241, "step": 18730 }, { "epoch": 143.05343511450383, "grad_norm": 1.6503639221191406, "learning_rate": 0.0001, "loss": 0.0242, "step": 18740 }, { "epoch": 143.12977099236642, "grad_norm": 1.6259106397628784, "learning_rate": 0.0001, "loss": 0.023, "step": 18750 }, { "epoch": 143.206106870229, "grad_norm": 1.7541943788528442, "learning_rate": 0.0001, "loss": 0.0238, "step": 18760 }, { "epoch": 143.2824427480916, "grad_norm": 1.605751395225525, "learning_rate": 0.0001, "loss": 0.0237, "step": 18770 }, { "epoch": 143.3587786259542, "grad_norm": 1.6891995668411255, "learning_rate": 0.0001, "loss": 0.0236, "step": 18780 }, { "epoch": 143.4351145038168, "grad_norm": 1.6178981065750122, "learning_rate": 0.0001, "loss": 0.0243, "step": 18790 }, { "epoch": 143.5114503816794, "grad_norm": 2.0948095321655273, "learning_rate": 0.0001, "loss": 0.0242, "step": 18800 }, { "epoch": 143.58778625954199, "grad_norm": 2.835303783416748, "learning_rate": 0.0001, "loss": 0.0239, "step": 18810 }, { "epoch": 143.66412213740458, "grad_norm": 2.3677756786346436, "learning_rate": 0.0001, "loss": 0.0243, "step": 18820 }, { "epoch": 143.74045801526717, "grad_norm": 1.7373228073120117, "learning_rate": 0.0001, "loss": 0.0234, "step": 18830 }, { "epoch": 143.81679389312978, "grad_norm": 1.4445258378982544, "learning_rate": 0.0001, "loss": 0.0229, "step": 18840 }, { "epoch": 143.89312977099237, "grad_norm": 2.0503900051116943, "learning_rate": 0.0001, "loss": 0.0247, "step": 18850 }, { "epoch": 143.96946564885496, "grad_norm": 1.3873450756072998, "learning_rate": 0.0001, "loss": 0.0236, "step": 18860 }, { "epoch": 144.04580152671755, "grad_norm": 2.2544305324554443, "learning_rate": 0.0001, "loss": 0.0235, "step": 18870 }, { "epoch": 144.12213740458014, "grad_norm": 1.6000431776046753, "learning_rate": 0.0001, "loss": 0.0236, "step": 18880 }, { "epoch": 144.19847328244273, "grad_norm": 2.246922731399536, "learning_rate": 0.0001, "loss": 0.023, "step": 18890 }, { "epoch": 144.27480916030535, "grad_norm": 2.287295341491699, "learning_rate": 0.0001, "loss": 0.0227, "step": 18900 }, { "epoch": 144.35114503816794, "grad_norm": 2.469330310821533, "learning_rate": 0.0001, "loss": 0.0224, "step": 18910 }, { "epoch": 144.42748091603053, "grad_norm": 2.708854913711548, "learning_rate": 0.0001, "loss": 0.0226, "step": 18920 }, { "epoch": 144.50381679389312, "grad_norm": 2.2957966327667236, "learning_rate": 0.0001, "loss": 0.0219, "step": 18930 }, { "epoch": 144.5801526717557, "grad_norm": 1.8328197002410889, "learning_rate": 0.0001, "loss": 0.0217, "step": 18940 }, { "epoch": 144.65648854961833, "grad_norm": 1.8786776065826416, "learning_rate": 0.0001, "loss": 0.0223, "step": 18950 }, { "epoch": 144.73282442748092, "grad_norm": 2.409547805786133, "learning_rate": 0.0001, "loss": 0.0226, "step": 18960 }, { "epoch": 144.8091603053435, "grad_norm": 1.57089102268219, "learning_rate": 0.0001, "loss": 0.022, "step": 18970 }, { "epoch": 144.8854961832061, "grad_norm": 1.8170711994171143, "learning_rate": 0.0001, "loss": 0.0224, "step": 18980 }, { "epoch": 144.9618320610687, "grad_norm": 2.880756378173828, "learning_rate": 0.0001, "loss": 0.0232, "step": 18990 }, { "epoch": 145.0381679389313, "grad_norm": 1.6972260475158691, "learning_rate": 0.0001, "loss": 0.024, "step": 19000 }, { "epoch": 145.1145038167939, "grad_norm": 1.5849835872650146, "learning_rate": 0.0001, "loss": 0.0228, "step": 19010 }, { "epoch": 145.1908396946565, "grad_norm": 1.578866720199585, "learning_rate": 0.0001, "loss": 0.0235, "step": 19020 }, { "epoch": 145.26717557251908, "grad_norm": 1.698745846748352, "learning_rate": 0.0001, "loss": 0.0229, "step": 19030 }, { "epoch": 145.34351145038167, "grad_norm": 1.7406078577041626, "learning_rate": 0.0001, "loss": 0.0234, "step": 19040 }, { "epoch": 145.4198473282443, "grad_norm": 1.6648285388946533, "learning_rate": 0.0001, "loss": 0.0245, "step": 19050 }, { "epoch": 145.49618320610688, "grad_norm": 1.4576486349105835, "learning_rate": 0.0001, "loss": 0.0251, "step": 19060 }, { "epoch": 145.57251908396947, "grad_norm": 1.8121185302734375, "learning_rate": 0.0001, "loss": 0.0252, "step": 19070 }, { "epoch": 145.64885496183206, "grad_norm": 2.1975667476654053, "learning_rate": 0.0001, "loss": 0.0241, "step": 19080 }, { "epoch": 145.72519083969465, "grad_norm": 1.8991353511810303, "learning_rate": 0.0001, "loss": 0.0241, "step": 19090 }, { "epoch": 145.80152671755727, "grad_norm": 2.0721771717071533, "learning_rate": 0.0001, "loss": 0.0231, "step": 19100 }, { "epoch": 145.87786259541986, "grad_norm": 1.947696566581726, "learning_rate": 0.0001, "loss": 0.0224, "step": 19110 }, { "epoch": 145.95419847328245, "grad_norm": 2.1143808364868164, "learning_rate": 0.0001, "loss": 0.0224, "step": 19120 }, { "epoch": 146.03053435114504, "grad_norm": 2.339362382888794, "learning_rate": 0.0001, "loss": 0.0223, "step": 19130 }, { "epoch": 146.10687022900763, "grad_norm": 2.1764614582061768, "learning_rate": 0.0001, "loss": 0.0237, "step": 19140 }, { "epoch": 146.18320610687022, "grad_norm": 1.529078483581543, "learning_rate": 0.0001, "loss": 0.0239, "step": 19150 }, { "epoch": 146.25954198473283, "grad_norm": 2.6940696239471436, "learning_rate": 0.0001, "loss": 0.0239, "step": 19160 }, { "epoch": 146.33587786259542, "grad_norm": 2.0226027965545654, "learning_rate": 0.0001, "loss": 0.0247, "step": 19170 }, { "epoch": 146.41221374045801, "grad_norm": 2.378397226333618, "learning_rate": 0.0001, "loss": 0.0237, "step": 19180 }, { "epoch": 146.4885496183206, "grad_norm": 1.8274277448654175, "learning_rate": 0.0001, "loss": 0.0231, "step": 19190 }, { "epoch": 146.5648854961832, "grad_norm": 1.6872711181640625, "learning_rate": 0.0001, "loss": 0.0234, "step": 19200 }, { "epoch": 146.6412213740458, "grad_norm": 2.088191509246826, "learning_rate": 0.0001, "loss": 0.0236, "step": 19210 }, { "epoch": 146.7175572519084, "grad_norm": 2.5711617469787598, "learning_rate": 0.0001, "loss": 0.0228, "step": 19220 }, { "epoch": 146.793893129771, "grad_norm": 2.0668835639953613, "learning_rate": 0.0001, "loss": 0.0235, "step": 19230 }, { "epoch": 146.87022900763358, "grad_norm": 1.6526426076889038, "learning_rate": 0.0001, "loss": 0.0229, "step": 19240 }, { "epoch": 146.94656488549617, "grad_norm": 1.8869959115982056, "learning_rate": 0.0001, "loss": 0.0231, "step": 19250 }, { "epoch": 147.0229007633588, "grad_norm": 1.6258431673049927, "learning_rate": 0.0001, "loss": 0.0237, "step": 19260 }, { "epoch": 147.09923664122138, "grad_norm": 1.6479649543762207, "learning_rate": 0.0001, "loss": 0.0238, "step": 19270 }, { "epoch": 147.17557251908397, "grad_norm": 1.809597134590149, "learning_rate": 0.0001, "loss": 0.0235, "step": 19280 }, { "epoch": 147.25190839694656, "grad_norm": 1.5854109525680542, "learning_rate": 0.0001, "loss": 0.0247, "step": 19290 }, { "epoch": 147.32824427480915, "grad_norm": 1.8167163133621216, "learning_rate": 0.0001, "loss": 0.0245, "step": 19300 }, { "epoch": 147.40458015267177, "grad_norm": 2.206284999847412, "learning_rate": 0.0001, "loss": 0.0239, "step": 19310 }, { "epoch": 147.48091603053436, "grad_norm": 2.0223848819732666, "learning_rate": 0.0001, "loss": 0.0244, "step": 19320 }, { "epoch": 147.55725190839695, "grad_norm": 1.9250297546386719, "learning_rate": 0.0001, "loss": 0.0247, "step": 19330 }, { "epoch": 147.63358778625954, "grad_norm": 1.931164026260376, "learning_rate": 0.0001, "loss": 0.0239, "step": 19340 }, { "epoch": 147.70992366412213, "grad_norm": 1.7324875593185425, "learning_rate": 0.0001, "loss": 0.023, "step": 19350 }, { "epoch": 147.78625954198472, "grad_norm": 1.7411240339279175, "learning_rate": 0.0001, "loss": 0.0236, "step": 19360 }, { "epoch": 147.86259541984734, "grad_norm": 1.3598190546035767, "learning_rate": 0.0001, "loss": 0.0234, "step": 19370 }, { "epoch": 147.93893129770993, "grad_norm": 2.1158676147460938, "learning_rate": 0.0001, "loss": 0.0241, "step": 19380 }, { "epoch": 148.01526717557252, "grad_norm": 1.4090447425842285, "learning_rate": 0.0001, "loss": 0.0229, "step": 19390 }, { "epoch": 148.0916030534351, "grad_norm": 1.9963891506195068, "learning_rate": 0.0001, "loss": 0.0248, "step": 19400 }, { "epoch": 148.1679389312977, "grad_norm": 2.64150071144104, "learning_rate": 0.0001, "loss": 0.0264, "step": 19410 }, { "epoch": 148.24427480916032, "grad_norm": 3.252351999282837, "learning_rate": 0.0001, "loss": 0.0253, "step": 19420 }, { "epoch": 148.3206106870229, "grad_norm": 1.8860291242599487, "learning_rate": 0.0001, "loss": 0.0237, "step": 19430 }, { "epoch": 148.3969465648855, "grad_norm": 1.4900681972503662, "learning_rate": 0.0001, "loss": 0.0236, "step": 19440 }, { "epoch": 148.4732824427481, "grad_norm": 2.0659985542297363, "learning_rate": 0.0001, "loss": 0.0238, "step": 19450 }, { "epoch": 148.54961832061068, "grad_norm": 2.494392156600952, "learning_rate": 0.0001, "loss": 0.0228, "step": 19460 }, { "epoch": 148.6259541984733, "grad_norm": 2.268549680709839, "learning_rate": 0.0001, "loss": 0.0233, "step": 19470 }, { "epoch": 148.70229007633588, "grad_norm": 2.5438458919525146, "learning_rate": 0.0001, "loss": 0.0231, "step": 19480 }, { "epoch": 148.77862595419847, "grad_norm": 2.1818320751190186, "learning_rate": 0.0001, "loss": 0.0232, "step": 19490 }, { "epoch": 148.85496183206106, "grad_norm": 1.710768699645996, "learning_rate": 0.0001, "loss": 0.0231, "step": 19500 }, { "epoch": 148.93129770992365, "grad_norm": 1.6776199340820312, "learning_rate": 0.0001, "loss": 0.0228, "step": 19510 }, { "epoch": 149.00763358778627, "grad_norm": 1.3429304361343384, "learning_rate": 0.0001, "loss": 0.0229, "step": 19520 }, { "epoch": 149.08396946564886, "grad_norm": 2.375135898590088, "learning_rate": 0.0001, "loss": 0.0231, "step": 19530 }, { "epoch": 149.16030534351145, "grad_norm": 2.2844555377960205, "learning_rate": 0.0001, "loss": 0.0238, "step": 19540 }, { "epoch": 149.23664122137404, "grad_norm": 2.0137836933135986, "learning_rate": 0.0001, "loss": 0.0239, "step": 19550 }, { "epoch": 149.31297709923663, "grad_norm": 2.776219367980957, "learning_rate": 0.0001, "loss": 0.0236, "step": 19560 }, { "epoch": 149.38931297709922, "grad_norm": 2.254399061203003, "learning_rate": 0.0001, "loss": 0.0233, "step": 19570 }, { "epoch": 149.46564885496184, "grad_norm": 1.4407480955123901, "learning_rate": 0.0001, "loss": 0.0218, "step": 19580 }, { "epoch": 149.54198473282443, "grad_norm": 2.179612159729004, "learning_rate": 0.0001, "loss": 0.0215, "step": 19590 }, { "epoch": 149.61832061068702, "grad_norm": 1.7531105279922485, "learning_rate": 0.0001, "loss": 0.0222, "step": 19600 }, { "epoch": 149.6946564885496, "grad_norm": 1.5660240650177002, "learning_rate": 0.0001, "loss": 0.0223, "step": 19610 }, { "epoch": 149.7709923664122, "grad_norm": 1.850452184677124, "learning_rate": 0.0001, "loss": 0.0224, "step": 19620 }, { "epoch": 149.84732824427482, "grad_norm": 2.3790791034698486, "learning_rate": 0.0001, "loss": 0.0223, "step": 19630 }, { "epoch": 149.9236641221374, "grad_norm": 2.053863763809204, "learning_rate": 0.0001, "loss": 0.0232, "step": 19640 }, { "epoch": 150.0, "grad_norm": 1.5652869939804077, "learning_rate": 0.0001, "loss": 0.0228, "step": 19650 }, { "epoch": 150.0763358778626, "grad_norm": 1.6206421852111816, "learning_rate": 0.0001, "loss": 0.023, "step": 19660 }, { "epoch": 150.15267175572518, "grad_norm": 1.5756727457046509, "learning_rate": 0.0001, "loss": 0.023, "step": 19670 }, { "epoch": 150.2290076335878, "grad_norm": 1.8529952764511108, "learning_rate": 0.0001, "loss": 0.0229, "step": 19680 }, { "epoch": 150.3053435114504, "grad_norm": 1.234471321105957, "learning_rate": 0.0001, "loss": 0.0217, "step": 19690 }, { "epoch": 150.38167938931298, "grad_norm": 2.271970510482788, "learning_rate": 0.0001, "loss": 0.0237, "step": 19700 }, { "epoch": 150.45801526717557, "grad_norm": 1.1196787357330322, "learning_rate": 0.0001, "loss": 0.0228, "step": 19710 }, { "epoch": 150.53435114503816, "grad_norm": 1.6805298328399658, "learning_rate": 0.0001, "loss": 0.024, "step": 19720 }, { "epoch": 150.61068702290078, "grad_norm": 2.0974154472351074, "learning_rate": 0.0001, "loss": 0.024, "step": 19730 }, { "epoch": 150.68702290076337, "grad_norm": 2.247795343399048, "learning_rate": 0.0001, "loss": 0.0241, "step": 19740 }, { "epoch": 150.76335877862596, "grad_norm": 1.7499595880508423, "learning_rate": 0.0001, "loss": 0.0235, "step": 19750 }, { "epoch": 150.83969465648855, "grad_norm": 1.6256053447723389, "learning_rate": 0.0001, "loss": 0.0237, "step": 19760 }, { "epoch": 150.91603053435114, "grad_norm": 2.3754379749298096, "learning_rate": 0.0001, "loss": 0.0241, "step": 19770 }, { "epoch": 150.99236641221373, "grad_norm": 2.241572618484497, "learning_rate": 0.0001, "loss": 0.0238, "step": 19780 }, { "epoch": 151.06870229007635, "grad_norm": 1.6448408365249634, "learning_rate": 0.0001, "loss": 0.0237, "step": 19790 }, { "epoch": 151.14503816793894, "grad_norm": 1.9836511611938477, "learning_rate": 0.0001, "loss": 0.0228, "step": 19800 }, { "epoch": 151.22137404580153, "grad_norm": 2.2543461322784424, "learning_rate": 0.0001, "loss": 0.0232, "step": 19810 }, { "epoch": 151.29770992366412, "grad_norm": 2.1773831844329834, "learning_rate": 0.0001, "loss": 0.0225, "step": 19820 }, { "epoch": 151.3740458015267, "grad_norm": 1.829363226890564, "learning_rate": 0.0001, "loss": 0.0225, "step": 19830 }, { "epoch": 151.45038167938932, "grad_norm": 1.7351843118667603, "learning_rate": 0.0001, "loss": 0.0226, "step": 19840 }, { "epoch": 151.5267175572519, "grad_norm": 1.6356474161148071, "learning_rate": 0.0001, "loss": 0.0224, "step": 19850 }, { "epoch": 151.6030534351145, "grad_norm": 2.028261423110962, "learning_rate": 0.0001, "loss": 0.024, "step": 19860 }, { "epoch": 151.6793893129771, "grad_norm": 1.9098550081253052, "learning_rate": 0.0001, "loss": 0.0235, "step": 19870 }, { "epoch": 151.75572519083968, "grad_norm": 1.846701979637146, "learning_rate": 0.0001, "loss": 0.0231, "step": 19880 }, { "epoch": 151.8320610687023, "grad_norm": 2.1825153827667236, "learning_rate": 0.0001, "loss": 0.023, "step": 19890 }, { "epoch": 151.9083969465649, "grad_norm": 2.628314256668091, "learning_rate": 0.0001, "loss": 0.0228, "step": 19900 }, { "epoch": 151.98473282442748, "grad_norm": 2.650766372680664, "learning_rate": 0.0001, "loss": 0.0237, "step": 19910 }, { "epoch": 152.06106870229007, "grad_norm": 1.9852737188339233, "learning_rate": 0.0001, "loss": 0.0233, "step": 19920 }, { "epoch": 152.13740458015266, "grad_norm": 1.9590682983398438, "learning_rate": 0.0001, "loss": 0.0226, "step": 19930 }, { "epoch": 152.21374045801528, "grad_norm": 2.0994811058044434, "learning_rate": 0.0001, "loss": 0.023, "step": 19940 }, { "epoch": 152.29007633587787, "grad_norm": 2.1326661109924316, "learning_rate": 0.0001, "loss": 0.0224, "step": 19950 }, { "epoch": 152.36641221374046, "grad_norm": 1.7707079648971558, "learning_rate": 0.0001, "loss": 0.0225, "step": 19960 }, { "epoch": 152.44274809160305, "grad_norm": 2.825021505355835, "learning_rate": 0.0001, "loss": 0.0234, "step": 19970 }, { "epoch": 152.51908396946564, "grad_norm": 1.7318092584609985, "learning_rate": 0.0001, "loss": 0.0235, "step": 19980 }, { "epoch": 152.59541984732823, "grad_norm": 2.2735652923583984, "learning_rate": 0.0001, "loss": 0.0231, "step": 19990 }, { "epoch": 152.67175572519085, "grad_norm": 1.54047691822052, "learning_rate": 0.0001, "loss": 0.0236, "step": 20000 }, { "epoch": 152.74809160305344, "grad_norm": 1.3645058870315552, "learning_rate": 0.0001, "loss": 0.0223, "step": 20010 }, { "epoch": 152.82442748091603, "grad_norm": 1.7466679811477661, "learning_rate": 0.0001, "loss": 0.0228, "step": 20020 }, { "epoch": 152.90076335877862, "grad_norm": 1.6853116750717163, "learning_rate": 0.0001, "loss": 0.023, "step": 20030 }, { "epoch": 152.9770992366412, "grad_norm": 2.1515796184539795, "learning_rate": 0.0001, "loss": 0.0235, "step": 20040 }, { "epoch": 153.05343511450383, "grad_norm": 2.3406577110290527, "learning_rate": 0.0001, "loss": 0.023, "step": 20050 }, { "epoch": 153.12977099236642, "grad_norm": 1.9794905185699463, "learning_rate": 0.0001, "loss": 0.0221, "step": 20060 }, { "epoch": 153.206106870229, "grad_norm": 1.5245537757873535, "learning_rate": 0.0001, "loss": 0.022, "step": 20070 }, { "epoch": 153.2824427480916, "grad_norm": 1.549617886543274, "learning_rate": 0.0001, "loss": 0.0222, "step": 20080 }, { "epoch": 153.3587786259542, "grad_norm": 1.6137843132019043, "learning_rate": 0.0001, "loss": 0.0221, "step": 20090 }, { "epoch": 153.4351145038168, "grad_norm": 1.9210678339004517, "learning_rate": 0.0001, "loss": 0.0218, "step": 20100 }, { "epoch": 153.5114503816794, "grad_norm": 1.4846371412277222, "learning_rate": 0.0001, "loss": 0.022, "step": 20110 }, { "epoch": 153.58778625954199, "grad_norm": 1.796353816986084, "learning_rate": 0.0001, "loss": 0.0218, "step": 20120 }, { "epoch": 153.66412213740458, "grad_norm": 1.6151292324066162, "learning_rate": 0.0001, "loss": 0.0226, "step": 20130 }, { "epoch": 153.74045801526717, "grad_norm": 1.6031830310821533, "learning_rate": 0.0001, "loss": 0.0229, "step": 20140 }, { "epoch": 153.81679389312978, "grad_norm": 1.3880336284637451, "learning_rate": 0.0001, "loss": 0.0221, "step": 20150 }, { "epoch": 153.89312977099237, "grad_norm": 1.5706098079681396, "learning_rate": 0.0001, "loss": 0.0223, "step": 20160 }, { "epoch": 153.96946564885496, "grad_norm": 2.195071220397949, "learning_rate": 0.0001, "loss": 0.0237, "step": 20170 }, { "epoch": 154.04580152671755, "grad_norm": 1.8442630767822266, "learning_rate": 0.0001, "loss": 0.0227, "step": 20180 }, { "epoch": 154.12213740458014, "grad_norm": 1.3034796714782715, "learning_rate": 0.0001, "loss": 0.0217, "step": 20190 }, { "epoch": 154.19847328244273, "grad_norm": 2.2276995182037354, "learning_rate": 0.0001, "loss": 0.0221, "step": 20200 }, { "epoch": 154.27480916030535, "grad_norm": 1.7623169422149658, "learning_rate": 0.0001, "loss": 0.0227, "step": 20210 }, { "epoch": 154.35114503816794, "grad_norm": 2.0394818782806396, "learning_rate": 0.0001, "loss": 0.0231, "step": 20220 }, { "epoch": 154.42748091603053, "grad_norm": 1.5255910158157349, "learning_rate": 0.0001, "loss": 0.0232, "step": 20230 }, { "epoch": 154.50381679389312, "grad_norm": 1.6409101486206055, "learning_rate": 0.0001, "loss": 0.0227, "step": 20240 }, { "epoch": 154.5801526717557, "grad_norm": 1.5602061748504639, "learning_rate": 0.0001, "loss": 0.0231, "step": 20250 }, { "epoch": 154.65648854961833, "grad_norm": 1.6521594524383545, "learning_rate": 0.0001, "loss": 0.0224, "step": 20260 }, { "epoch": 154.73282442748092, "grad_norm": 1.66873300075531, "learning_rate": 0.0001, "loss": 0.0223, "step": 20270 }, { "epoch": 154.8091603053435, "grad_norm": 1.3387658596038818, "learning_rate": 0.0001, "loss": 0.0222, "step": 20280 }, { "epoch": 154.8854961832061, "grad_norm": 1.9882229566574097, "learning_rate": 0.0001, "loss": 0.0236, "step": 20290 }, { "epoch": 154.9618320610687, "grad_norm": 1.2342780828475952, "learning_rate": 0.0001, "loss": 0.0231, "step": 20300 }, { "epoch": 155.0381679389313, "grad_norm": 1.7621792554855347, "learning_rate": 0.0001, "loss": 0.0232, "step": 20310 }, { "epoch": 155.1145038167939, "grad_norm": 1.2797856330871582, "learning_rate": 0.0001, "loss": 0.0226, "step": 20320 }, { "epoch": 155.1908396946565, "grad_norm": 2.030310869216919, "learning_rate": 0.0001, "loss": 0.0226, "step": 20330 }, { "epoch": 155.26717557251908, "grad_norm": 1.7472466230392456, "learning_rate": 0.0001, "loss": 0.0231, "step": 20340 }, { "epoch": 155.34351145038167, "grad_norm": 1.750090479850769, "learning_rate": 0.0001, "loss": 0.0238, "step": 20350 }, { "epoch": 155.4198473282443, "grad_norm": 1.9415348768234253, "learning_rate": 0.0001, "loss": 0.0228, "step": 20360 }, { "epoch": 155.49618320610688, "grad_norm": 1.8177266120910645, "learning_rate": 0.0001, "loss": 0.0229, "step": 20370 }, { "epoch": 155.57251908396947, "grad_norm": 1.4320719242095947, "learning_rate": 0.0001, "loss": 0.0231, "step": 20380 }, { "epoch": 155.64885496183206, "grad_norm": 1.2968498468399048, "learning_rate": 0.0001, "loss": 0.0228, "step": 20390 }, { "epoch": 155.72519083969465, "grad_norm": 2.284210681915283, "learning_rate": 0.0001, "loss": 0.0237, "step": 20400 }, { "epoch": 155.80152671755727, "grad_norm": 2.2397711277008057, "learning_rate": 0.0001, "loss": 0.0229, "step": 20410 }, { "epoch": 155.87786259541986, "grad_norm": 2.091787338256836, "learning_rate": 0.0001, "loss": 0.0228, "step": 20420 }, { "epoch": 155.95419847328245, "grad_norm": 2.1657309532165527, "learning_rate": 0.0001, "loss": 0.0226, "step": 20430 }, { "epoch": 156.03053435114504, "grad_norm": 1.774595856666565, "learning_rate": 0.0001, "loss": 0.0225, "step": 20440 }, { "epoch": 156.10687022900763, "grad_norm": 1.9972580671310425, "learning_rate": 0.0001, "loss": 0.0224, "step": 20450 }, { "epoch": 156.18320610687022, "grad_norm": 2.9125568866729736, "learning_rate": 0.0001, "loss": 0.0218, "step": 20460 }, { "epoch": 156.25954198473283, "grad_norm": 2.082122802734375, "learning_rate": 0.0001, "loss": 0.0221, "step": 20470 }, { "epoch": 156.33587786259542, "grad_norm": 1.6598306894302368, "learning_rate": 0.0001, "loss": 0.0218, "step": 20480 }, { "epoch": 156.41221374045801, "grad_norm": 2.2801973819732666, "learning_rate": 0.0001, "loss": 0.0215, "step": 20490 }, { "epoch": 156.4885496183206, "grad_norm": 1.3664933443069458, "learning_rate": 0.0001, "loss": 0.0221, "step": 20500 }, { "epoch": 156.5648854961832, "grad_norm": 1.5850743055343628, "learning_rate": 0.0001, "loss": 0.0216, "step": 20510 }, { "epoch": 156.6412213740458, "grad_norm": 1.5081183910369873, "learning_rate": 0.0001, "loss": 0.0213, "step": 20520 }, { "epoch": 156.7175572519084, "grad_norm": 2.1421873569488525, "learning_rate": 0.0001, "loss": 0.0231, "step": 20530 }, { "epoch": 156.793893129771, "grad_norm": 1.6029245853424072, "learning_rate": 0.0001, "loss": 0.0223, "step": 20540 }, { "epoch": 156.87022900763358, "grad_norm": 1.8846198320388794, "learning_rate": 0.0001, "loss": 0.0226, "step": 20550 }, { "epoch": 156.94656488549617, "grad_norm": 2.1587564945220947, "learning_rate": 0.0001, "loss": 0.0231, "step": 20560 }, { "epoch": 157.0229007633588, "grad_norm": 1.8997907638549805, "learning_rate": 0.0001, "loss": 0.0224, "step": 20570 }, { "epoch": 157.09923664122138, "grad_norm": 1.443826675415039, "learning_rate": 0.0001, "loss": 0.022, "step": 20580 }, { "epoch": 157.17557251908397, "grad_norm": 2.1900951862335205, "learning_rate": 0.0001, "loss": 0.0221, "step": 20590 }, { "epoch": 157.25190839694656, "grad_norm": 1.265499472618103, "learning_rate": 0.0001, "loss": 0.0224, "step": 20600 }, { "epoch": 157.32824427480915, "grad_norm": 2.348393678665161, "learning_rate": 0.0001, "loss": 0.0232, "step": 20610 }, { "epoch": 157.40458015267177, "grad_norm": 1.51902174949646, "learning_rate": 0.0001, "loss": 0.0237, "step": 20620 }, { "epoch": 157.48091603053436, "grad_norm": 2.2306675910949707, "learning_rate": 0.0001, "loss": 0.0231, "step": 20630 }, { "epoch": 157.55725190839695, "grad_norm": 2.557908773422241, "learning_rate": 0.0001, "loss": 0.0237, "step": 20640 }, { "epoch": 157.63358778625954, "grad_norm": 1.6571712493896484, "learning_rate": 0.0001, "loss": 0.023, "step": 20650 }, { "epoch": 157.70992366412213, "grad_norm": 1.6288750171661377, "learning_rate": 0.0001, "loss": 0.0225, "step": 20660 }, { "epoch": 157.78625954198472, "grad_norm": 1.533493161201477, "learning_rate": 0.0001, "loss": 0.0225, "step": 20670 }, { "epoch": 157.86259541984734, "grad_norm": 2.320960760116577, "learning_rate": 0.0001, "loss": 0.0224, "step": 20680 }, { "epoch": 157.93893129770993, "grad_norm": 1.8434555530548096, "learning_rate": 0.0001, "loss": 0.023, "step": 20690 }, { "epoch": 158.01526717557252, "grad_norm": 1.495469570159912, "learning_rate": 0.0001, "loss": 0.0223, "step": 20700 }, { "epoch": 158.0916030534351, "grad_norm": 1.4455885887145996, "learning_rate": 0.0001, "loss": 0.0214, "step": 20710 }, { "epoch": 158.1679389312977, "grad_norm": 1.5085335969924927, "learning_rate": 0.0001, "loss": 0.0218, "step": 20720 }, { "epoch": 158.24427480916032, "grad_norm": 1.5770752429962158, "learning_rate": 0.0001, "loss": 0.022, "step": 20730 }, { "epoch": 158.3206106870229, "grad_norm": 1.6041951179504395, "learning_rate": 0.0001, "loss": 0.0225, "step": 20740 }, { "epoch": 158.3969465648855, "grad_norm": 1.7670633792877197, "learning_rate": 0.0001, "loss": 0.0225, "step": 20750 }, { "epoch": 158.4732824427481, "grad_norm": 2.1739494800567627, "learning_rate": 0.0001, "loss": 0.0225, "step": 20760 }, { "epoch": 158.54961832061068, "grad_norm": 1.4148677587509155, "learning_rate": 0.0001, "loss": 0.0231, "step": 20770 }, { "epoch": 158.6259541984733, "grad_norm": 1.540802001953125, "learning_rate": 0.0001, "loss": 0.0227, "step": 20780 }, { "epoch": 158.70229007633588, "grad_norm": 1.977429986000061, "learning_rate": 0.0001, "loss": 0.0223, "step": 20790 }, { "epoch": 158.77862595419847, "grad_norm": 2.5174083709716797, "learning_rate": 0.0001, "loss": 0.0221, "step": 20800 }, { "epoch": 158.85496183206106, "grad_norm": 1.3660293817520142, "learning_rate": 0.0001, "loss": 0.0217, "step": 20810 }, { "epoch": 158.93129770992365, "grad_norm": 2.028463125228882, "learning_rate": 0.0001, "loss": 0.0222, "step": 20820 }, { "epoch": 159.00763358778627, "grad_norm": 1.826317310333252, "learning_rate": 0.0001, "loss": 0.0219, "step": 20830 }, { "epoch": 159.08396946564886, "grad_norm": 1.1808154582977295, "learning_rate": 0.0001, "loss": 0.0219, "step": 20840 }, { "epoch": 159.16030534351145, "grad_norm": 1.8712854385375977, "learning_rate": 0.0001, "loss": 0.0225, "step": 20850 }, { "epoch": 159.23664122137404, "grad_norm": 1.5041232109069824, "learning_rate": 0.0001, "loss": 0.0227, "step": 20860 }, { "epoch": 159.31297709923663, "grad_norm": 2.377897262573242, "learning_rate": 0.0001, "loss": 0.0225, "step": 20870 }, { "epoch": 159.38931297709922, "grad_norm": 1.590222716331482, "learning_rate": 0.0001, "loss": 0.0218, "step": 20880 }, { "epoch": 159.46564885496184, "grad_norm": 2.4272360801696777, "learning_rate": 0.0001, "loss": 0.0225, "step": 20890 }, { "epoch": 159.54198473282443, "grad_norm": 1.7964502573013306, "learning_rate": 0.0001, "loss": 0.0223, "step": 20900 }, { "epoch": 159.61832061068702, "grad_norm": 1.7343199253082275, "learning_rate": 0.0001, "loss": 0.0224, "step": 20910 }, { "epoch": 159.6946564885496, "grad_norm": 1.8019856214523315, "learning_rate": 0.0001, "loss": 0.0231, "step": 20920 }, { "epoch": 159.7709923664122, "grad_norm": 2.7756104469299316, "learning_rate": 0.0001, "loss": 0.0227, "step": 20930 }, { "epoch": 159.84732824427482, "grad_norm": 2.416621685028076, "learning_rate": 0.0001, "loss": 0.0227, "step": 20940 }, { "epoch": 159.9236641221374, "grad_norm": 1.1915258169174194, "learning_rate": 0.0001, "loss": 0.0229, "step": 20950 }, { "epoch": 160.0, "grad_norm": 2.0961689949035645, "learning_rate": 0.0001, "loss": 0.0223, "step": 20960 }, { "epoch": 160.0763358778626, "grad_norm": 1.644853949546814, "learning_rate": 0.0001, "loss": 0.0226, "step": 20970 }, { "epoch": 160.15267175572518, "grad_norm": 1.400006890296936, "learning_rate": 0.0001, "loss": 0.0226, "step": 20980 }, { "epoch": 160.2290076335878, "grad_norm": 1.3436979055404663, "learning_rate": 0.0001, "loss": 0.0225, "step": 20990 }, { "epoch": 160.3053435114504, "grad_norm": 1.8488281965255737, "learning_rate": 0.0001, "loss": 0.0228, "step": 21000 }, { "epoch": 160.38167938931298, "grad_norm": 1.4491729736328125, "learning_rate": 0.0001, "loss": 0.0219, "step": 21010 }, { "epoch": 160.45801526717557, "grad_norm": 2.552615165710449, "learning_rate": 0.0001, "loss": 0.0227, "step": 21020 }, { "epoch": 160.53435114503816, "grad_norm": 1.6306467056274414, "learning_rate": 0.0001, "loss": 0.0229, "step": 21030 }, { "epoch": 160.61068702290078, "grad_norm": 2.7090377807617188, "learning_rate": 0.0001, "loss": 0.0227, "step": 21040 }, { "epoch": 160.68702290076337, "grad_norm": 2.738842010498047, "learning_rate": 0.0001, "loss": 0.0225, "step": 21050 }, { "epoch": 160.76335877862596, "grad_norm": 2.80767822265625, "learning_rate": 0.0001, "loss": 0.0239, "step": 21060 }, { "epoch": 160.83969465648855, "grad_norm": 2.1634647846221924, "learning_rate": 0.0001, "loss": 0.0229, "step": 21070 }, { "epoch": 160.91603053435114, "grad_norm": 1.7446473836898804, "learning_rate": 0.0001, "loss": 0.0222, "step": 21080 }, { "epoch": 160.99236641221373, "grad_norm": 1.482653260231018, "learning_rate": 0.0001, "loss": 0.0221, "step": 21090 }, { "epoch": 161.06870229007635, "grad_norm": 1.5611882209777832, "learning_rate": 0.0001, "loss": 0.0216, "step": 21100 }, { "epoch": 161.14503816793894, "grad_norm": 1.810501217842102, "learning_rate": 0.0001, "loss": 0.0222, "step": 21110 }, { "epoch": 161.22137404580153, "grad_norm": 2.6905362606048584, "learning_rate": 0.0001, "loss": 0.0231, "step": 21120 }, { "epoch": 161.29770992366412, "grad_norm": 1.4370319843292236, "learning_rate": 0.0001, "loss": 0.022, "step": 21130 }, { "epoch": 161.3740458015267, "grad_norm": 1.767755389213562, "learning_rate": 0.0001, "loss": 0.022, "step": 21140 }, { "epoch": 161.45038167938932, "grad_norm": 2.2656195163726807, "learning_rate": 0.0001, "loss": 0.0235, "step": 21150 }, { "epoch": 161.5267175572519, "grad_norm": 2.036336660385132, "learning_rate": 0.0001, "loss": 0.0233, "step": 21160 }, { "epoch": 161.6030534351145, "grad_norm": 2.494558334350586, "learning_rate": 0.0001, "loss": 0.0226, "step": 21170 }, { "epoch": 161.6793893129771, "grad_norm": 1.7989283800125122, "learning_rate": 0.0001, "loss": 0.0223, "step": 21180 }, { "epoch": 161.75572519083968, "grad_norm": 2.760239839553833, "learning_rate": 0.0001, "loss": 0.0225, "step": 21190 }, { "epoch": 161.8320610687023, "grad_norm": 2.308076858520508, "learning_rate": 0.0001, "loss": 0.023, "step": 21200 }, { "epoch": 161.9083969465649, "grad_norm": 1.7670156955718994, "learning_rate": 0.0001, "loss": 0.022, "step": 21210 }, { "epoch": 161.98473282442748, "grad_norm": 2.290025472640991, "learning_rate": 0.0001, "loss": 0.0225, "step": 21220 }, { "epoch": 162.06106870229007, "grad_norm": 1.7133904695510864, "learning_rate": 0.0001, "loss": 0.0228, "step": 21230 }, { "epoch": 162.13740458015266, "grad_norm": 1.542707920074463, "learning_rate": 0.0001, "loss": 0.0224, "step": 21240 }, { "epoch": 162.21374045801528, "grad_norm": 1.9998873472213745, "learning_rate": 0.0001, "loss": 0.0224, "step": 21250 }, { "epoch": 162.29007633587787, "grad_norm": 1.7766517400741577, "learning_rate": 0.0001, "loss": 0.0224, "step": 21260 }, { "epoch": 162.36641221374046, "grad_norm": 2.1525485515594482, "learning_rate": 0.0001, "loss": 0.023, "step": 21270 }, { "epoch": 162.44274809160305, "grad_norm": 1.7814714908599854, "learning_rate": 0.0001, "loss": 0.0231, "step": 21280 }, { "epoch": 162.51908396946564, "grad_norm": 2.245404005050659, "learning_rate": 0.0001, "loss": 0.0232, "step": 21290 }, { "epoch": 162.59541984732823, "grad_norm": 2.645280361175537, "learning_rate": 0.0001, "loss": 0.0226, "step": 21300 }, { "epoch": 162.67175572519085, "grad_norm": 1.35831880569458, "learning_rate": 0.0001, "loss": 0.0223, "step": 21310 }, { "epoch": 162.74809160305344, "grad_norm": 1.8595210313796997, "learning_rate": 0.0001, "loss": 0.0224, "step": 21320 }, { "epoch": 162.82442748091603, "grad_norm": 2.0239992141723633, "learning_rate": 0.0001, "loss": 0.0221, "step": 21330 }, { "epoch": 162.90076335877862, "grad_norm": 1.6626815795898438, "learning_rate": 0.0001, "loss": 0.0224, "step": 21340 }, { "epoch": 162.9770992366412, "grad_norm": 1.6344040632247925, "learning_rate": 0.0001, "loss": 0.0218, "step": 21350 }, { "epoch": 163.05343511450383, "grad_norm": 1.9138458967208862, "learning_rate": 0.0001, "loss": 0.0219, "step": 21360 }, { "epoch": 163.12977099236642, "grad_norm": 2.348830223083496, "learning_rate": 0.0001, "loss": 0.0219, "step": 21370 }, { "epoch": 163.206106870229, "grad_norm": 1.7952173948287964, "learning_rate": 0.0001, "loss": 0.0217, "step": 21380 }, { "epoch": 163.2824427480916, "grad_norm": 2.696326732635498, "learning_rate": 0.0001, "loss": 0.0216, "step": 21390 }, { "epoch": 163.3587786259542, "grad_norm": 1.8900479078292847, "learning_rate": 0.0001, "loss": 0.0212, "step": 21400 }, { "epoch": 163.4351145038168, "grad_norm": 1.844935417175293, "learning_rate": 0.0001, "loss": 0.0218, "step": 21410 }, { "epoch": 163.5114503816794, "grad_norm": 1.733061671257019, "learning_rate": 0.0001, "loss": 0.0213, "step": 21420 }, { "epoch": 163.58778625954199, "grad_norm": 1.7184141874313354, "learning_rate": 0.0001, "loss": 0.0205, "step": 21430 }, { "epoch": 163.66412213740458, "grad_norm": 2.2622625827789307, "learning_rate": 0.0001, "loss": 0.0219, "step": 21440 }, { "epoch": 163.74045801526717, "grad_norm": 1.667421579360962, "learning_rate": 0.0001, "loss": 0.0217, "step": 21450 }, { "epoch": 163.81679389312978, "grad_norm": 1.911989688873291, "learning_rate": 0.0001, "loss": 0.0217, "step": 21460 }, { "epoch": 163.89312977099237, "grad_norm": 2.333657741546631, "learning_rate": 0.0001, "loss": 0.0219, "step": 21470 }, { "epoch": 163.96946564885496, "grad_norm": 1.6159148216247559, "learning_rate": 0.0001, "loss": 0.0219, "step": 21480 }, { "epoch": 164.04580152671755, "grad_norm": 2.282236099243164, "learning_rate": 0.0001, "loss": 0.0223, "step": 21490 }, { "epoch": 164.12213740458014, "grad_norm": 1.9245388507843018, "learning_rate": 0.0001, "loss": 0.0226, "step": 21500 }, { "epoch": 164.19847328244273, "grad_norm": 1.9243812561035156, "learning_rate": 0.0001, "loss": 0.0226, "step": 21510 }, { "epoch": 164.27480916030535, "grad_norm": 1.4781380891799927, "learning_rate": 0.0001, "loss": 0.0227, "step": 21520 }, { "epoch": 164.35114503816794, "grad_norm": 1.6523112058639526, "learning_rate": 0.0001, "loss": 0.0226, "step": 21530 }, { "epoch": 164.42748091603053, "grad_norm": 2.3633415699005127, "learning_rate": 0.0001, "loss": 0.0231, "step": 21540 }, { "epoch": 164.50381679389312, "grad_norm": 2.046403646469116, "learning_rate": 0.0001, "loss": 0.0237, "step": 21550 }, { "epoch": 164.5801526717557, "grad_norm": 2.2380237579345703, "learning_rate": 0.0001, "loss": 0.0238, "step": 21560 }, { "epoch": 164.65648854961833, "grad_norm": 2.1071581840515137, "learning_rate": 0.0001, "loss": 0.0228, "step": 21570 }, { "epoch": 164.73282442748092, "grad_norm": 1.3401384353637695, "learning_rate": 0.0001, "loss": 0.0224, "step": 21580 }, { "epoch": 164.8091603053435, "grad_norm": 1.095198392868042, "learning_rate": 0.0001, "loss": 0.022, "step": 21590 }, { "epoch": 164.8854961832061, "grad_norm": 1.9901129007339478, "learning_rate": 0.0001, "loss": 0.0227, "step": 21600 }, { "epoch": 164.9618320610687, "grad_norm": 2.3970062732696533, "learning_rate": 0.0001, "loss": 0.0248, "step": 21610 }, { "epoch": 165.0381679389313, "grad_norm": 1.7832609415054321, "learning_rate": 0.0001, "loss": 0.0239, "step": 21620 }, { "epoch": 165.1145038167939, "grad_norm": 2.1922905445098877, "learning_rate": 0.0001, "loss": 0.024, "step": 21630 }, { "epoch": 165.1908396946565, "grad_norm": 1.3919329643249512, "learning_rate": 0.0001, "loss": 0.0241, "step": 21640 }, { "epoch": 165.26717557251908, "grad_norm": 2.056129217147827, "learning_rate": 0.0001, "loss": 0.0241, "step": 21650 }, { "epoch": 165.34351145038167, "grad_norm": 1.7111437320709229, "learning_rate": 0.0001, "loss": 0.023, "step": 21660 }, { "epoch": 165.4198473282443, "grad_norm": 1.5150258541107178, "learning_rate": 0.0001, "loss": 0.022, "step": 21670 }, { "epoch": 165.49618320610688, "grad_norm": 1.390462875366211, "learning_rate": 0.0001, "loss": 0.0226, "step": 21680 }, { "epoch": 165.57251908396947, "grad_norm": 1.2673635482788086, "learning_rate": 0.0001, "loss": 0.0221, "step": 21690 }, { "epoch": 165.64885496183206, "grad_norm": 2.0648419857025146, "learning_rate": 0.0001, "loss": 0.022, "step": 21700 }, { "epoch": 165.72519083969465, "grad_norm": 1.5635472536087036, "learning_rate": 0.0001, "loss": 0.022, "step": 21710 }, { "epoch": 165.80152671755727, "grad_norm": 1.3016839027404785, "learning_rate": 0.0001, "loss": 0.0222, "step": 21720 }, { "epoch": 165.87786259541986, "grad_norm": 1.7816557884216309, "learning_rate": 0.0001, "loss": 0.0228, "step": 21730 }, { "epoch": 165.95419847328245, "grad_norm": 1.739176630973816, "learning_rate": 0.0001, "loss": 0.0227, "step": 21740 }, { "epoch": 166.03053435114504, "grad_norm": 1.503270149230957, "learning_rate": 0.0001, "loss": 0.0233, "step": 21750 }, { "epoch": 166.10687022900763, "grad_norm": 2.0339412689208984, "learning_rate": 0.0001, "loss": 0.023, "step": 21760 }, { "epoch": 166.18320610687022, "grad_norm": 1.8399406671524048, "learning_rate": 0.0001, "loss": 0.023, "step": 21770 }, { "epoch": 166.25954198473283, "grad_norm": 1.8863929510116577, "learning_rate": 0.0001, "loss": 0.022, "step": 21780 }, { "epoch": 166.33587786259542, "grad_norm": 2.3053321838378906, "learning_rate": 0.0001, "loss": 0.0215, "step": 21790 }, { "epoch": 166.41221374045801, "grad_norm": 2.5267534255981445, "learning_rate": 0.0001, "loss": 0.022, "step": 21800 }, { "epoch": 166.4885496183206, "grad_norm": 2.394163131713867, "learning_rate": 0.0001, "loss": 0.0222, "step": 21810 }, { "epoch": 166.5648854961832, "grad_norm": 3.061138391494751, "learning_rate": 0.0001, "loss": 0.0213, "step": 21820 }, { "epoch": 166.6412213740458, "grad_norm": 1.828086256980896, "learning_rate": 0.0001, "loss": 0.023, "step": 21830 }, { "epoch": 166.7175572519084, "grad_norm": 1.8556013107299805, "learning_rate": 0.0001, "loss": 0.0223, "step": 21840 }, { "epoch": 166.793893129771, "grad_norm": 1.7313706874847412, "learning_rate": 0.0001, "loss": 0.0216, "step": 21850 }, { "epoch": 166.87022900763358, "grad_norm": 1.4094111919403076, "learning_rate": 0.0001, "loss": 0.0227, "step": 21860 }, { "epoch": 166.94656488549617, "grad_norm": 1.6860069036483765, "learning_rate": 0.0001, "loss": 0.0224, "step": 21870 }, { "epoch": 167.0229007633588, "grad_norm": 1.41201913356781, "learning_rate": 0.0001, "loss": 0.0224, "step": 21880 }, { "epoch": 167.09923664122138, "grad_norm": 1.3667361736297607, "learning_rate": 0.0001, "loss": 0.0219, "step": 21890 }, { "epoch": 167.17557251908397, "grad_norm": 1.8994598388671875, "learning_rate": 0.0001, "loss": 0.0222, "step": 21900 }, { "epoch": 167.25190839694656, "grad_norm": 2.6653950214385986, "learning_rate": 0.0001, "loss": 0.023, "step": 21910 }, { "epoch": 167.32824427480915, "grad_norm": 2.0293736457824707, "learning_rate": 0.0001, "loss": 0.0234, "step": 21920 }, { "epoch": 167.40458015267177, "grad_norm": 1.954944372177124, "learning_rate": 0.0001, "loss": 0.0227, "step": 21930 }, { "epoch": 167.48091603053436, "grad_norm": 2.2231149673461914, "learning_rate": 0.0001, "loss": 0.0225, "step": 21940 }, { "epoch": 167.55725190839695, "grad_norm": 1.1785309314727783, "learning_rate": 0.0001, "loss": 0.022, "step": 21950 }, { "epoch": 167.63358778625954, "grad_norm": 1.629575490951538, "learning_rate": 0.0001, "loss": 0.0223, "step": 21960 }, { "epoch": 167.70992366412213, "grad_norm": 2.1284749507904053, "learning_rate": 0.0001, "loss": 0.023, "step": 21970 }, { "epoch": 167.78625954198472, "grad_norm": 1.9314961433410645, "learning_rate": 0.0001, "loss": 0.0239, "step": 21980 }, { "epoch": 167.86259541984734, "grad_norm": 2.5389418601989746, "learning_rate": 0.0001, "loss": 0.0244, "step": 21990 }, { "epoch": 167.93893129770993, "grad_norm": 2.509272336959839, "learning_rate": 0.0001, "loss": 0.0248, "step": 22000 }, { "epoch": 168.01526717557252, "grad_norm": 1.6489572525024414, "learning_rate": 0.0001, "loss": 0.0242, "step": 22010 }, { "epoch": 168.0916030534351, "grad_norm": 1.427587628364563, "learning_rate": 0.0001, "loss": 0.0243, "step": 22020 }, { "epoch": 168.1679389312977, "grad_norm": 1.9076489210128784, "learning_rate": 0.0001, "loss": 0.0243, "step": 22030 }, { "epoch": 168.24427480916032, "grad_norm": 1.904399037361145, "learning_rate": 0.0001, "loss": 0.0235, "step": 22040 }, { "epoch": 168.3206106870229, "grad_norm": 1.1019223928451538, "learning_rate": 0.0001, "loss": 0.0236, "step": 22050 }, { "epoch": 168.3969465648855, "grad_norm": 1.6104487180709839, "learning_rate": 0.0001, "loss": 0.0233, "step": 22060 }, { "epoch": 168.4732824427481, "grad_norm": 2.576910972595215, "learning_rate": 0.0001, "loss": 0.023, "step": 22070 }, { "epoch": 168.54961832061068, "grad_norm": 1.751979112625122, "learning_rate": 0.0001, "loss": 0.0228, "step": 22080 }, { "epoch": 168.6259541984733, "grad_norm": 2.6988778114318848, "learning_rate": 0.0001, "loss": 0.0229, "step": 22090 }, { "epoch": 168.70229007633588, "grad_norm": 1.7181799411773682, "learning_rate": 0.0001, "loss": 0.0225, "step": 22100 }, { "epoch": 168.77862595419847, "grad_norm": 1.520654559135437, "learning_rate": 0.0001, "loss": 0.0225, "step": 22110 }, { "epoch": 168.85496183206106, "grad_norm": 1.7298856973648071, "learning_rate": 0.0001, "loss": 0.0214, "step": 22120 }, { "epoch": 168.93129770992365, "grad_norm": 1.430615782737732, "learning_rate": 0.0001, "loss": 0.0213, "step": 22130 }, { "epoch": 169.00763358778627, "grad_norm": 1.4948440790176392, "learning_rate": 0.0001, "loss": 0.0217, "step": 22140 }, { "epoch": 169.08396946564886, "grad_norm": 1.4207515716552734, "learning_rate": 0.0001, "loss": 0.022, "step": 22150 }, { "epoch": 169.16030534351145, "grad_norm": 1.6001836061477661, "learning_rate": 0.0001, "loss": 0.0238, "step": 22160 }, { "epoch": 169.23664122137404, "grad_norm": 2.164001703262329, "learning_rate": 0.0001, "loss": 0.0235, "step": 22170 }, { "epoch": 169.31297709923663, "grad_norm": 1.8282979726791382, "learning_rate": 0.0001, "loss": 0.0229, "step": 22180 }, { "epoch": 169.38931297709922, "grad_norm": 1.739997386932373, "learning_rate": 0.0001, "loss": 0.0224, "step": 22190 }, { "epoch": 169.46564885496184, "grad_norm": 1.5654529333114624, "learning_rate": 0.0001, "loss": 0.0222, "step": 22200 }, { "epoch": 169.54198473282443, "grad_norm": 1.3176194429397583, "learning_rate": 0.0001, "loss": 0.022, "step": 22210 }, { "epoch": 169.61832061068702, "grad_norm": 1.371198296546936, "learning_rate": 0.0001, "loss": 0.0223, "step": 22220 }, { "epoch": 169.6946564885496, "grad_norm": 2.4342150688171387, "learning_rate": 0.0001, "loss": 0.022, "step": 22230 }, { "epoch": 169.7709923664122, "grad_norm": 1.6372140645980835, "learning_rate": 0.0001, "loss": 0.0221, "step": 22240 }, { "epoch": 169.84732824427482, "grad_norm": 1.6444579362869263, "learning_rate": 0.0001, "loss": 0.0213, "step": 22250 }, { "epoch": 169.9236641221374, "grad_norm": 2.4476587772369385, "learning_rate": 0.0001, "loss": 0.0216, "step": 22260 }, { "epoch": 170.0, "grad_norm": 1.7223793268203735, "learning_rate": 0.0001, "loss": 0.0213, "step": 22270 }, { "epoch": 170.0763358778626, "grad_norm": 2.118220329284668, "learning_rate": 0.0001, "loss": 0.0217, "step": 22280 }, { "epoch": 170.15267175572518, "grad_norm": 1.9058126211166382, "learning_rate": 0.0001, "loss": 0.0216, "step": 22290 }, { "epoch": 170.2290076335878, "grad_norm": 1.134217619895935, "learning_rate": 0.0001, "loss": 0.0217, "step": 22300 }, { "epoch": 170.3053435114504, "grad_norm": 2.00246262550354, "learning_rate": 0.0001, "loss": 0.0221, "step": 22310 }, { "epoch": 170.38167938931298, "grad_norm": 1.365354061126709, "learning_rate": 0.0001, "loss": 0.022, "step": 22320 }, { "epoch": 170.45801526717557, "grad_norm": 1.4453214406967163, "learning_rate": 0.0001, "loss": 0.0216, "step": 22330 }, { "epoch": 170.53435114503816, "grad_norm": 2.0261266231536865, "learning_rate": 0.0001, "loss": 0.0227, "step": 22340 }, { "epoch": 170.61068702290078, "grad_norm": 1.546951174736023, "learning_rate": 0.0001, "loss": 0.0223, "step": 22350 }, { "epoch": 170.68702290076337, "grad_norm": 1.628455400466919, "learning_rate": 0.0001, "loss": 0.0213, "step": 22360 }, { "epoch": 170.76335877862596, "grad_norm": 1.4186100959777832, "learning_rate": 0.0001, "loss": 0.0221, "step": 22370 }, { "epoch": 170.83969465648855, "grad_norm": 1.2605143785476685, "learning_rate": 0.0001, "loss": 0.0224, "step": 22380 }, { "epoch": 170.91603053435114, "grad_norm": 1.5181063413619995, "learning_rate": 0.0001, "loss": 0.0233, "step": 22390 }, { "epoch": 170.99236641221373, "grad_norm": 2.6096911430358887, "learning_rate": 0.0001, "loss": 0.0227, "step": 22400 }, { "epoch": 171.06870229007635, "grad_norm": 2.668086051940918, "learning_rate": 0.0001, "loss": 0.0225, "step": 22410 }, { "epoch": 171.14503816793894, "grad_norm": 1.961531400680542, "learning_rate": 0.0001, "loss": 0.0214, "step": 22420 }, { "epoch": 171.22137404580153, "grad_norm": 1.5727512836456299, "learning_rate": 0.0001, "loss": 0.0212, "step": 22430 }, { "epoch": 171.29770992366412, "grad_norm": 1.389510989189148, "learning_rate": 0.0001, "loss": 0.0212, "step": 22440 }, { "epoch": 171.3740458015267, "grad_norm": 1.7460689544677734, "learning_rate": 0.0001, "loss": 0.0204, "step": 22450 }, { "epoch": 171.45038167938932, "grad_norm": 1.6446739435195923, "learning_rate": 0.0001, "loss": 0.0211, "step": 22460 }, { "epoch": 171.5267175572519, "grad_norm": 1.565048098564148, "learning_rate": 0.0001, "loss": 0.0217, "step": 22470 }, { "epoch": 171.6030534351145, "grad_norm": 1.8469401597976685, "learning_rate": 0.0001, "loss": 0.0215, "step": 22480 }, { "epoch": 171.6793893129771, "grad_norm": 1.573217511177063, "learning_rate": 0.0001, "loss": 0.0217, "step": 22490 }, { "epoch": 171.75572519083968, "grad_norm": 1.511260747909546, "learning_rate": 0.0001, "loss": 0.0212, "step": 22500 }, { "epoch": 171.8320610687023, "grad_norm": 1.7348198890686035, "learning_rate": 0.0001, "loss": 0.022, "step": 22510 }, { "epoch": 171.9083969465649, "grad_norm": 1.547853946685791, "learning_rate": 0.0001, "loss": 0.0224, "step": 22520 }, { "epoch": 171.98473282442748, "grad_norm": 2.024946451187134, "learning_rate": 0.0001, "loss": 0.0224, "step": 22530 }, { "epoch": 172.06106870229007, "grad_norm": 1.4796260595321655, "learning_rate": 0.0001, "loss": 0.0218, "step": 22540 }, { "epoch": 172.13740458015266, "grad_norm": 1.6689313650131226, "learning_rate": 0.0001, "loss": 0.0219, "step": 22550 }, { "epoch": 172.21374045801528, "grad_norm": 1.9619717597961426, "learning_rate": 0.0001, "loss": 0.0224, "step": 22560 }, { "epoch": 172.29007633587787, "grad_norm": 2.747021436691284, "learning_rate": 0.0001, "loss": 0.0228, "step": 22570 }, { "epoch": 172.36641221374046, "grad_norm": 2.550711154937744, "learning_rate": 0.0001, "loss": 0.0223, "step": 22580 }, { "epoch": 172.44274809160305, "grad_norm": 1.4402483701705933, "learning_rate": 0.0001, "loss": 0.022, "step": 22590 }, { "epoch": 172.51908396946564, "grad_norm": 2.287869930267334, "learning_rate": 0.0001, "loss": 0.0225, "step": 22600 }, { "epoch": 172.59541984732823, "grad_norm": 2.5432255268096924, "learning_rate": 0.0001, "loss": 0.0219, "step": 22610 }, { "epoch": 172.67175572519085, "grad_norm": 2.0096254348754883, "learning_rate": 0.0001, "loss": 0.0214, "step": 22620 }, { "epoch": 172.74809160305344, "grad_norm": 1.5181543827056885, "learning_rate": 0.0001, "loss": 0.0216, "step": 22630 }, { "epoch": 172.82442748091603, "grad_norm": 1.4245482683181763, "learning_rate": 0.0001, "loss": 0.0215, "step": 22640 }, { "epoch": 172.90076335877862, "grad_norm": 1.5430949926376343, "learning_rate": 0.0001, "loss": 0.0216, "step": 22650 }, { "epoch": 172.9770992366412, "grad_norm": 1.6375384330749512, "learning_rate": 0.0001, "loss": 0.0219, "step": 22660 }, { "epoch": 173.05343511450383, "grad_norm": 1.7349551916122437, "learning_rate": 0.0001, "loss": 0.0212, "step": 22670 }, { "epoch": 173.12977099236642, "grad_norm": 2.2003519535064697, "learning_rate": 0.0001, "loss": 0.0203, "step": 22680 }, { "epoch": 173.206106870229, "grad_norm": 1.6164220571517944, "learning_rate": 0.0001, "loss": 0.0216, "step": 22690 }, { "epoch": 173.2824427480916, "grad_norm": 2.0463509559631348, "learning_rate": 0.0001, "loss": 0.0223, "step": 22700 }, { "epoch": 173.3587786259542, "grad_norm": 1.869471788406372, "learning_rate": 0.0001, "loss": 0.0219, "step": 22710 }, { "epoch": 173.4351145038168, "grad_norm": 2.151671886444092, "learning_rate": 0.0001, "loss": 0.0214, "step": 22720 }, { "epoch": 173.5114503816794, "grad_norm": 2.2877933979034424, "learning_rate": 0.0001, "loss": 0.0216, "step": 22730 }, { "epoch": 173.58778625954199, "grad_norm": 1.6533466577529907, "learning_rate": 0.0001, "loss": 0.0225, "step": 22740 }, { "epoch": 173.66412213740458, "grad_norm": 1.9397984743118286, "learning_rate": 0.0001, "loss": 0.0221, "step": 22750 }, { "epoch": 173.74045801526717, "grad_norm": 2.33534574508667, "learning_rate": 0.0001, "loss": 0.0223, "step": 22760 }, { "epoch": 173.81679389312978, "grad_norm": 2.064439058303833, "learning_rate": 0.0001, "loss": 0.0218, "step": 22770 }, { "epoch": 173.89312977099237, "grad_norm": 1.972825527191162, "learning_rate": 0.0001, "loss": 0.0222, "step": 22780 }, { "epoch": 173.96946564885496, "grad_norm": 2.2120039463043213, "learning_rate": 0.0001, "loss": 0.0219, "step": 22790 }, { "epoch": 174.04580152671755, "grad_norm": 1.8022205829620361, "learning_rate": 0.0001, "loss": 0.021, "step": 22800 }, { "epoch": 174.12213740458014, "grad_norm": 1.8382627964019775, "learning_rate": 0.0001, "loss": 0.0211, "step": 22810 }, { "epoch": 174.19847328244273, "grad_norm": 1.8781518936157227, "learning_rate": 0.0001, "loss": 0.0212, "step": 22820 }, { "epoch": 174.27480916030535, "grad_norm": 2.3066890239715576, "learning_rate": 0.0001, "loss": 0.0212, "step": 22830 }, { "epoch": 174.35114503816794, "grad_norm": 1.4360506534576416, "learning_rate": 0.0001, "loss": 0.0206, "step": 22840 }, { "epoch": 174.42748091603053, "grad_norm": 1.4023793935775757, "learning_rate": 0.0001, "loss": 0.021, "step": 22850 }, { "epoch": 174.50381679389312, "grad_norm": 2.0375921726226807, "learning_rate": 0.0001, "loss": 0.0212, "step": 22860 }, { "epoch": 174.5801526717557, "grad_norm": 1.5942081212997437, "learning_rate": 0.0001, "loss": 0.0218, "step": 22870 }, { "epoch": 174.65648854961833, "grad_norm": 1.7026386260986328, "learning_rate": 0.0001, "loss": 0.0214, "step": 22880 }, { "epoch": 174.73282442748092, "grad_norm": 2.544931650161743, "learning_rate": 0.0001, "loss": 0.0219, "step": 22890 }, { "epoch": 174.8091603053435, "grad_norm": 1.5849616527557373, "learning_rate": 0.0001, "loss": 0.022, "step": 22900 }, { "epoch": 174.8854961832061, "grad_norm": 1.585086464881897, "learning_rate": 0.0001, "loss": 0.0217, "step": 22910 }, { "epoch": 174.9618320610687, "grad_norm": 2.366029739379883, "learning_rate": 0.0001, "loss": 0.0214, "step": 22920 }, { "epoch": 175.0381679389313, "grad_norm": 1.6543370485305786, "learning_rate": 0.0001, "loss": 0.0212, "step": 22930 }, { "epoch": 175.1145038167939, "grad_norm": 1.9989891052246094, "learning_rate": 0.0001, "loss": 0.0208, "step": 22940 }, { "epoch": 175.1908396946565, "grad_norm": 1.4155999422073364, "learning_rate": 0.0001, "loss": 0.0209, "step": 22950 }, { "epoch": 175.26717557251908, "grad_norm": 2.2338099479675293, "learning_rate": 0.0001, "loss": 0.021, "step": 22960 }, { "epoch": 175.34351145038167, "grad_norm": 2.3135874271392822, "learning_rate": 0.0001, "loss": 0.0213, "step": 22970 }, { "epoch": 175.4198473282443, "grad_norm": 2.8875181674957275, "learning_rate": 0.0001, "loss": 0.0214, "step": 22980 }, { "epoch": 175.49618320610688, "grad_norm": 1.6061732769012451, "learning_rate": 0.0001, "loss": 0.022, "step": 22990 }, { "epoch": 175.57251908396947, "grad_norm": 1.8339366912841797, "learning_rate": 0.0001, "loss": 0.0212, "step": 23000 }, { "epoch": 175.64885496183206, "grad_norm": 1.622044563293457, "learning_rate": 0.0001, "loss": 0.021, "step": 23010 }, { "epoch": 175.72519083969465, "grad_norm": 1.9814164638519287, "learning_rate": 0.0001, "loss": 0.0205, "step": 23020 }, { "epoch": 175.80152671755727, "grad_norm": 1.9272867441177368, "learning_rate": 0.0001, "loss": 0.0207, "step": 23030 }, { "epoch": 175.87786259541986, "grad_norm": 1.56826651096344, "learning_rate": 0.0001, "loss": 0.021, "step": 23040 }, { "epoch": 175.95419847328245, "grad_norm": 1.667075514793396, "learning_rate": 0.0001, "loss": 0.0208, "step": 23050 }, { "epoch": 176.03053435114504, "grad_norm": 2.202023983001709, "learning_rate": 0.0001, "loss": 0.0209, "step": 23060 }, { "epoch": 176.10687022900763, "grad_norm": 2.3043127059936523, "learning_rate": 0.0001, "loss": 0.022, "step": 23070 }, { "epoch": 176.18320610687022, "grad_norm": 1.4306933879852295, "learning_rate": 0.0001, "loss": 0.0215, "step": 23080 }, { "epoch": 176.25954198473283, "grad_norm": 1.4740246534347534, "learning_rate": 0.0001, "loss": 0.0212, "step": 23090 }, { "epoch": 176.33587786259542, "grad_norm": 1.8373068571090698, "learning_rate": 0.0001, "loss": 0.0213, "step": 23100 }, { "epoch": 176.41221374045801, "grad_norm": 2.69724178314209, "learning_rate": 0.0001, "loss": 0.0228, "step": 23110 }, { "epoch": 176.4885496183206, "grad_norm": 2.1299679279327393, "learning_rate": 0.0001, "loss": 0.0233, "step": 23120 }, { "epoch": 176.5648854961832, "grad_norm": 1.546917200088501, "learning_rate": 0.0001, "loss": 0.0223, "step": 23130 }, { "epoch": 176.6412213740458, "grad_norm": 1.3498526811599731, "learning_rate": 0.0001, "loss": 0.0219, "step": 23140 }, { "epoch": 176.7175572519084, "grad_norm": 2.324483633041382, "learning_rate": 0.0001, "loss": 0.0228, "step": 23150 }, { "epoch": 176.793893129771, "grad_norm": 2.326476812362671, "learning_rate": 0.0001, "loss": 0.0233, "step": 23160 }, { "epoch": 176.87022900763358, "grad_norm": 2.141373634338379, "learning_rate": 0.0001, "loss": 0.0236, "step": 23170 }, { "epoch": 176.94656488549617, "grad_norm": 2.209378957748413, "learning_rate": 0.0001, "loss": 0.0237, "step": 23180 }, { "epoch": 177.0229007633588, "grad_norm": 2.175377607345581, "learning_rate": 0.0001, "loss": 0.0227, "step": 23190 }, { "epoch": 177.09923664122138, "grad_norm": 1.8794366121292114, "learning_rate": 0.0001, "loss": 0.0226, "step": 23200 }, { "epoch": 177.17557251908397, "grad_norm": 1.5162209272384644, "learning_rate": 0.0001, "loss": 0.0222, "step": 23210 }, { "epoch": 177.25190839694656, "grad_norm": 1.9812172651290894, "learning_rate": 0.0001, "loss": 0.0221, "step": 23220 }, { "epoch": 177.32824427480915, "grad_norm": 1.625001072883606, "learning_rate": 0.0001, "loss": 0.0223, "step": 23230 }, { "epoch": 177.40458015267177, "grad_norm": 1.55943763256073, "learning_rate": 0.0001, "loss": 0.0226, "step": 23240 }, { "epoch": 177.48091603053436, "grad_norm": 1.9534077644348145, "learning_rate": 0.0001, "loss": 0.0225, "step": 23250 }, { "epoch": 177.55725190839695, "grad_norm": 1.4065423011779785, "learning_rate": 0.0001, "loss": 0.0211, "step": 23260 }, { "epoch": 177.63358778625954, "grad_norm": 1.3742746114730835, "learning_rate": 0.0001, "loss": 0.0209, "step": 23270 }, { "epoch": 177.70992366412213, "grad_norm": 2.6688413619995117, "learning_rate": 0.0001, "loss": 0.021, "step": 23280 }, { "epoch": 177.78625954198472, "grad_norm": 1.9326279163360596, "learning_rate": 0.0001, "loss": 0.022, "step": 23290 }, { "epoch": 177.86259541984734, "grad_norm": 1.9034980535507202, "learning_rate": 0.0001, "loss": 0.0215, "step": 23300 }, { "epoch": 177.93893129770993, "grad_norm": 1.8423033952713013, "learning_rate": 0.0001, "loss": 0.0217, "step": 23310 }, { "epoch": 178.01526717557252, "grad_norm": 1.480318307876587, "learning_rate": 0.0001, "loss": 0.0219, "step": 23320 }, { "epoch": 178.0916030534351, "grad_norm": 2.290743112564087, "learning_rate": 0.0001, "loss": 0.0217, "step": 23330 }, { "epoch": 178.1679389312977, "grad_norm": 1.7768945693969727, "learning_rate": 0.0001, "loss": 0.0216, "step": 23340 }, { "epoch": 178.24427480916032, "grad_norm": 1.2703638076782227, "learning_rate": 0.0001, "loss": 0.0213, "step": 23350 }, { "epoch": 178.3206106870229, "grad_norm": 2.2103254795074463, "learning_rate": 0.0001, "loss": 0.0211, "step": 23360 }, { "epoch": 178.3969465648855, "grad_norm": 2.2665324211120605, "learning_rate": 0.0001, "loss": 0.0226, "step": 23370 }, { "epoch": 178.4732824427481, "grad_norm": 1.7662855386734009, "learning_rate": 0.0001, "loss": 0.0228, "step": 23380 }, { "epoch": 178.54961832061068, "grad_norm": 1.8775403499603271, "learning_rate": 0.0001, "loss": 0.0223, "step": 23390 }, { "epoch": 178.6259541984733, "grad_norm": 2.603980541229248, "learning_rate": 0.0001, "loss": 0.0221, "step": 23400 }, { "epoch": 178.70229007633588, "grad_norm": 1.7662676572799683, "learning_rate": 0.0001, "loss": 0.0208, "step": 23410 }, { "epoch": 178.77862595419847, "grad_norm": 1.6137055158615112, "learning_rate": 0.0001, "loss": 0.0206, "step": 23420 }, { "epoch": 178.85496183206106, "grad_norm": 1.455832839012146, "learning_rate": 0.0001, "loss": 0.0203, "step": 23430 }, { "epoch": 178.93129770992365, "grad_norm": 2.500965118408203, "learning_rate": 0.0001, "loss": 0.0207, "step": 23440 }, { "epoch": 179.00763358778627, "grad_norm": 2.27103328704834, "learning_rate": 0.0001, "loss": 0.021, "step": 23450 }, { "epoch": 179.08396946564886, "grad_norm": 1.6171780824661255, "learning_rate": 0.0001, "loss": 0.0206, "step": 23460 }, { "epoch": 179.16030534351145, "grad_norm": 2.8489954471588135, "learning_rate": 0.0001, "loss": 0.0213, "step": 23470 }, { "epoch": 179.23664122137404, "grad_norm": 2.7722225189208984, "learning_rate": 0.0001, "loss": 0.0209, "step": 23480 }, { "epoch": 179.31297709923663, "grad_norm": 2.169916868209839, "learning_rate": 0.0001, "loss": 0.0204, "step": 23490 }, { "epoch": 179.38931297709922, "grad_norm": 1.378861427307129, "learning_rate": 0.0001, "loss": 0.0202, "step": 23500 }, { "epoch": 179.46564885496184, "grad_norm": 2.1137077808380127, "learning_rate": 0.0001, "loss": 0.0204, "step": 23510 }, { "epoch": 179.54198473282443, "grad_norm": 1.859334111213684, "learning_rate": 0.0001, "loss": 0.0207, "step": 23520 }, { "epoch": 179.61832061068702, "grad_norm": 1.510838508605957, "learning_rate": 0.0001, "loss": 0.0208, "step": 23530 }, { "epoch": 179.6946564885496, "grad_norm": 1.5819664001464844, "learning_rate": 0.0001, "loss": 0.0207, "step": 23540 }, { "epoch": 179.7709923664122, "grad_norm": 1.4531619548797607, "learning_rate": 0.0001, "loss": 0.0211, "step": 23550 }, { "epoch": 179.84732824427482, "grad_norm": 2.183025360107422, "learning_rate": 0.0001, "loss": 0.0213, "step": 23560 }, { "epoch": 179.9236641221374, "grad_norm": 2.865668773651123, "learning_rate": 0.0001, "loss": 0.0215, "step": 23570 }, { "epoch": 180.0, "grad_norm": 1.5833503007888794, "learning_rate": 0.0001, "loss": 0.0217, "step": 23580 }, { "epoch": 180.0763358778626, "grad_norm": 1.6836386919021606, "learning_rate": 0.0001, "loss": 0.0216, "step": 23590 }, { "epoch": 180.15267175572518, "grad_norm": 1.4694405794143677, "learning_rate": 0.0001, "loss": 0.0214, "step": 23600 }, { "epoch": 180.2290076335878, "grad_norm": 1.8470240831375122, "learning_rate": 0.0001, "loss": 0.022, "step": 23610 }, { "epoch": 180.3053435114504, "grad_norm": 1.85581374168396, "learning_rate": 0.0001, "loss": 0.0214, "step": 23620 }, { "epoch": 180.38167938931298, "grad_norm": 2.663604497909546, "learning_rate": 0.0001, "loss": 0.022, "step": 23630 }, { "epoch": 180.45801526717557, "grad_norm": 2.7016007900238037, "learning_rate": 0.0001, "loss": 0.0218, "step": 23640 }, { "epoch": 180.53435114503816, "grad_norm": 2.4148244857788086, "learning_rate": 0.0001, "loss": 0.0212, "step": 23650 }, { "epoch": 180.61068702290078, "grad_norm": 1.561735987663269, "learning_rate": 0.0001, "loss": 0.0213, "step": 23660 }, { "epoch": 180.68702290076337, "grad_norm": 1.6925123929977417, "learning_rate": 0.0001, "loss": 0.0214, "step": 23670 }, { "epoch": 180.76335877862596, "grad_norm": 1.4903180599212646, "learning_rate": 0.0001, "loss": 0.0218, "step": 23680 }, { "epoch": 180.83969465648855, "grad_norm": 2.0233962535858154, "learning_rate": 0.0001, "loss": 0.0211, "step": 23690 }, { "epoch": 180.91603053435114, "grad_norm": 1.3126225471496582, "learning_rate": 0.0001, "loss": 0.0217, "step": 23700 }, { "epoch": 180.99236641221373, "grad_norm": 1.6055541038513184, "learning_rate": 0.0001, "loss": 0.0214, "step": 23710 }, { "epoch": 181.06870229007635, "grad_norm": 2.7059760093688965, "learning_rate": 0.0001, "loss": 0.0216, "step": 23720 }, { "epoch": 181.14503816793894, "grad_norm": 1.8951843976974487, "learning_rate": 0.0001, "loss": 0.0211, "step": 23730 }, { "epoch": 181.22137404580153, "grad_norm": 2.5263521671295166, "learning_rate": 0.0001, "loss": 0.021, "step": 23740 }, { "epoch": 181.29770992366412, "grad_norm": 2.1200926303863525, "learning_rate": 0.0001, "loss": 0.021, "step": 23750 }, { "epoch": 181.3740458015267, "grad_norm": 1.5636577606201172, "learning_rate": 0.0001, "loss": 0.0207, "step": 23760 }, { "epoch": 181.45038167938932, "grad_norm": 1.3529049158096313, "learning_rate": 0.0001, "loss": 0.0213, "step": 23770 }, { "epoch": 181.5267175572519, "grad_norm": 1.8406412601470947, "learning_rate": 0.0001, "loss": 0.0213, "step": 23780 }, { "epoch": 181.6030534351145, "grad_norm": 2.2109076976776123, "learning_rate": 0.0001, "loss": 0.0211, "step": 23790 }, { "epoch": 181.6793893129771, "grad_norm": 1.3595234155654907, "learning_rate": 0.0001, "loss": 0.021, "step": 23800 }, { "epoch": 181.75572519083968, "grad_norm": 1.4716556072235107, "learning_rate": 0.0001, "loss": 0.0206, "step": 23810 }, { "epoch": 181.8320610687023, "grad_norm": 1.8774707317352295, "learning_rate": 0.0001, "loss": 0.0207, "step": 23820 }, { "epoch": 181.9083969465649, "grad_norm": 2.6850697994232178, "learning_rate": 0.0001, "loss": 0.0216, "step": 23830 }, { "epoch": 181.98473282442748, "grad_norm": 2.300858974456787, "learning_rate": 0.0001, "loss": 0.0221, "step": 23840 }, { "epoch": 182.06106870229007, "grad_norm": 1.2819257974624634, "learning_rate": 0.0001, "loss": 0.021, "step": 23850 }, { "epoch": 182.13740458015266, "grad_norm": 1.6633937358856201, "learning_rate": 0.0001, "loss": 0.0206, "step": 23860 }, { "epoch": 182.21374045801528, "grad_norm": 1.3736001253128052, "learning_rate": 0.0001, "loss": 0.022, "step": 23870 }, { "epoch": 182.29007633587787, "grad_norm": 1.6166458129882812, "learning_rate": 0.0001, "loss": 0.0218, "step": 23880 }, { "epoch": 182.36641221374046, "grad_norm": 1.8772823810577393, "learning_rate": 0.0001, "loss": 0.0223, "step": 23890 }, { "epoch": 182.44274809160305, "grad_norm": 1.9029546976089478, "learning_rate": 0.0001, "loss": 0.0221, "step": 23900 }, { "epoch": 182.51908396946564, "grad_norm": 1.2890727519989014, "learning_rate": 0.0001, "loss": 0.0215, "step": 23910 }, { "epoch": 182.59541984732823, "grad_norm": 1.6802728176116943, "learning_rate": 0.0001, "loss": 0.0215, "step": 23920 }, { "epoch": 182.67175572519085, "grad_norm": 1.8943941593170166, "learning_rate": 0.0001, "loss": 0.0216, "step": 23930 }, { "epoch": 182.74809160305344, "grad_norm": 1.7797926664352417, "learning_rate": 0.0001, "loss": 0.0215, "step": 23940 }, { "epoch": 182.82442748091603, "grad_norm": 1.8922474384307861, "learning_rate": 0.0001, "loss": 0.0219, "step": 23950 }, { "epoch": 182.90076335877862, "grad_norm": 1.7902382612228394, "learning_rate": 0.0001, "loss": 0.0213, "step": 23960 }, { "epoch": 182.9770992366412, "grad_norm": 1.6689289808273315, "learning_rate": 0.0001, "loss": 0.0218, "step": 23970 }, { "epoch": 183.05343511450383, "grad_norm": 1.4063385725021362, "learning_rate": 0.0001, "loss": 0.021, "step": 23980 }, { "epoch": 183.12977099236642, "grad_norm": 1.792421579360962, "learning_rate": 0.0001, "loss": 0.0214, "step": 23990 }, { "epoch": 183.206106870229, "grad_norm": 2.3790440559387207, "learning_rate": 0.0001, "loss": 0.0225, "step": 24000 }, { "epoch": 183.2824427480916, "grad_norm": 2.1628336906433105, "learning_rate": 0.0001, "loss": 0.0219, "step": 24010 }, { "epoch": 183.3587786259542, "grad_norm": 1.854961633682251, "learning_rate": 0.0001, "loss": 0.0209, "step": 24020 }, { "epoch": 183.4351145038168, "grad_norm": 2.478804111480713, "learning_rate": 0.0001, "loss": 0.0209, "step": 24030 }, { "epoch": 183.5114503816794, "grad_norm": 1.7606444358825684, "learning_rate": 0.0001, "loss": 0.0211, "step": 24040 }, { "epoch": 183.58778625954199, "grad_norm": 1.737505316734314, "learning_rate": 0.0001, "loss": 0.0211, "step": 24050 }, { "epoch": 183.66412213740458, "grad_norm": 1.560318112373352, "learning_rate": 0.0001, "loss": 0.0208, "step": 24060 }, { "epoch": 183.74045801526717, "grad_norm": 1.8272325992584229, "learning_rate": 0.0001, "loss": 0.0207, "step": 24070 }, { "epoch": 183.81679389312978, "grad_norm": 2.232839822769165, "learning_rate": 0.0001, "loss": 0.0212, "step": 24080 }, { "epoch": 183.89312977099237, "grad_norm": 2.4001588821411133, "learning_rate": 0.0001, "loss": 0.0214, "step": 24090 }, { "epoch": 183.96946564885496, "grad_norm": 2.1796329021453857, "learning_rate": 0.0001, "loss": 0.021, "step": 24100 }, { "epoch": 184.04580152671755, "grad_norm": 1.3878755569458008, "learning_rate": 0.0001, "loss": 0.0205, "step": 24110 }, { "epoch": 184.12213740458014, "grad_norm": 2.000319480895996, "learning_rate": 0.0001, "loss": 0.0211, "step": 24120 }, { "epoch": 184.19847328244273, "grad_norm": 1.5013810396194458, "learning_rate": 0.0001, "loss": 0.0208, "step": 24130 }, { "epoch": 184.27480916030535, "grad_norm": 2.19740629196167, "learning_rate": 0.0001, "loss": 0.0206, "step": 24140 }, { "epoch": 184.35114503816794, "grad_norm": 3.1286392211914062, "learning_rate": 0.0001, "loss": 0.0208, "step": 24150 }, { "epoch": 184.42748091603053, "grad_norm": 1.886335849761963, "learning_rate": 0.0001, "loss": 0.0202, "step": 24160 }, { "epoch": 184.50381679389312, "grad_norm": 2.492257595062256, "learning_rate": 0.0001, "loss": 0.0204, "step": 24170 }, { "epoch": 184.5801526717557, "grad_norm": 1.531848430633545, "learning_rate": 0.0001, "loss": 0.0202, "step": 24180 }, { "epoch": 184.65648854961833, "grad_norm": 1.8774722814559937, "learning_rate": 0.0001, "loss": 0.0202, "step": 24190 }, { "epoch": 184.73282442748092, "grad_norm": 2.464479923248291, "learning_rate": 0.0001, "loss": 0.0202, "step": 24200 }, { "epoch": 184.8091603053435, "grad_norm": 2.6645684242248535, "learning_rate": 0.0001, "loss": 0.021, "step": 24210 }, { "epoch": 184.8854961832061, "grad_norm": 1.7915068864822388, "learning_rate": 0.0001, "loss": 0.0202, "step": 24220 }, { "epoch": 184.9618320610687, "grad_norm": 2.258918046951294, "learning_rate": 0.0001, "loss": 0.0204, "step": 24230 }, { "epoch": 185.0381679389313, "grad_norm": 1.690462350845337, "learning_rate": 0.0001, "loss": 0.0205, "step": 24240 }, { "epoch": 185.1145038167939, "grad_norm": 1.5370103120803833, "learning_rate": 0.0001, "loss": 0.0214, "step": 24250 }, { "epoch": 185.1908396946565, "grad_norm": 1.6130129098892212, "learning_rate": 0.0001, "loss": 0.0208, "step": 24260 }, { "epoch": 185.26717557251908, "grad_norm": 1.588241457939148, "learning_rate": 0.0001, "loss": 0.0216, "step": 24270 }, { "epoch": 185.34351145038167, "grad_norm": 1.302205204963684, "learning_rate": 0.0001, "loss": 0.0223, "step": 24280 }, { "epoch": 185.4198473282443, "grad_norm": 2.306002140045166, "learning_rate": 0.0001, "loss": 0.0212, "step": 24290 }, { "epoch": 185.49618320610688, "grad_norm": 1.6836116313934326, "learning_rate": 0.0001, "loss": 0.0214, "step": 24300 }, { "epoch": 185.57251908396947, "grad_norm": 1.4346277713775635, "learning_rate": 0.0001, "loss": 0.0206, "step": 24310 }, { "epoch": 185.64885496183206, "grad_norm": 1.6901466846466064, "learning_rate": 0.0001, "loss": 0.0208, "step": 24320 }, { "epoch": 185.72519083969465, "grad_norm": 1.3910927772521973, "learning_rate": 0.0001, "loss": 0.0208, "step": 24330 }, { "epoch": 185.80152671755727, "grad_norm": 1.7080031633377075, "learning_rate": 0.0001, "loss": 0.0213, "step": 24340 }, { "epoch": 185.87786259541986, "grad_norm": 1.3614143133163452, "learning_rate": 0.0001, "loss": 0.022, "step": 24350 }, { "epoch": 185.95419847328245, "grad_norm": 2.126539468765259, "learning_rate": 0.0001, "loss": 0.0227, "step": 24360 }, { "epoch": 186.03053435114504, "grad_norm": 1.9021193981170654, "learning_rate": 0.0001, "loss": 0.0223, "step": 24370 }, { "epoch": 186.10687022900763, "grad_norm": 1.7754918336868286, "learning_rate": 0.0001, "loss": 0.0224, "step": 24380 }, { "epoch": 186.18320610687022, "grad_norm": 1.606283187866211, "learning_rate": 0.0001, "loss": 0.0227, "step": 24390 }, { "epoch": 186.25954198473283, "grad_norm": 1.380035400390625, "learning_rate": 0.0001, "loss": 0.0219, "step": 24400 }, { "epoch": 186.33587786259542, "grad_norm": 1.7168049812316895, "learning_rate": 0.0001, "loss": 0.0223, "step": 24410 }, { "epoch": 186.41221374045801, "grad_norm": 1.7254729270935059, "learning_rate": 0.0001, "loss": 0.0217, "step": 24420 }, { "epoch": 186.4885496183206, "grad_norm": 2.422802686691284, "learning_rate": 0.0001, "loss": 0.0218, "step": 24430 }, { "epoch": 186.5648854961832, "grad_norm": 1.254399061203003, "learning_rate": 0.0001, "loss": 0.0213, "step": 24440 }, { "epoch": 186.6412213740458, "grad_norm": 1.4924052953720093, "learning_rate": 0.0001, "loss": 0.0212, "step": 24450 }, { "epoch": 186.7175572519084, "grad_norm": 1.3528138399124146, "learning_rate": 0.0001, "loss": 0.0207, "step": 24460 }, { "epoch": 186.793893129771, "grad_norm": 1.336328387260437, "learning_rate": 0.0001, "loss": 0.0219, "step": 24470 }, { "epoch": 186.87022900763358, "grad_norm": 1.2630469799041748, "learning_rate": 0.0001, "loss": 0.0215, "step": 24480 }, { "epoch": 186.94656488549617, "grad_norm": 1.2479759454727173, "learning_rate": 0.0001, "loss": 0.0218, "step": 24490 }, { "epoch": 187.0229007633588, "grad_norm": 1.6441534757614136, "learning_rate": 0.0001, "loss": 0.0215, "step": 24500 }, { "epoch": 187.09923664122138, "grad_norm": 1.8178633451461792, "learning_rate": 0.0001, "loss": 0.0224, "step": 24510 }, { "epoch": 187.17557251908397, "grad_norm": 1.921492338180542, "learning_rate": 0.0001, "loss": 0.022, "step": 24520 }, { "epoch": 187.25190839694656, "grad_norm": 2.5878827571868896, "learning_rate": 0.0001, "loss": 0.0222, "step": 24530 }, { "epoch": 187.32824427480915, "grad_norm": 2.204683542251587, "learning_rate": 0.0001, "loss": 0.0227, "step": 24540 }, { "epoch": 187.40458015267177, "grad_norm": 2.1959681510925293, "learning_rate": 0.0001, "loss": 0.0223, "step": 24550 }, { "epoch": 187.48091603053436, "grad_norm": 1.4407042264938354, "learning_rate": 0.0001, "loss": 0.022, "step": 24560 }, { "epoch": 187.55725190839695, "grad_norm": 1.972283124923706, "learning_rate": 0.0001, "loss": 0.0214, "step": 24570 }, { "epoch": 187.63358778625954, "grad_norm": 1.8289546966552734, "learning_rate": 0.0001, "loss": 0.0213, "step": 24580 }, { "epoch": 187.70992366412213, "grad_norm": 2.017331600189209, "learning_rate": 0.0001, "loss": 0.0213, "step": 24590 }, { "epoch": 187.78625954198472, "grad_norm": 2.2262496948242188, "learning_rate": 0.0001, "loss": 0.0203, "step": 24600 }, { "epoch": 187.86259541984734, "grad_norm": 2.3340606689453125, "learning_rate": 0.0001, "loss": 0.0216, "step": 24610 }, { "epoch": 187.93893129770993, "grad_norm": 1.9121818542480469, "learning_rate": 0.0001, "loss": 0.0223, "step": 24620 }, { "epoch": 188.01526717557252, "grad_norm": 1.7811102867126465, "learning_rate": 0.0001, "loss": 0.0211, "step": 24630 }, { "epoch": 188.0916030534351, "grad_norm": 1.7377312183380127, "learning_rate": 0.0001, "loss": 0.0212, "step": 24640 }, { "epoch": 188.1679389312977, "grad_norm": 1.299376130104065, "learning_rate": 0.0001, "loss": 0.0212, "step": 24650 }, { "epoch": 188.24427480916032, "grad_norm": 1.6490294933319092, "learning_rate": 0.0001, "loss": 0.021, "step": 24660 }, { "epoch": 188.3206106870229, "grad_norm": 1.6752101182937622, "learning_rate": 0.0001, "loss": 0.0215, "step": 24670 }, { "epoch": 188.3969465648855, "grad_norm": 2.590914249420166, "learning_rate": 0.0001, "loss": 0.0215, "step": 24680 }, { "epoch": 188.4732824427481, "grad_norm": 2.3224422931671143, "learning_rate": 0.0001, "loss": 0.0216, "step": 24690 }, { "epoch": 188.54961832061068, "grad_norm": 2.181428909301758, "learning_rate": 0.0001, "loss": 0.0216, "step": 24700 }, { "epoch": 188.6259541984733, "grad_norm": 2.440789222717285, "learning_rate": 0.0001, "loss": 0.0212, "step": 24710 }, { "epoch": 188.70229007633588, "grad_norm": 2.673232078552246, "learning_rate": 0.0001, "loss": 0.0207, "step": 24720 }, { "epoch": 188.77862595419847, "grad_norm": 2.2461915016174316, "learning_rate": 0.0001, "loss": 0.0205, "step": 24730 }, { "epoch": 188.85496183206106, "grad_norm": 1.8265876770019531, "learning_rate": 0.0001, "loss": 0.0206, "step": 24740 }, { "epoch": 188.93129770992365, "grad_norm": 1.4479213953018188, "learning_rate": 0.0001, "loss": 0.0206, "step": 24750 }, { "epoch": 189.00763358778627, "grad_norm": 1.1734102964401245, "learning_rate": 0.0001, "loss": 0.0201, "step": 24760 }, { "epoch": 189.08396946564886, "grad_norm": 1.923027753829956, "learning_rate": 0.0001, "loss": 0.0199, "step": 24770 }, { "epoch": 189.16030534351145, "grad_norm": 1.862060785293579, "learning_rate": 0.0001, "loss": 0.0214, "step": 24780 }, { "epoch": 189.23664122137404, "grad_norm": 2.493471622467041, "learning_rate": 0.0001, "loss": 0.0207, "step": 24790 }, { "epoch": 189.31297709923663, "grad_norm": 3.2096986770629883, "learning_rate": 0.0001, "loss": 0.0214, "step": 24800 }, { "epoch": 189.38931297709922, "grad_norm": 2.176710367202759, "learning_rate": 0.0001, "loss": 0.0207, "step": 24810 }, { "epoch": 189.46564885496184, "grad_norm": 1.9514966011047363, "learning_rate": 0.0001, "loss": 0.0204, "step": 24820 }, { "epoch": 189.54198473282443, "grad_norm": 2.0065982341766357, "learning_rate": 0.0001, "loss": 0.0204, "step": 24830 }, { "epoch": 189.61832061068702, "grad_norm": 1.3153396844863892, "learning_rate": 0.0001, "loss": 0.0202, "step": 24840 }, { "epoch": 189.6946564885496, "grad_norm": 1.6591753959655762, "learning_rate": 0.0001, "loss": 0.0196, "step": 24850 }, { "epoch": 189.7709923664122, "grad_norm": 1.143673062324524, "learning_rate": 0.0001, "loss": 0.0202, "step": 24860 }, { "epoch": 189.84732824427482, "grad_norm": 1.6715487241744995, "learning_rate": 0.0001, "loss": 0.0211, "step": 24870 }, { "epoch": 189.9236641221374, "grad_norm": 1.1547070741653442, "learning_rate": 0.0001, "loss": 0.0215, "step": 24880 }, { "epoch": 190.0, "grad_norm": 1.4401967525482178, "learning_rate": 0.0001, "loss": 0.0227, "step": 24890 }, { "epoch": 190.0763358778626, "grad_norm": 1.8273587226867676, "learning_rate": 0.0001, "loss": 0.0219, "step": 24900 }, { "epoch": 190.15267175572518, "grad_norm": 1.6947829723358154, "learning_rate": 0.0001, "loss": 0.0218, "step": 24910 }, { "epoch": 190.2290076335878, "grad_norm": 1.7719875574111938, "learning_rate": 0.0001, "loss": 0.022, "step": 24920 }, { "epoch": 190.3053435114504, "grad_norm": 2.1691079139709473, "learning_rate": 0.0001, "loss": 0.023, "step": 24930 }, { "epoch": 190.38167938931298, "grad_norm": 1.657763123512268, "learning_rate": 0.0001, "loss": 0.0232, "step": 24940 }, { "epoch": 190.45801526717557, "grad_norm": 1.569589614868164, "learning_rate": 0.0001, "loss": 0.0223, "step": 24950 }, { "epoch": 190.53435114503816, "grad_norm": 1.2969566583633423, "learning_rate": 0.0001, "loss": 0.0229, "step": 24960 }, { "epoch": 190.61068702290078, "grad_norm": 1.3462532758712769, "learning_rate": 0.0001, "loss": 0.0225, "step": 24970 }, { "epoch": 190.68702290076337, "grad_norm": 1.9644362926483154, "learning_rate": 0.0001, "loss": 0.0228, "step": 24980 }, { "epoch": 190.76335877862596, "grad_norm": 2.270869016647339, "learning_rate": 0.0001, "loss": 0.0225, "step": 24990 }, { "epoch": 190.83969465648855, "grad_norm": 2.2564215660095215, "learning_rate": 0.0001, "loss": 0.0221, "step": 25000 }, { "epoch": 190.91603053435114, "grad_norm": 1.883233666419983, "learning_rate": 0.0001, "loss": 0.0211, "step": 25010 }, { "epoch": 190.99236641221373, "grad_norm": 1.7191903591156006, "learning_rate": 0.0001, "loss": 0.0214, "step": 25020 }, { "epoch": 191.06870229007635, "grad_norm": 1.6659576892852783, "learning_rate": 0.0001, "loss": 0.0216, "step": 25030 }, { "epoch": 191.14503816793894, "grad_norm": 2.0483558177948, "learning_rate": 0.0001, "loss": 0.0215, "step": 25040 }, { "epoch": 191.22137404580153, "grad_norm": 1.5138111114501953, "learning_rate": 0.0001, "loss": 0.0212, "step": 25050 }, { "epoch": 191.29770992366412, "grad_norm": 1.171167016029358, "learning_rate": 0.0001, "loss": 0.0207, "step": 25060 }, { "epoch": 191.3740458015267, "grad_norm": 2.313232421875, "learning_rate": 0.0001, "loss": 0.0205, "step": 25070 }, { "epoch": 191.45038167938932, "grad_norm": 1.543483018875122, "learning_rate": 0.0001, "loss": 0.021, "step": 25080 }, { "epoch": 191.5267175572519, "grad_norm": 1.5995535850524902, "learning_rate": 0.0001, "loss": 0.0212, "step": 25090 }, { "epoch": 191.6030534351145, "grad_norm": 2.0863986015319824, "learning_rate": 0.0001, "loss": 0.0212, "step": 25100 }, { "epoch": 191.6793893129771, "grad_norm": 2.122938394546509, "learning_rate": 0.0001, "loss": 0.0205, "step": 25110 }, { "epoch": 191.75572519083968, "grad_norm": 1.776602864265442, "learning_rate": 0.0001, "loss": 0.0207, "step": 25120 }, { "epoch": 191.8320610687023, "grad_norm": 1.1836775541305542, "learning_rate": 0.0001, "loss": 0.0201, "step": 25130 }, { "epoch": 191.9083969465649, "grad_norm": 1.4137580394744873, "learning_rate": 0.0001, "loss": 0.0201, "step": 25140 }, { "epoch": 191.98473282442748, "grad_norm": 2.294917345046997, "learning_rate": 0.0001, "loss": 0.02, "step": 25150 }, { "epoch": 192.06106870229007, "grad_norm": 1.5571357011795044, "learning_rate": 0.0001, "loss": 0.0217, "step": 25160 }, { "epoch": 192.13740458015266, "grad_norm": 2.2405998706817627, "learning_rate": 0.0001, "loss": 0.0212, "step": 25170 }, { "epoch": 192.21374045801528, "grad_norm": 2.066162586212158, "learning_rate": 0.0001, "loss": 0.0211, "step": 25180 }, { "epoch": 192.29007633587787, "grad_norm": 1.5227230787277222, "learning_rate": 0.0001, "loss": 0.0205, "step": 25190 }, { "epoch": 192.36641221374046, "grad_norm": 1.65030837059021, "learning_rate": 0.0001, "loss": 0.0207, "step": 25200 }, { "epoch": 192.44274809160305, "grad_norm": 1.3429454565048218, "learning_rate": 0.0001, "loss": 0.0208, "step": 25210 }, { "epoch": 192.51908396946564, "grad_norm": 1.2663654088974, "learning_rate": 0.0001, "loss": 0.0204, "step": 25220 }, { "epoch": 192.59541984732823, "grad_norm": 1.6129311323165894, "learning_rate": 0.0001, "loss": 0.0209, "step": 25230 }, { "epoch": 192.67175572519085, "grad_norm": 1.387220025062561, "learning_rate": 0.0001, "loss": 0.0209, "step": 25240 }, { "epoch": 192.74809160305344, "grad_norm": 1.9901195764541626, "learning_rate": 0.0001, "loss": 0.0214, "step": 25250 }, { "epoch": 192.82442748091603, "grad_norm": 1.8975510597229004, "learning_rate": 0.0001, "loss": 0.0204, "step": 25260 }, { "epoch": 192.90076335877862, "grad_norm": 1.4419692754745483, "learning_rate": 0.0001, "loss": 0.0204, "step": 25270 }, { "epoch": 192.9770992366412, "grad_norm": 1.9346991777420044, "learning_rate": 0.0001, "loss": 0.02, "step": 25280 }, { "epoch": 193.05343511450383, "grad_norm": 1.4313932657241821, "learning_rate": 0.0001, "loss": 0.0202, "step": 25290 }, { "epoch": 193.12977099236642, "grad_norm": 1.8662762641906738, "learning_rate": 0.0001, "loss": 0.0209, "step": 25300 }, { "epoch": 193.206106870229, "grad_norm": 2.3276939392089844, "learning_rate": 0.0001, "loss": 0.0203, "step": 25310 }, { "epoch": 193.2824427480916, "grad_norm": 2.22287654876709, "learning_rate": 0.0001, "loss": 0.0199, "step": 25320 }, { "epoch": 193.3587786259542, "grad_norm": 2.3649494647979736, "learning_rate": 0.0001, "loss": 0.0199, "step": 25330 }, { "epoch": 193.4351145038168, "grad_norm": 1.414961338043213, "learning_rate": 0.0001, "loss": 0.0203, "step": 25340 }, { "epoch": 193.5114503816794, "grad_norm": 1.7812360525131226, "learning_rate": 0.0001, "loss": 0.0199, "step": 25350 }, { "epoch": 193.58778625954199, "grad_norm": 1.7873692512512207, "learning_rate": 0.0001, "loss": 0.02, "step": 25360 }, { "epoch": 193.66412213740458, "grad_norm": 1.5823055505752563, "learning_rate": 0.0001, "loss": 0.0199, "step": 25370 }, { "epoch": 193.74045801526717, "grad_norm": 1.7832744121551514, "learning_rate": 0.0001, "loss": 0.0202, "step": 25380 }, { "epoch": 193.81679389312978, "grad_norm": 1.9247503280639648, "learning_rate": 0.0001, "loss": 0.0204, "step": 25390 }, { "epoch": 193.89312977099237, "grad_norm": 2.0992431640625, "learning_rate": 0.0001, "loss": 0.0207, "step": 25400 }, { "epoch": 193.96946564885496, "grad_norm": 1.88157320022583, "learning_rate": 0.0001, "loss": 0.0202, "step": 25410 }, { "epoch": 194.04580152671755, "grad_norm": 1.518349051475525, "learning_rate": 0.0001, "loss": 0.0201, "step": 25420 }, { "epoch": 194.12213740458014, "grad_norm": 1.4374821186065674, "learning_rate": 0.0001, "loss": 0.0217, "step": 25430 }, { "epoch": 194.19847328244273, "grad_norm": 1.2250518798828125, "learning_rate": 0.0001, "loss": 0.0203, "step": 25440 }, { "epoch": 194.27480916030535, "grad_norm": 1.5717436075210571, "learning_rate": 0.0001, "loss": 0.0204, "step": 25450 }, { "epoch": 194.35114503816794, "grad_norm": 1.7351469993591309, "learning_rate": 0.0001, "loss": 0.0209, "step": 25460 }, { "epoch": 194.42748091603053, "grad_norm": 1.3131842613220215, "learning_rate": 0.0001, "loss": 0.0212, "step": 25470 }, { "epoch": 194.50381679389312, "grad_norm": 2.356367826461792, "learning_rate": 0.0001, "loss": 0.0219, "step": 25480 }, { "epoch": 194.5801526717557, "grad_norm": 1.9607226848602295, "learning_rate": 0.0001, "loss": 0.0216, "step": 25490 }, { "epoch": 194.65648854961833, "grad_norm": 2.4615838527679443, "learning_rate": 0.0001, "loss": 0.0213, "step": 25500 }, { "epoch": 194.73282442748092, "grad_norm": 2.120248317718506, "learning_rate": 0.0001, "loss": 0.0204, "step": 25510 }, { "epoch": 194.8091603053435, "grad_norm": 1.7007299661636353, "learning_rate": 0.0001, "loss": 0.0207, "step": 25520 }, { "epoch": 194.8854961832061, "grad_norm": 1.5624656677246094, "learning_rate": 0.0001, "loss": 0.0206, "step": 25530 }, { "epoch": 194.9618320610687, "grad_norm": 1.7941778898239136, "learning_rate": 0.0001, "loss": 0.021, "step": 25540 }, { "epoch": 195.0381679389313, "grad_norm": 1.9770199060440063, "learning_rate": 0.0001, "loss": 0.0211, "step": 25550 }, { "epoch": 195.1145038167939, "grad_norm": 1.516528844833374, "learning_rate": 0.0001, "loss": 0.0216, "step": 25560 }, { "epoch": 195.1908396946565, "grad_norm": 1.610478162765503, "learning_rate": 0.0001, "loss": 0.0211, "step": 25570 }, { "epoch": 195.26717557251908, "grad_norm": 1.2712883949279785, "learning_rate": 0.0001, "loss": 0.0205, "step": 25580 }, { "epoch": 195.34351145038167, "grad_norm": 1.493096113204956, "learning_rate": 0.0001, "loss": 0.0212, "step": 25590 }, { "epoch": 195.4198473282443, "grad_norm": 1.931639313697815, "learning_rate": 0.0001, "loss": 0.0213, "step": 25600 }, { "epoch": 195.49618320610688, "grad_norm": 2.3051040172576904, "learning_rate": 0.0001, "loss": 0.0219, "step": 25610 }, { "epoch": 195.57251908396947, "grad_norm": 1.6023287773132324, "learning_rate": 0.0001, "loss": 0.0228, "step": 25620 }, { "epoch": 195.64885496183206, "grad_norm": 1.2486482858657837, "learning_rate": 0.0001, "loss": 0.0214, "step": 25630 }, { "epoch": 195.72519083969465, "grad_norm": 1.8908517360687256, "learning_rate": 0.0001, "loss": 0.0209, "step": 25640 }, { "epoch": 195.80152671755727, "grad_norm": 1.412025809288025, "learning_rate": 0.0001, "loss": 0.0204, "step": 25650 }, { "epoch": 195.87786259541986, "grad_norm": 1.629518747329712, "learning_rate": 0.0001, "loss": 0.0214, "step": 25660 }, { "epoch": 195.95419847328245, "grad_norm": 2.008986711502075, "learning_rate": 0.0001, "loss": 0.0208, "step": 25670 }, { "epoch": 196.03053435114504, "grad_norm": 1.2423771619796753, "learning_rate": 0.0001, "loss": 0.0212, "step": 25680 }, { "epoch": 196.10687022900763, "grad_norm": 1.4652312994003296, "learning_rate": 0.0001, "loss": 0.0222, "step": 25690 }, { "epoch": 196.18320610687022, "grad_norm": 1.533582329750061, "learning_rate": 0.0001, "loss": 0.0219, "step": 25700 }, { "epoch": 196.25954198473283, "grad_norm": 1.978154182434082, "learning_rate": 0.0001, "loss": 0.0219, "step": 25710 }, { "epoch": 196.33587786259542, "grad_norm": 1.9732271432876587, "learning_rate": 0.0001, "loss": 0.0217, "step": 25720 }, { "epoch": 196.41221374045801, "grad_norm": 1.5821125507354736, "learning_rate": 0.0001, "loss": 0.0212, "step": 25730 }, { "epoch": 196.4885496183206, "grad_norm": 2.410003423690796, "learning_rate": 0.0001, "loss": 0.0213, "step": 25740 }, { "epoch": 196.5648854961832, "grad_norm": 1.6089009046554565, "learning_rate": 0.0001, "loss": 0.021, "step": 25750 }, { "epoch": 196.6412213740458, "grad_norm": 1.318726658821106, "learning_rate": 0.0001, "loss": 0.0203, "step": 25760 }, { "epoch": 196.7175572519084, "grad_norm": 1.9084997177124023, "learning_rate": 0.0001, "loss": 0.0207, "step": 25770 }, { "epoch": 196.793893129771, "grad_norm": 1.4985430240631104, "learning_rate": 0.0001, "loss": 0.0212, "step": 25780 }, { "epoch": 196.87022900763358, "grad_norm": 1.7243964672088623, "learning_rate": 0.0001, "loss": 0.0215, "step": 25790 }, { "epoch": 196.94656488549617, "grad_norm": 1.864140272140503, "learning_rate": 0.0001, "loss": 0.0217, "step": 25800 }, { "epoch": 197.0229007633588, "grad_norm": 1.953851580619812, "learning_rate": 0.0001, "loss": 0.0209, "step": 25810 }, { "epoch": 197.09923664122138, "grad_norm": 2.1577510833740234, "learning_rate": 0.0001, "loss": 0.021, "step": 25820 }, { "epoch": 197.17557251908397, "grad_norm": 2.3518190383911133, "learning_rate": 0.0001, "loss": 0.0206, "step": 25830 }, { "epoch": 197.25190839694656, "grad_norm": 1.9763453006744385, "learning_rate": 0.0001, "loss": 0.0203, "step": 25840 }, { "epoch": 197.32824427480915, "grad_norm": 2.6582396030426025, "learning_rate": 0.0001, "loss": 0.0197, "step": 25850 }, { "epoch": 197.40458015267177, "grad_norm": 2.5200035572052, "learning_rate": 0.0001, "loss": 0.0202, "step": 25860 }, { "epoch": 197.48091603053436, "grad_norm": 1.2994946241378784, "learning_rate": 0.0001, "loss": 0.0201, "step": 25870 }, { "epoch": 197.55725190839695, "grad_norm": 1.3008157014846802, "learning_rate": 0.0001, "loss": 0.0203, "step": 25880 }, { "epoch": 197.63358778625954, "grad_norm": 1.3639748096466064, "learning_rate": 0.0001, "loss": 0.0211, "step": 25890 }, { "epoch": 197.70992366412213, "grad_norm": 1.9963198900222778, "learning_rate": 0.0001, "loss": 0.0204, "step": 25900 }, { "epoch": 197.78625954198472, "grad_norm": 1.6243650913238525, "learning_rate": 0.0001, "loss": 0.0211, "step": 25910 }, { "epoch": 197.86259541984734, "grad_norm": 1.4914608001708984, "learning_rate": 0.0001, "loss": 0.0209, "step": 25920 }, { "epoch": 197.93893129770993, "grad_norm": 1.4614256620407104, "learning_rate": 0.0001, "loss": 0.0208, "step": 25930 }, { "epoch": 198.01526717557252, "grad_norm": 1.4422334432601929, "learning_rate": 0.0001, "loss": 0.0214, "step": 25940 }, { "epoch": 198.0916030534351, "grad_norm": 1.2510730028152466, "learning_rate": 0.0001, "loss": 0.0209, "step": 25950 }, { "epoch": 198.1679389312977, "grad_norm": 1.4194636344909668, "learning_rate": 0.0001, "loss": 0.0213, "step": 25960 }, { "epoch": 198.24427480916032, "grad_norm": 1.8513966798782349, "learning_rate": 0.0001, "loss": 0.0211, "step": 25970 }, { "epoch": 198.3206106870229, "grad_norm": 1.262325406074524, "learning_rate": 0.0001, "loss": 0.0208, "step": 25980 }, { "epoch": 198.3969465648855, "grad_norm": 1.9727262258529663, "learning_rate": 0.0001, "loss": 0.0215, "step": 25990 }, { "epoch": 198.4732824427481, "grad_norm": 1.3845431804656982, "learning_rate": 0.0001, "loss": 0.0214, "step": 26000 }, { "epoch": 198.54961832061068, "grad_norm": 1.967891812324524, "learning_rate": 0.0001, "loss": 0.022, "step": 26010 }, { "epoch": 198.6259541984733, "grad_norm": 1.1907190084457397, "learning_rate": 0.0001, "loss": 0.0219, "step": 26020 }, { "epoch": 198.70229007633588, "grad_norm": 1.6858340501785278, "learning_rate": 0.0001, "loss": 0.0221, "step": 26030 }, { "epoch": 198.77862595419847, "grad_norm": 2.2342660427093506, "learning_rate": 0.0001, "loss": 0.0209, "step": 26040 }, { "epoch": 198.85496183206106, "grad_norm": 1.4175772666931152, "learning_rate": 0.0001, "loss": 0.0213, "step": 26050 }, { "epoch": 198.93129770992365, "grad_norm": 1.6306406259536743, "learning_rate": 0.0001, "loss": 0.0205, "step": 26060 }, { "epoch": 199.00763358778627, "grad_norm": 1.4600783586502075, "learning_rate": 0.0001, "loss": 0.0207, "step": 26070 }, { "epoch": 199.08396946564886, "grad_norm": 1.2376774549484253, "learning_rate": 0.0001, "loss": 0.0211, "step": 26080 }, { "epoch": 199.16030534351145, "grad_norm": 1.5132362842559814, "learning_rate": 0.0001, "loss": 0.021, "step": 26090 }, { "epoch": 199.23664122137404, "grad_norm": 1.3213590383529663, "learning_rate": 0.0001, "loss": 0.021, "step": 26100 }, { "epoch": 199.31297709923663, "grad_norm": 2.2441680431365967, "learning_rate": 0.0001, "loss": 0.0219, "step": 26110 }, { "epoch": 199.38931297709922, "grad_norm": 2.480471611022949, "learning_rate": 0.0001, "loss": 0.0225, "step": 26120 }, { "epoch": 199.46564885496184, "grad_norm": 2.5266854763031006, "learning_rate": 0.0001, "loss": 0.0221, "step": 26130 }, { "epoch": 199.54198473282443, "grad_norm": 1.9838755130767822, "learning_rate": 0.0001, "loss": 0.022, "step": 26140 }, { "epoch": 199.61832061068702, "grad_norm": 2.5255990028381348, "learning_rate": 0.0001, "loss": 0.0207, "step": 26150 }, { "epoch": 199.6946564885496, "grad_norm": 1.6005539894104004, "learning_rate": 0.0001, "loss": 0.021, "step": 26160 }, { "epoch": 199.7709923664122, "grad_norm": 2.531935691833496, "learning_rate": 0.0001, "loss": 0.021, "step": 26170 }, { "epoch": 199.84732824427482, "grad_norm": 1.5619628429412842, "learning_rate": 0.0001, "loss": 0.0211, "step": 26180 }, { "epoch": 199.9236641221374, "grad_norm": 2.2166407108306885, "learning_rate": 0.0001, "loss": 0.02, "step": 26190 }, { "epoch": 200.0, "grad_norm": 2.0505411624908447, "learning_rate": 0.0001, "loss": 0.0202, "step": 26200 }, { "epoch": 200.0763358778626, "grad_norm": 2.0971944332122803, "learning_rate": 0.0001, "loss": 0.0202, "step": 26210 }, { "epoch": 200.15267175572518, "grad_norm": 2.3252522945404053, "learning_rate": 0.0001, "loss": 0.021, "step": 26220 }, { "epoch": 200.2290076335878, "grad_norm": 2.115175485610962, "learning_rate": 0.0001, "loss": 0.0208, "step": 26230 }, { "epoch": 200.3053435114504, "grad_norm": 2.524609088897705, "learning_rate": 0.0001, "loss": 0.0206, "step": 26240 }, { "epoch": 200.38167938931298, "grad_norm": 2.633477210998535, "learning_rate": 0.0001, "loss": 0.0213, "step": 26250 }, { "epoch": 200.45801526717557, "grad_norm": 1.4181874990463257, "learning_rate": 0.0001, "loss": 0.021, "step": 26260 }, { "epoch": 200.53435114503816, "grad_norm": 1.8097254037857056, "learning_rate": 0.0001, "loss": 0.0203, "step": 26270 }, { "epoch": 200.61068702290078, "grad_norm": 1.5539164543151855, "learning_rate": 0.0001, "loss": 0.0202, "step": 26280 }, { "epoch": 200.68702290076337, "grad_norm": 1.2554845809936523, "learning_rate": 0.0001, "loss": 0.0209, "step": 26290 }, { "epoch": 200.76335877862596, "grad_norm": 1.7749508619308472, "learning_rate": 0.0001, "loss": 0.0212, "step": 26300 }, { "epoch": 200.83969465648855, "grad_norm": 1.5821936130523682, "learning_rate": 0.0001, "loss": 0.0216, "step": 26310 }, { "epoch": 200.91603053435114, "grad_norm": 1.7453250885009766, "learning_rate": 0.0001, "loss": 0.0215, "step": 26320 }, { "epoch": 200.99236641221373, "grad_norm": 2.285177230834961, "learning_rate": 0.0001, "loss": 0.0216, "step": 26330 }, { "epoch": 201.06870229007635, "grad_norm": 2.4239068031311035, "learning_rate": 0.0001, "loss": 0.021, "step": 26340 }, { "epoch": 201.14503816793894, "grad_norm": 1.8135358095169067, "learning_rate": 0.0001, "loss": 0.0209, "step": 26350 }, { "epoch": 201.22137404580153, "grad_norm": 1.6893008947372437, "learning_rate": 0.0001, "loss": 0.0201, "step": 26360 }, { "epoch": 201.29770992366412, "grad_norm": 2.8020081520080566, "learning_rate": 0.0001, "loss": 0.0243, "step": 26370 }, { "epoch": 201.3740458015267, "grad_norm": 4.101394176483154, "learning_rate": 0.0001, "loss": 0.0211, "step": 26380 }, { "epoch": 201.45038167938932, "grad_norm": 3.393603801727295, "learning_rate": 0.0001, "loss": 0.0234, "step": 26390 }, { "epoch": 201.5267175572519, "grad_norm": 2.3763480186462402, "learning_rate": 0.0001, "loss": 0.021, "step": 26400 }, { "epoch": 201.6030534351145, "grad_norm": 2.610578775405884, "learning_rate": 0.0001, "loss": 0.0205, "step": 26410 }, { "epoch": 201.6793893129771, "grad_norm": 2.456936836242676, "learning_rate": 0.0001, "loss": 0.0194, "step": 26420 }, { "epoch": 201.75572519083968, "grad_norm": 1.879046082496643, "learning_rate": 0.0001, "loss": 0.0194, "step": 26430 }, { "epoch": 201.8320610687023, "grad_norm": 2.101832389831543, "learning_rate": 0.0001, "loss": 0.0202, "step": 26440 }, { "epoch": 201.9083969465649, "grad_norm": 1.4844917058944702, "learning_rate": 0.0001, "loss": 0.0192, "step": 26450 }, { "epoch": 201.98473282442748, "grad_norm": 1.9252665042877197, "learning_rate": 0.0001, "loss": 0.0194, "step": 26460 }, { "epoch": 202.06106870229007, "grad_norm": 1.6406190395355225, "learning_rate": 0.0001, "loss": 0.0198, "step": 26470 }, { "epoch": 202.13740458015266, "grad_norm": 1.3238013982772827, "learning_rate": 0.0001, "loss": 0.0194, "step": 26480 }, { "epoch": 202.21374045801528, "grad_norm": 2.061901569366455, "learning_rate": 0.0001, "loss": 0.0198, "step": 26490 }, { "epoch": 202.29007633587787, "grad_norm": 2.479733467102051, "learning_rate": 0.0001, "loss": 0.0196, "step": 26500 }, { "epoch": 202.36641221374046, "grad_norm": 1.633816123008728, "learning_rate": 0.0001, "loss": 0.0201, "step": 26510 }, { "epoch": 202.44274809160305, "grad_norm": 1.4716978073120117, "learning_rate": 0.0001, "loss": 0.0195, "step": 26520 }, { "epoch": 202.51908396946564, "grad_norm": 2.78371000289917, "learning_rate": 0.0001, "loss": 0.0206, "step": 26530 }, { "epoch": 202.59541984732823, "grad_norm": 2.4901034832000732, "learning_rate": 0.0001, "loss": 0.0202, "step": 26540 }, { "epoch": 202.67175572519085, "grad_norm": 1.668656349182129, "learning_rate": 0.0001, "loss": 0.02, "step": 26550 }, { "epoch": 202.74809160305344, "grad_norm": 3.108130693435669, "learning_rate": 0.0001, "loss": 0.0199, "step": 26560 }, { "epoch": 202.82442748091603, "grad_norm": 2.5266544818878174, "learning_rate": 0.0001, "loss": 0.0207, "step": 26570 }, { "epoch": 202.90076335877862, "grad_norm": 2.0689167976379395, "learning_rate": 0.0001, "loss": 0.0195, "step": 26580 }, { "epoch": 202.9770992366412, "grad_norm": 2.028278112411499, "learning_rate": 0.0001, "loss": 0.0193, "step": 26590 }, { "epoch": 203.05343511450383, "grad_norm": 1.8265827894210815, "learning_rate": 0.0001, "loss": 0.0197, "step": 26600 }, { "epoch": 203.12977099236642, "grad_norm": 2.688021421432495, "learning_rate": 0.0001, "loss": 0.02, "step": 26610 }, { "epoch": 203.206106870229, "grad_norm": 1.6163108348846436, "learning_rate": 0.0001, "loss": 0.02, "step": 26620 }, { "epoch": 203.2824427480916, "grad_norm": 2.3607163429260254, "learning_rate": 0.0001, "loss": 0.0196, "step": 26630 }, { "epoch": 203.3587786259542, "grad_norm": 1.654289722442627, "learning_rate": 0.0001, "loss": 0.0192, "step": 26640 }, { "epoch": 203.4351145038168, "grad_norm": 1.5952820777893066, "learning_rate": 0.0001, "loss": 0.0192, "step": 26650 }, { "epoch": 203.5114503816794, "grad_norm": 1.5866838693618774, "learning_rate": 0.0001, "loss": 0.019, "step": 26660 }, { "epoch": 203.58778625954199, "grad_norm": 1.5747735500335693, "learning_rate": 0.0001, "loss": 0.0198, "step": 26670 }, { "epoch": 203.66412213740458, "grad_norm": 2.2824018001556396, "learning_rate": 0.0001, "loss": 0.0198, "step": 26680 }, { "epoch": 203.74045801526717, "grad_norm": 1.6620080471038818, "learning_rate": 0.0001, "loss": 0.0199, "step": 26690 }, { "epoch": 203.81679389312978, "grad_norm": 1.4043896198272705, "learning_rate": 0.0001, "loss": 0.02, "step": 26700 }, { "epoch": 203.89312977099237, "grad_norm": 1.6406891345977783, "learning_rate": 0.0001, "loss": 0.02, "step": 26710 }, { "epoch": 203.96946564885496, "grad_norm": 1.983680248260498, "learning_rate": 0.0001, "loss": 0.0201, "step": 26720 }, { "epoch": 204.04580152671755, "grad_norm": 1.3972786664962769, "learning_rate": 0.0001, "loss": 0.0201, "step": 26730 }, { "epoch": 204.12213740458014, "grad_norm": 2.4201183319091797, "learning_rate": 0.0001, "loss": 0.0209, "step": 26740 }, { "epoch": 204.19847328244273, "grad_norm": 2.542529821395874, "learning_rate": 0.0001, "loss": 0.0212, "step": 26750 }, { "epoch": 204.27480916030535, "grad_norm": 1.6794860363006592, "learning_rate": 0.0001, "loss": 0.0199, "step": 26760 }, { "epoch": 204.35114503816794, "grad_norm": 1.4297575950622559, "learning_rate": 0.0001, "loss": 0.0203, "step": 26770 }, { "epoch": 204.42748091603053, "grad_norm": 2.0521326065063477, "learning_rate": 0.0001, "loss": 0.0204, "step": 26780 }, { "epoch": 204.50381679389312, "grad_norm": 2.1617515087127686, "learning_rate": 0.0001, "loss": 0.0207, "step": 26790 }, { "epoch": 204.5801526717557, "grad_norm": 2.3127293586730957, "learning_rate": 0.0001, "loss": 0.0204, "step": 26800 }, { "epoch": 204.65648854961833, "grad_norm": 2.0551235675811768, "learning_rate": 0.0001, "loss": 0.0213, "step": 26810 }, { "epoch": 204.73282442748092, "grad_norm": 2.7467575073242188, "learning_rate": 0.0001, "loss": 0.0205, "step": 26820 }, { "epoch": 204.8091603053435, "grad_norm": 1.4230047464370728, "learning_rate": 0.0001, "loss": 0.0213, "step": 26830 }, { "epoch": 204.8854961832061, "grad_norm": 1.6353328227996826, "learning_rate": 0.0001, "loss": 0.0207, "step": 26840 }, { "epoch": 204.9618320610687, "grad_norm": 2.271705150604248, "learning_rate": 0.0001, "loss": 0.0211, "step": 26850 }, { "epoch": 205.0381679389313, "grad_norm": 2.158505439758301, "learning_rate": 0.0001, "loss": 0.0209, "step": 26860 }, { "epoch": 205.1145038167939, "grad_norm": 1.4265183210372925, "learning_rate": 0.0001, "loss": 0.0207, "step": 26870 }, { "epoch": 205.1908396946565, "grad_norm": 1.7563475370407104, "learning_rate": 0.0001, "loss": 0.0202, "step": 26880 }, { "epoch": 205.26717557251908, "grad_norm": 1.5354632139205933, "learning_rate": 0.0001, "loss": 0.0207, "step": 26890 }, { "epoch": 205.34351145038167, "grad_norm": 1.3487199544906616, "learning_rate": 0.0001, "loss": 0.02, "step": 26900 }, { "epoch": 205.4198473282443, "grad_norm": 1.8316227197647095, "learning_rate": 0.0001, "loss": 0.0206, "step": 26910 }, { "epoch": 205.49618320610688, "grad_norm": 1.5117100477218628, "learning_rate": 0.0001, "loss": 0.0217, "step": 26920 }, { "epoch": 205.57251908396947, "grad_norm": 2.0004680156707764, "learning_rate": 0.0001, "loss": 0.0223, "step": 26930 }, { "epoch": 205.64885496183206, "grad_norm": 1.6143884658813477, "learning_rate": 0.0001, "loss": 0.0221, "step": 26940 }, { "epoch": 205.72519083969465, "grad_norm": 1.9144108295440674, "learning_rate": 0.0001, "loss": 0.0216, "step": 26950 }, { "epoch": 205.80152671755727, "grad_norm": 1.4758294820785522, "learning_rate": 0.0001, "loss": 0.0209, "step": 26960 }, { "epoch": 205.87786259541986, "grad_norm": 1.335221290588379, "learning_rate": 0.0001, "loss": 0.0209, "step": 26970 }, { "epoch": 205.95419847328245, "grad_norm": 1.6360411643981934, "learning_rate": 0.0001, "loss": 0.0208, "step": 26980 }, { "epoch": 206.03053435114504, "grad_norm": 1.5708142518997192, "learning_rate": 0.0001, "loss": 0.0209, "step": 26990 }, { "epoch": 206.10687022900763, "grad_norm": 2.533008098602295, "learning_rate": 0.0001, "loss": 0.021, "step": 27000 }, { "epoch": 206.18320610687022, "grad_norm": 1.8295315504074097, "learning_rate": 0.0001, "loss": 0.0214, "step": 27010 }, { "epoch": 206.25954198473283, "grad_norm": 1.9898263216018677, "learning_rate": 0.0001, "loss": 0.0207, "step": 27020 }, { "epoch": 206.33587786259542, "grad_norm": 2.1131770610809326, "learning_rate": 0.0001, "loss": 0.0207, "step": 27030 }, { "epoch": 206.41221374045801, "grad_norm": 1.8620904684066772, "learning_rate": 0.0001, "loss": 0.0202, "step": 27040 }, { "epoch": 206.4885496183206, "grad_norm": 1.9555779695510864, "learning_rate": 0.0001, "loss": 0.02, "step": 27050 }, { "epoch": 206.5648854961832, "grad_norm": 1.714107632637024, "learning_rate": 0.0001, "loss": 0.0196, "step": 27060 }, { "epoch": 206.6412213740458, "grad_norm": 1.637515664100647, "learning_rate": 0.0001, "loss": 0.0196, "step": 27070 }, { "epoch": 206.7175572519084, "grad_norm": 1.733701467514038, "learning_rate": 0.0001, "loss": 0.0197, "step": 27080 }, { "epoch": 206.793893129771, "grad_norm": 1.8132163286209106, "learning_rate": 0.0001, "loss": 0.0205, "step": 27090 }, { "epoch": 206.87022900763358, "grad_norm": 1.6314431428909302, "learning_rate": 0.0001, "loss": 0.0203, "step": 27100 }, { "epoch": 206.94656488549617, "grad_norm": 2.592583179473877, "learning_rate": 0.0001, "loss": 0.0206, "step": 27110 }, { "epoch": 207.0229007633588, "grad_norm": 1.3299411535263062, "learning_rate": 0.0001, "loss": 0.0204, "step": 27120 }, { "epoch": 207.09923664122138, "grad_norm": 2.624154806137085, "learning_rate": 0.0001, "loss": 0.0206, "step": 27130 }, { "epoch": 207.17557251908397, "grad_norm": 2.8226912021636963, "learning_rate": 0.0001, "loss": 0.0214, "step": 27140 }, { "epoch": 207.25190839694656, "grad_norm": 1.45262610912323, "learning_rate": 0.0001, "loss": 0.0204, "step": 27150 }, { "epoch": 207.32824427480915, "grad_norm": 1.874950647354126, "learning_rate": 0.0001, "loss": 0.0197, "step": 27160 }, { "epoch": 207.40458015267177, "grad_norm": 1.8700401782989502, "learning_rate": 0.0001, "loss": 0.0208, "step": 27170 }, { "epoch": 207.48091603053436, "grad_norm": 1.719325304031372, "learning_rate": 0.0001, "loss": 0.0205, "step": 27180 }, { "epoch": 207.55725190839695, "grad_norm": 1.9141298532485962, "learning_rate": 0.0001, "loss": 0.0204, "step": 27190 }, { "epoch": 207.63358778625954, "grad_norm": 1.4990153312683105, "learning_rate": 0.0001, "loss": 0.0203, "step": 27200 }, { "epoch": 207.70992366412213, "grad_norm": 1.4991377592086792, "learning_rate": 0.0001, "loss": 0.0205, "step": 27210 }, { "epoch": 207.78625954198472, "grad_norm": 1.3223329782485962, "learning_rate": 0.0001, "loss": 0.0214, "step": 27220 }, { "epoch": 207.86259541984734, "grad_norm": 2.068871021270752, "learning_rate": 0.0001, "loss": 0.0209, "step": 27230 }, { "epoch": 207.93893129770993, "grad_norm": 2.445418357849121, "learning_rate": 0.0001, "loss": 0.0208, "step": 27240 }, { "epoch": 208.01526717557252, "grad_norm": 1.5885541439056396, "learning_rate": 0.0001, "loss": 0.0203, "step": 27250 }, { "epoch": 208.0916030534351, "grad_norm": 2.366514205932617, "learning_rate": 0.0001, "loss": 0.0204, "step": 27260 }, { "epoch": 208.1679389312977, "grad_norm": 2.0146467685699463, "learning_rate": 0.0001, "loss": 0.0202, "step": 27270 }, { "epoch": 208.24427480916032, "grad_norm": 1.3837987184524536, "learning_rate": 0.0001, "loss": 0.0214, "step": 27280 }, { "epoch": 208.3206106870229, "grad_norm": 1.3823758363723755, "learning_rate": 0.0001, "loss": 0.0209, "step": 27290 }, { "epoch": 208.3969465648855, "grad_norm": 1.1021032333374023, "learning_rate": 0.0001, "loss": 0.0205, "step": 27300 }, { "epoch": 208.4732824427481, "grad_norm": 1.5127923488616943, "learning_rate": 0.0001, "loss": 0.0211, "step": 27310 }, { "epoch": 208.54961832061068, "grad_norm": 2.3904812335968018, "learning_rate": 0.0001, "loss": 0.021, "step": 27320 }, { "epoch": 208.6259541984733, "grad_norm": 1.343209147453308, "learning_rate": 0.0001, "loss": 0.0213, "step": 27330 }, { "epoch": 208.70229007633588, "grad_norm": 1.5305612087249756, "learning_rate": 0.0001, "loss": 0.0207, "step": 27340 }, { "epoch": 208.77862595419847, "grad_norm": 1.8190057277679443, "learning_rate": 0.0001, "loss": 0.0207, "step": 27350 }, { "epoch": 208.85496183206106, "grad_norm": 1.931143879890442, "learning_rate": 0.0001, "loss": 0.0211, "step": 27360 }, { "epoch": 208.93129770992365, "grad_norm": 2.1505002975463867, "learning_rate": 0.0001, "loss": 0.0214, "step": 27370 }, { "epoch": 209.00763358778627, "grad_norm": 1.4709300994873047, "learning_rate": 0.0001, "loss": 0.0206, "step": 27380 }, { "epoch": 209.08396946564886, "grad_norm": 1.973481297492981, "learning_rate": 0.0001, "loss": 0.0212, "step": 27390 }, { "epoch": 209.16030534351145, "grad_norm": 2.176262855529785, "learning_rate": 0.0001, "loss": 0.0209, "step": 27400 }, { "epoch": 209.23664122137404, "grad_norm": 2.652657985687256, "learning_rate": 0.0001, "loss": 0.0209, "step": 27410 }, { "epoch": 209.31297709923663, "grad_norm": 1.6883821487426758, "learning_rate": 0.0001, "loss": 0.0208, "step": 27420 }, { "epoch": 209.38931297709922, "grad_norm": 2.7805442810058594, "learning_rate": 0.0001, "loss": 0.0206, "step": 27430 }, { "epoch": 209.46564885496184, "grad_norm": 1.8260077238082886, "learning_rate": 0.0001, "loss": 0.0219, "step": 27440 }, { "epoch": 209.54198473282443, "grad_norm": 2.0365915298461914, "learning_rate": 0.0001, "loss": 0.0224, "step": 27450 }, { "epoch": 209.61832061068702, "grad_norm": 1.9132717847824097, "learning_rate": 0.0001, "loss": 0.0212, "step": 27460 }, { "epoch": 209.6946564885496, "grad_norm": 2.2025821208953857, "learning_rate": 0.0001, "loss": 0.0209, "step": 27470 }, { "epoch": 209.7709923664122, "grad_norm": 1.6673763990402222, "learning_rate": 0.0001, "loss": 0.021, "step": 27480 }, { "epoch": 209.84732824427482, "grad_norm": 2.1249051094055176, "learning_rate": 0.0001, "loss": 0.0202, "step": 27490 }, { "epoch": 209.9236641221374, "grad_norm": 1.3058847188949585, "learning_rate": 0.0001, "loss": 0.0199, "step": 27500 }, { "epoch": 210.0, "grad_norm": 2.1890225410461426, "learning_rate": 0.0001, "loss": 0.0199, "step": 27510 }, { "epoch": 210.0763358778626, "grad_norm": 1.5765471458435059, "learning_rate": 0.0001, "loss": 0.0204, "step": 27520 }, { "epoch": 210.15267175572518, "grad_norm": 1.7995563745498657, "learning_rate": 0.0001, "loss": 0.02, "step": 27530 }, { "epoch": 210.2290076335878, "grad_norm": 1.7902801036834717, "learning_rate": 0.0001, "loss": 0.0207, "step": 27540 }, { "epoch": 210.3053435114504, "grad_norm": 2.5412657260894775, "learning_rate": 0.0001, "loss": 0.0208, "step": 27550 }, { "epoch": 210.38167938931298, "grad_norm": 1.8451683521270752, "learning_rate": 0.0001, "loss": 0.0204, "step": 27560 }, { "epoch": 210.45801526717557, "grad_norm": 1.9085943698883057, "learning_rate": 0.0001, "loss": 0.0199, "step": 27570 }, { "epoch": 210.53435114503816, "grad_norm": 1.653106451034546, "learning_rate": 0.0001, "loss": 0.0207, "step": 27580 }, { "epoch": 210.61068702290078, "grad_norm": 1.7299699783325195, "learning_rate": 0.0001, "loss": 0.0202, "step": 27590 }, { "epoch": 210.68702290076337, "grad_norm": 1.4467660188674927, "learning_rate": 0.0001, "loss": 0.021, "step": 27600 }, { "epoch": 210.76335877862596, "grad_norm": 1.83334481716156, "learning_rate": 0.0001, "loss": 0.0212, "step": 27610 }, { "epoch": 210.83969465648855, "grad_norm": 1.6279795169830322, "learning_rate": 0.0001, "loss": 0.0203, "step": 27620 }, { "epoch": 210.91603053435114, "grad_norm": 2.063584804534912, "learning_rate": 0.0001, "loss": 0.021, "step": 27630 }, { "epoch": 210.99236641221373, "grad_norm": 1.5662627220153809, "learning_rate": 0.0001, "loss": 0.0203, "step": 27640 }, { "epoch": 211.06870229007635, "grad_norm": 2.275730848312378, "learning_rate": 0.0001, "loss": 0.0202, "step": 27650 }, { "epoch": 211.14503816793894, "grad_norm": 1.693025827407837, "learning_rate": 0.0001, "loss": 0.0205, "step": 27660 }, { "epoch": 211.22137404580153, "grad_norm": 1.6300643682479858, "learning_rate": 0.0001, "loss": 0.0207, "step": 27670 }, { "epoch": 211.29770992366412, "grad_norm": 1.8936760425567627, "learning_rate": 0.0001, "loss": 0.0214, "step": 27680 }, { "epoch": 211.3740458015267, "grad_norm": 2.1798923015594482, "learning_rate": 0.0001, "loss": 0.0215, "step": 27690 }, { "epoch": 211.45038167938932, "grad_norm": 1.7693767547607422, "learning_rate": 0.0001, "loss": 0.0213, "step": 27700 }, { "epoch": 211.5267175572519, "grad_norm": 1.5978628396987915, "learning_rate": 0.0001, "loss": 0.0207, "step": 27710 }, { "epoch": 211.6030534351145, "grad_norm": 1.5217336416244507, "learning_rate": 0.0001, "loss": 0.0202, "step": 27720 }, { "epoch": 211.6793893129771, "grad_norm": 1.68646240234375, "learning_rate": 0.0001, "loss": 0.0198, "step": 27730 }, { "epoch": 211.75572519083968, "grad_norm": 2.1971261501312256, "learning_rate": 0.0001, "loss": 0.0213, "step": 27740 }, { "epoch": 211.8320610687023, "grad_norm": 1.5170910358428955, "learning_rate": 0.0001, "loss": 0.0208, "step": 27750 }, { "epoch": 211.9083969465649, "grad_norm": 2.242201089859009, "learning_rate": 0.0001, "loss": 0.0209, "step": 27760 }, { "epoch": 211.98473282442748, "grad_norm": 1.576088547706604, "learning_rate": 0.0001, "loss": 0.0204, "step": 27770 }, { "epoch": 212.06106870229007, "grad_norm": 1.311453104019165, "learning_rate": 0.0001, "loss": 0.0204, "step": 27780 }, { "epoch": 212.13740458015266, "grad_norm": 1.8859491348266602, "learning_rate": 0.0001, "loss": 0.0207, "step": 27790 }, { "epoch": 212.21374045801528, "grad_norm": 2.2696444988250732, "learning_rate": 0.0001, "loss": 0.022, "step": 27800 }, { "epoch": 212.29007633587787, "grad_norm": 1.8434621095657349, "learning_rate": 0.0001, "loss": 0.0218, "step": 27810 }, { "epoch": 212.36641221374046, "grad_norm": 1.5585198402404785, "learning_rate": 0.0001, "loss": 0.0212, "step": 27820 }, { "epoch": 212.44274809160305, "grad_norm": 2.306368827819824, "learning_rate": 0.0001, "loss": 0.0206, "step": 27830 }, { "epoch": 212.51908396946564, "grad_norm": 1.69815194606781, "learning_rate": 0.0001, "loss": 0.0205, "step": 27840 }, { "epoch": 212.59541984732823, "grad_norm": 1.673020601272583, "learning_rate": 0.0001, "loss": 0.0204, "step": 27850 }, { "epoch": 212.67175572519085, "grad_norm": 1.5570849180221558, "learning_rate": 0.0001, "loss": 0.0196, "step": 27860 }, { "epoch": 212.74809160305344, "grad_norm": 1.4678293466567993, "learning_rate": 0.0001, "loss": 0.0199, "step": 27870 }, { "epoch": 212.82442748091603, "grad_norm": 1.788081169128418, "learning_rate": 0.0001, "loss": 0.0199, "step": 27880 }, { "epoch": 212.90076335877862, "grad_norm": 1.457740068435669, "learning_rate": 0.0001, "loss": 0.0211, "step": 27890 }, { "epoch": 212.9770992366412, "grad_norm": 2.3756232261657715, "learning_rate": 0.0001, "loss": 0.0203, "step": 27900 }, { "epoch": 213.05343511450383, "grad_norm": 1.7439686059951782, "learning_rate": 0.0001, "loss": 0.0204, "step": 27910 }, { "epoch": 213.12977099236642, "grad_norm": 1.316028356552124, "learning_rate": 0.0001, "loss": 0.0202, "step": 27920 }, { "epoch": 213.206106870229, "grad_norm": 2.7539069652557373, "learning_rate": 0.0001, "loss": 0.02, "step": 27930 }, { "epoch": 213.2824427480916, "grad_norm": 2.437589645385742, "learning_rate": 0.0001, "loss": 0.0204, "step": 27940 }, { "epoch": 213.3587786259542, "grad_norm": 1.4251718521118164, "learning_rate": 0.0001, "loss": 0.02, "step": 27950 }, { "epoch": 213.4351145038168, "grad_norm": 1.7641780376434326, "learning_rate": 0.0001, "loss": 0.0202, "step": 27960 }, { "epoch": 213.5114503816794, "grad_norm": 1.3660343885421753, "learning_rate": 0.0001, "loss": 0.0206, "step": 27970 }, { "epoch": 213.58778625954199, "grad_norm": 1.4448049068450928, "learning_rate": 0.0001, "loss": 0.0208, "step": 27980 }, { "epoch": 213.66412213740458, "grad_norm": 1.6302624940872192, "learning_rate": 0.0001, "loss": 0.0206, "step": 27990 }, { "epoch": 213.74045801526717, "grad_norm": 1.6430072784423828, "learning_rate": 0.0001, "loss": 0.0203, "step": 28000 }, { "epoch": 213.81679389312978, "grad_norm": 1.5200703144073486, "learning_rate": 0.0001, "loss": 0.0205, "step": 28010 }, { "epoch": 213.89312977099237, "grad_norm": 1.426211953163147, "learning_rate": 0.0001, "loss": 0.0207, "step": 28020 }, { "epoch": 213.96946564885496, "grad_norm": 1.6046974658966064, "learning_rate": 0.0001, "loss": 0.0198, "step": 28030 }, { "epoch": 214.04580152671755, "grad_norm": 1.2850210666656494, "learning_rate": 0.0001, "loss": 0.0202, "step": 28040 }, { "epoch": 214.12213740458014, "grad_norm": 1.252676248550415, "learning_rate": 0.0001, "loss": 0.0201, "step": 28050 }, { "epoch": 214.19847328244273, "grad_norm": 1.2549070119857788, "learning_rate": 0.0001, "loss": 0.0195, "step": 28060 }, { "epoch": 214.27480916030535, "grad_norm": 2.0331053733825684, "learning_rate": 0.0001, "loss": 0.0202, "step": 28070 }, { "epoch": 214.35114503816794, "grad_norm": 2.446725368499756, "learning_rate": 0.0001, "loss": 0.0202, "step": 28080 }, { "epoch": 214.42748091603053, "grad_norm": 1.335269570350647, "learning_rate": 0.0001, "loss": 0.0201, "step": 28090 }, { "epoch": 214.50381679389312, "grad_norm": 2.0982000827789307, "learning_rate": 0.0001, "loss": 0.02, "step": 28100 }, { "epoch": 214.5801526717557, "grad_norm": 1.572153091430664, "learning_rate": 0.0001, "loss": 0.0201, "step": 28110 }, { "epoch": 214.65648854961833, "grad_norm": 1.6363229751586914, "learning_rate": 0.0001, "loss": 0.0195, "step": 28120 }, { "epoch": 214.73282442748092, "grad_norm": 1.3012871742248535, "learning_rate": 0.0001, "loss": 0.0191, "step": 28130 }, { "epoch": 214.8091603053435, "grad_norm": 2.4825353622436523, "learning_rate": 0.0001, "loss": 0.0193, "step": 28140 }, { "epoch": 214.8854961832061, "grad_norm": 2.3354010581970215, "learning_rate": 0.0001, "loss": 0.0191, "step": 28150 }, { "epoch": 214.9618320610687, "grad_norm": 1.6258134841918945, "learning_rate": 0.0001, "loss": 0.0196, "step": 28160 }, { "epoch": 215.0381679389313, "grad_norm": 1.460593581199646, "learning_rate": 0.0001, "loss": 0.02, "step": 28170 }, { "epoch": 215.1145038167939, "grad_norm": 1.3480240106582642, "learning_rate": 0.0001, "loss": 0.0197, "step": 28180 }, { "epoch": 215.1908396946565, "grad_norm": 2.0152087211608887, "learning_rate": 0.0001, "loss": 0.0206, "step": 28190 }, { "epoch": 215.26717557251908, "grad_norm": 1.54742431640625, "learning_rate": 0.0001, "loss": 0.0204, "step": 28200 }, { "epoch": 215.34351145038167, "grad_norm": 1.345720648765564, "learning_rate": 0.0001, "loss": 0.0209, "step": 28210 }, { "epoch": 215.4198473282443, "grad_norm": 2.147447347640991, "learning_rate": 0.0001, "loss": 0.021, "step": 28220 }, { "epoch": 215.49618320610688, "grad_norm": 1.4371258020401, "learning_rate": 0.0001, "loss": 0.0217, "step": 28230 }, { "epoch": 215.57251908396947, "grad_norm": 1.5566226243972778, "learning_rate": 0.0001, "loss": 0.0218, "step": 28240 }, { "epoch": 215.64885496183206, "grad_norm": 1.7352979183197021, "learning_rate": 0.0001, "loss": 0.0211, "step": 28250 }, { "epoch": 215.72519083969465, "grad_norm": 1.7985960245132446, "learning_rate": 0.0001, "loss": 0.0212, "step": 28260 }, { "epoch": 215.80152671755727, "grad_norm": 2.2911436557769775, "learning_rate": 0.0001, "loss": 0.0204, "step": 28270 }, { "epoch": 215.87786259541986, "grad_norm": 1.3819313049316406, "learning_rate": 0.0001, "loss": 0.0198, "step": 28280 }, { "epoch": 215.95419847328245, "grad_norm": 1.9344269037246704, "learning_rate": 0.0001, "loss": 0.0197, "step": 28290 }, { "epoch": 216.03053435114504, "grad_norm": 2.1088671684265137, "learning_rate": 0.0001, "loss": 0.0198, "step": 28300 }, { "epoch": 216.10687022900763, "grad_norm": 1.4003080129623413, "learning_rate": 0.0001, "loss": 0.0201, "step": 28310 }, { "epoch": 216.18320610687022, "grad_norm": 1.954113245010376, "learning_rate": 0.0001, "loss": 0.0199, "step": 28320 }, { "epoch": 216.25954198473283, "grad_norm": 1.5311298370361328, "learning_rate": 0.0001, "loss": 0.0197, "step": 28330 }, { "epoch": 216.33587786259542, "grad_norm": 1.8769078254699707, "learning_rate": 0.0001, "loss": 0.02, "step": 28340 }, { "epoch": 216.41221374045801, "grad_norm": 1.5963352918624878, "learning_rate": 0.0001, "loss": 0.0207, "step": 28350 }, { "epoch": 216.4885496183206, "grad_norm": 1.9980591535568237, "learning_rate": 0.0001, "loss": 0.02, "step": 28360 }, { "epoch": 216.5648854961832, "grad_norm": 1.5913280248641968, "learning_rate": 0.0001, "loss": 0.0198, "step": 28370 }, { "epoch": 216.6412213740458, "grad_norm": 2.276242256164551, "learning_rate": 0.0001, "loss": 0.0206, "step": 28380 }, { "epoch": 216.7175572519084, "grad_norm": 2.182882308959961, "learning_rate": 0.0001, "loss": 0.0208, "step": 28390 }, { "epoch": 216.793893129771, "grad_norm": 1.9120475053787231, "learning_rate": 0.0001, "loss": 0.0216, "step": 28400 }, { "epoch": 216.87022900763358, "grad_norm": 2.075226068496704, "learning_rate": 0.0001, "loss": 0.0211, "step": 28410 }, { "epoch": 216.94656488549617, "grad_norm": 1.5452800989151, "learning_rate": 0.0001, "loss": 0.0208, "step": 28420 }, { "epoch": 217.0229007633588, "grad_norm": 1.9438483715057373, "learning_rate": 0.0001, "loss": 0.0204, "step": 28430 }, { "epoch": 217.09923664122138, "grad_norm": 1.5433224439620972, "learning_rate": 0.0001, "loss": 0.0203, "step": 28440 }, { "epoch": 217.17557251908397, "grad_norm": 2.0382652282714844, "learning_rate": 0.0001, "loss": 0.0202, "step": 28450 }, { "epoch": 217.25190839694656, "grad_norm": 2.2394495010375977, "learning_rate": 0.0001, "loss": 0.0204, "step": 28460 }, { "epoch": 217.32824427480915, "grad_norm": 1.3152090311050415, "learning_rate": 0.0001, "loss": 0.0195, "step": 28470 }, { "epoch": 217.40458015267177, "grad_norm": 1.7599700689315796, "learning_rate": 0.0001, "loss": 0.0196, "step": 28480 }, { "epoch": 217.48091603053436, "grad_norm": 1.3538343906402588, "learning_rate": 0.0001, "loss": 0.0194, "step": 28490 }, { "epoch": 217.55725190839695, "grad_norm": 1.4631637334823608, "learning_rate": 0.0001, "loss": 0.0204, "step": 28500 }, { "epoch": 217.63358778625954, "grad_norm": 2.3358054161071777, "learning_rate": 0.0001, "loss": 0.021, "step": 28510 }, { "epoch": 217.70992366412213, "grad_norm": 1.619722843170166, "learning_rate": 0.0001, "loss": 0.0213, "step": 28520 }, { "epoch": 217.78625954198472, "grad_norm": 1.0547622442245483, "learning_rate": 0.0001, "loss": 0.0221, "step": 28530 }, { "epoch": 217.86259541984734, "grad_norm": 1.3186780214309692, "learning_rate": 0.0001, "loss": 0.0215, "step": 28540 }, { "epoch": 217.93893129770993, "grad_norm": 2.051070213317871, "learning_rate": 0.0001, "loss": 0.0216, "step": 28550 }, { "epoch": 218.01526717557252, "grad_norm": 1.4979239702224731, "learning_rate": 0.0001, "loss": 0.0224, "step": 28560 }, { "epoch": 218.0916030534351, "grad_norm": 1.05502450466156, "learning_rate": 0.0001, "loss": 0.0217, "step": 28570 }, { "epoch": 218.1679389312977, "grad_norm": 1.5763208866119385, "learning_rate": 0.0001, "loss": 0.0213, "step": 28580 }, { "epoch": 218.24427480916032, "grad_norm": 1.359687089920044, "learning_rate": 0.0001, "loss": 0.0222, "step": 28590 }, { "epoch": 218.3206106870229, "grad_norm": 1.8959916830062866, "learning_rate": 0.0001, "loss": 0.0219, "step": 28600 }, { "epoch": 218.3969465648855, "grad_norm": 1.3633463382720947, "learning_rate": 0.0001, "loss": 0.0214, "step": 28610 }, { "epoch": 218.4732824427481, "grad_norm": 1.262142539024353, "learning_rate": 0.0001, "loss": 0.021, "step": 28620 }, { "epoch": 218.54961832061068, "grad_norm": 1.4329779148101807, "learning_rate": 0.0001, "loss": 0.0216, "step": 28630 }, { "epoch": 218.6259541984733, "grad_norm": 2.13632869720459, "learning_rate": 0.0001, "loss": 0.0214, "step": 28640 }, { "epoch": 218.70229007633588, "grad_norm": 1.529595971107483, "learning_rate": 0.0001, "loss": 0.0216, "step": 28650 }, { "epoch": 218.77862595419847, "grad_norm": 1.5816082954406738, "learning_rate": 0.0001, "loss": 0.0208, "step": 28660 }, { "epoch": 218.85496183206106, "grad_norm": 2.2113466262817383, "learning_rate": 0.0001, "loss": 0.0206, "step": 28670 }, { "epoch": 218.93129770992365, "grad_norm": 1.807401180267334, "learning_rate": 0.0001, "loss": 0.0204, "step": 28680 }, { "epoch": 219.00763358778627, "grad_norm": 2.241530179977417, "learning_rate": 0.0001, "loss": 0.0203, "step": 28690 }, { "epoch": 219.08396946564886, "grad_norm": 1.7928316593170166, "learning_rate": 0.0001, "loss": 0.0198, "step": 28700 }, { "epoch": 219.16030534351145, "grad_norm": 2.3072621822357178, "learning_rate": 0.0001, "loss": 0.0215, "step": 28710 }, { "epoch": 219.23664122137404, "grad_norm": 1.7990803718566895, "learning_rate": 0.0001, "loss": 0.0207, "step": 28720 }, { "epoch": 219.31297709923663, "grad_norm": 1.1606738567352295, "learning_rate": 0.0001, "loss": 0.0201, "step": 28730 }, { "epoch": 219.38931297709922, "grad_norm": 1.2079836130142212, "learning_rate": 0.0001, "loss": 0.0191, "step": 28740 }, { "epoch": 219.46564885496184, "grad_norm": 1.3179863691329956, "learning_rate": 0.0001, "loss": 0.0197, "step": 28750 }, { "epoch": 219.54198473282443, "grad_norm": 1.242875576019287, "learning_rate": 0.0001, "loss": 0.0204, "step": 28760 }, { "epoch": 219.61832061068702, "grad_norm": 1.257846474647522, "learning_rate": 0.0001, "loss": 0.02, "step": 28770 }, { "epoch": 219.6946564885496, "grad_norm": 2.403452157974243, "learning_rate": 0.0001, "loss": 0.021, "step": 28780 }, { "epoch": 219.7709923664122, "grad_norm": 1.7474406957626343, "learning_rate": 0.0001, "loss": 0.0206, "step": 28790 }, { "epoch": 219.84732824427482, "grad_norm": 1.6901497840881348, "learning_rate": 0.0001, "loss": 0.0204, "step": 28800 }, { "epoch": 219.9236641221374, "grad_norm": 1.1231333017349243, "learning_rate": 0.0001, "loss": 0.0205, "step": 28810 }, { "epoch": 220.0, "grad_norm": 2.144416332244873, "learning_rate": 0.0001, "loss": 0.0205, "step": 28820 }, { "epoch": 220.0763358778626, "grad_norm": 1.3721556663513184, "learning_rate": 0.0001, "loss": 0.0202, "step": 28830 }, { "epoch": 220.15267175572518, "grad_norm": 2.124152421951294, "learning_rate": 0.0001, "loss": 0.0204, "step": 28840 }, { "epoch": 220.2290076335878, "grad_norm": 1.6405678987503052, "learning_rate": 0.0001, "loss": 0.0204, "step": 28850 }, { "epoch": 220.3053435114504, "grad_norm": 2.204699993133545, "learning_rate": 0.0001, "loss": 0.0199, "step": 28860 }, { "epoch": 220.38167938931298, "grad_norm": 2.3953118324279785, "learning_rate": 0.0001, "loss": 0.0196, "step": 28870 }, { "epoch": 220.45801526717557, "grad_norm": 1.7726300954818726, "learning_rate": 0.0001, "loss": 0.0201, "step": 28880 }, { "epoch": 220.53435114503816, "grad_norm": 1.2733145952224731, "learning_rate": 0.0001, "loss": 0.0197, "step": 28890 }, { "epoch": 220.61068702290078, "grad_norm": 1.4659887552261353, "learning_rate": 0.0001, "loss": 0.0195, "step": 28900 }, { "epoch": 220.68702290076337, "grad_norm": 2.003513813018799, "learning_rate": 0.0001, "loss": 0.0203, "step": 28910 }, { "epoch": 220.76335877862596, "grad_norm": 1.7781622409820557, "learning_rate": 0.0001, "loss": 0.0206, "step": 28920 }, { "epoch": 220.83969465648855, "grad_norm": 1.9630175828933716, "learning_rate": 0.0001, "loss": 0.0206, "step": 28930 }, { "epoch": 220.91603053435114, "grad_norm": 1.593380093574524, "learning_rate": 0.0001, "loss": 0.0203, "step": 28940 }, { "epoch": 220.99236641221373, "grad_norm": 1.740178108215332, "learning_rate": 0.0001, "loss": 0.0209, "step": 28950 }, { "epoch": 221.06870229007635, "grad_norm": 1.4656589031219482, "learning_rate": 0.0001, "loss": 0.0201, "step": 28960 }, { "epoch": 221.14503816793894, "grad_norm": 2.1234357357025146, "learning_rate": 0.0001, "loss": 0.0209, "step": 28970 }, { "epoch": 221.22137404580153, "grad_norm": 1.7884838581085205, "learning_rate": 0.0001, "loss": 0.0208, "step": 28980 }, { "epoch": 221.29770992366412, "grad_norm": 1.7491395473480225, "learning_rate": 0.0001, "loss": 0.02, "step": 28990 }, { "epoch": 221.3740458015267, "grad_norm": 1.2766375541687012, "learning_rate": 0.0001, "loss": 0.0201, "step": 29000 }, { "epoch": 221.45038167938932, "grad_norm": 1.3567336797714233, "learning_rate": 0.0001, "loss": 0.0203, "step": 29010 }, { "epoch": 221.5267175572519, "grad_norm": 1.1771137714385986, "learning_rate": 0.0001, "loss": 0.0198, "step": 29020 }, { "epoch": 221.6030534351145, "grad_norm": 1.243779182434082, "learning_rate": 0.0001, "loss": 0.0197, "step": 29030 }, { "epoch": 221.6793893129771, "grad_norm": 1.3523719310760498, "learning_rate": 0.0001, "loss": 0.0203, "step": 29040 }, { "epoch": 221.75572519083968, "grad_norm": 1.4425909519195557, "learning_rate": 0.0001, "loss": 0.0205, "step": 29050 }, { "epoch": 221.8320610687023, "grad_norm": 1.3716975450515747, "learning_rate": 0.0001, "loss": 0.0203, "step": 29060 }, { "epoch": 221.9083969465649, "grad_norm": 1.3359591960906982, "learning_rate": 0.0001, "loss": 0.02, "step": 29070 }, { "epoch": 221.98473282442748, "grad_norm": 1.536983609199524, "learning_rate": 0.0001, "loss": 0.02, "step": 29080 }, { "epoch": 222.06106870229007, "grad_norm": 2.1527931690216064, "learning_rate": 0.0001, "loss": 0.02, "step": 29090 }, { "epoch": 222.13740458015266, "grad_norm": 1.40152108669281, "learning_rate": 0.0001, "loss": 0.02, "step": 29100 }, { "epoch": 222.21374045801528, "grad_norm": 1.6010959148406982, "learning_rate": 0.0001, "loss": 0.0197, "step": 29110 }, { "epoch": 222.29007633587787, "grad_norm": 1.9295575618743896, "learning_rate": 0.0001, "loss": 0.0203, "step": 29120 }, { "epoch": 222.36641221374046, "grad_norm": 1.7209798097610474, "learning_rate": 0.0001, "loss": 0.0204, "step": 29130 }, { "epoch": 222.44274809160305, "grad_norm": 1.4537174701690674, "learning_rate": 0.0001, "loss": 0.0199, "step": 29140 }, { "epoch": 222.51908396946564, "grad_norm": 1.3006707429885864, "learning_rate": 0.0001, "loss": 0.0195, "step": 29150 }, { "epoch": 222.59541984732823, "grad_norm": 1.3730390071868896, "learning_rate": 0.0001, "loss": 0.0199, "step": 29160 }, { "epoch": 222.67175572519085, "grad_norm": 1.5900877714157104, "learning_rate": 0.0001, "loss": 0.0199, "step": 29170 }, { "epoch": 222.74809160305344, "grad_norm": 1.5293546915054321, "learning_rate": 0.0001, "loss": 0.0199, "step": 29180 }, { "epoch": 222.82442748091603, "grad_norm": 2.1309456825256348, "learning_rate": 0.0001, "loss": 0.0198, "step": 29190 }, { "epoch": 222.90076335877862, "grad_norm": 1.4649724960327148, "learning_rate": 0.0001, "loss": 0.0195, "step": 29200 }, { "epoch": 222.9770992366412, "grad_norm": 1.634232521057129, "learning_rate": 0.0001, "loss": 0.0199, "step": 29210 }, { "epoch": 223.05343511450383, "grad_norm": 0.9574413299560547, "learning_rate": 0.0001, "loss": 0.0192, "step": 29220 }, { "epoch": 223.12977099236642, "grad_norm": 2.3139431476593018, "learning_rate": 0.0001, "loss": 0.02, "step": 29230 }, { "epoch": 223.206106870229, "grad_norm": 1.1041975021362305, "learning_rate": 0.0001, "loss": 0.0206, "step": 29240 }, { "epoch": 223.2824427480916, "grad_norm": 1.5777868032455444, "learning_rate": 0.0001, "loss": 0.0206, "step": 29250 }, { "epoch": 223.3587786259542, "grad_norm": 1.7810662984848022, "learning_rate": 0.0001, "loss": 0.0212, "step": 29260 }, { "epoch": 223.4351145038168, "grad_norm": 1.388418197631836, "learning_rate": 0.0001, "loss": 0.0203, "step": 29270 }, { "epoch": 223.5114503816794, "grad_norm": 2.0364205837249756, "learning_rate": 0.0001, "loss": 0.0215, "step": 29280 }, { "epoch": 223.58778625954199, "grad_norm": 2.080505132675171, "learning_rate": 0.0001, "loss": 0.0213, "step": 29290 }, { "epoch": 223.66412213740458, "grad_norm": 2.0756709575653076, "learning_rate": 0.0001, "loss": 0.0201, "step": 29300 }, { "epoch": 223.74045801526717, "grad_norm": 1.6004127264022827, "learning_rate": 0.0001, "loss": 0.0203, "step": 29310 }, { "epoch": 223.81679389312978, "grad_norm": 1.9730418920516968, "learning_rate": 0.0001, "loss": 0.0199, "step": 29320 }, { "epoch": 223.89312977099237, "grad_norm": 1.8248792886734009, "learning_rate": 0.0001, "loss": 0.0197, "step": 29330 }, { "epoch": 223.96946564885496, "grad_norm": 1.8192909955978394, "learning_rate": 0.0001, "loss": 0.0191, "step": 29340 }, { "epoch": 224.04580152671755, "grad_norm": 2.043994665145874, "learning_rate": 0.0001, "loss": 0.0189, "step": 29350 }, { "epoch": 224.12213740458014, "grad_norm": 2.196943521499634, "learning_rate": 0.0001, "loss": 0.0194, "step": 29360 }, { "epoch": 224.19847328244273, "grad_norm": 1.468881368637085, "learning_rate": 0.0001, "loss": 0.019, "step": 29370 }, { "epoch": 224.27480916030535, "grad_norm": 1.5317951440811157, "learning_rate": 0.0001, "loss": 0.0188, "step": 29380 }, { "epoch": 224.35114503816794, "grad_norm": 1.3502023220062256, "learning_rate": 0.0001, "loss": 0.0191, "step": 29390 }, { "epoch": 224.42748091603053, "grad_norm": 2.252312421798706, "learning_rate": 0.0001, "loss": 0.0185, "step": 29400 }, { "epoch": 224.50381679389312, "grad_norm": 2.2818620204925537, "learning_rate": 0.0001, "loss": 0.0195, "step": 29410 }, { "epoch": 224.5801526717557, "grad_norm": 2.147808790206909, "learning_rate": 0.0001, "loss": 0.0198, "step": 29420 }, { "epoch": 224.65648854961833, "grad_norm": 1.7961788177490234, "learning_rate": 0.0001, "loss": 0.0199, "step": 29430 }, { "epoch": 224.73282442748092, "grad_norm": 1.2758413553237915, "learning_rate": 0.0001, "loss": 0.0195, "step": 29440 }, { "epoch": 224.8091603053435, "grad_norm": 2.1053171157836914, "learning_rate": 0.0001, "loss": 0.0196, "step": 29450 }, { "epoch": 224.8854961832061, "grad_norm": 1.73026442527771, "learning_rate": 0.0001, "loss": 0.0192, "step": 29460 }, { "epoch": 224.9618320610687, "grad_norm": 2.1647210121154785, "learning_rate": 0.0001, "loss": 0.0197, "step": 29470 }, { "epoch": 225.0381679389313, "grad_norm": 1.9947857856750488, "learning_rate": 0.0001, "loss": 0.0191, "step": 29480 }, { "epoch": 225.1145038167939, "grad_norm": 2.2772207260131836, "learning_rate": 0.0001, "loss": 0.0193, "step": 29490 }, { "epoch": 225.1908396946565, "grad_norm": 1.8945339918136597, "learning_rate": 0.0001, "loss": 0.0189, "step": 29500 }, { "epoch": 225.26717557251908, "grad_norm": 1.2361266613006592, "learning_rate": 0.0001, "loss": 0.0191, "step": 29510 }, { "epoch": 225.34351145038167, "grad_norm": 1.133845329284668, "learning_rate": 0.0001, "loss": 0.0194, "step": 29520 }, { "epoch": 225.4198473282443, "grad_norm": 1.1246668100357056, "learning_rate": 0.0001, "loss": 0.0189, "step": 29530 }, { "epoch": 225.49618320610688, "grad_norm": 1.625278115272522, "learning_rate": 0.0001, "loss": 0.0193, "step": 29540 }, { "epoch": 225.57251908396947, "grad_norm": 1.5951130390167236, "learning_rate": 0.0001, "loss": 0.0197, "step": 29550 }, { "epoch": 225.64885496183206, "grad_norm": 2.0909759998321533, "learning_rate": 0.0001, "loss": 0.0197, "step": 29560 }, { "epoch": 225.72519083969465, "grad_norm": 1.1762276887893677, "learning_rate": 0.0001, "loss": 0.0195, "step": 29570 }, { "epoch": 225.80152671755727, "grad_norm": 1.4834656715393066, "learning_rate": 0.0001, "loss": 0.0198, "step": 29580 }, { "epoch": 225.87786259541986, "grad_norm": 1.9039586782455444, "learning_rate": 0.0001, "loss": 0.0205, "step": 29590 }, { "epoch": 225.95419847328245, "grad_norm": 1.6116594076156616, "learning_rate": 0.0001, "loss": 0.0203, "step": 29600 }, { "epoch": 226.03053435114504, "grad_norm": 2.4152326583862305, "learning_rate": 0.0001, "loss": 0.0201, "step": 29610 }, { "epoch": 226.10687022900763, "grad_norm": 1.493388295173645, "learning_rate": 0.0001, "loss": 0.0197, "step": 29620 }, { "epoch": 226.18320610687022, "grad_norm": 1.4094836711883545, "learning_rate": 0.0001, "loss": 0.0204, "step": 29630 }, { "epoch": 226.25954198473283, "grad_norm": 1.2348333597183228, "learning_rate": 0.0001, "loss": 0.0205, "step": 29640 }, { "epoch": 226.33587786259542, "grad_norm": 1.1133959293365479, "learning_rate": 0.0001, "loss": 0.0203, "step": 29650 }, { "epoch": 226.41221374045801, "grad_norm": 1.2334685325622559, "learning_rate": 0.0001, "loss": 0.0202, "step": 29660 }, { "epoch": 226.4885496183206, "grad_norm": 1.439401626586914, "learning_rate": 0.0001, "loss": 0.0197, "step": 29670 }, { "epoch": 226.5648854961832, "grad_norm": 1.8774590492248535, "learning_rate": 0.0001, "loss": 0.0197, "step": 29680 }, { "epoch": 226.6412213740458, "grad_norm": 1.7551984786987305, "learning_rate": 0.0001, "loss": 0.0199, "step": 29690 }, { "epoch": 226.7175572519084, "grad_norm": 1.9186248779296875, "learning_rate": 0.0001, "loss": 0.0201, "step": 29700 }, { "epoch": 226.793893129771, "grad_norm": 2.1258385181427, "learning_rate": 0.0001, "loss": 0.0196, "step": 29710 }, { "epoch": 226.87022900763358, "grad_norm": 1.8435635566711426, "learning_rate": 0.0001, "loss": 0.0198, "step": 29720 }, { "epoch": 226.94656488549617, "grad_norm": 1.2327368259429932, "learning_rate": 0.0001, "loss": 0.0192, "step": 29730 }, { "epoch": 227.0229007633588, "grad_norm": 2.217848300933838, "learning_rate": 0.0001, "loss": 0.0195, "step": 29740 }, { "epoch": 227.09923664122138, "grad_norm": 2.4791369438171387, "learning_rate": 0.0001, "loss": 0.0204, "step": 29750 }, { "epoch": 227.17557251908397, "grad_norm": 1.6651208400726318, "learning_rate": 0.0001, "loss": 0.0197, "step": 29760 }, { "epoch": 227.25190839694656, "grad_norm": 1.7731866836547852, "learning_rate": 0.0001, "loss": 0.0198, "step": 29770 }, { "epoch": 227.32824427480915, "grad_norm": 1.517991542816162, "learning_rate": 0.0001, "loss": 0.0192, "step": 29780 }, { "epoch": 227.40458015267177, "grad_norm": 1.721442699432373, "learning_rate": 0.0001, "loss": 0.0192, "step": 29790 }, { "epoch": 227.48091603053436, "grad_norm": 1.165387749671936, "learning_rate": 0.0001, "loss": 0.0192, "step": 29800 }, { "epoch": 227.55725190839695, "grad_norm": 1.2192063331604004, "learning_rate": 0.0001, "loss": 0.0192, "step": 29810 }, { "epoch": 227.63358778625954, "grad_norm": 1.8976054191589355, "learning_rate": 0.0001, "loss": 0.0196, "step": 29820 }, { "epoch": 227.70992366412213, "grad_norm": 1.7328932285308838, "learning_rate": 0.0001, "loss": 0.0211, "step": 29830 }, { "epoch": 227.78625954198472, "grad_norm": 1.9554579257965088, "learning_rate": 0.0001, "loss": 0.0215, "step": 29840 }, { "epoch": 227.86259541984734, "grad_norm": 1.6914628744125366, "learning_rate": 0.0001, "loss": 0.0204, "step": 29850 }, { "epoch": 227.93893129770993, "grad_norm": 1.5431219339370728, "learning_rate": 0.0001, "loss": 0.0206, "step": 29860 }, { "epoch": 228.01526717557252, "grad_norm": 1.7145832777023315, "learning_rate": 0.0001, "loss": 0.0197, "step": 29870 }, { "epoch": 228.0916030534351, "grad_norm": 1.5168288946151733, "learning_rate": 0.0001, "loss": 0.0209, "step": 29880 }, { "epoch": 228.1679389312977, "grad_norm": 1.4598968029022217, "learning_rate": 0.0001, "loss": 0.0204, "step": 29890 }, { "epoch": 228.24427480916032, "grad_norm": 1.6064609289169312, "learning_rate": 0.0001, "loss": 0.0202, "step": 29900 }, { "epoch": 228.3206106870229, "grad_norm": 1.2886296510696411, "learning_rate": 0.0001, "loss": 0.0208, "step": 29910 }, { "epoch": 228.3969465648855, "grad_norm": 2.6080002784729004, "learning_rate": 0.0001, "loss": 0.0208, "step": 29920 }, { "epoch": 228.4732824427481, "grad_norm": 2.8673348426818848, "learning_rate": 0.0001, "loss": 0.0208, "step": 29930 }, { "epoch": 228.54961832061068, "grad_norm": 2.3542916774749756, "learning_rate": 0.0001, "loss": 0.0198, "step": 29940 }, { "epoch": 228.6259541984733, "grad_norm": 1.7710041999816895, "learning_rate": 0.0001, "loss": 0.0194, "step": 29950 }, { "epoch": 228.70229007633588, "grad_norm": 1.3628404140472412, "learning_rate": 0.0001, "loss": 0.0193, "step": 29960 }, { "epoch": 228.77862595419847, "grad_norm": 2.5420050621032715, "learning_rate": 0.0001, "loss": 0.0191, "step": 29970 }, { "epoch": 228.85496183206106, "grad_norm": 1.5470178127288818, "learning_rate": 0.0001, "loss": 0.019, "step": 29980 }, { "epoch": 228.93129770992365, "grad_norm": 1.640442132949829, "learning_rate": 0.0001, "loss": 0.0184, "step": 29990 }, { "epoch": 229.00763358778627, "grad_norm": 1.58949875831604, "learning_rate": 0.0001, "loss": 0.019, "step": 30000 }, { "epoch": 229.00763358778627, "step": 30000, "total_flos": 0.0, "train_loss": 0.021208243355651697, "train_runtime": 7418.3209, "train_samples_per_second": 258.819, "train_steps_per_second": 4.044 } ], "logging_steps": 10, "max_steps": 30000, "num_input_tokens_seen": 0, "num_train_epochs": 230, "save_steps": 5000, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 64, "trial_name": null, "trial_params": null }