{ "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": 147.54278564453125, "learning_rate": 0.0001, "loss": 19.7957, "step": 10 }, { "epoch": 0.15267175572519084, "grad_norm": 71.73938751220703, "learning_rate": 0.0001, "loss": 4.927, "step": 20 }, { "epoch": 0.22900763358778625, "grad_norm": 61.56806564331055, "learning_rate": 0.0001, "loss": 3.6859, "step": 30 }, { "epoch": 0.3053435114503817, "grad_norm": 62.87024688720703, "learning_rate": 0.0001, "loss": 3.2496, "step": 40 }, { "epoch": 0.3816793893129771, "grad_norm": 65.14897918701172, "learning_rate": 0.0001, "loss": 2.9549, "step": 50 }, { "epoch": 0.4580152671755725, "grad_norm": 48.867881774902344, "learning_rate": 0.0001, "loss": 2.6972, "step": 60 }, { "epoch": 0.5343511450381679, "grad_norm": 46.04783248901367, "learning_rate": 0.0001, "loss": 2.4911, "step": 70 }, { "epoch": 0.6106870229007634, "grad_norm": 43.08463668823242, "learning_rate": 0.0001, "loss": 2.2699, "step": 80 }, { "epoch": 0.6870229007633588, "grad_norm": 50.57072830200195, "learning_rate": 0.0001, "loss": 2.1785, "step": 90 }, { "epoch": 0.7633587786259542, "grad_norm": 56.5272102355957, "learning_rate": 0.0001, "loss": 2.1087, "step": 100 }, { "epoch": 0.8396946564885496, "grad_norm": 45.83732604980469, "learning_rate": 0.0001, "loss": 1.9286, "step": 110 }, { "epoch": 0.916030534351145, "grad_norm": 40.292457580566406, "learning_rate": 0.0001, "loss": 1.854, "step": 120 }, { "epoch": 0.9923664122137404, "grad_norm": 40.04655456542969, "learning_rate": 0.0001, "loss": 1.7605, "step": 130 }, { "epoch": 1.0687022900763359, "grad_norm": 40.1998405456543, "learning_rate": 0.0001, "loss": 1.6288, "step": 140 }, { "epoch": 1.1450381679389312, "grad_norm": 48.981353759765625, "learning_rate": 0.0001, "loss": 1.5736, "step": 150 }, { "epoch": 1.2213740458015268, "grad_norm": 41.25627899169922, "learning_rate": 0.0001, "loss": 1.5328, "step": 160 }, { "epoch": 1.297709923664122, "grad_norm": 41.42210388183594, "learning_rate": 0.0001, "loss": 1.4441, "step": 170 }, { "epoch": 1.3740458015267176, "grad_norm": 39.97264862060547, "learning_rate": 0.0001, "loss": 1.3678, "step": 180 }, { "epoch": 1.450381679389313, "grad_norm": 34.61440658569336, "learning_rate": 0.0001, "loss": 1.2902, "step": 190 }, { "epoch": 1.5267175572519083, "grad_norm": 37.50465393066406, "learning_rate": 0.0001, "loss": 1.224, "step": 200 }, { "epoch": 1.6030534351145038, "grad_norm": 38.53634262084961, "learning_rate": 0.0001, "loss": 1.1735, "step": 210 }, { "epoch": 1.6793893129770994, "grad_norm": 32.820858001708984, "learning_rate": 0.0001, "loss": 1.1058, "step": 220 }, { "epoch": 1.7557251908396947, "grad_norm": 39.280033111572266, "learning_rate": 0.0001, "loss": 1.0663, "step": 230 }, { "epoch": 1.83206106870229, "grad_norm": 29.400344848632812, "learning_rate": 0.0001, "loss": 1.0092, "step": 240 }, { "epoch": 1.9083969465648853, "grad_norm": 32.804386138916016, "learning_rate": 0.0001, "loss": 0.9342, "step": 250 }, { "epoch": 1.984732824427481, "grad_norm": 31.14279556274414, "learning_rate": 0.0001, "loss": 0.889, "step": 260 }, { "epoch": 2.0610687022900764, "grad_norm": 33.04987335205078, "learning_rate": 0.0001, "loss": 0.8557, "step": 270 }, { "epoch": 2.1374045801526718, "grad_norm": 31.740598678588867, "learning_rate": 0.0001, "loss": 0.8187, "step": 280 }, { "epoch": 2.213740458015267, "grad_norm": 32.1165771484375, "learning_rate": 0.0001, "loss": 0.7714, "step": 290 }, { "epoch": 2.2900763358778624, "grad_norm": 25.50832748413086, "learning_rate": 0.0001, "loss": 0.73, "step": 300 }, { "epoch": 2.366412213740458, "grad_norm": 29.073759078979492, "learning_rate": 0.0001, "loss": 0.7096, "step": 310 }, { "epoch": 2.4427480916030535, "grad_norm": 26.61600685119629, "learning_rate": 0.0001, "loss": 0.6736, "step": 320 }, { "epoch": 2.519083969465649, "grad_norm": 24.383235931396484, "learning_rate": 0.0001, "loss": 0.6319, "step": 330 }, { "epoch": 2.595419847328244, "grad_norm": 28.688480377197266, "learning_rate": 0.0001, "loss": 0.6124, "step": 340 }, { "epoch": 2.67175572519084, "grad_norm": 26.218429565429688, "learning_rate": 0.0001, "loss": 0.5945, "step": 350 }, { "epoch": 2.7480916030534353, "grad_norm": 25.172529220581055, "learning_rate": 0.0001, "loss": 0.5655, "step": 360 }, { "epoch": 2.8244274809160306, "grad_norm": 23.70870018005371, "learning_rate": 0.0001, "loss": 0.5332, "step": 370 }, { "epoch": 2.900763358778626, "grad_norm": 21.387413024902344, "learning_rate": 0.0001, "loss": 0.5179, "step": 380 }, { "epoch": 2.9770992366412212, "grad_norm": 21.752134323120117, "learning_rate": 0.0001, "loss": 0.4902, "step": 390 }, { "epoch": 3.053435114503817, "grad_norm": 23.66169548034668, "learning_rate": 0.0001, "loss": 0.47, "step": 400 }, { "epoch": 3.1297709923664123, "grad_norm": 21.347278594970703, "learning_rate": 0.0001, "loss": 0.4497, "step": 410 }, { "epoch": 3.2061068702290076, "grad_norm": 22.956825256347656, "learning_rate": 0.0001, "loss": 0.4358, "step": 420 }, { "epoch": 3.282442748091603, "grad_norm": 19.53047752380371, "learning_rate": 0.0001, "loss": 0.429, "step": 430 }, { "epoch": 3.3587786259541983, "grad_norm": 22.486127853393555, "learning_rate": 0.0001, "loss": 0.4032, "step": 440 }, { "epoch": 3.435114503816794, "grad_norm": 22.76004981994629, "learning_rate": 0.0001, "loss": 0.3955, "step": 450 }, { "epoch": 3.5114503816793894, "grad_norm": 18.542287826538086, "learning_rate": 0.0001, "loss": 0.3817, "step": 460 }, { "epoch": 3.5877862595419847, "grad_norm": 20.056743621826172, "learning_rate": 0.0001, "loss": 0.3798, "step": 470 }, { "epoch": 3.66412213740458, "grad_norm": 17.211408615112305, "learning_rate": 0.0001, "loss": 0.3627, "step": 480 }, { "epoch": 3.7404580152671754, "grad_norm": 19.42917251586914, "learning_rate": 0.0001, "loss": 0.347, "step": 490 }, { "epoch": 3.816793893129771, "grad_norm": 18.256202697753906, "learning_rate": 0.0001, "loss": 0.3365, "step": 500 }, { "epoch": 3.8931297709923665, "grad_norm": 20.695405960083008, "learning_rate": 0.0001, "loss": 0.3331, "step": 510 }, { "epoch": 3.969465648854962, "grad_norm": 16.734952926635742, "learning_rate": 0.0001, "loss": 0.3184, "step": 520 }, { "epoch": 4.0458015267175576, "grad_norm": 14.16760540008545, "learning_rate": 0.0001, "loss": 0.3101, "step": 530 }, { "epoch": 4.122137404580153, "grad_norm": 16.39317512512207, "learning_rate": 0.0001, "loss": 0.303, "step": 540 }, { "epoch": 4.198473282442748, "grad_norm": 15.19994831085205, "learning_rate": 0.0001, "loss": 0.2981, "step": 550 }, { "epoch": 4.2748091603053435, "grad_norm": 14.27464771270752, "learning_rate": 0.0001, "loss": 0.2854, "step": 560 }, { "epoch": 4.351145038167939, "grad_norm": 13.44527530670166, "learning_rate": 0.0001, "loss": 0.2857, "step": 570 }, { "epoch": 4.427480916030534, "grad_norm": 16.66502571105957, "learning_rate": 0.0001, "loss": 0.2904, "step": 580 }, { "epoch": 4.5038167938931295, "grad_norm": 15.269689559936523, "learning_rate": 0.0001, "loss": 0.2815, "step": 590 }, { "epoch": 4.580152671755725, "grad_norm": 17.810136795043945, "learning_rate": 0.0001, "loss": 0.2689, "step": 600 }, { "epoch": 4.65648854961832, "grad_norm": 16.39440155029297, "learning_rate": 0.0001, "loss": 0.2664, "step": 610 }, { "epoch": 4.732824427480916, "grad_norm": 14.118106842041016, "learning_rate": 0.0001, "loss": 0.2582, "step": 620 }, { "epoch": 4.809160305343512, "grad_norm": 14.661361694335938, "learning_rate": 0.0001, "loss": 0.2548, "step": 630 }, { "epoch": 4.885496183206107, "grad_norm": 14.211882591247559, "learning_rate": 0.0001, "loss": 0.2517, "step": 640 }, { "epoch": 4.961832061068702, "grad_norm": 13.827794075012207, "learning_rate": 0.0001, "loss": 0.2501, "step": 650 }, { "epoch": 5.038167938931298, "grad_norm": 13.910344123840332, "learning_rate": 0.0001, "loss": 0.2422, "step": 660 }, { "epoch": 5.114503816793893, "grad_norm": 13.647887229919434, "learning_rate": 0.0001, "loss": 0.2341, "step": 670 }, { "epoch": 5.190839694656488, "grad_norm": 15.487009048461914, "learning_rate": 0.0001, "loss": 0.2334, "step": 680 }, { "epoch": 5.267175572519084, "grad_norm": 13.726089477539062, "learning_rate": 0.0001, "loss": 0.2278, "step": 690 }, { "epoch": 5.34351145038168, "grad_norm": 13.865954399108887, "learning_rate": 0.0001, "loss": 0.2249, "step": 700 }, { "epoch": 5.419847328244275, "grad_norm": 14.773253440856934, "learning_rate": 0.0001, "loss": 0.2291, "step": 710 }, { "epoch": 5.4961832061068705, "grad_norm": 13.660510063171387, "learning_rate": 0.0001, "loss": 0.2194, "step": 720 }, { "epoch": 5.572519083969466, "grad_norm": 13.792906761169434, "learning_rate": 0.0001, "loss": 0.2155, "step": 730 }, { "epoch": 5.648854961832061, "grad_norm": 12.688262939453125, "learning_rate": 0.0001, "loss": 0.2169, "step": 740 }, { "epoch": 5.7251908396946565, "grad_norm": 13.384363174438477, "learning_rate": 0.0001, "loss": 0.2143, "step": 750 }, { "epoch": 5.801526717557252, "grad_norm": 10.732817649841309, "learning_rate": 0.0001, "loss": 0.207, "step": 760 }, { "epoch": 5.877862595419847, "grad_norm": 10.621391296386719, "learning_rate": 0.0001, "loss": 0.1995, "step": 770 }, { "epoch": 5.9541984732824424, "grad_norm": 13.053667068481445, "learning_rate": 0.0001, "loss": 0.2021, "step": 780 }, { "epoch": 6.030534351145038, "grad_norm": 13.367091178894043, "learning_rate": 0.0001, "loss": 0.2054, "step": 790 }, { "epoch": 6.106870229007634, "grad_norm": 13.424927711486816, "learning_rate": 0.0001, "loss": 0.1988, "step": 800 }, { "epoch": 6.183206106870229, "grad_norm": 12.64376163482666, "learning_rate": 0.0001, "loss": 0.1956, "step": 810 }, { "epoch": 6.259541984732825, "grad_norm": 10.219240188598633, "learning_rate": 0.0001, "loss": 0.1904, "step": 820 }, { "epoch": 6.33587786259542, "grad_norm": 12.870089530944824, "learning_rate": 0.0001, "loss": 0.1904, "step": 830 }, { "epoch": 6.412213740458015, "grad_norm": 12.174644470214844, "learning_rate": 0.0001, "loss": 0.1903, "step": 840 }, { "epoch": 6.488549618320611, "grad_norm": 13.894688606262207, "learning_rate": 0.0001, "loss": 0.1879, "step": 850 }, { "epoch": 6.564885496183206, "grad_norm": 11.980137825012207, "learning_rate": 0.0001, "loss": 0.1864, "step": 860 }, { "epoch": 6.641221374045801, "grad_norm": 11.165816307067871, "learning_rate": 0.0001, "loss": 0.1808, "step": 870 }, { "epoch": 6.717557251908397, "grad_norm": 12.184906005859375, "learning_rate": 0.0001, "loss": 0.1843, "step": 880 }, { "epoch": 6.793893129770993, "grad_norm": 11.783286094665527, "learning_rate": 0.0001, "loss": 0.1779, "step": 890 }, { "epoch": 6.870229007633588, "grad_norm": 11.963661193847656, "learning_rate": 0.0001, "loss": 0.1774, "step": 900 }, { "epoch": 6.9465648854961835, "grad_norm": 11.186521530151367, "learning_rate": 0.0001, "loss": 0.1696, "step": 910 }, { "epoch": 7.022900763358779, "grad_norm": 10.506477355957031, "learning_rate": 0.0001, "loss": 0.1721, "step": 920 }, { "epoch": 7.099236641221374, "grad_norm": 11.150827407836914, "learning_rate": 0.0001, "loss": 0.177, "step": 930 }, { "epoch": 7.175572519083969, "grad_norm": 12.628408432006836, "learning_rate": 0.0001, "loss": 0.1828, "step": 940 }, { "epoch": 7.251908396946565, "grad_norm": 11.720745086669922, "learning_rate": 0.0001, "loss": 0.173, "step": 950 }, { "epoch": 7.32824427480916, "grad_norm": 9.607105255126953, "learning_rate": 0.0001, "loss": 0.1701, "step": 960 }, { "epoch": 7.404580152671755, "grad_norm": 11.940749168395996, "learning_rate": 0.0001, "loss": 0.1673, "step": 970 }, { "epoch": 7.480916030534351, "grad_norm": 10.820718765258789, "learning_rate": 0.0001, "loss": 0.1691, "step": 980 }, { "epoch": 7.557251908396947, "grad_norm": 11.825104713439941, "learning_rate": 0.0001, "loss": 0.162, "step": 990 }, { "epoch": 7.633587786259542, "grad_norm": 9.767478942871094, "learning_rate": 0.0001, "loss": 0.1683, "step": 1000 }, { "epoch": 7.709923664122138, "grad_norm": 11.077762603759766, "learning_rate": 0.0001, "loss": 0.1629, "step": 1010 }, { "epoch": 7.786259541984733, "grad_norm": 10.987869262695312, "learning_rate": 0.0001, "loss": 0.1638, "step": 1020 }, { "epoch": 7.862595419847328, "grad_norm": 10.713665008544922, "learning_rate": 0.0001, "loss": 0.1619, "step": 1030 }, { "epoch": 7.938931297709924, "grad_norm": 10.556421279907227, "learning_rate": 0.0001, "loss": 0.1598, "step": 1040 }, { "epoch": 8.01526717557252, "grad_norm": 9.643098831176758, "learning_rate": 0.0001, "loss": 0.1552, "step": 1050 }, { "epoch": 8.091603053435115, "grad_norm": 11.182514190673828, "learning_rate": 0.0001, "loss": 0.1525, "step": 1060 }, { "epoch": 8.16793893129771, "grad_norm": 8.844498634338379, "learning_rate": 0.0001, "loss": 0.1517, "step": 1070 }, { "epoch": 8.244274809160306, "grad_norm": 8.784748077392578, "learning_rate": 0.0001, "loss": 0.1506, "step": 1080 }, { "epoch": 8.320610687022901, "grad_norm": 10.931597709655762, "learning_rate": 0.0001, "loss": 0.1512, "step": 1090 }, { "epoch": 8.396946564885496, "grad_norm": 11.355035781860352, "learning_rate": 0.0001, "loss": 0.1496, "step": 1100 }, { "epoch": 8.473282442748092, "grad_norm": 10.003264427185059, "learning_rate": 0.0001, "loss": 0.1506, "step": 1110 }, { "epoch": 8.549618320610687, "grad_norm": 11.08468246459961, "learning_rate": 0.0001, "loss": 0.1488, "step": 1120 }, { "epoch": 8.625954198473282, "grad_norm": 11.967672348022461, "learning_rate": 0.0001, "loss": 0.1457, "step": 1130 }, { "epoch": 8.702290076335878, "grad_norm": 10.759592056274414, "learning_rate": 0.0001, "loss": 0.1441, "step": 1140 }, { "epoch": 8.778625954198473, "grad_norm": 11.397623062133789, "learning_rate": 0.0001, "loss": 0.144, "step": 1150 }, { "epoch": 8.854961832061068, "grad_norm": 9.075875282287598, "learning_rate": 0.0001, "loss": 0.137, "step": 1160 }, { "epoch": 8.931297709923664, "grad_norm": 9.723676681518555, "learning_rate": 0.0001, "loss": 0.1398, "step": 1170 }, { "epoch": 9.007633587786259, "grad_norm": 9.697829246520996, "learning_rate": 0.0001, "loss": 0.1385, "step": 1180 }, { "epoch": 9.083969465648854, "grad_norm": 8.504887580871582, "learning_rate": 0.0001, "loss": 0.1366, "step": 1190 }, { "epoch": 9.16030534351145, "grad_norm": 11.053720474243164, "learning_rate": 0.0001, "loss": 0.1343, "step": 1200 }, { "epoch": 9.236641221374045, "grad_norm": 8.731850624084473, "learning_rate": 0.0001, "loss": 0.1367, "step": 1210 }, { "epoch": 9.312977099236642, "grad_norm": 8.588528633117676, "learning_rate": 0.0001, "loss": 0.1346, "step": 1220 }, { "epoch": 9.389312977099237, "grad_norm": 9.84349536895752, "learning_rate": 0.0001, "loss": 0.1322, "step": 1230 }, { "epoch": 9.465648854961833, "grad_norm": 10.479976654052734, "learning_rate": 0.0001, "loss": 0.132, "step": 1240 }, { "epoch": 9.541984732824428, "grad_norm": 8.264710426330566, "learning_rate": 0.0001, "loss": 0.13, "step": 1250 }, { "epoch": 9.618320610687023, "grad_norm": 10.63433837890625, "learning_rate": 0.0001, "loss": 0.1336, "step": 1260 }, { "epoch": 9.694656488549619, "grad_norm": 11.238446235656738, "learning_rate": 0.0001, "loss": 0.1367, "step": 1270 }, { "epoch": 9.770992366412214, "grad_norm": 9.607017517089844, "learning_rate": 0.0001, "loss": 0.1339, "step": 1280 }, { "epoch": 9.84732824427481, "grad_norm": 8.53625774383545, "learning_rate": 0.0001, "loss": 0.1311, "step": 1290 }, { "epoch": 9.923664122137405, "grad_norm": 9.626749038696289, "learning_rate": 0.0001, "loss": 0.1287, "step": 1300 }, { "epoch": 10.0, "grad_norm": 9.240878105163574, "learning_rate": 0.0001, "loss": 0.1282, "step": 1310 }, { "epoch": 10.076335877862595, "grad_norm": 9.571104049682617, "learning_rate": 0.0001, "loss": 0.1255, "step": 1320 }, { "epoch": 10.15267175572519, "grad_norm": 8.48743724822998, "learning_rate": 0.0001, "loss": 0.1299, "step": 1330 }, { "epoch": 10.229007633587786, "grad_norm": 8.605777740478516, "learning_rate": 0.0001, "loss": 0.1248, "step": 1340 }, { "epoch": 10.305343511450381, "grad_norm": 8.421809196472168, "learning_rate": 0.0001, "loss": 0.1212, "step": 1350 }, { "epoch": 10.381679389312977, "grad_norm": 8.071803092956543, "learning_rate": 0.0001, "loss": 0.1194, "step": 1360 }, { "epoch": 10.458015267175572, "grad_norm": 12.13585090637207, "learning_rate": 0.0001, "loss": 0.1212, "step": 1370 }, { "epoch": 10.534351145038167, "grad_norm": 10.415532112121582, "learning_rate": 0.0001, "loss": 0.1239, "step": 1380 }, { "epoch": 10.610687022900763, "grad_norm": 8.911483764648438, "learning_rate": 0.0001, "loss": 0.1238, "step": 1390 }, { "epoch": 10.68702290076336, "grad_norm": 9.024186134338379, "learning_rate": 0.0001, "loss": 0.1264, "step": 1400 }, { "epoch": 10.763358778625955, "grad_norm": 8.595063209533691, "learning_rate": 0.0001, "loss": 0.1203, "step": 1410 }, { "epoch": 10.83969465648855, "grad_norm": 8.59050464630127, "learning_rate": 0.0001, "loss": 0.1223, "step": 1420 }, { "epoch": 10.916030534351146, "grad_norm": 8.435136795043945, "learning_rate": 0.0001, "loss": 0.1223, "step": 1430 }, { "epoch": 10.992366412213741, "grad_norm": 7.092011451721191, "learning_rate": 0.0001, "loss": 0.1201, "step": 1440 }, { "epoch": 11.068702290076336, "grad_norm": 6.8233418464660645, "learning_rate": 0.0001, "loss": 0.1185, "step": 1450 }, { "epoch": 11.145038167938932, "grad_norm": 8.655424118041992, "learning_rate": 0.0001, "loss": 0.1166, "step": 1460 }, { "epoch": 11.221374045801527, "grad_norm": 8.150492668151855, "learning_rate": 0.0001, "loss": 0.1169, "step": 1470 }, { "epoch": 11.297709923664122, "grad_norm": 8.02965259552002, "learning_rate": 0.0001, "loss": 0.1125, "step": 1480 }, { "epoch": 11.374045801526718, "grad_norm": 8.532236099243164, "learning_rate": 0.0001, "loss": 0.1115, "step": 1490 }, { "epoch": 11.450381679389313, "grad_norm": 8.7094087600708, "learning_rate": 0.0001, "loss": 0.1141, "step": 1500 }, { "epoch": 11.526717557251908, "grad_norm": 9.984124183654785, "learning_rate": 0.0001, "loss": 0.1158, "step": 1510 }, { "epoch": 11.603053435114504, "grad_norm": 9.179110527038574, "learning_rate": 0.0001, "loss": 0.1136, "step": 1520 }, { "epoch": 11.679389312977099, "grad_norm": 9.337518692016602, "learning_rate": 0.0001, "loss": 0.1109, "step": 1530 }, { "epoch": 11.755725190839694, "grad_norm": 8.761415481567383, "learning_rate": 0.0001, "loss": 0.112, "step": 1540 }, { "epoch": 11.83206106870229, "grad_norm": 7.958017349243164, "learning_rate": 0.0001, "loss": 0.112, "step": 1550 }, { "epoch": 11.908396946564885, "grad_norm": 8.387868881225586, "learning_rate": 0.0001, "loss": 0.1084, "step": 1560 }, { "epoch": 11.98473282442748, "grad_norm": 9.487008094787598, "learning_rate": 0.0001, "loss": 0.1064, "step": 1570 }, { "epoch": 12.061068702290076, "grad_norm": 8.329371452331543, "learning_rate": 0.0001, "loss": 0.1018, "step": 1580 }, { "epoch": 12.137404580152673, "grad_norm": 6.717339515686035, "learning_rate": 0.0001, "loss": 0.103, "step": 1590 }, { "epoch": 12.213740458015268, "grad_norm": 7.178629398345947, "learning_rate": 0.0001, "loss": 0.1035, "step": 1600 }, { "epoch": 12.290076335877863, "grad_norm": 7.426730632781982, "learning_rate": 0.0001, "loss": 0.1023, "step": 1610 }, { "epoch": 12.366412213740459, "grad_norm": 7.084171772003174, "learning_rate": 0.0001, "loss": 0.1067, "step": 1620 }, { "epoch": 12.442748091603054, "grad_norm": 7.888678073883057, "learning_rate": 0.0001, "loss": 0.1038, "step": 1630 }, { "epoch": 12.51908396946565, "grad_norm": 8.083460807800293, "learning_rate": 0.0001, "loss": 0.1034, "step": 1640 }, { "epoch": 12.595419847328245, "grad_norm": 8.23158073425293, "learning_rate": 0.0001, "loss": 0.1029, "step": 1650 }, { "epoch": 12.67175572519084, "grad_norm": 8.143179893493652, "learning_rate": 0.0001, "loss": 0.0998, "step": 1660 }, { "epoch": 12.748091603053435, "grad_norm": 9.177613258361816, "learning_rate": 0.0001, "loss": 0.1033, "step": 1670 }, { "epoch": 12.82442748091603, "grad_norm": 7.70025634765625, "learning_rate": 0.0001, "loss": 0.1029, "step": 1680 }, { "epoch": 12.900763358778626, "grad_norm": 7.22683048248291, "learning_rate": 0.0001, "loss": 0.1024, "step": 1690 }, { "epoch": 12.977099236641221, "grad_norm": 8.746634483337402, "learning_rate": 0.0001, "loss": 0.0999, "step": 1700 }, { "epoch": 13.053435114503817, "grad_norm": 7.535231590270996, "learning_rate": 0.0001, "loss": 0.1025, "step": 1710 }, { "epoch": 13.129770992366412, "grad_norm": 8.186327934265137, "learning_rate": 0.0001, "loss": 0.0994, "step": 1720 }, { "epoch": 13.206106870229007, "grad_norm": 7.263393402099609, "learning_rate": 0.0001, "loss": 0.1009, "step": 1730 }, { "epoch": 13.282442748091603, "grad_norm": 7.788677215576172, "learning_rate": 0.0001, "loss": 0.0969, "step": 1740 }, { "epoch": 13.358778625954198, "grad_norm": 7.482095241546631, "learning_rate": 0.0001, "loss": 0.0972, "step": 1750 }, { "epoch": 13.435114503816793, "grad_norm": 7.282449245452881, "learning_rate": 0.0001, "loss": 0.0997, "step": 1760 }, { "epoch": 13.511450381679388, "grad_norm": 7.1699442863464355, "learning_rate": 0.0001, "loss": 0.0992, "step": 1770 }, { "epoch": 13.587786259541986, "grad_norm": 6.661755561828613, "learning_rate": 0.0001, "loss": 0.0979, "step": 1780 }, { "epoch": 13.664122137404581, "grad_norm": 7.164150714874268, "learning_rate": 0.0001, "loss": 0.1003, "step": 1790 }, { "epoch": 13.740458015267176, "grad_norm": 8.030363082885742, "learning_rate": 0.0001, "loss": 0.0967, "step": 1800 }, { "epoch": 13.816793893129772, "grad_norm": 8.771184921264648, "learning_rate": 0.0001, "loss": 0.0949, "step": 1810 }, { "epoch": 13.893129770992367, "grad_norm": 9.721538543701172, "learning_rate": 0.0001, "loss": 0.0959, "step": 1820 }, { "epoch": 13.969465648854962, "grad_norm": 6.878361701965332, "learning_rate": 0.0001, "loss": 0.0954, "step": 1830 }, { "epoch": 14.045801526717558, "grad_norm": 7.426727294921875, "learning_rate": 0.0001, "loss": 0.0946, "step": 1840 }, { "epoch": 14.122137404580153, "grad_norm": 6.9039411544799805, "learning_rate": 0.0001, "loss": 0.0955, "step": 1850 }, { "epoch": 14.198473282442748, "grad_norm": 8.596633911132812, "learning_rate": 0.0001, "loss": 0.0948, "step": 1860 }, { "epoch": 14.274809160305344, "grad_norm": 7.096273899078369, "learning_rate": 0.0001, "loss": 0.0946, "step": 1870 }, { "epoch": 14.351145038167939, "grad_norm": 6.933183193206787, "learning_rate": 0.0001, "loss": 0.0944, "step": 1880 }, { "epoch": 14.427480916030534, "grad_norm": 6.845614910125732, "learning_rate": 0.0001, "loss": 0.093, "step": 1890 }, { "epoch": 14.50381679389313, "grad_norm": 7.387820243835449, "learning_rate": 0.0001, "loss": 0.0922, "step": 1900 }, { "epoch": 14.580152671755725, "grad_norm": 7.310410022735596, "learning_rate": 0.0001, "loss": 0.0936, "step": 1910 }, { "epoch": 14.65648854961832, "grad_norm": 6.819015979766846, "learning_rate": 0.0001, "loss": 0.0949, "step": 1920 }, { "epoch": 14.732824427480915, "grad_norm": 7.081634521484375, "learning_rate": 0.0001, "loss": 0.092, "step": 1930 }, { "epoch": 14.80916030534351, "grad_norm": 6.262160301208496, "learning_rate": 0.0001, "loss": 0.0936, "step": 1940 }, { "epoch": 14.885496183206106, "grad_norm": 7.995691299438477, "learning_rate": 0.0001, "loss": 0.0979, "step": 1950 }, { "epoch": 14.961832061068701, "grad_norm": 6.448323726654053, "learning_rate": 0.0001, "loss": 0.0968, "step": 1960 }, { "epoch": 15.038167938931299, "grad_norm": 6.8430562019348145, "learning_rate": 0.0001, "loss": 0.0949, "step": 1970 }, { "epoch": 15.114503816793894, "grad_norm": 5.865932941436768, "learning_rate": 0.0001, "loss": 0.0967, "step": 1980 }, { "epoch": 15.19083969465649, "grad_norm": 6.833093166351318, "learning_rate": 0.0001, "loss": 0.0934, "step": 1990 }, { "epoch": 15.267175572519085, "grad_norm": 5.82744026184082, "learning_rate": 0.0001, "loss": 0.0938, "step": 2000 }, { "epoch": 15.34351145038168, "grad_norm": 6.319075107574463, "learning_rate": 0.0001, "loss": 0.0943, "step": 2010 }, { "epoch": 15.419847328244275, "grad_norm": 6.780836582183838, "learning_rate": 0.0001, "loss": 0.09, "step": 2020 }, { "epoch": 15.49618320610687, "grad_norm": 8.43004035949707, "learning_rate": 0.0001, "loss": 0.0882, "step": 2030 }, { "epoch": 15.572519083969466, "grad_norm": 6.8405022621154785, "learning_rate": 0.0001, "loss": 0.087, "step": 2040 }, { "epoch": 15.648854961832061, "grad_norm": 6.8986406326293945, "learning_rate": 0.0001, "loss": 0.0854, "step": 2050 }, { "epoch": 15.725190839694656, "grad_norm": 8.16067123413086, "learning_rate": 0.0001, "loss": 0.0848, "step": 2060 }, { "epoch": 15.801526717557252, "grad_norm": 5.7431559562683105, "learning_rate": 0.0001, "loss": 0.0844, "step": 2070 }, { "epoch": 15.877862595419847, "grad_norm": 6.054393768310547, "learning_rate": 0.0001, "loss": 0.0844, "step": 2080 }, { "epoch": 15.954198473282442, "grad_norm": 5.676413059234619, "learning_rate": 0.0001, "loss": 0.0867, "step": 2090 }, { "epoch": 16.03053435114504, "grad_norm": 6.817693710327148, "learning_rate": 0.0001, "loss": 0.0851, "step": 2100 }, { "epoch": 16.106870229007633, "grad_norm": 6.005424499511719, "learning_rate": 0.0001, "loss": 0.0822, "step": 2110 }, { "epoch": 16.18320610687023, "grad_norm": 5.7066731452941895, "learning_rate": 0.0001, "loss": 0.0829, "step": 2120 }, { "epoch": 16.259541984732824, "grad_norm": 6.415604114532471, "learning_rate": 0.0001, "loss": 0.0835, "step": 2130 }, { "epoch": 16.33587786259542, "grad_norm": 5.839663028717041, "learning_rate": 0.0001, "loss": 0.084, "step": 2140 }, { "epoch": 16.412213740458014, "grad_norm": 6.5181884765625, "learning_rate": 0.0001, "loss": 0.0846, "step": 2150 }, { "epoch": 16.48854961832061, "grad_norm": 5.807124614715576, "learning_rate": 0.0001, "loss": 0.0836, "step": 2160 }, { "epoch": 16.564885496183205, "grad_norm": 6.861081600189209, "learning_rate": 0.0001, "loss": 0.0835, "step": 2170 }, { "epoch": 16.641221374045802, "grad_norm": 6.403948783874512, "learning_rate": 0.0001, "loss": 0.0839, "step": 2180 }, { "epoch": 16.717557251908396, "grad_norm": 6.196493148803711, "learning_rate": 0.0001, "loss": 0.0823, "step": 2190 }, { "epoch": 16.793893129770993, "grad_norm": 5.618916988372803, "learning_rate": 0.0001, "loss": 0.0835, "step": 2200 }, { "epoch": 16.870229007633586, "grad_norm": 6.0217461585998535, "learning_rate": 0.0001, "loss": 0.0813, "step": 2210 }, { "epoch": 16.946564885496183, "grad_norm": 7.073050022125244, "learning_rate": 0.0001, "loss": 0.0793, "step": 2220 }, { "epoch": 17.022900763358777, "grad_norm": 6.759532451629639, "learning_rate": 0.0001, "loss": 0.0826, "step": 2230 }, { "epoch": 17.099236641221374, "grad_norm": 6.551732540130615, "learning_rate": 0.0001, "loss": 0.0786, "step": 2240 }, { "epoch": 17.17557251908397, "grad_norm": 6.542884349822998, "learning_rate": 0.0001, "loss": 0.0809, "step": 2250 }, { "epoch": 17.251908396946565, "grad_norm": 6.336164474487305, "learning_rate": 0.0001, "loss": 0.0797, "step": 2260 }, { "epoch": 17.328244274809162, "grad_norm": 5.427951335906982, "learning_rate": 0.0001, "loss": 0.0787, "step": 2270 }, { "epoch": 17.404580152671755, "grad_norm": 6.742742538452148, "learning_rate": 0.0001, "loss": 0.0773, "step": 2280 }, { "epoch": 17.480916030534353, "grad_norm": 6.507199287414551, "learning_rate": 0.0001, "loss": 0.0778, "step": 2290 }, { "epoch": 17.557251908396946, "grad_norm": 6.252775192260742, "learning_rate": 0.0001, "loss": 0.0777, "step": 2300 }, { "epoch": 17.633587786259543, "grad_norm": 5.499303817749023, "learning_rate": 0.0001, "loss": 0.0792, "step": 2310 }, { "epoch": 17.709923664122137, "grad_norm": 5.665761470794678, "learning_rate": 0.0001, "loss": 0.0775, "step": 2320 }, { "epoch": 17.786259541984734, "grad_norm": 7.090462684631348, "learning_rate": 0.0001, "loss": 0.0793, "step": 2330 }, { "epoch": 17.862595419847327, "grad_norm": 6.212080478668213, "learning_rate": 0.0001, "loss": 0.0788, "step": 2340 }, { "epoch": 17.938931297709924, "grad_norm": 5.4240007400512695, "learning_rate": 0.0001, "loss": 0.0767, "step": 2350 }, { "epoch": 18.015267175572518, "grad_norm": 6.398306846618652, "learning_rate": 0.0001, "loss": 0.0776, "step": 2360 }, { "epoch": 18.091603053435115, "grad_norm": 5.8842363357543945, "learning_rate": 0.0001, "loss": 0.0787, "step": 2370 }, { "epoch": 18.16793893129771, "grad_norm": 6.69719123840332, "learning_rate": 0.0001, "loss": 0.0786, "step": 2380 }, { "epoch": 18.244274809160306, "grad_norm": 6.55148983001709, "learning_rate": 0.0001, "loss": 0.0754, "step": 2390 }, { "epoch": 18.3206106870229, "grad_norm": 6.76693868637085, "learning_rate": 0.0001, "loss": 0.0738, "step": 2400 }, { "epoch": 18.396946564885496, "grad_norm": 4.890024185180664, "learning_rate": 0.0001, "loss": 0.0711, "step": 2410 }, { "epoch": 18.47328244274809, "grad_norm": 6.348754405975342, "learning_rate": 0.0001, "loss": 0.0715, "step": 2420 }, { "epoch": 18.549618320610687, "grad_norm": 5.022323131561279, "learning_rate": 0.0001, "loss": 0.073, "step": 2430 }, { "epoch": 18.625954198473284, "grad_norm": 5.908594608306885, "learning_rate": 0.0001, "loss": 0.0719, "step": 2440 }, { "epoch": 18.702290076335878, "grad_norm": 6.1747283935546875, "learning_rate": 0.0001, "loss": 0.0736, "step": 2450 }, { "epoch": 18.778625954198475, "grad_norm": 5.470703125, "learning_rate": 0.0001, "loss": 0.0703, "step": 2460 }, { "epoch": 18.85496183206107, "grad_norm": 5.558078765869141, "learning_rate": 0.0001, "loss": 0.0734, "step": 2470 }, { "epoch": 18.931297709923665, "grad_norm": 4.72965669631958, "learning_rate": 0.0001, "loss": 0.0727, "step": 2480 }, { "epoch": 19.00763358778626, "grad_norm": 5.651520252227783, "learning_rate": 0.0001, "loss": 0.0707, "step": 2490 }, { "epoch": 19.083969465648856, "grad_norm": 5.9236321449279785, "learning_rate": 0.0001, "loss": 0.0726, "step": 2500 }, { "epoch": 19.16030534351145, "grad_norm": 5.708179473876953, "learning_rate": 0.0001, "loss": 0.0731, "step": 2510 }, { "epoch": 19.236641221374047, "grad_norm": 5.24141788482666, "learning_rate": 0.0001, "loss": 0.0716, "step": 2520 }, { "epoch": 19.31297709923664, "grad_norm": 5.064179420471191, "learning_rate": 0.0001, "loss": 0.0701, "step": 2530 }, { "epoch": 19.389312977099237, "grad_norm": 6.088762283325195, "learning_rate": 0.0001, "loss": 0.0727, "step": 2540 }, { "epoch": 19.46564885496183, "grad_norm": 5.805668830871582, "learning_rate": 0.0001, "loss": 0.0728, "step": 2550 }, { "epoch": 19.541984732824428, "grad_norm": 6.495599746704102, "learning_rate": 0.0001, "loss": 0.0742, "step": 2560 }, { "epoch": 19.61832061068702, "grad_norm": 5.799531936645508, "learning_rate": 0.0001, "loss": 0.0737, "step": 2570 }, { "epoch": 19.69465648854962, "grad_norm": 5.999574184417725, "learning_rate": 0.0001, "loss": 0.0731, "step": 2580 }, { "epoch": 19.770992366412212, "grad_norm": 5.355017185211182, "learning_rate": 0.0001, "loss": 0.0716, "step": 2590 }, { "epoch": 19.84732824427481, "grad_norm": 5.2799296379089355, "learning_rate": 0.0001, "loss": 0.0693, "step": 2600 }, { "epoch": 19.923664122137403, "grad_norm": 5.468225002288818, "learning_rate": 0.0001, "loss": 0.0684, "step": 2610 }, { "epoch": 20.0, "grad_norm": 5.757409572601318, "learning_rate": 0.0001, "loss": 0.0712, "step": 2620 }, { "epoch": 20.076335877862597, "grad_norm": 5.860946178436279, "learning_rate": 0.0001, "loss": 0.0726, "step": 2630 }, { "epoch": 20.15267175572519, "grad_norm": 5.043482780456543, "learning_rate": 0.0001, "loss": 0.0707, "step": 2640 }, { "epoch": 20.229007633587788, "grad_norm": 4.968205451965332, "learning_rate": 0.0001, "loss": 0.0708, "step": 2650 }, { "epoch": 20.30534351145038, "grad_norm": 4.688232421875, "learning_rate": 0.0001, "loss": 0.0713, "step": 2660 }, { "epoch": 20.38167938931298, "grad_norm": 4.850985527038574, "learning_rate": 0.0001, "loss": 0.0714, "step": 2670 }, { "epoch": 20.458015267175572, "grad_norm": 5.335032939910889, "learning_rate": 0.0001, "loss": 0.0713, "step": 2680 }, { "epoch": 20.53435114503817, "grad_norm": 5.385645866394043, "learning_rate": 0.0001, "loss": 0.0709, "step": 2690 }, { "epoch": 20.610687022900763, "grad_norm": 5.8719072341918945, "learning_rate": 0.0001, "loss": 0.0705, "step": 2700 }, { "epoch": 20.68702290076336, "grad_norm": 5.424786567687988, "learning_rate": 0.0001, "loss": 0.0699, "step": 2710 }, { "epoch": 20.763358778625953, "grad_norm": 4.713037490844727, "learning_rate": 0.0001, "loss": 0.07, "step": 2720 }, { "epoch": 20.83969465648855, "grad_norm": 5.032795429229736, "learning_rate": 0.0001, "loss": 0.0703, "step": 2730 }, { "epoch": 20.916030534351144, "grad_norm": 5.427506446838379, "learning_rate": 0.0001, "loss": 0.0722, "step": 2740 }, { "epoch": 20.99236641221374, "grad_norm": 5.326523780822754, "learning_rate": 0.0001, "loss": 0.07, "step": 2750 }, { "epoch": 21.068702290076335, "grad_norm": 4.823108673095703, "learning_rate": 0.0001, "loss": 0.0697, "step": 2760 }, { "epoch": 21.14503816793893, "grad_norm": 5.59124231338501, "learning_rate": 0.0001, "loss": 0.0702, "step": 2770 }, { "epoch": 21.221374045801525, "grad_norm": 5.108730792999268, "learning_rate": 0.0001, "loss": 0.0685, "step": 2780 }, { "epoch": 21.297709923664122, "grad_norm": 4.880346775054932, "learning_rate": 0.0001, "loss": 0.0657, "step": 2790 }, { "epoch": 21.374045801526716, "grad_norm": 4.915987968444824, "learning_rate": 0.0001, "loss": 0.0669, "step": 2800 }, { "epoch": 21.450381679389313, "grad_norm": 5.375945568084717, "learning_rate": 0.0001, "loss": 0.0655, "step": 2810 }, { "epoch": 21.52671755725191, "grad_norm": 5.481439590454102, "learning_rate": 0.0001, "loss": 0.0656, "step": 2820 }, { "epoch": 21.603053435114504, "grad_norm": 4.7082109451293945, "learning_rate": 0.0001, "loss": 0.0652, "step": 2830 }, { "epoch": 21.6793893129771, "grad_norm": 4.177405834197998, "learning_rate": 0.0001, "loss": 0.0649, "step": 2840 }, { "epoch": 21.755725190839694, "grad_norm": 5.380243301391602, "learning_rate": 0.0001, "loss": 0.0657, "step": 2850 }, { "epoch": 21.83206106870229, "grad_norm": 6.463586330413818, "learning_rate": 0.0001, "loss": 0.0663, "step": 2860 }, { "epoch": 21.908396946564885, "grad_norm": 5.532231330871582, "learning_rate": 0.0001, "loss": 0.0654, "step": 2870 }, { "epoch": 21.984732824427482, "grad_norm": 4.868259906768799, "learning_rate": 0.0001, "loss": 0.0656, "step": 2880 }, { "epoch": 22.061068702290076, "grad_norm": 4.536988735198975, "learning_rate": 0.0001, "loss": 0.0607, "step": 2890 }, { "epoch": 22.137404580152673, "grad_norm": 5.517280101776123, "learning_rate": 0.0001, "loss": 0.0631, "step": 2900 }, { "epoch": 22.213740458015266, "grad_norm": 5.0161051750183105, "learning_rate": 0.0001, "loss": 0.0638, "step": 2910 }, { "epoch": 22.290076335877863, "grad_norm": 5.1960530281066895, "learning_rate": 0.0001, "loss": 0.0657, "step": 2920 }, { "epoch": 22.366412213740457, "grad_norm": 5.15121603012085, "learning_rate": 0.0001, "loss": 0.0648, "step": 2930 }, { "epoch": 22.442748091603054, "grad_norm": 4.680413246154785, "learning_rate": 0.0001, "loss": 0.0625, "step": 2940 }, { "epoch": 22.519083969465647, "grad_norm": 5.452278137207031, "learning_rate": 0.0001, "loss": 0.0627, "step": 2950 }, { "epoch": 22.595419847328245, "grad_norm": 4.433237075805664, "learning_rate": 0.0001, "loss": 0.0643, "step": 2960 }, { "epoch": 22.671755725190838, "grad_norm": 4.72567892074585, "learning_rate": 0.0001, "loss": 0.0609, "step": 2970 }, { "epoch": 22.748091603053435, "grad_norm": 5.55199670791626, "learning_rate": 0.0001, "loss": 0.0653, "step": 2980 }, { "epoch": 22.82442748091603, "grad_norm": 5.086281776428223, "learning_rate": 0.0001, "loss": 0.0653, "step": 2990 }, { "epoch": 22.900763358778626, "grad_norm": 5.643373966217041, "learning_rate": 0.0001, "loss": 0.0633, "step": 3000 }, { "epoch": 22.977099236641223, "grad_norm": 4.682735443115234, "learning_rate": 0.0001, "loss": 0.0642, "step": 3010 }, { "epoch": 23.053435114503817, "grad_norm": 5.168717861175537, "learning_rate": 0.0001, "loss": 0.0605, "step": 3020 }, { "epoch": 23.129770992366414, "grad_norm": 5.326900482177734, "learning_rate": 0.0001, "loss": 0.0599, "step": 3030 }, { "epoch": 23.206106870229007, "grad_norm": 5.583173751831055, "learning_rate": 0.0001, "loss": 0.0605, "step": 3040 }, { "epoch": 23.282442748091604, "grad_norm": 4.181484222412109, "learning_rate": 0.0001, "loss": 0.0603, "step": 3050 }, { "epoch": 23.358778625954198, "grad_norm": 4.950572490692139, "learning_rate": 0.0001, "loss": 0.0605, "step": 3060 }, { "epoch": 23.435114503816795, "grad_norm": 5.002340793609619, "learning_rate": 0.0001, "loss": 0.0607, "step": 3070 }, { "epoch": 23.51145038167939, "grad_norm": 5.474300384521484, "learning_rate": 0.0001, "loss": 0.0594, "step": 3080 }, { "epoch": 23.587786259541986, "grad_norm": 4.613079071044922, "learning_rate": 0.0001, "loss": 0.0591, "step": 3090 }, { "epoch": 23.66412213740458, "grad_norm": 4.939960479736328, "learning_rate": 0.0001, "loss": 0.0593, "step": 3100 }, { "epoch": 23.740458015267176, "grad_norm": 4.263094425201416, "learning_rate": 0.0001, "loss": 0.0603, "step": 3110 }, { "epoch": 23.81679389312977, "grad_norm": 4.624768257141113, "learning_rate": 0.0001, "loss": 0.06, "step": 3120 }, { "epoch": 23.893129770992367, "grad_norm": 4.506699085235596, "learning_rate": 0.0001, "loss": 0.0618, "step": 3130 }, { "epoch": 23.96946564885496, "grad_norm": 4.287389755249023, "learning_rate": 0.0001, "loss": 0.0619, "step": 3140 }, { "epoch": 24.045801526717558, "grad_norm": 3.7751388549804688, "learning_rate": 0.0001, "loss": 0.0619, "step": 3150 }, { "epoch": 24.12213740458015, "grad_norm": 4.7489800453186035, "learning_rate": 0.0001, "loss": 0.0608, "step": 3160 }, { "epoch": 24.198473282442748, "grad_norm": 4.210694313049316, "learning_rate": 0.0001, "loss": 0.0621, "step": 3170 }, { "epoch": 24.274809160305345, "grad_norm": 5.282393932342529, "learning_rate": 0.0001, "loss": 0.063, "step": 3180 }, { "epoch": 24.35114503816794, "grad_norm": 4.7373247146606445, "learning_rate": 0.0001, "loss": 0.0624, "step": 3190 }, { "epoch": 24.427480916030536, "grad_norm": 4.494520664215088, "learning_rate": 0.0001, "loss": 0.0616, "step": 3200 }, { "epoch": 24.50381679389313, "grad_norm": 5.217311859130859, "learning_rate": 0.0001, "loss": 0.0615, "step": 3210 }, { "epoch": 24.580152671755727, "grad_norm": 4.188553333282471, "learning_rate": 0.0001, "loss": 0.0603, "step": 3220 }, { "epoch": 24.65648854961832, "grad_norm": 4.3841552734375, "learning_rate": 0.0001, "loss": 0.0615, "step": 3230 }, { "epoch": 24.732824427480917, "grad_norm": 3.733138084411621, "learning_rate": 0.0001, "loss": 0.0647, "step": 3240 }, { "epoch": 24.80916030534351, "grad_norm": 4.873394012451172, "learning_rate": 0.0001, "loss": 0.0637, "step": 3250 }, { "epoch": 24.885496183206108, "grad_norm": 4.455073833465576, "learning_rate": 0.0001, "loss": 0.0635, "step": 3260 }, { "epoch": 24.9618320610687, "grad_norm": 5.123472213745117, "learning_rate": 0.0001, "loss": 0.0628, "step": 3270 }, { "epoch": 25.0381679389313, "grad_norm": 4.326816558837891, "learning_rate": 0.0001, "loss": 0.0592, "step": 3280 }, { "epoch": 25.114503816793892, "grad_norm": 3.9683542251586914, "learning_rate": 0.0001, "loss": 0.0588, "step": 3290 }, { "epoch": 25.19083969465649, "grad_norm": 3.9156711101531982, "learning_rate": 0.0001, "loss": 0.0571, "step": 3300 }, { "epoch": 25.267175572519083, "grad_norm": 4.442856788635254, "learning_rate": 0.0001, "loss": 0.0599, "step": 3310 }, { "epoch": 25.34351145038168, "grad_norm": 3.9976418018341064, "learning_rate": 0.0001, "loss": 0.0593, "step": 3320 }, { "epoch": 25.419847328244273, "grad_norm": 4.44735050201416, "learning_rate": 0.0001, "loss": 0.0576, "step": 3330 }, { "epoch": 25.49618320610687, "grad_norm": 4.586538791656494, "learning_rate": 0.0001, "loss": 0.0587, "step": 3340 }, { "epoch": 25.572519083969464, "grad_norm": 4.757920742034912, "learning_rate": 0.0001, "loss": 0.0575, "step": 3350 }, { "epoch": 25.64885496183206, "grad_norm": 4.8472795486450195, "learning_rate": 0.0001, "loss": 0.0571, "step": 3360 }, { "epoch": 25.725190839694655, "grad_norm": 4.526644706726074, "learning_rate": 0.0001, "loss": 0.0567, "step": 3370 }, { "epoch": 25.801526717557252, "grad_norm": 4.256525993347168, "learning_rate": 0.0001, "loss": 0.0549, "step": 3380 }, { "epoch": 25.87786259541985, "grad_norm": 4.208540916442871, "learning_rate": 0.0001, "loss": 0.0525, "step": 3390 }, { "epoch": 25.954198473282442, "grad_norm": 4.10109281539917, "learning_rate": 0.0001, "loss": 0.0545, "step": 3400 }, { "epoch": 26.03053435114504, "grad_norm": 3.969860076904297, "learning_rate": 0.0001, "loss": 0.0544, "step": 3410 }, { "epoch": 26.106870229007633, "grad_norm": 4.043691635131836, "learning_rate": 0.0001, "loss": 0.0539, "step": 3420 }, { "epoch": 26.18320610687023, "grad_norm": 4.085604190826416, "learning_rate": 0.0001, "loss": 0.0541, "step": 3430 }, { "epoch": 26.259541984732824, "grad_norm": 4.927974700927734, "learning_rate": 0.0001, "loss": 0.0549, "step": 3440 }, { "epoch": 26.33587786259542, "grad_norm": 4.501067161560059, "learning_rate": 0.0001, "loss": 0.0547, "step": 3450 }, { "epoch": 26.412213740458014, "grad_norm": 4.468747138977051, "learning_rate": 0.0001, "loss": 0.0549, "step": 3460 }, { "epoch": 26.48854961832061, "grad_norm": 4.049109935760498, "learning_rate": 0.0001, "loss": 0.053, "step": 3470 }, { "epoch": 26.564885496183205, "grad_norm": 3.4813642501831055, "learning_rate": 0.0001, "loss": 0.0534, "step": 3480 }, { "epoch": 26.641221374045802, "grad_norm": 4.2048773765563965, "learning_rate": 0.0001, "loss": 0.0542, "step": 3490 }, { "epoch": 26.717557251908396, "grad_norm": 4.252817153930664, "learning_rate": 0.0001, "loss": 0.0555, "step": 3500 }, { "epoch": 26.793893129770993, "grad_norm": 4.301164627075195, "learning_rate": 0.0001, "loss": 0.0584, "step": 3510 }, { "epoch": 26.870229007633586, "grad_norm": 4.7255120277404785, "learning_rate": 0.0001, "loss": 0.0567, "step": 3520 }, { "epoch": 26.946564885496183, "grad_norm": 4.322248458862305, "learning_rate": 0.0001, "loss": 0.0548, "step": 3530 }, { "epoch": 27.022900763358777, "grad_norm": 4.003017425537109, "learning_rate": 0.0001, "loss": 0.0573, "step": 3540 }, { "epoch": 27.099236641221374, "grad_norm": 3.9209835529327393, "learning_rate": 0.0001, "loss": 0.0549, "step": 3550 }, { "epoch": 27.17557251908397, "grad_norm": 4.5803680419921875, "learning_rate": 0.0001, "loss": 0.0552, "step": 3560 }, { "epoch": 27.251908396946565, "grad_norm": 4.183896064758301, "learning_rate": 0.0001, "loss": 0.0546, "step": 3570 }, { "epoch": 27.328244274809162, "grad_norm": 3.7178878784179688, "learning_rate": 0.0001, "loss": 0.0567, "step": 3580 }, { "epoch": 27.404580152671755, "grad_norm": 4.429042339324951, "learning_rate": 0.0001, "loss": 0.0586, "step": 3590 }, { "epoch": 27.480916030534353, "grad_norm": 3.8084299564361572, "learning_rate": 0.0001, "loss": 0.0562, "step": 3600 }, { "epoch": 27.557251908396946, "grad_norm": 3.723543405532837, "learning_rate": 0.0001, "loss": 0.0568, "step": 3610 }, { "epoch": 27.633587786259543, "grad_norm": 4.730576515197754, "learning_rate": 0.0001, "loss": 0.0539, "step": 3620 }, { "epoch": 27.709923664122137, "grad_norm": 4.801916122436523, "learning_rate": 0.0001, "loss": 0.0571, "step": 3630 }, { "epoch": 27.786259541984734, "grad_norm": 4.130457878112793, "learning_rate": 0.0001, "loss": 0.054, "step": 3640 }, { "epoch": 27.862595419847327, "grad_norm": 4.568282127380371, "learning_rate": 0.0001, "loss": 0.0537, "step": 3650 }, { "epoch": 27.938931297709924, "grad_norm": 4.068840503692627, "learning_rate": 0.0001, "loss": 0.0537, "step": 3660 }, { "epoch": 28.015267175572518, "grad_norm": 4.095743179321289, "learning_rate": 0.0001, "loss": 0.0526, "step": 3670 }, { "epoch": 28.091603053435115, "grad_norm": 4.103592395782471, "learning_rate": 0.0001, "loss": 0.054, "step": 3680 }, { "epoch": 28.16793893129771, "grad_norm": 4.134469509124756, "learning_rate": 0.0001, "loss": 0.0537, "step": 3690 }, { "epoch": 28.244274809160306, "grad_norm": 4.22285795211792, "learning_rate": 0.0001, "loss": 0.0531, "step": 3700 }, { "epoch": 28.3206106870229, "grad_norm": 4.049163341522217, "learning_rate": 0.0001, "loss": 0.0536, "step": 3710 }, { "epoch": 28.396946564885496, "grad_norm": 3.8612587451934814, "learning_rate": 0.0001, "loss": 0.0553, "step": 3720 }, { "epoch": 28.47328244274809, "grad_norm": 3.1168410778045654, "learning_rate": 0.0001, "loss": 0.0526, "step": 3730 }, { "epoch": 28.549618320610687, "grad_norm": 4.589408874511719, "learning_rate": 0.0001, "loss": 0.0548, "step": 3740 }, { "epoch": 28.625954198473284, "grad_norm": 3.6628568172454834, "learning_rate": 0.0001, "loss": 0.0533, "step": 3750 }, { "epoch": 28.702290076335878, "grad_norm": 4.062976360321045, "learning_rate": 0.0001, "loss": 0.053, "step": 3760 }, { "epoch": 28.778625954198475, "grad_norm": 3.74629282951355, "learning_rate": 0.0001, "loss": 0.0537, "step": 3770 }, { "epoch": 28.85496183206107, "grad_norm": 3.840147018432617, "learning_rate": 0.0001, "loss": 0.0545, "step": 3780 }, { "epoch": 28.931297709923665, "grad_norm": 3.969365358352661, "learning_rate": 0.0001, "loss": 0.0524, "step": 3790 }, { "epoch": 29.00763358778626, "grad_norm": 4.0320587158203125, "learning_rate": 0.0001, "loss": 0.0524, "step": 3800 }, { "epoch": 29.083969465648856, "grad_norm": 3.81132173538208, "learning_rate": 0.0001, "loss": 0.0508, "step": 3810 }, { "epoch": 29.16030534351145, "grad_norm": 4.175119876861572, "learning_rate": 0.0001, "loss": 0.0528, "step": 3820 }, { "epoch": 29.236641221374047, "grad_norm": 3.947319746017456, "learning_rate": 0.0001, "loss": 0.0509, "step": 3830 }, { "epoch": 29.31297709923664, "grad_norm": 3.350123405456543, "learning_rate": 0.0001, "loss": 0.0519, "step": 3840 }, { "epoch": 29.389312977099237, "grad_norm": 3.9435768127441406, "learning_rate": 0.0001, "loss": 0.0538, "step": 3850 }, { "epoch": 29.46564885496183, "grad_norm": 4.297461032867432, "learning_rate": 0.0001, "loss": 0.0538, "step": 3860 }, { "epoch": 29.541984732824428, "grad_norm": 3.6668527126312256, "learning_rate": 0.0001, "loss": 0.0508, "step": 3870 }, { "epoch": 29.61832061068702, "grad_norm": 3.2527763843536377, "learning_rate": 0.0001, "loss": 0.0534, "step": 3880 }, { "epoch": 29.69465648854962, "grad_norm": 3.243529796600342, "learning_rate": 0.0001, "loss": 0.0523, "step": 3890 }, { "epoch": 29.770992366412212, "grad_norm": 3.7658450603485107, "learning_rate": 0.0001, "loss": 0.0528, "step": 3900 }, { "epoch": 29.84732824427481, "grad_norm": 3.597898006439209, "learning_rate": 0.0001, "loss": 0.0514, "step": 3910 }, { "epoch": 29.923664122137403, "grad_norm": 3.5481040477752686, "learning_rate": 0.0001, "loss": 0.0527, "step": 3920 }, { "epoch": 30.0, "grad_norm": 3.540281295776367, "learning_rate": 0.0001, "loss": 0.0495, "step": 3930 }, { "epoch": 30.076335877862597, "grad_norm": 4.231480598449707, "learning_rate": 0.0001, "loss": 0.0493, "step": 3940 }, { "epoch": 30.15267175572519, "grad_norm": 3.5927789211273193, "learning_rate": 0.0001, "loss": 0.0476, "step": 3950 }, { "epoch": 30.229007633587788, "grad_norm": 3.3124194145202637, "learning_rate": 0.0001, "loss": 0.0471, "step": 3960 }, { "epoch": 30.30534351145038, "grad_norm": 3.7947354316711426, "learning_rate": 0.0001, "loss": 0.0469, "step": 3970 }, { "epoch": 30.38167938931298, "grad_norm": 3.476174831390381, "learning_rate": 0.0001, "loss": 0.0498, "step": 3980 }, { "epoch": 30.458015267175572, "grad_norm": 4.287493705749512, "learning_rate": 0.0001, "loss": 0.0487, "step": 3990 }, { "epoch": 30.53435114503817, "grad_norm": 3.682419776916504, "learning_rate": 0.0001, "loss": 0.0499, "step": 4000 }, { "epoch": 30.610687022900763, "grad_norm": 3.0311787128448486, "learning_rate": 0.0001, "loss": 0.0503, "step": 4010 }, { "epoch": 30.68702290076336, "grad_norm": 3.6606760025024414, "learning_rate": 0.0001, "loss": 0.0512, "step": 4020 }, { "epoch": 30.763358778625953, "grad_norm": 3.471261978149414, "learning_rate": 0.0001, "loss": 0.0505, "step": 4030 }, { "epoch": 30.83969465648855, "grad_norm": 3.528874158859253, "learning_rate": 0.0001, "loss": 0.0513, "step": 4040 }, { "epoch": 30.916030534351144, "grad_norm": 3.5980167388916016, "learning_rate": 0.0001, "loss": 0.052, "step": 4050 }, { "epoch": 30.99236641221374, "grad_norm": 4.1895060539245605, "learning_rate": 0.0001, "loss": 0.0521, "step": 4060 }, { "epoch": 31.068702290076335, "grad_norm": 3.693962335586548, "learning_rate": 0.0001, "loss": 0.0502, "step": 4070 }, { "epoch": 31.14503816793893, "grad_norm": 4.285224437713623, "learning_rate": 0.0001, "loss": 0.0492, "step": 4080 }, { "epoch": 31.221374045801525, "grad_norm": 3.7632713317871094, "learning_rate": 0.0001, "loss": 0.0496, "step": 4090 }, { "epoch": 31.297709923664122, "grad_norm": 2.991529941558838, "learning_rate": 0.0001, "loss": 0.0468, "step": 4100 }, { "epoch": 31.374045801526716, "grad_norm": 3.3071091175079346, "learning_rate": 0.0001, "loss": 0.0467, "step": 4110 }, { "epoch": 31.450381679389313, "grad_norm": 2.811474084854126, "learning_rate": 0.0001, "loss": 0.0491, "step": 4120 }, { "epoch": 31.52671755725191, "grad_norm": 3.7419955730438232, "learning_rate": 0.0001, "loss": 0.0494, "step": 4130 }, { "epoch": 31.603053435114504, "grad_norm": 3.7809200286865234, "learning_rate": 0.0001, "loss": 0.0499, "step": 4140 }, { "epoch": 31.6793893129771, "grad_norm": 3.6535890102386475, "learning_rate": 0.0001, "loss": 0.052, "step": 4150 }, { "epoch": 31.755725190839694, "grad_norm": 3.422175884246826, "learning_rate": 0.0001, "loss": 0.051, "step": 4160 }, { "epoch": 31.83206106870229, "grad_norm": 3.3907699584960938, "learning_rate": 0.0001, "loss": 0.0501, "step": 4170 }, { "epoch": 31.908396946564885, "grad_norm": 4.447734355926514, "learning_rate": 0.0001, "loss": 0.0499, "step": 4180 }, { "epoch": 31.984732824427482, "grad_norm": 2.9836788177490234, "learning_rate": 0.0001, "loss": 0.0492, "step": 4190 }, { "epoch": 32.06106870229008, "grad_norm": 3.3494157791137695, "learning_rate": 0.0001, "loss": 0.0479, "step": 4200 }, { "epoch": 32.13740458015267, "grad_norm": 3.3947997093200684, "learning_rate": 0.0001, "loss": 0.0508, "step": 4210 }, { "epoch": 32.213740458015266, "grad_norm": 3.8189189434051514, "learning_rate": 0.0001, "loss": 0.0522, "step": 4220 }, { "epoch": 32.29007633587786, "grad_norm": 3.805128335952759, "learning_rate": 0.0001, "loss": 0.0519, "step": 4230 }, { "epoch": 32.36641221374046, "grad_norm": 3.8691458702087402, "learning_rate": 0.0001, "loss": 0.0502, "step": 4240 }, { "epoch": 32.44274809160305, "grad_norm": 3.853379726409912, "learning_rate": 0.0001, "loss": 0.05, "step": 4250 }, { "epoch": 32.51908396946565, "grad_norm": 4.042487144470215, "learning_rate": 0.0001, "loss": 0.0471, "step": 4260 }, { "epoch": 32.595419847328245, "grad_norm": 3.5378873348236084, "learning_rate": 0.0001, "loss": 0.0478, "step": 4270 }, { "epoch": 32.67175572519084, "grad_norm": 3.0878140926361084, "learning_rate": 0.0001, "loss": 0.0483, "step": 4280 }, { "epoch": 32.74809160305343, "grad_norm": 3.7951457500457764, "learning_rate": 0.0001, "loss": 0.0479, "step": 4290 }, { "epoch": 32.82442748091603, "grad_norm": 3.221129894256592, "learning_rate": 0.0001, "loss": 0.0483, "step": 4300 }, { "epoch": 32.900763358778626, "grad_norm": 3.624239921569824, "learning_rate": 0.0001, "loss": 0.0467, "step": 4310 }, { "epoch": 32.97709923664122, "grad_norm": 3.4339449405670166, "learning_rate": 0.0001, "loss": 0.0483, "step": 4320 }, { "epoch": 33.05343511450382, "grad_norm": 3.3092167377471924, "learning_rate": 0.0001, "loss": 0.0479, "step": 4330 }, { "epoch": 33.12977099236641, "grad_norm": 3.404426097869873, "learning_rate": 0.0001, "loss": 0.0482, "step": 4340 }, { "epoch": 33.20610687022901, "grad_norm": 3.984555721282959, "learning_rate": 0.0001, "loss": 0.0492, "step": 4350 }, { "epoch": 33.282442748091604, "grad_norm": 3.168691396713257, "learning_rate": 0.0001, "loss": 0.0503, "step": 4360 }, { "epoch": 33.3587786259542, "grad_norm": 3.513151168823242, "learning_rate": 0.0001, "loss": 0.0506, "step": 4370 }, { "epoch": 33.43511450381679, "grad_norm": 3.3079819679260254, "learning_rate": 0.0001, "loss": 0.0499, "step": 4380 }, { "epoch": 33.51145038167939, "grad_norm": 3.8096835613250732, "learning_rate": 0.0001, "loss": 0.05, "step": 4390 }, { "epoch": 33.587786259541986, "grad_norm": 3.628011465072632, "learning_rate": 0.0001, "loss": 0.0496, "step": 4400 }, { "epoch": 33.66412213740458, "grad_norm": 3.034440279006958, "learning_rate": 0.0001, "loss": 0.0471, "step": 4410 }, { "epoch": 33.74045801526717, "grad_norm": 3.472968816757202, "learning_rate": 0.0001, "loss": 0.0462, "step": 4420 }, { "epoch": 33.81679389312977, "grad_norm": 2.8505799770355225, "learning_rate": 0.0001, "loss": 0.0473, "step": 4430 }, { "epoch": 33.89312977099237, "grad_norm": 3.215486526489258, "learning_rate": 0.0001, "loss": 0.047, "step": 4440 }, { "epoch": 33.969465648854964, "grad_norm": 2.8934333324432373, "learning_rate": 0.0001, "loss": 0.0476, "step": 4450 }, { "epoch": 34.045801526717554, "grad_norm": 3.2487425804138184, "learning_rate": 0.0001, "loss": 0.0468, "step": 4460 }, { "epoch": 34.12213740458015, "grad_norm": 3.060657262802124, "learning_rate": 0.0001, "loss": 0.0455, "step": 4470 }, { "epoch": 34.19847328244275, "grad_norm": 3.015436887741089, "learning_rate": 0.0001, "loss": 0.0469, "step": 4480 }, { "epoch": 34.274809160305345, "grad_norm": 3.276743173599243, "learning_rate": 0.0001, "loss": 0.0481, "step": 4490 }, { "epoch": 34.35114503816794, "grad_norm": 4.280351638793945, "learning_rate": 0.0001, "loss": 0.0484, "step": 4500 }, { "epoch": 34.42748091603053, "grad_norm": 3.036823034286499, "learning_rate": 0.0001, "loss": 0.0481, "step": 4510 }, { "epoch": 34.50381679389313, "grad_norm": 3.4289803504943848, "learning_rate": 0.0001, "loss": 0.049, "step": 4520 }, { "epoch": 34.58015267175573, "grad_norm": 3.3398303985595703, "learning_rate": 0.0001, "loss": 0.0473, "step": 4530 }, { "epoch": 34.656488549618324, "grad_norm": 2.9010024070739746, "learning_rate": 0.0001, "loss": 0.0467, "step": 4540 }, { "epoch": 34.732824427480914, "grad_norm": 3.8902783393859863, "learning_rate": 0.0001, "loss": 0.046, "step": 4550 }, { "epoch": 34.80916030534351, "grad_norm": 3.792538642883301, "learning_rate": 0.0001, "loss": 0.046, "step": 4560 }, { "epoch": 34.88549618320611, "grad_norm": 3.5714380741119385, "learning_rate": 0.0001, "loss": 0.0455, "step": 4570 }, { "epoch": 34.961832061068705, "grad_norm": 3.2313506603240967, "learning_rate": 0.0001, "loss": 0.0462, "step": 4580 }, { "epoch": 35.038167938931295, "grad_norm": 3.000964403152466, "learning_rate": 0.0001, "loss": 0.0455, "step": 4590 }, { "epoch": 35.11450381679389, "grad_norm": 2.9698078632354736, "learning_rate": 0.0001, "loss": 0.0444, "step": 4600 }, { "epoch": 35.19083969465649, "grad_norm": 3.3333258628845215, "learning_rate": 0.0001, "loss": 0.0454, "step": 4610 }, { "epoch": 35.267175572519086, "grad_norm": 3.0106120109558105, "learning_rate": 0.0001, "loss": 0.0437, "step": 4620 }, { "epoch": 35.343511450381676, "grad_norm": 3.292477607727051, "learning_rate": 0.0001, "loss": 0.0449, "step": 4630 }, { "epoch": 35.41984732824427, "grad_norm": 3.483128070831299, "learning_rate": 0.0001, "loss": 0.0459, "step": 4640 }, { "epoch": 35.49618320610687, "grad_norm": 3.788200855255127, "learning_rate": 0.0001, "loss": 0.0451, "step": 4650 }, { "epoch": 35.57251908396947, "grad_norm": 4.046688079833984, "learning_rate": 0.0001, "loss": 0.0463, "step": 4660 }, { "epoch": 35.64885496183206, "grad_norm": 3.902345657348633, "learning_rate": 0.0001, "loss": 0.0486, "step": 4670 }, { "epoch": 35.725190839694655, "grad_norm": 3.4997637271881104, "learning_rate": 0.0001, "loss": 0.0471, "step": 4680 }, { "epoch": 35.80152671755725, "grad_norm": 2.962519407272339, "learning_rate": 0.0001, "loss": 0.0462, "step": 4690 }, { "epoch": 35.87786259541985, "grad_norm": 3.5871753692626953, "learning_rate": 0.0001, "loss": 0.0446, "step": 4700 }, { "epoch": 35.954198473282446, "grad_norm": 3.3213613033294678, "learning_rate": 0.0001, "loss": 0.0469, "step": 4710 }, { "epoch": 36.030534351145036, "grad_norm": 2.788958787918091, "learning_rate": 0.0001, "loss": 0.0451, "step": 4720 }, { "epoch": 36.10687022900763, "grad_norm": 3.1710762977600098, "learning_rate": 0.0001, "loss": 0.0458, "step": 4730 }, { "epoch": 36.18320610687023, "grad_norm": 3.177168130874634, "learning_rate": 0.0001, "loss": 0.0473, "step": 4740 }, { "epoch": 36.25954198473283, "grad_norm": 3.2222483158111572, "learning_rate": 0.0001, "loss": 0.0457, "step": 4750 }, { "epoch": 36.33587786259542, "grad_norm": 3.2382943630218506, "learning_rate": 0.0001, "loss": 0.0434, "step": 4760 }, { "epoch": 36.412213740458014, "grad_norm": 3.0911316871643066, "learning_rate": 0.0001, "loss": 0.0454, "step": 4770 }, { "epoch": 36.48854961832061, "grad_norm": 3.1496739387512207, "learning_rate": 0.0001, "loss": 0.044, "step": 4780 }, { "epoch": 36.56488549618321, "grad_norm": 3.526561737060547, "learning_rate": 0.0001, "loss": 0.0432, "step": 4790 }, { "epoch": 36.6412213740458, "grad_norm": 3.1194629669189453, "learning_rate": 0.0001, "loss": 0.0451, "step": 4800 }, { "epoch": 36.717557251908396, "grad_norm": 3.173083543777466, "learning_rate": 0.0001, "loss": 0.045, "step": 4810 }, { "epoch": 36.79389312977099, "grad_norm": 3.493943214416504, "learning_rate": 0.0001, "loss": 0.0449, "step": 4820 }, { "epoch": 36.87022900763359, "grad_norm": 3.088474988937378, "learning_rate": 0.0001, "loss": 0.0461, "step": 4830 }, { "epoch": 36.94656488549618, "grad_norm": 2.6097235679626465, "learning_rate": 0.0001, "loss": 0.0435, "step": 4840 }, { "epoch": 37.02290076335878, "grad_norm": 3.166834831237793, "learning_rate": 0.0001, "loss": 0.0441, "step": 4850 }, { "epoch": 37.099236641221374, "grad_norm": 2.667574882507324, "learning_rate": 0.0001, "loss": 0.0422, "step": 4860 }, { "epoch": 37.17557251908397, "grad_norm": 3.3252201080322266, "learning_rate": 0.0001, "loss": 0.0448, "step": 4870 }, { "epoch": 37.25190839694657, "grad_norm": 3.2053420543670654, "learning_rate": 0.0001, "loss": 0.043, "step": 4880 }, { "epoch": 37.32824427480916, "grad_norm": 3.097738742828369, "learning_rate": 0.0001, "loss": 0.0433, "step": 4890 }, { "epoch": 37.404580152671755, "grad_norm": 3.6360034942626953, "learning_rate": 0.0001, "loss": 0.0438, "step": 4900 }, { "epoch": 37.48091603053435, "grad_norm": 2.698472738265991, "learning_rate": 0.0001, "loss": 0.0439, "step": 4910 }, { "epoch": 37.55725190839695, "grad_norm": 2.8787572383880615, "learning_rate": 0.0001, "loss": 0.0456, "step": 4920 }, { "epoch": 37.63358778625954, "grad_norm": 2.7771525382995605, "learning_rate": 0.0001, "loss": 0.0442, "step": 4930 }, { "epoch": 37.70992366412214, "grad_norm": 2.9571940898895264, "learning_rate": 0.0001, "loss": 0.0468, "step": 4940 }, { "epoch": 37.786259541984734, "grad_norm": 2.300699234008789, "learning_rate": 0.0001, "loss": 0.0454, "step": 4950 }, { "epoch": 37.86259541984733, "grad_norm": 2.900625467300415, "learning_rate": 0.0001, "loss": 0.0439, "step": 4960 }, { "epoch": 37.93893129770992, "grad_norm": 2.1741316318511963, "learning_rate": 0.0001, "loss": 0.0451, "step": 4970 }, { "epoch": 38.01526717557252, "grad_norm": 2.9241783618927, "learning_rate": 0.0001, "loss": 0.0439, "step": 4980 }, { "epoch": 38.091603053435115, "grad_norm": 3.2482235431671143, "learning_rate": 0.0001, "loss": 0.0441, "step": 4990 }, { "epoch": 38.16793893129771, "grad_norm": 3.062605142593384, "learning_rate": 0.0001, "loss": 0.0443, "step": 5000 }, { "epoch": 38.2442748091603, "grad_norm": 2.650078773498535, "learning_rate": 0.0001, "loss": 0.0442, "step": 5010 }, { "epoch": 38.3206106870229, "grad_norm": 3.7150070667266846, "learning_rate": 0.0001, "loss": 0.0438, "step": 5020 }, { "epoch": 38.396946564885496, "grad_norm": 2.959179401397705, "learning_rate": 0.0001, "loss": 0.0436, "step": 5030 }, { "epoch": 38.47328244274809, "grad_norm": 2.5778098106384277, "learning_rate": 0.0001, "loss": 0.0436, "step": 5040 }, { "epoch": 38.54961832061069, "grad_norm": 3.093268871307373, "learning_rate": 0.0001, "loss": 0.0432, "step": 5050 }, { "epoch": 38.62595419847328, "grad_norm": 3.1361584663391113, "learning_rate": 0.0001, "loss": 0.0421, "step": 5060 }, { "epoch": 38.70229007633588, "grad_norm": 2.9501729011535645, "learning_rate": 0.0001, "loss": 0.0414, "step": 5070 }, { "epoch": 38.778625954198475, "grad_norm": 3.0983991622924805, "learning_rate": 0.0001, "loss": 0.0432, "step": 5080 }, { "epoch": 38.85496183206107, "grad_norm": 2.4670023918151855, "learning_rate": 0.0001, "loss": 0.0444, "step": 5090 }, { "epoch": 38.93129770992366, "grad_norm": 2.834796667098999, "learning_rate": 0.0001, "loss": 0.043, "step": 5100 }, { "epoch": 39.00763358778626, "grad_norm": 3.014303684234619, "learning_rate": 0.0001, "loss": 0.0415, "step": 5110 }, { "epoch": 39.083969465648856, "grad_norm": 2.756037473678589, "learning_rate": 0.0001, "loss": 0.0414, "step": 5120 }, { "epoch": 39.16030534351145, "grad_norm": 2.93886661529541, "learning_rate": 0.0001, "loss": 0.0412, "step": 5130 }, { "epoch": 39.23664122137404, "grad_norm": 3.496835947036743, "learning_rate": 0.0001, "loss": 0.0422, "step": 5140 }, { "epoch": 39.31297709923664, "grad_norm": 3.027367115020752, "learning_rate": 0.0001, "loss": 0.0435, "step": 5150 }, { "epoch": 39.38931297709924, "grad_norm": 2.4400932788848877, "learning_rate": 0.0001, "loss": 0.0419, "step": 5160 }, { "epoch": 39.465648854961835, "grad_norm": 2.4248390197753906, "learning_rate": 0.0001, "loss": 0.0414, "step": 5170 }, { "epoch": 39.541984732824424, "grad_norm": 2.6410133838653564, "learning_rate": 0.0001, "loss": 0.042, "step": 5180 }, { "epoch": 39.61832061068702, "grad_norm": 2.902656316757202, "learning_rate": 0.0001, "loss": 0.0409, "step": 5190 }, { "epoch": 39.69465648854962, "grad_norm": 2.4277093410491943, "learning_rate": 0.0001, "loss": 0.0409, "step": 5200 }, { "epoch": 39.770992366412216, "grad_norm": 2.583878993988037, "learning_rate": 0.0001, "loss": 0.043, "step": 5210 }, { "epoch": 39.847328244274806, "grad_norm": 3.0592598915100098, "learning_rate": 0.0001, "loss": 0.0415, "step": 5220 }, { "epoch": 39.9236641221374, "grad_norm": 2.895296573638916, "learning_rate": 0.0001, "loss": 0.0426, "step": 5230 }, { "epoch": 40.0, "grad_norm": 3.176464557647705, "learning_rate": 0.0001, "loss": 0.0431, "step": 5240 }, { "epoch": 40.0763358778626, "grad_norm": 2.7505714893341064, "learning_rate": 0.0001, "loss": 0.0416, "step": 5250 }, { "epoch": 40.152671755725194, "grad_norm": 3.103447914123535, "learning_rate": 0.0001, "loss": 0.0416, "step": 5260 }, { "epoch": 40.229007633587784, "grad_norm": 3.119067430496216, "learning_rate": 0.0001, "loss": 0.0411, "step": 5270 }, { "epoch": 40.30534351145038, "grad_norm": 2.657464027404785, "learning_rate": 0.0001, "loss": 0.0422, "step": 5280 }, { "epoch": 40.38167938931298, "grad_norm": 2.6710398197174072, "learning_rate": 0.0001, "loss": 0.0409, "step": 5290 }, { "epoch": 40.458015267175576, "grad_norm": 2.7847416400909424, "learning_rate": 0.0001, "loss": 0.0397, "step": 5300 }, { "epoch": 40.534351145038165, "grad_norm": 2.7041256427764893, "learning_rate": 0.0001, "loss": 0.0406, "step": 5310 }, { "epoch": 40.61068702290076, "grad_norm": 3.177058219909668, "learning_rate": 0.0001, "loss": 0.0409, "step": 5320 }, { "epoch": 40.68702290076336, "grad_norm": 2.38445782661438, "learning_rate": 0.0001, "loss": 0.0409, "step": 5330 }, { "epoch": 40.76335877862596, "grad_norm": 2.6212852001190186, "learning_rate": 0.0001, "loss": 0.0413, "step": 5340 }, { "epoch": 40.83969465648855, "grad_norm": 2.713756561279297, "learning_rate": 0.0001, "loss": 0.0401, "step": 5350 }, { "epoch": 40.916030534351144, "grad_norm": 2.616422653198242, "learning_rate": 0.0001, "loss": 0.0411, "step": 5360 }, { "epoch": 40.99236641221374, "grad_norm": 2.6061697006225586, "learning_rate": 0.0001, "loss": 0.0421, "step": 5370 }, { "epoch": 41.06870229007634, "grad_norm": 2.431854248046875, "learning_rate": 0.0001, "loss": 0.0405, "step": 5380 }, { "epoch": 41.14503816793893, "grad_norm": 2.64780330657959, "learning_rate": 0.0001, "loss": 0.044, "step": 5390 }, { "epoch": 41.221374045801525, "grad_norm": 2.9676668643951416, "learning_rate": 0.0001, "loss": 0.0443, "step": 5400 }, { "epoch": 41.29770992366412, "grad_norm": 2.6115572452545166, "learning_rate": 0.0001, "loss": 0.0451, "step": 5410 }, { "epoch": 41.37404580152672, "grad_norm": 2.560312509536743, "learning_rate": 0.0001, "loss": 0.0433, "step": 5420 }, { "epoch": 41.45038167938931, "grad_norm": 2.2086288928985596, "learning_rate": 0.0001, "loss": 0.0421, "step": 5430 }, { "epoch": 41.52671755725191, "grad_norm": 3.0233829021453857, "learning_rate": 0.0001, "loss": 0.0427, "step": 5440 }, { "epoch": 41.603053435114504, "grad_norm": 2.2433292865753174, "learning_rate": 0.0001, "loss": 0.0421, "step": 5450 }, { "epoch": 41.6793893129771, "grad_norm": 2.742871046066284, "learning_rate": 0.0001, "loss": 0.0417, "step": 5460 }, { "epoch": 41.7557251908397, "grad_norm": 2.698446035385132, "learning_rate": 0.0001, "loss": 0.0417, "step": 5470 }, { "epoch": 41.83206106870229, "grad_norm": 3.030923843383789, "learning_rate": 0.0001, "loss": 0.0423, "step": 5480 }, { "epoch": 41.908396946564885, "grad_norm": 2.591278553009033, "learning_rate": 0.0001, "loss": 0.0425, "step": 5490 }, { "epoch": 41.98473282442748, "grad_norm": 2.784003734588623, "learning_rate": 0.0001, "loss": 0.0407, "step": 5500 }, { "epoch": 42.06106870229008, "grad_norm": 2.5421814918518066, "learning_rate": 0.0001, "loss": 0.0408, "step": 5510 }, { "epoch": 42.13740458015267, "grad_norm": 2.6267828941345215, "learning_rate": 0.0001, "loss": 0.042, "step": 5520 }, { "epoch": 42.213740458015266, "grad_norm": 3.1032094955444336, "learning_rate": 0.0001, "loss": 0.0421, "step": 5530 }, { "epoch": 42.29007633587786, "grad_norm": 2.7401843070983887, "learning_rate": 0.0001, "loss": 0.0417, "step": 5540 }, { "epoch": 42.36641221374046, "grad_norm": 2.7974212169647217, "learning_rate": 0.0001, "loss": 0.0419, "step": 5550 }, { "epoch": 42.44274809160305, "grad_norm": 2.6444106101989746, "learning_rate": 0.0001, "loss": 0.0419, "step": 5560 }, { "epoch": 42.51908396946565, "grad_norm": 2.4664876461029053, "learning_rate": 0.0001, "loss": 0.0418, "step": 5570 }, { "epoch": 42.595419847328245, "grad_norm": 2.339710235595703, "learning_rate": 0.0001, "loss": 0.0408, "step": 5580 }, { "epoch": 42.67175572519084, "grad_norm": 2.4500250816345215, "learning_rate": 0.0001, "loss": 0.0409, "step": 5590 }, { "epoch": 42.74809160305343, "grad_norm": 2.691645860671997, "learning_rate": 0.0001, "loss": 0.0402, "step": 5600 }, { "epoch": 42.82442748091603, "grad_norm": 2.8897862434387207, "learning_rate": 0.0001, "loss": 0.0424, "step": 5610 }, { "epoch": 42.900763358778626, "grad_norm": 3.1596169471740723, "learning_rate": 0.0001, "loss": 0.0413, "step": 5620 }, { "epoch": 42.97709923664122, "grad_norm": 2.5730175971984863, "learning_rate": 0.0001, "loss": 0.0426, "step": 5630 }, { "epoch": 43.05343511450382, "grad_norm": 3.1214640140533447, "learning_rate": 0.0001, "loss": 0.0411, "step": 5640 }, { "epoch": 43.12977099236641, "grad_norm": 3.0774435997009277, "learning_rate": 0.0001, "loss": 0.04, "step": 5650 }, { "epoch": 43.20610687022901, "grad_norm": 2.4855499267578125, "learning_rate": 0.0001, "loss": 0.0407, "step": 5660 }, { "epoch": 43.282442748091604, "grad_norm": 2.361482858657837, "learning_rate": 0.0001, "loss": 0.0398, "step": 5670 }, { "epoch": 43.3587786259542, "grad_norm": 3.246044397354126, "learning_rate": 0.0001, "loss": 0.0414, "step": 5680 }, { "epoch": 43.43511450381679, "grad_norm": 2.2616066932678223, "learning_rate": 0.0001, "loss": 0.0397, "step": 5690 }, { "epoch": 43.51145038167939, "grad_norm": 2.5731356143951416, "learning_rate": 0.0001, "loss": 0.0402, "step": 5700 }, { "epoch": 43.587786259541986, "grad_norm": 3.4312613010406494, "learning_rate": 0.0001, "loss": 0.0421, "step": 5710 }, { "epoch": 43.66412213740458, "grad_norm": 3.3782730102539062, "learning_rate": 0.0001, "loss": 0.0434, "step": 5720 }, { "epoch": 43.74045801526717, "grad_norm": 3.0172417163848877, "learning_rate": 0.0001, "loss": 0.0406, "step": 5730 }, { "epoch": 43.81679389312977, "grad_norm": 2.5727875232696533, "learning_rate": 0.0001, "loss": 0.0403, "step": 5740 }, { "epoch": 43.89312977099237, "grad_norm": 2.5881786346435547, "learning_rate": 0.0001, "loss": 0.04, "step": 5750 }, { "epoch": 43.969465648854964, "grad_norm": 2.500659465789795, "learning_rate": 0.0001, "loss": 0.0395, "step": 5760 }, { "epoch": 44.045801526717554, "grad_norm": 2.85809063911438, "learning_rate": 0.0001, "loss": 0.04, "step": 5770 }, { "epoch": 44.12213740458015, "grad_norm": 2.7044007778167725, "learning_rate": 0.0001, "loss": 0.0389, "step": 5780 }, { "epoch": 44.19847328244275, "grad_norm": 2.6302437782287598, "learning_rate": 0.0001, "loss": 0.0411, "step": 5790 }, { "epoch": 44.274809160305345, "grad_norm": 2.4910266399383545, "learning_rate": 0.0001, "loss": 0.0384, "step": 5800 }, { "epoch": 44.35114503816794, "grad_norm": 2.453784465789795, "learning_rate": 0.0001, "loss": 0.0427, "step": 5810 }, { "epoch": 44.42748091603053, "grad_norm": 2.1327147483825684, "learning_rate": 0.0001, "loss": 0.0413, "step": 5820 }, { "epoch": 44.50381679389313, "grad_norm": 3.083108901977539, "learning_rate": 0.0001, "loss": 0.0406, "step": 5830 }, { "epoch": 44.58015267175573, "grad_norm": 2.251417875289917, "learning_rate": 0.0001, "loss": 0.04, "step": 5840 }, { "epoch": 44.656488549618324, "grad_norm": 2.787566900253296, "learning_rate": 0.0001, "loss": 0.039, "step": 5850 }, { "epoch": 44.732824427480914, "grad_norm": 2.512261390686035, "learning_rate": 0.0001, "loss": 0.0402, "step": 5860 }, { "epoch": 44.80916030534351, "grad_norm": 2.1494898796081543, "learning_rate": 0.0001, "loss": 0.0388, "step": 5870 }, { "epoch": 44.88549618320611, "grad_norm": 2.544969320297241, "learning_rate": 0.0001, "loss": 0.0373, "step": 5880 }, { "epoch": 44.961832061068705, "grad_norm": 2.20096755027771, "learning_rate": 0.0001, "loss": 0.0367, "step": 5890 }, { "epoch": 45.038167938931295, "grad_norm": 2.5262482166290283, "learning_rate": 0.0001, "loss": 0.0375, "step": 5900 }, { "epoch": 45.11450381679389, "grad_norm": 2.6007280349731445, "learning_rate": 0.0001, "loss": 0.0395, "step": 5910 }, { "epoch": 45.19083969465649, "grad_norm": 2.765237808227539, "learning_rate": 0.0001, "loss": 0.0391, "step": 5920 }, { "epoch": 45.267175572519086, "grad_norm": 2.7387094497680664, "learning_rate": 0.0001, "loss": 0.0407, "step": 5930 }, { "epoch": 45.343511450381676, "grad_norm": 2.384263038635254, "learning_rate": 0.0001, "loss": 0.0389, "step": 5940 }, { "epoch": 45.41984732824427, "grad_norm": 2.556217670440674, "learning_rate": 0.0001, "loss": 0.0383, "step": 5950 }, { "epoch": 45.49618320610687, "grad_norm": 2.3596951961517334, "learning_rate": 0.0001, "loss": 0.0392, "step": 5960 }, { "epoch": 45.57251908396947, "grad_norm": 2.4544594287872314, "learning_rate": 0.0001, "loss": 0.0386, "step": 5970 }, { "epoch": 45.64885496183206, "grad_norm": 2.33474063873291, "learning_rate": 0.0001, "loss": 0.0393, "step": 5980 }, { "epoch": 45.725190839694655, "grad_norm": 2.6560633182525635, "learning_rate": 0.0001, "loss": 0.0397, "step": 5990 }, { "epoch": 45.80152671755725, "grad_norm": 3.5202627182006836, "learning_rate": 0.0001, "loss": 0.0399, "step": 6000 }, { "epoch": 45.87786259541985, "grad_norm": 2.146144151687622, "learning_rate": 0.0001, "loss": 0.0394, "step": 6010 }, { "epoch": 45.954198473282446, "grad_norm": 2.597266435623169, "learning_rate": 0.0001, "loss": 0.0381, "step": 6020 }, { "epoch": 46.030534351145036, "grad_norm": 2.517627477645874, "learning_rate": 0.0001, "loss": 0.0403, "step": 6030 }, { "epoch": 46.10687022900763, "grad_norm": 2.548168182373047, "learning_rate": 0.0001, "loss": 0.038, "step": 6040 }, { "epoch": 46.18320610687023, "grad_norm": 2.6763503551483154, "learning_rate": 0.0001, "loss": 0.0385, "step": 6050 }, { "epoch": 46.25954198473283, "grad_norm": 2.0182878971099854, "learning_rate": 0.0001, "loss": 0.0377, "step": 6060 }, { "epoch": 46.33587786259542, "grad_norm": 2.5702624320983887, "learning_rate": 0.0001, "loss": 0.0382, "step": 6070 }, { "epoch": 46.412213740458014, "grad_norm": 2.560995578765869, "learning_rate": 0.0001, "loss": 0.0388, "step": 6080 }, { "epoch": 46.48854961832061, "grad_norm": 2.580677032470703, "learning_rate": 0.0001, "loss": 0.0398, "step": 6090 }, { "epoch": 46.56488549618321, "grad_norm": 2.2916321754455566, "learning_rate": 0.0001, "loss": 0.0392, "step": 6100 }, { "epoch": 46.6412213740458, "grad_norm": 3.1075310707092285, "learning_rate": 0.0001, "loss": 0.0393, "step": 6110 }, { "epoch": 46.717557251908396, "grad_norm": 2.3877203464508057, "learning_rate": 0.0001, "loss": 0.0408, "step": 6120 }, { "epoch": 46.79389312977099, "grad_norm": 2.422913074493408, "learning_rate": 0.0001, "loss": 0.0399, "step": 6130 }, { "epoch": 46.87022900763359, "grad_norm": 2.407735824584961, "learning_rate": 0.0001, "loss": 0.039, "step": 6140 }, { "epoch": 46.94656488549618, "grad_norm": 2.28184175491333, "learning_rate": 0.0001, "loss": 0.0393, "step": 6150 }, { "epoch": 47.02290076335878, "grad_norm": 3.450758934020996, "learning_rate": 0.0001, "loss": 0.0401, "step": 6160 }, { "epoch": 47.099236641221374, "grad_norm": 2.561671257019043, "learning_rate": 0.0001, "loss": 0.0399, "step": 6170 }, { "epoch": 47.17557251908397, "grad_norm": 2.6351301670074463, "learning_rate": 0.0001, "loss": 0.0395, "step": 6180 }, { "epoch": 47.25190839694657, "grad_norm": 2.262587308883667, "learning_rate": 0.0001, "loss": 0.039, "step": 6190 }, { "epoch": 47.32824427480916, "grad_norm": 2.262640953063965, "learning_rate": 0.0001, "loss": 0.038, "step": 6200 }, { "epoch": 47.404580152671755, "grad_norm": 3.016984462738037, "learning_rate": 0.0001, "loss": 0.0384, "step": 6210 }, { "epoch": 47.48091603053435, "grad_norm": 2.5919418334960938, "learning_rate": 0.0001, "loss": 0.0394, "step": 6220 }, { "epoch": 47.55725190839695, "grad_norm": 2.7218074798583984, "learning_rate": 0.0001, "loss": 0.0408, "step": 6230 }, { "epoch": 47.63358778625954, "grad_norm": 2.6902222633361816, "learning_rate": 0.0001, "loss": 0.0392, "step": 6240 }, { "epoch": 47.70992366412214, "grad_norm": 2.5885958671569824, "learning_rate": 0.0001, "loss": 0.0405, "step": 6250 }, { "epoch": 47.786259541984734, "grad_norm": 2.6370084285736084, "learning_rate": 0.0001, "loss": 0.0392, "step": 6260 }, { "epoch": 47.86259541984733, "grad_norm": 2.484933853149414, "learning_rate": 0.0001, "loss": 0.0392, "step": 6270 }, { "epoch": 47.93893129770992, "grad_norm": 2.566596746444702, "learning_rate": 0.0001, "loss": 0.0377, "step": 6280 }, { "epoch": 48.01526717557252, "grad_norm": 2.4079723358154297, "learning_rate": 0.0001, "loss": 0.0372, "step": 6290 }, { "epoch": 48.091603053435115, "grad_norm": 2.116542339324951, "learning_rate": 0.0001, "loss": 0.0362, "step": 6300 }, { "epoch": 48.16793893129771, "grad_norm": 2.5575928688049316, "learning_rate": 0.0001, "loss": 0.0388, "step": 6310 }, { "epoch": 48.2442748091603, "grad_norm": 2.520078420639038, "learning_rate": 0.0001, "loss": 0.0373, "step": 6320 }, { "epoch": 48.3206106870229, "grad_norm": 2.1232094764709473, "learning_rate": 0.0001, "loss": 0.0373, "step": 6330 }, { "epoch": 48.396946564885496, "grad_norm": 2.458665609359741, "learning_rate": 0.0001, "loss": 0.0369, "step": 6340 }, { "epoch": 48.47328244274809, "grad_norm": 2.0642988681793213, "learning_rate": 0.0001, "loss": 0.0369, "step": 6350 }, { "epoch": 48.54961832061069, "grad_norm": 2.617170810699463, "learning_rate": 0.0001, "loss": 0.0381, "step": 6360 }, { "epoch": 48.62595419847328, "grad_norm": 2.2888572216033936, "learning_rate": 0.0001, "loss": 0.039, "step": 6370 }, { "epoch": 48.70229007633588, "grad_norm": 2.777254343032837, "learning_rate": 0.0001, "loss": 0.0395, "step": 6380 }, { "epoch": 48.778625954198475, "grad_norm": 3.3420934677124023, "learning_rate": 0.0001, "loss": 0.0381, "step": 6390 }, { "epoch": 48.85496183206107, "grad_norm": 2.1780619621276855, "learning_rate": 0.0001, "loss": 0.037, "step": 6400 }, { "epoch": 48.93129770992366, "grad_norm": 2.4814014434814453, "learning_rate": 0.0001, "loss": 0.0369, "step": 6410 }, { "epoch": 49.00763358778626, "grad_norm": 2.3347229957580566, "learning_rate": 0.0001, "loss": 0.0356, "step": 6420 }, { "epoch": 49.083969465648856, "grad_norm": 2.5095677375793457, "learning_rate": 0.0001, "loss": 0.0377, "step": 6430 }, { "epoch": 49.16030534351145, "grad_norm": 2.5740671157836914, "learning_rate": 0.0001, "loss": 0.0358, "step": 6440 }, { "epoch": 49.23664122137404, "grad_norm": 2.3154501914978027, "learning_rate": 0.0001, "loss": 0.0373, "step": 6450 }, { "epoch": 49.31297709923664, "grad_norm": 2.4934825897216797, "learning_rate": 0.0001, "loss": 0.0359, "step": 6460 }, { "epoch": 49.38931297709924, "grad_norm": 2.797114849090576, "learning_rate": 0.0001, "loss": 0.0376, "step": 6470 }, { "epoch": 49.465648854961835, "grad_norm": 2.615638017654419, "learning_rate": 0.0001, "loss": 0.036, "step": 6480 }, { "epoch": 49.541984732824424, "grad_norm": 2.262054920196533, "learning_rate": 0.0001, "loss": 0.0379, "step": 6490 }, { "epoch": 49.61832061068702, "grad_norm": 2.8484206199645996, "learning_rate": 0.0001, "loss": 0.0376, "step": 6500 }, { "epoch": 49.69465648854962, "grad_norm": 2.2632737159729004, "learning_rate": 0.0001, "loss": 0.0373, "step": 6510 }, { "epoch": 49.770992366412216, "grad_norm": 2.5639185905456543, "learning_rate": 0.0001, "loss": 0.0378, "step": 6520 }, { "epoch": 49.847328244274806, "grad_norm": 2.259215831756592, "learning_rate": 0.0001, "loss": 0.0368, "step": 6530 }, { "epoch": 49.9236641221374, "grad_norm": 2.5548930168151855, "learning_rate": 0.0001, "loss": 0.0369, "step": 6540 }, { "epoch": 50.0, "grad_norm": 2.0138444900512695, "learning_rate": 0.0001, "loss": 0.0377, "step": 6550 }, { "epoch": 50.0763358778626, "grad_norm": 2.6028785705566406, "learning_rate": 0.0001, "loss": 0.0393, "step": 6560 }, { "epoch": 50.152671755725194, "grad_norm": 3.247865915298462, "learning_rate": 0.0001, "loss": 0.0404, "step": 6570 }, { "epoch": 50.229007633587784, "grad_norm": 2.6646339893341064, "learning_rate": 0.0001, "loss": 0.0376, "step": 6580 }, { "epoch": 50.30534351145038, "grad_norm": 2.5719568729400635, "learning_rate": 0.0001, "loss": 0.0378, "step": 6590 }, { "epoch": 50.38167938931298, "grad_norm": 2.3843116760253906, "learning_rate": 0.0001, "loss": 0.0381, "step": 6600 }, { "epoch": 50.458015267175576, "grad_norm": 2.444913387298584, "learning_rate": 0.0001, "loss": 0.0354, "step": 6610 }, { "epoch": 50.534351145038165, "grad_norm": 2.471081018447876, "learning_rate": 0.0001, "loss": 0.0357, "step": 6620 }, { "epoch": 50.61068702290076, "grad_norm": 2.4210753440856934, "learning_rate": 0.0001, "loss": 0.0359, "step": 6630 }, { "epoch": 50.68702290076336, "grad_norm": 2.2749478816986084, "learning_rate": 0.0001, "loss": 0.0367, "step": 6640 }, { "epoch": 50.76335877862596, "grad_norm": 1.976814866065979, "learning_rate": 0.0001, "loss": 0.0381, "step": 6650 }, { "epoch": 50.83969465648855, "grad_norm": 2.850905656814575, "learning_rate": 0.0001, "loss": 0.0375, "step": 6660 }, { "epoch": 50.916030534351144, "grad_norm": 2.4471330642700195, "learning_rate": 0.0001, "loss": 0.0416, "step": 6670 }, { "epoch": 50.99236641221374, "grad_norm": 2.158838987350464, "learning_rate": 0.0001, "loss": 0.0367, "step": 6680 }, { "epoch": 51.06870229007634, "grad_norm": 2.4949498176574707, "learning_rate": 0.0001, "loss": 0.0358, "step": 6690 }, { "epoch": 51.14503816793893, "grad_norm": 2.1088666915893555, "learning_rate": 0.0001, "loss": 0.0364, "step": 6700 }, { "epoch": 51.221374045801525, "grad_norm": 2.2596609592437744, "learning_rate": 0.0001, "loss": 0.0366, "step": 6710 }, { "epoch": 51.29770992366412, "grad_norm": 2.850332736968994, "learning_rate": 0.0001, "loss": 0.036, "step": 6720 }, { "epoch": 51.37404580152672, "grad_norm": 2.4383704662323, "learning_rate": 0.0001, "loss": 0.037, "step": 6730 }, { "epoch": 51.45038167938931, "grad_norm": 2.0408389568328857, "learning_rate": 0.0001, "loss": 0.0375, "step": 6740 }, { "epoch": 51.52671755725191, "grad_norm": 2.2277939319610596, "learning_rate": 0.0001, "loss": 0.0374, "step": 6750 }, { "epoch": 51.603053435114504, "grad_norm": 2.130634069442749, "learning_rate": 0.0001, "loss": 0.0367, "step": 6760 }, { "epoch": 51.6793893129771, "grad_norm": 2.1904540061950684, "learning_rate": 0.0001, "loss": 0.037, "step": 6770 }, { "epoch": 51.7557251908397, "grad_norm": 2.602275848388672, "learning_rate": 0.0001, "loss": 0.0374, "step": 6780 }, { "epoch": 51.83206106870229, "grad_norm": 2.5852606296539307, "learning_rate": 0.0001, "loss": 0.0388, "step": 6790 }, { "epoch": 51.908396946564885, "grad_norm": 2.526120901107788, "learning_rate": 0.0001, "loss": 0.0376, "step": 6800 }, { "epoch": 51.98473282442748, "grad_norm": 2.2341883182525635, "learning_rate": 0.0001, "loss": 0.0367, "step": 6810 }, { "epoch": 52.06106870229008, "grad_norm": 1.9465210437774658, "learning_rate": 0.0001, "loss": 0.0367, "step": 6820 }, { "epoch": 52.13740458015267, "grad_norm": 2.279557228088379, "learning_rate": 0.0001, "loss": 0.0378, "step": 6830 }, { "epoch": 52.213740458015266, "grad_norm": 2.4117469787597656, "learning_rate": 0.0001, "loss": 0.0388, "step": 6840 }, { "epoch": 52.29007633587786, "grad_norm": 1.7653770446777344, "learning_rate": 0.0001, "loss": 0.0375, "step": 6850 }, { "epoch": 52.36641221374046, "grad_norm": 2.6824357509613037, "learning_rate": 0.0001, "loss": 0.0359, "step": 6860 }, { "epoch": 52.44274809160305, "grad_norm": 2.5237090587615967, "learning_rate": 0.0001, "loss": 0.0368, "step": 6870 }, { "epoch": 52.51908396946565, "grad_norm": 1.9553059339523315, "learning_rate": 0.0001, "loss": 0.0347, "step": 6880 }, { "epoch": 52.595419847328245, "grad_norm": 2.4210915565490723, "learning_rate": 0.0001, "loss": 0.0364, "step": 6890 }, { "epoch": 52.67175572519084, "grad_norm": 2.4859330654144287, "learning_rate": 0.0001, "loss": 0.0359, "step": 6900 }, { "epoch": 52.74809160305343, "grad_norm": 2.990481376647949, "learning_rate": 0.0001, "loss": 0.0356, "step": 6910 }, { "epoch": 52.82442748091603, "grad_norm": 2.0916519165039062, "learning_rate": 0.0001, "loss": 0.0346, "step": 6920 }, { "epoch": 52.900763358778626, "grad_norm": 2.243943452835083, "learning_rate": 0.0001, "loss": 0.0359, "step": 6930 }, { "epoch": 52.97709923664122, "grad_norm": 2.7910895347595215, "learning_rate": 0.0001, "loss": 0.0361, "step": 6940 }, { "epoch": 53.05343511450382, "grad_norm": 2.399101972579956, "learning_rate": 0.0001, "loss": 0.0367, "step": 6950 }, { "epoch": 53.12977099236641, "grad_norm": 2.9079952239990234, "learning_rate": 0.0001, "loss": 0.0366, "step": 6960 }, { "epoch": 53.20610687022901, "grad_norm": 2.5616681575775146, "learning_rate": 0.0001, "loss": 0.0366, "step": 6970 }, { "epoch": 53.282442748091604, "grad_norm": 2.0228450298309326, "learning_rate": 0.0001, "loss": 0.0353, "step": 6980 }, { "epoch": 53.3587786259542, "grad_norm": 2.528608560562134, "learning_rate": 0.0001, "loss": 0.0363, "step": 6990 }, { "epoch": 53.43511450381679, "grad_norm": 2.5268826484680176, "learning_rate": 0.0001, "loss": 0.0359, "step": 7000 }, { "epoch": 53.51145038167939, "grad_norm": 3.4130141735076904, "learning_rate": 0.0001, "loss": 0.0368, "step": 7010 }, { "epoch": 53.587786259541986, "grad_norm": 2.201747179031372, "learning_rate": 0.0001, "loss": 0.0369, "step": 7020 }, { "epoch": 53.66412213740458, "grad_norm": 2.412712335586548, "learning_rate": 0.0001, "loss": 0.0349, "step": 7030 }, { "epoch": 53.74045801526717, "grad_norm": 2.5812766551971436, "learning_rate": 0.0001, "loss": 0.0355, "step": 7040 }, { "epoch": 53.81679389312977, "grad_norm": 2.0464043617248535, "learning_rate": 0.0001, "loss": 0.0353, "step": 7050 }, { "epoch": 53.89312977099237, "grad_norm": 1.8771175146102905, "learning_rate": 0.0001, "loss": 0.0333, "step": 7060 }, { "epoch": 53.969465648854964, "grad_norm": 1.974566102027893, "learning_rate": 0.0001, "loss": 0.0356, "step": 7070 }, { "epoch": 54.045801526717554, "grad_norm": 1.9990216493606567, "learning_rate": 0.0001, "loss": 0.0345, "step": 7080 }, { "epoch": 54.12213740458015, "grad_norm": 2.4714646339416504, "learning_rate": 0.0001, "loss": 0.0356, "step": 7090 }, { "epoch": 54.19847328244275, "grad_norm": 1.8311411142349243, "learning_rate": 0.0001, "loss": 0.0352, "step": 7100 }, { "epoch": 54.274809160305345, "grad_norm": 2.9636130332946777, "learning_rate": 0.0001, "loss": 0.0334, "step": 7110 }, { "epoch": 54.35114503816794, "grad_norm": 1.998831033706665, "learning_rate": 0.0001, "loss": 0.0349, "step": 7120 }, { "epoch": 54.42748091603053, "grad_norm": 2.3731849193573, "learning_rate": 0.0001, "loss": 0.0345, "step": 7130 }, { "epoch": 54.50381679389313, "grad_norm": 2.6849381923675537, "learning_rate": 0.0001, "loss": 0.0338, "step": 7140 }, { "epoch": 54.58015267175573, "grad_norm": 2.025912284851074, "learning_rate": 0.0001, "loss": 0.0345, "step": 7150 }, { "epoch": 54.656488549618324, "grad_norm": 1.978691577911377, "learning_rate": 0.0001, "loss": 0.0336, "step": 7160 }, { "epoch": 54.732824427480914, "grad_norm": 2.3142426013946533, "learning_rate": 0.0001, "loss": 0.034, "step": 7170 }, { "epoch": 54.80916030534351, "grad_norm": 2.588580846786499, "learning_rate": 0.0001, "loss": 0.0349, "step": 7180 }, { "epoch": 54.88549618320611, "grad_norm": 2.1301097869873047, "learning_rate": 0.0001, "loss": 0.0356, "step": 7190 }, { "epoch": 54.961832061068705, "grad_norm": 2.1245341300964355, "learning_rate": 0.0001, "loss": 0.035, "step": 7200 }, { "epoch": 55.038167938931295, "grad_norm": 2.371931552886963, "learning_rate": 0.0001, "loss": 0.0367, "step": 7210 }, { "epoch": 55.11450381679389, "grad_norm": 2.509474992752075, "learning_rate": 0.0001, "loss": 0.0358, "step": 7220 }, { "epoch": 55.19083969465649, "grad_norm": 2.3232922554016113, "learning_rate": 0.0001, "loss": 0.0373, "step": 7230 }, { "epoch": 55.267175572519086, "grad_norm": 2.3745455741882324, "learning_rate": 0.0001, "loss": 0.0354, "step": 7240 }, { "epoch": 55.343511450381676, "grad_norm": 2.462066411972046, "learning_rate": 0.0001, "loss": 0.0366, "step": 7250 }, { "epoch": 55.41984732824427, "grad_norm": 1.8799290657043457, "learning_rate": 0.0001, "loss": 0.0337, "step": 7260 }, { "epoch": 55.49618320610687, "grad_norm": 1.992353081703186, "learning_rate": 0.0001, "loss": 0.0338, "step": 7270 }, { "epoch": 55.57251908396947, "grad_norm": 2.2094357013702393, "learning_rate": 0.0001, "loss": 0.0332, "step": 7280 }, { "epoch": 55.64885496183206, "grad_norm": 2.4099769592285156, "learning_rate": 0.0001, "loss": 0.0353, "step": 7290 }, { "epoch": 55.725190839694655, "grad_norm": 2.0105793476104736, "learning_rate": 0.0001, "loss": 0.0327, "step": 7300 }, { "epoch": 55.80152671755725, "grad_norm": 2.256767511367798, "learning_rate": 0.0001, "loss": 0.0335, "step": 7310 }, { "epoch": 55.87786259541985, "grad_norm": 2.0141239166259766, "learning_rate": 0.0001, "loss": 0.0336, "step": 7320 }, { "epoch": 55.954198473282446, "grad_norm": 2.021768808364868, "learning_rate": 0.0001, "loss": 0.0341, "step": 7330 }, { "epoch": 56.030534351145036, "grad_norm": 1.9869009256362915, "learning_rate": 0.0001, "loss": 0.0347, "step": 7340 }, { "epoch": 56.10687022900763, "grad_norm": 2.307389974594116, "learning_rate": 0.0001, "loss": 0.0357, "step": 7350 }, { "epoch": 56.18320610687023, "grad_norm": 2.200424909591675, "learning_rate": 0.0001, "loss": 0.0357, "step": 7360 }, { "epoch": 56.25954198473283, "grad_norm": 2.077596664428711, "learning_rate": 0.0001, "loss": 0.0368, "step": 7370 }, { "epoch": 56.33587786259542, "grad_norm": 2.098054885864258, "learning_rate": 0.0001, "loss": 0.0366, "step": 7380 }, { "epoch": 56.412213740458014, "grad_norm": 2.385228157043457, "learning_rate": 0.0001, "loss": 0.0356, "step": 7390 }, { "epoch": 56.48854961832061, "grad_norm": 2.8476154804229736, "learning_rate": 0.0001, "loss": 0.0348, "step": 7400 }, { "epoch": 56.56488549618321, "grad_norm": 2.106201648712158, "learning_rate": 0.0001, "loss": 0.0352, "step": 7410 }, { "epoch": 56.6412213740458, "grad_norm": 2.2042043209075928, "learning_rate": 0.0001, "loss": 0.0352, "step": 7420 }, { "epoch": 56.717557251908396, "grad_norm": 2.1945607662200928, "learning_rate": 0.0001, "loss": 0.0356, "step": 7430 }, { "epoch": 56.79389312977099, "grad_norm": 2.37185001373291, "learning_rate": 0.0001, "loss": 0.0355, "step": 7440 }, { "epoch": 56.87022900763359, "grad_norm": 2.4911396503448486, "learning_rate": 0.0001, "loss": 0.0358, "step": 7450 }, { "epoch": 56.94656488549618, "grad_norm": 2.899531602859497, "learning_rate": 0.0001, "loss": 0.0332, "step": 7460 }, { "epoch": 57.02290076335878, "grad_norm": 2.395420789718628, "learning_rate": 0.0001, "loss": 0.0339, "step": 7470 }, { "epoch": 57.099236641221374, "grad_norm": 1.8638272285461426, "learning_rate": 0.0001, "loss": 0.0343, "step": 7480 }, { "epoch": 57.17557251908397, "grad_norm": 2.553999423980713, "learning_rate": 0.0001, "loss": 0.0344, "step": 7490 }, { "epoch": 57.25190839694657, "grad_norm": 2.0195605754852295, "learning_rate": 0.0001, "loss": 0.034, "step": 7500 }, { "epoch": 57.32824427480916, "grad_norm": 1.9203416109085083, "learning_rate": 0.0001, "loss": 0.0339, "step": 7510 }, { "epoch": 57.404580152671755, "grad_norm": 2.2202751636505127, "learning_rate": 0.0001, "loss": 0.0343, "step": 7520 }, { "epoch": 57.48091603053435, "grad_norm": 1.7887217998504639, "learning_rate": 0.0001, "loss": 0.0337, "step": 7530 }, { "epoch": 57.55725190839695, "grad_norm": 2.249825954437256, "learning_rate": 0.0001, "loss": 0.0338, "step": 7540 }, { "epoch": 57.63358778625954, "grad_norm": 2.2510952949523926, "learning_rate": 0.0001, "loss": 0.0347, "step": 7550 }, { "epoch": 57.70992366412214, "grad_norm": 2.1027770042419434, "learning_rate": 0.0001, "loss": 0.0345, "step": 7560 }, { "epoch": 57.786259541984734, "grad_norm": 1.955045223236084, "learning_rate": 0.0001, "loss": 0.035, "step": 7570 }, { "epoch": 57.86259541984733, "grad_norm": 2.7056710720062256, "learning_rate": 0.0001, "loss": 0.0336, "step": 7580 }, { "epoch": 57.93893129770992, "grad_norm": 2.3583877086639404, "learning_rate": 0.0001, "loss": 0.0351, "step": 7590 }, { "epoch": 58.01526717557252, "grad_norm": 2.0031862258911133, "learning_rate": 0.0001, "loss": 0.0363, "step": 7600 }, { "epoch": 58.091603053435115, "grad_norm": 3.1002542972564697, "learning_rate": 0.0001, "loss": 0.0359, "step": 7610 }, { "epoch": 58.16793893129771, "grad_norm": 2.3348140716552734, "learning_rate": 0.0001, "loss": 0.0357, "step": 7620 }, { "epoch": 58.2442748091603, "grad_norm": 2.2558951377868652, "learning_rate": 0.0001, "loss": 0.0346, "step": 7630 }, { "epoch": 58.3206106870229, "grad_norm": 2.0427920818328857, "learning_rate": 0.0001, "loss": 0.034, "step": 7640 }, { "epoch": 58.396946564885496, "grad_norm": 2.4306886196136475, "learning_rate": 0.0001, "loss": 0.0334, "step": 7650 }, { "epoch": 58.47328244274809, "grad_norm": 1.654780626296997, "learning_rate": 0.0001, "loss": 0.0337, "step": 7660 }, { "epoch": 58.54961832061069, "grad_norm": 2.358701229095459, "learning_rate": 0.0001, "loss": 0.0339, "step": 7670 }, { "epoch": 58.62595419847328, "grad_norm": 2.0678493976593018, "learning_rate": 0.0001, "loss": 0.0343, "step": 7680 }, { "epoch": 58.70229007633588, "grad_norm": 2.375300884246826, "learning_rate": 0.0001, "loss": 0.0339, "step": 7690 }, { "epoch": 58.778625954198475, "grad_norm": 2.1627490520477295, "learning_rate": 0.0001, "loss": 0.0337, "step": 7700 }, { "epoch": 58.85496183206107, "grad_norm": 1.9372777938842773, "learning_rate": 0.0001, "loss": 0.0352, "step": 7710 }, { "epoch": 58.93129770992366, "grad_norm": 2.1236376762390137, "learning_rate": 0.0001, "loss": 0.0341, "step": 7720 }, { "epoch": 59.00763358778626, "grad_norm": 2.2824532985687256, "learning_rate": 0.0001, "loss": 0.0337, "step": 7730 }, { "epoch": 59.083969465648856, "grad_norm": 1.9238686561584473, "learning_rate": 0.0001, "loss": 0.0352, "step": 7740 }, { "epoch": 59.16030534351145, "grad_norm": 2.1847448348999023, "learning_rate": 0.0001, "loss": 0.0341, "step": 7750 }, { "epoch": 59.23664122137404, "grad_norm": 2.187697172164917, "learning_rate": 0.0001, "loss": 0.0356, "step": 7760 }, { "epoch": 59.31297709923664, "grad_norm": 1.8435051441192627, "learning_rate": 0.0001, "loss": 0.0336, "step": 7770 }, { "epoch": 59.38931297709924, "grad_norm": 2.5332248210906982, "learning_rate": 0.0001, "loss": 0.0353, "step": 7780 }, { "epoch": 59.465648854961835, "grad_norm": 2.6252660751342773, "learning_rate": 0.0001, "loss": 0.0354, "step": 7790 }, { "epoch": 59.541984732824424, "grad_norm": 1.8487701416015625, "learning_rate": 0.0001, "loss": 0.0345, "step": 7800 }, { "epoch": 59.61832061068702, "grad_norm": 2.389089584350586, "learning_rate": 0.0001, "loss": 0.0351, "step": 7810 }, { "epoch": 59.69465648854962, "grad_norm": 2.7237071990966797, "learning_rate": 0.0001, "loss": 0.0348, "step": 7820 }, { "epoch": 59.770992366412216, "grad_norm": 1.9663394689559937, "learning_rate": 0.0001, "loss": 0.0348, "step": 7830 }, { "epoch": 59.847328244274806, "grad_norm": 2.227137804031372, "learning_rate": 0.0001, "loss": 0.0342, "step": 7840 }, { "epoch": 59.9236641221374, "grad_norm": 2.5541322231292725, "learning_rate": 0.0001, "loss": 0.0362, "step": 7850 }, { "epoch": 60.0, "grad_norm": 4.0418829917907715, "learning_rate": 0.0001, "loss": 0.0382, "step": 7860 }, { "epoch": 60.0763358778626, "grad_norm": 2.747990608215332, "learning_rate": 0.0001, "loss": 0.0363, "step": 7870 }, { "epoch": 60.152671755725194, "grad_norm": 3.020023822784424, "learning_rate": 0.0001, "loss": 0.0356, "step": 7880 }, { "epoch": 60.229007633587784, "grad_norm": 3.0537092685699463, "learning_rate": 0.0001, "loss": 0.0338, "step": 7890 }, { "epoch": 60.30534351145038, "grad_norm": 2.2654378414154053, "learning_rate": 0.0001, "loss": 0.0334, "step": 7900 }, { "epoch": 60.38167938931298, "grad_norm": 2.439066171646118, "learning_rate": 0.0001, "loss": 0.0329, "step": 7910 }, { "epoch": 60.458015267175576, "grad_norm": 2.3557534217834473, "learning_rate": 0.0001, "loss": 0.0319, "step": 7920 }, { "epoch": 60.534351145038165, "grad_norm": 2.9069554805755615, "learning_rate": 0.0001, "loss": 0.0326, "step": 7930 }, { "epoch": 60.61068702290076, "grad_norm": 2.6401805877685547, "learning_rate": 0.0001, "loss": 0.0325, "step": 7940 }, { "epoch": 60.68702290076336, "grad_norm": 2.161010980606079, "learning_rate": 0.0001, "loss": 0.0323, "step": 7950 }, { "epoch": 60.76335877862596, "grad_norm": 2.290264129638672, "learning_rate": 0.0001, "loss": 0.0333, "step": 7960 }, { "epoch": 60.83969465648855, "grad_norm": 2.4023067951202393, "learning_rate": 0.0001, "loss": 0.0316, "step": 7970 }, { "epoch": 60.916030534351144, "grad_norm": 1.601913332939148, "learning_rate": 0.0001, "loss": 0.0305, "step": 7980 }, { "epoch": 60.99236641221374, "grad_norm": 2.1671409606933594, "learning_rate": 0.0001, "loss": 0.032, "step": 7990 }, { "epoch": 61.06870229007634, "grad_norm": 2.8818705081939697, "learning_rate": 0.0001, "loss": 0.0338, "step": 8000 }, { "epoch": 61.14503816793893, "grad_norm": 1.930097222328186, "learning_rate": 0.0001, "loss": 0.0342, "step": 8010 }, { "epoch": 61.221374045801525, "grad_norm": 2.2678327560424805, "learning_rate": 0.0001, "loss": 0.0341, "step": 8020 }, { "epoch": 61.29770992366412, "grad_norm": 1.7841767072677612, "learning_rate": 0.0001, "loss": 0.0346, "step": 8030 }, { "epoch": 61.37404580152672, "grad_norm": 2.5320069789886475, "learning_rate": 0.0001, "loss": 0.0352, "step": 8040 }, { "epoch": 61.45038167938931, "grad_norm": 2.353942394256592, "learning_rate": 0.0001, "loss": 0.0344, "step": 8050 }, { "epoch": 61.52671755725191, "grad_norm": 2.1045732498168945, "learning_rate": 0.0001, "loss": 0.0338, "step": 8060 }, { "epoch": 61.603053435114504, "grad_norm": 2.218026638031006, "learning_rate": 0.0001, "loss": 0.034, "step": 8070 }, { "epoch": 61.6793893129771, "grad_norm": 1.682723879814148, "learning_rate": 0.0001, "loss": 0.032, "step": 8080 }, { "epoch": 61.7557251908397, "grad_norm": 2.0885679721832275, "learning_rate": 0.0001, "loss": 0.0325, "step": 8090 }, { "epoch": 61.83206106870229, "grad_norm": 1.9043333530426025, "learning_rate": 0.0001, "loss": 0.0329, "step": 8100 }, { "epoch": 61.908396946564885, "grad_norm": 2.2113804817199707, "learning_rate": 0.0001, "loss": 0.033, "step": 8110 }, { "epoch": 61.98473282442748, "grad_norm": 2.2499167919158936, "learning_rate": 0.0001, "loss": 0.0337, "step": 8120 }, { "epoch": 62.06106870229008, "grad_norm": 1.9069730043411255, "learning_rate": 0.0001, "loss": 0.0333, "step": 8130 }, { "epoch": 62.13740458015267, "grad_norm": 1.799684762954712, "learning_rate": 0.0001, "loss": 0.033, "step": 8140 }, { "epoch": 62.213740458015266, "grad_norm": 2.0678136348724365, "learning_rate": 0.0001, "loss": 0.0334, "step": 8150 }, { "epoch": 62.29007633587786, "grad_norm": 2.008470296859741, "learning_rate": 0.0001, "loss": 0.0331, "step": 8160 }, { "epoch": 62.36641221374046, "grad_norm": 2.4037301540374756, "learning_rate": 0.0001, "loss": 0.0325, "step": 8170 }, { "epoch": 62.44274809160305, "grad_norm": 2.2827162742614746, "learning_rate": 0.0001, "loss": 0.0325, "step": 8180 }, { "epoch": 62.51908396946565, "grad_norm": 1.6348804235458374, "learning_rate": 0.0001, "loss": 0.0324, "step": 8190 }, { "epoch": 62.595419847328245, "grad_norm": 2.037236213684082, "learning_rate": 0.0001, "loss": 0.0315, "step": 8200 }, { "epoch": 62.67175572519084, "grad_norm": 2.1576948165893555, "learning_rate": 0.0001, "loss": 0.0323, "step": 8210 }, { "epoch": 62.74809160305343, "grad_norm": 2.1814310550689697, "learning_rate": 0.0001, "loss": 0.0329, "step": 8220 }, { "epoch": 62.82442748091603, "grad_norm": 2.5274770259857178, "learning_rate": 0.0001, "loss": 0.0327, "step": 8230 }, { "epoch": 62.900763358778626, "grad_norm": 2.0999045372009277, "learning_rate": 0.0001, "loss": 0.0341, "step": 8240 }, { "epoch": 62.97709923664122, "grad_norm": 1.4280180931091309, "learning_rate": 0.0001, "loss": 0.0332, "step": 8250 }, { "epoch": 63.05343511450382, "grad_norm": 1.9395331144332886, "learning_rate": 0.0001, "loss": 0.0318, "step": 8260 }, { "epoch": 63.12977099236641, "grad_norm": 2.1114752292633057, "learning_rate": 0.0001, "loss": 0.0334, "step": 8270 }, { "epoch": 63.20610687022901, "grad_norm": 2.326430082321167, "learning_rate": 0.0001, "loss": 0.0322, "step": 8280 }, { "epoch": 63.282442748091604, "grad_norm": 2.0694119930267334, "learning_rate": 0.0001, "loss": 0.0343, "step": 8290 }, { "epoch": 63.3587786259542, "grad_norm": 2.2314443588256836, "learning_rate": 0.0001, "loss": 0.0334, "step": 8300 }, { "epoch": 63.43511450381679, "grad_norm": 1.8394598960876465, "learning_rate": 0.0001, "loss": 0.0328, "step": 8310 }, { "epoch": 63.51145038167939, "grad_norm": 2.205113410949707, "learning_rate": 0.0001, "loss": 0.0333, "step": 8320 }, { "epoch": 63.587786259541986, "grad_norm": 2.261225700378418, "learning_rate": 0.0001, "loss": 0.0325, "step": 8330 }, { "epoch": 63.66412213740458, "grad_norm": 1.9365214109420776, "learning_rate": 0.0001, "loss": 0.0335, "step": 8340 }, { "epoch": 63.74045801526717, "grad_norm": 2.075115442276001, "learning_rate": 0.0001, "loss": 0.0334, "step": 8350 }, { "epoch": 63.81679389312977, "grad_norm": 2.4412317276000977, "learning_rate": 0.0001, "loss": 0.0334, "step": 8360 }, { "epoch": 63.89312977099237, "grad_norm": 2.647555351257324, "learning_rate": 0.0001, "loss": 0.0334, "step": 8370 }, { "epoch": 63.969465648854964, "grad_norm": 2.438798427581787, "learning_rate": 0.0001, "loss": 0.0331, "step": 8380 }, { "epoch": 64.04580152671755, "grad_norm": 1.7357386350631714, "learning_rate": 0.0001, "loss": 0.0316, "step": 8390 }, { "epoch": 64.12213740458016, "grad_norm": 2.051335334777832, "learning_rate": 0.0001, "loss": 0.0314, "step": 8400 }, { "epoch": 64.19847328244275, "grad_norm": 1.8672370910644531, "learning_rate": 0.0001, "loss": 0.0309, "step": 8410 }, { "epoch": 64.27480916030534, "grad_norm": 1.8209877014160156, "learning_rate": 0.0001, "loss": 0.0303, "step": 8420 }, { "epoch": 64.35114503816794, "grad_norm": 1.8265125751495361, "learning_rate": 0.0001, "loss": 0.0307, "step": 8430 }, { "epoch": 64.42748091603053, "grad_norm": 1.9087167978286743, "learning_rate": 0.0001, "loss": 0.0306, "step": 8440 }, { "epoch": 64.50381679389314, "grad_norm": 1.8512228727340698, "learning_rate": 0.0001, "loss": 0.0299, "step": 8450 }, { "epoch": 64.58015267175573, "grad_norm": 1.8608756065368652, "learning_rate": 0.0001, "loss": 0.0314, "step": 8460 }, { "epoch": 64.65648854961832, "grad_norm": 1.8747302293777466, "learning_rate": 0.0001, "loss": 0.0326, "step": 8470 }, { "epoch": 64.73282442748092, "grad_norm": 2.105095386505127, "learning_rate": 0.0001, "loss": 0.0331, "step": 8480 }, { "epoch": 64.80916030534351, "grad_norm": 1.8910152912139893, "learning_rate": 0.0001, "loss": 0.0342, "step": 8490 }, { "epoch": 64.8854961832061, "grad_norm": 2.2022881507873535, "learning_rate": 0.0001, "loss": 0.0322, "step": 8500 }, { "epoch": 64.9618320610687, "grad_norm": 1.7072248458862305, "learning_rate": 0.0001, "loss": 0.0319, "step": 8510 }, { "epoch": 65.0381679389313, "grad_norm": 1.8535088300704956, "learning_rate": 0.0001, "loss": 0.0328, "step": 8520 }, { "epoch": 65.1145038167939, "grad_norm": 1.9654542207717896, "learning_rate": 0.0001, "loss": 0.0328, "step": 8530 }, { "epoch": 65.19083969465649, "grad_norm": 1.87727952003479, "learning_rate": 0.0001, "loss": 0.031, "step": 8540 }, { "epoch": 65.26717557251908, "grad_norm": 1.6790430545806885, "learning_rate": 0.0001, "loss": 0.0319, "step": 8550 }, { "epoch": 65.34351145038168, "grad_norm": 1.8585304021835327, "learning_rate": 0.0001, "loss": 0.0316, "step": 8560 }, { "epoch": 65.41984732824427, "grad_norm": 2.380606174468994, "learning_rate": 0.0001, "loss": 0.0326, "step": 8570 }, { "epoch": 65.49618320610686, "grad_norm": 2.1912593841552734, "learning_rate": 0.0001, "loss": 0.0321, "step": 8580 }, { "epoch": 65.57251908396947, "grad_norm": 2.1981465816497803, "learning_rate": 0.0001, "loss": 0.0329, "step": 8590 }, { "epoch": 65.64885496183206, "grad_norm": 2.0238940715789795, "learning_rate": 0.0001, "loss": 0.0317, "step": 8600 }, { "epoch": 65.72519083969466, "grad_norm": 2.425171136856079, "learning_rate": 0.0001, "loss": 0.0318, "step": 8610 }, { "epoch": 65.80152671755725, "grad_norm": 2.2386350631713867, "learning_rate": 0.0001, "loss": 0.0317, "step": 8620 }, { "epoch": 65.87786259541984, "grad_norm": 2.1742913722991943, "learning_rate": 0.0001, "loss": 0.0314, "step": 8630 }, { "epoch": 65.95419847328245, "grad_norm": 2.585299015045166, "learning_rate": 0.0001, "loss": 0.0331, "step": 8640 }, { "epoch": 66.03053435114504, "grad_norm": 2.399355888366699, "learning_rate": 0.0001, "loss": 0.033, "step": 8650 }, { "epoch": 66.10687022900764, "grad_norm": 2.236966133117676, "learning_rate": 0.0001, "loss": 0.0324, "step": 8660 }, { "epoch": 66.18320610687023, "grad_norm": 3.050044536590576, "learning_rate": 0.0001, "loss": 0.0334, "step": 8670 }, { "epoch": 66.25954198473282, "grad_norm": 1.7139815092086792, "learning_rate": 0.0001, "loss": 0.0329, "step": 8680 }, { "epoch": 66.33587786259542, "grad_norm": 2.8715596199035645, "learning_rate": 0.0001, "loss": 0.0335, "step": 8690 }, { "epoch": 66.41221374045801, "grad_norm": 1.9991495609283447, "learning_rate": 0.0001, "loss": 0.0321, "step": 8700 }, { "epoch": 66.4885496183206, "grad_norm": 1.7902460098266602, "learning_rate": 0.0001, "loss": 0.0312, "step": 8710 }, { "epoch": 66.56488549618321, "grad_norm": 2.018962860107422, "learning_rate": 0.0001, "loss": 0.0314, "step": 8720 }, { "epoch": 66.6412213740458, "grad_norm": 1.8428512811660767, "learning_rate": 0.0001, "loss": 0.0311, "step": 8730 }, { "epoch": 66.7175572519084, "grad_norm": 1.895655632019043, "learning_rate": 0.0001, "loss": 0.0316, "step": 8740 }, { "epoch": 66.79389312977099, "grad_norm": 2.034080982208252, "learning_rate": 0.0001, "loss": 0.032, "step": 8750 }, { "epoch": 66.87022900763358, "grad_norm": 1.9673740863800049, "learning_rate": 0.0001, "loss": 0.0329, "step": 8760 }, { "epoch": 66.94656488549619, "grad_norm": 1.9319981336593628, "learning_rate": 0.0001, "loss": 0.0318, "step": 8770 }, { "epoch": 67.02290076335878, "grad_norm": 2.0224506855010986, "learning_rate": 0.0001, "loss": 0.032, "step": 8780 }, { "epoch": 67.09923664122137, "grad_norm": 1.8986036777496338, "learning_rate": 0.0001, "loss": 0.0311, "step": 8790 }, { "epoch": 67.17557251908397, "grad_norm": 2.006720542907715, "learning_rate": 0.0001, "loss": 0.0316, "step": 8800 }, { "epoch": 67.25190839694656, "grad_norm": 1.9423151016235352, "learning_rate": 0.0001, "loss": 0.0312, "step": 8810 }, { "epoch": 67.32824427480917, "grad_norm": 2.5191385746002197, "learning_rate": 0.0001, "loss": 0.0309, "step": 8820 }, { "epoch": 67.40458015267176, "grad_norm": 2.1544747352600098, "learning_rate": 0.0001, "loss": 0.0322, "step": 8830 }, { "epoch": 67.48091603053435, "grad_norm": 2.0014283657073975, "learning_rate": 0.0001, "loss": 0.0312, "step": 8840 }, { "epoch": 67.55725190839695, "grad_norm": 2.2463228702545166, "learning_rate": 0.0001, "loss": 0.0301, "step": 8850 }, { "epoch": 67.63358778625954, "grad_norm": 1.7769800424575806, "learning_rate": 0.0001, "loss": 0.0316, "step": 8860 }, { "epoch": 67.70992366412214, "grad_norm": 2.3063852787017822, "learning_rate": 0.0001, "loss": 0.0298, "step": 8870 }, { "epoch": 67.78625954198473, "grad_norm": 2.0150914192199707, "learning_rate": 0.0001, "loss": 0.0307, "step": 8880 }, { "epoch": 67.86259541984732, "grad_norm": 1.940226435661316, "learning_rate": 0.0001, "loss": 0.0327, "step": 8890 }, { "epoch": 67.93893129770993, "grad_norm": 2.573270082473755, "learning_rate": 0.0001, "loss": 0.0341, "step": 8900 }, { "epoch": 68.01526717557252, "grad_norm": 2.1373555660247803, "learning_rate": 0.0001, "loss": 0.0334, "step": 8910 }, { "epoch": 68.09160305343511, "grad_norm": 1.7359657287597656, "learning_rate": 0.0001, "loss": 0.0318, "step": 8920 }, { "epoch": 68.16793893129771, "grad_norm": 1.9167956113815308, "learning_rate": 0.0001, "loss": 0.0309, "step": 8930 }, { "epoch": 68.2442748091603, "grad_norm": 2.2278623580932617, "learning_rate": 0.0001, "loss": 0.0314, "step": 8940 }, { "epoch": 68.3206106870229, "grad_norm": 1.8551290035247803, "learning_rate": 0.0001, "loss": 0.0306, "step": 8950 }, { "epoch": 68.3969465648855, "grad_norm": 1.9447733163833618, "learning_rate": 0.0001, "loss": 0.0316, "step": 8960 }, { "epoch": 68.47328244274809, "grad_norm": 2.2287023067474365, "learning_rate": 0.0001, "loss": 0.0311, "step": 8970 }, { "epoch": 68.54961832061069, "grad_norm": 1.998684048652649, "learning_rate": 0.0001, "loss": 0.0323, "step": 8980 }, { "epoch": 68.62595419847328, "grad_norm": 1.8546037673950195, "learning_rate": 0.0001, "loss": 0.0316, "step": 8990 }, { "epoch": 68.70229007633588, "grad_norm": 1.9047350883483887, "learning_rate": 0.0001, "loss": 0.0317, "step": 9000 }, { "epoch": 68.77862595419847, "grad_norm": 1.9886093139648438, "learning_rate": 0.0001, "loss": 0.0316, "step": 9010 }, { "epoch": 68.85496183206106, "grad_norm": 1.9599641561508179, "learning_rate": 0.0001, "loss": 0.0314, "step": 9020 }, { "epoch": 68.93129770992367, "grad_norm": 1.9493275880813599, "learning_rate": 0.0001, "loss": 0.033, "step": 9030 }, { "epoch": 69.00763358778626, "grad_norm": 1.77168869972229, "learning_rate": 0.0001, "loss": 0.031, "step": 9040 }, { "epoch": 69.08396946564885, "grad_norm": 2.1670947074890137, "learning_rate": 0.0001, "loss": 0.0312, "step": 9050 }, { "epoch": 69.16030534351145, "grad_norm": 1.5380451679229736, "learning_rate": 0.0001, "loss": 0.03, "step": 9060 }, { "epoch": 69.23664122137404, "grad_norm": 1.7930448055267334, "learning_rate": 0.0001, "loss": 0.0307, "step": 9070 }, { "epoch": 69.31297709923665, "grad_norm": 1.750120759010315, "learning_rate": 0.0001, "loss": 0.0302, "step": 9080 }, { "epoch": 69.38931297709924, "grad_norm": 2.3020830154418945, "learning_rate": 0.0001, "loss": 0.0308, "step": 9090 }, { "epoch": 69.46564885496183, "grad_norm": 2.584292411804199, "learning_rate": 0.0001, "loss": 0.0327, "step": 9100 }, { "epoch": 69.54198473282443, "grad_norm": 2.291311740875244, "learning_rate": 0.0001, "loss": 0.0316, "step": 9110 }, { "epoch": 69.61832061068702, "grad_norm": 2.032383441925049, "learning_rate": 0.0001, "loss": 0.0301, "step": 9120 }, { "epoch": 69.69465648854961, "grad_norm": 2.9219186305999756, "learning_rate": 0.0001, "loss": 0.0308, "step": 9130 }, { "epoch": 69.77099236641222, "grad_norm": 2.155080556869507, "learning_rate": 0.0001, "loss": 0.0313, "step": 9140 }, { "epoch": 69.8473282442748, "grad_norm": 2.1043732166290283, "learning_rate": 0.0001, "loss": 0.0312, "step": 9150 }, { "epoch": 69.92366412213741, "grad_norm": 1.7804378271102905, "learning_rate": 0.0001, "loss": 0.031, "step": 9160 }, { "epoch": 70.0, "grad_norm": 2.004148483276367, "learning_rate": 0.0001, "loss": 0.0317, "step": 9170 }, { "epoch": 70.07633587786259, "grad_norm": 2.1229279041290283, "learning_rate": 0.0001, "loss": 0.0347, "step": 9180 }, { "epoch": 70.1526717557252, "grad_norm": 1.818219542503357, "learning_rate": 0.0001, "loss": 0.034, "step": 9190 }, { "epoch": 70.22900763358778, "grad_norm": 2.327953577041626, "learning_rate": 0.0001, "loss": 0.0335, "step": 9200 }, { "epoch": 70.30534351145039, "grad_norm": 2.3311235904693604, "learning_rate": 0.0001, "loss": 0.0343, "step": 9210 }, { "epoch": 70.38167938931298, "grad_norm": 2.3924829959869385, "learning_rate": 0.0001, "loss": 0.0334, "step": 9220 }, { "epoch": 70.45801526717557, "grad_norm": 1.6162126064300537, "learning_rate": 0.0001, "loss": 0.0333, "step": 9230 }, { "epoch": 70.53435114503817, "grad_norm": 2.3845489025115967, "learning_rate": 0.0001, "loss": 0.0314, "step": 9240 }, { "epoch": 70.61068702290076, "grad_norm": 2.157707691192627, "learning_rate": 0.0001, "loss": 0.0331, "step": 9250 }, { "epoch": 70.68702290076335, "grad_norm": 2.0515224933624268, "learning_rate": 0.0001, "loss": 0.0326, "step": 9260 }, { "epoch": 70.76335877862596, "grad_norm": 1.9283429384231567, "learning_rate": 0.0001, "loss": 0.0309, "step": 9270 }, { "epoch": 70.83969465648855, "grad_norm": 2.5248777866363525, "learning_rate": 0.0001, "loss": 0.0305, "step": 9280 }, { "epoch": 70.91603053435115, "grad_norm": 2.0418145656585693, "learning_rate": 0.0001, "loss": 0.0301, "step": 9290 }, { "epoch": 70.99236641221374, "grad_norm": 1.9460296630859375, "learning_rate": 0.0001, "loss": 0.0298, "step": 9300 }, { "epoch": 71.06870229007633, "grad_norm": 2.006429672241211, "learning_rate": 0.0001, "loss": 0.0301, "step": 9310 }, { "epoch": 71.14503816793894, "grad_norm": 1.9456238746643066, "learning_rate": 0.0001, "loss": 0.0314, "step": 9320 }, { "epoch": 71.22137404580153, "grad_norm": 1.887656569480896, "learning_rate": 0.0001, "loss": 0.0303, "step": 9330 }, { "epoch": 71.29770992366412, "grad_norm": 1.971016764640808, "learning_rate": 0.0001, "loss": 0.0304, "step": 9340 }, { "epoch": 71.37404580152672, "grad_norm": 1.7095963954925537, "learning_rate": 0.0001, "loss": 0.0294, "step": 9350 }, { "epoch": 71.45038167938931, "grad_norm": 2.819554328918457, "learning_rate": 0.0001, "loss": 0.0309, "step": 9360 }, { "epoch": 71.52671755725191, "grad_norm": 3.2100112438201904, "learning_rate": 0.0001, "loss": 0.0316, "step": 9370 }, { "epoch": 71.6030534351145, "grad_norm": 2.3000850677490234, "learning_rate": 0.0001, "loss": 0.0316, "step": 9380 }, { "epoch": 71.6793893129771, "grad_norm": 2.5212085247039795, "learning_rate": 0.0001, "loss": 0.0302, "step": 9390 }, { "epoch": 71.7557251908397, "grad_norm": 2.1088664531707764, "learning_rate": 0.0001, "loss": 0.0297, "step": 9400 }, { "epoch": 71.83206106870229, "grad_norm": 2.43825626373291, "learning_rate": 0.0001, "loss": 0.0304, "step": 9410 }, { "epoch": 71.90839694656489, "grad_norm": 2.6283228397369385, "learning_rate": 0.0001, "loss": 0.0306, "step": 9420 }, { "epoch": 71.98473282442748, "grad_norm": 1.5210907459259033, "learning_rate": 0.0001, "loss": 0.0311, "step": 9430 }, { "epoch": 72.06106870229007, "grad_norm": 1.4175899028778076, "learning_rate": 0.0001, "loss": 0.0297, "step": 9440 }, { "epoch": 72.13740458015268, "grad_norm": 2.1701509952545166, "learning_rate": 0.0001, "loss": 0.0297, "step": 9450 }, { "epoch": 72.21374045801527, "grad_norm": 1.7659883499145508, "learning_rate": 0.0001, "loss": 0.0299, "step": 9460 }, { "epoch": 72.29007633587786, "grad_norm": 1.739682674407959, "learning_rate": 0.0001, "loss": 0.0301, "step": 9470 }, { "epoch": 72.36641221374046, "grad_norm": 1.4671120643615723, "learning_rate": 0.0001, "loss": 0.03, "step": 9480 }, { "epoch": 72.44274809160305, "grad_norm": 1.7680164575576782, "learning_rate": 0.0001, "loss": 0.0293, "step": 9490 }, { "epoch": 72.51908396946565, "grad_norm": 2.1287331581115723, "learning_rate": 0.0001, "loss": 0.0305, "step": 9500 }, { "epoch": 72.59541984732824, "grad_norm": 1.9648665189743042, "learning_rate": 0.0001, "loss": 0.0305, "step": 9510 }, { "epoch": 72.67175572519083, "grad_norm": 2.154679298400879, "learning_rate": 0.0001, "loss": 0.0311, "step": 9520 }, { "epoch": 72.74809160305344, "grad_norm": 2.0950827598571777, "learning_rate": 0.0001, "loss": 0.0311, "step": 9530 }, { "epoch": 72.82442748091603, "grad_norm": 2.2732393741607666, "learning_rate": 0.0001, "loss": 0.0323, "step": 9540 }, { "epoch": 72.90076335877862, "grad_norm": 2.792245626449585, "learning_rate": 0.0001, "loss": 0.0311, "step": 9550 }, { "epoch": 72.97709923664122, "grad_norm": 2.7146973609924316, "learning_rate": 0.0001, "loss": 0.0303, "step": 9560 }, { "epoch": 73.05343511450381, "grad_norm": 2.110018253326416, "learning_rate": 0.0001, "loss": 0.0307, "step": 9570 }, { "epoch": 73.12977099236642, "grad_norm": 1.7969670295715332, "learning_rate": 0.0001, "loss": 0.0309, "step": 9580 }, { "epoch": 73.20610687022901, "grad_norm": 2.1935744285583496, "learning_rate": 0.0001, "loss": 0.0302, "step": 9590 }, { "epoch": 73.2824427480916, "grad_norm": 1.5944437980651855, "learning_rate": 0.0001, "loss": 0.031, "step": 9600 }, { "epoch": 73.3587786259542, "grad_norm": 1.3954401016235352, "learning_rate": 0.0001, "loss": 0.0305, "step": 9610 }, { "epoch": 73.43511450381679, "grad_norm": 1.8972209692001343, "learning_rate": 0.0001, "loss": 0.0296, "step": 9620 }, { "epoch": 73.5114503816794, "grad_norm": 2.494072198867798, "learning_rate": 0.0001, "loss": 0.0297, "step": 9630 }, { "epoch": 73.58778625954199, "grad_norm": 1.6840403079986572, "learning_rate": 0.0001, "loss": 0.0308, "step": 9640 }, { "epoch": 73.66412213740458, "grad_norm": 2.718947410583496, "learning_rate": 0.0001, "loss": 0.03, "step": 9650 }, { "epoch": 73.74045801526718, "grad_norm": 1.9064067602157593, "learning_rate": 0.0001, "loss": 0.0302, "step": 9660 }, { "epoch": 73.81679389312977, "grad_norm": 2.1682868003845215, "learning_rate": 0.0001, "loss": 0.0295, "step": 9670 }, { "epoch": 73.89312977099236, "grad_norm": 1.914857268333435, "learning_rate": 0.0001, "loss": 0.0286, "step": 9680 }, { "epoch": 73.96946564885496, "grad_norm": 2.2185134887695312, "learning_rate": 0.0001, "loss": 0.0305, "step": 9690 }, { "epoch": 74.04580152671755, "grad_norm": 2.0108611583709717, "learning_rate": 0.0001, "loss": 0.0308, "step": 9700 }, { "epoch": 74.12213740458016, "grad_norm": 2.2451353073120117, "learning_rate": 0.0001, "loss": 0.0295, "step": 9710 }, { "epoch": 74.19847328244275, "grad_norm": 2.0373098850250244, "learning_rate": 0.0001, "loss": 0.0301, "step": 9720 }, { "epoch": 74.27480916030534, "grad_norm": 2.31498646736145, "learning_rate": 0.0001, "loss": 0.0299, "step": 9730 }, { "epoch": 74.35114503816794, "grad_norm": 1.8066850900650024, "learning_rate": 0.0001, "loss": 0.0284, "step": 9740 }, { "epoch": 74.42748091603053, "grad_norm": 2.9405715465545654, "learning_rate": 0.0001, "loss": 0.0308, "step": 9750 }, { "epoch": 74.50381679389314, "grad_norm": 2.295368194580078, "learning_rate": 0.0001, "loss": 0.0296, "step": 9760 }, { "epoch": 74.58015267175573, "grad_norm": 1.7089531421661377, "learning_rate": 0.0001, "loss": 0.0291, "step": 9770 }, { "epoch": 74.65648854961832, "grad_norm": 1.9439520835876465, "learning_rate": 0.0001, "loss": 0.0301, "step": 9780 }, { "epoch": 74.73282442748092, "grad_norm": 1.7480524778366089, "learning_rate": 0.0001, "loss": 0.0306, "step": 9790 }, { "epoch": 74.80916030534351, "grad_norm": 2.2082107067108154, "learning_rate": 0.0001, "loss": 0.0296, "step": 9800 }, { "epoch": 74.8854961832061, "grad_norm": 2.408975124359131, "learning_rate": 0.0001, "loss": 0.0297, "step": 9810 }, { "epoch": 74.9618320610687, "grad_norm": 1.8659499883651733, "learning_rate": 0.0001, "loss": 0.03, "step": 9820 }, { "epoch": 75.0381679389313, "grad_norm": 2.2608163356781006, "learning_rate": 0.0001, "loss": 0.0302, "step": 9830 }, { "epoch": 75.1145038167939, "grad_norm": 1.9717062711715698, "learning_rate": 0.0001, "loss": 0.0287, "step": 9840 }, { "epoch": 75.19083969465649, "grad_norm": 1.727089285850525, "learning_rate": 0.0001, "loss": 0.0304, "step": 9850 }, { "epoch": 75.26717557251908, "grad_norm": 1.6234052181243896, "learning_rate": 0.0001, "loss": 0.0289, "step": 9860 }, { "epoch": 75.34351145038168, "grad_norm": 1.8231574296951294, "learning_rate": 0.0001, "loss": 0.0295, "step": 9870 }, { "epoch": 75.41984732824427, "grad_norm": 2.127321243286133, "learning_rate": 0.0001, "loss": 0.0313, "step": 9880 }, { "epoch": 75.49618320610686, "grad_norm": 1.8653051853179932, "learning_rate": 0.0001, "loss": 0.0301, "step": 9890 }, { "epoch": 75.57251908396947, "grad_norm": 2.1792759895324707, "learning_rate": 0.0001, "loss": 0.0322, "step": 9900 }, { "epoch": 75.64885496183206, "grad_norm": 2.0072824954986572, "learning_rate": 0.0001, "loss": 0.031, "step": 9910 }, { "epoch": 75.72519083969466, "grad_norm": 2.4691367149353027, "learning_rate": 0.0001, "loss": 0.0306, "step": 9920 }, { "epoch": 75.80152671755725, "grad_norm": 2.0339834690093994, "learning_rate": 0.0001, "loss": 0.0296, "step": 9930 }, { "epoch": 75.87786259541984, "grad_norm": 2.1568753719329834, "learning_rate": 0.0001, "loss": 0.0306, "step": 9940 }, { "epoch": 75.95419847328245, "grad_norm": 2.5423388481140137, "learning_rate": 0.0001, "loss": 0.0319, "step": 9950 }, { "epoch": 76.03053435114504, "grad_norm": 1.82675302028656, "learning_rate": 0.0001, "loss": 0.031, "step": 9960 }, { "epoch": 76.10687022900764, "grad_norm": 2.37101149559021, "learning_rate": 0.0001, "loss": 0.0312, "step": 9970 }, { "epoch": 76.18320610687023, "grad_norm": 2.4987857341766357, "learning_rate": 0.0001, "loss": 0.0313, "step": 9980 }, { "epoch": 76.25954198473282, "grad_norm": 2.3724536895751953, "learning_rate": 0.0001, "loss": 0.0301, "step": 9990 }, { "epoch": 76.33587786259542, "grad_norm": 2.5122110843658447, "learning_rate": 0.0001, "loss": 0.0314, "step": 10000 }, { "epoch": 76.41221374045801, "grad_norm": 1.9837688207626343, "learning_rate": 0.0001, "loss": 0.0308, "step": 10010 }, { "epoch": 76.4885496183206, "grad_norm": 1.9591697454452515, "learning_rate": 0.0001, "loss": 0.0295, "step": 10020 }, { "epoch": 76.56488549618321, "grad_norm": 2.0250747203826904, "learning_rate": 0.0001, "loss": 0.029, "step": 10030 }, { "epoch": 76.6412213740458, "grad_norm": 2.62152361869812, "learning_rate": 0.0001, "loss": 0.0298, "step": 10040 }, { "epoch": 76.7175572519084, "grad_norm": 2.1159608364105225, "learning_rate": 0.0001, "loss": 0.0298, "step": 10050 }, { "epoch": 76.79389312977099, "grad_norm": 1.9883360862731934, "learning_rate": 0.0001, "loss": 0.0309, "step": 10060 }, { "epoch": 76.87022900763358, "grad_norm": 2.064776659011841, "learning_rate": 0.0001, "loss": 0.0305, "step": 10070 }, { "epoch": 76.94656488549619, "grad_norm": 1.912115216255188, "learning_rate": 0.0001, "loss": 0.0302, "step": 10080 }, { "epoch": 77.02290076335878, "grad_norm": 2.631666898727417, "learning_rate": 0.0001, "loss": 0.0295, "step": 10090 }, { "epoch": 77.09923664122137, "grad_norm": 2.4879024028778076, "learning_rate": 0.0001, "loss": 0.0288, "step": 10100 }, { "epoch": 77.17557251908397, "grad_norm": 2.1066629886627197, "learning_rate": 0.0001, "loss": 0.0279, "step": 10110 }, { "epoch": 77.25190839694656, "grad_norm": 1.953343391418457, "learning_rate": 0.0001, "loss": 0.029, "step": 10120 }, { "epoch": 77.32824427480917, "grad_norm": 2.2634828090667725, "learning_rate": 0.0001, "loss": 0.03, "step": 10130 }, { "epoch": 77.40458015267176, "grad_norm": 1.7603226900100708, "learning_rate": 0.0001, "loss": 0.0296, "step": 10140 }, { "epoch": 77.48091603053435, "grad_norm": 2.256755828857422, "learning_rate": 0.0001, "loss": 0.0293, "step": 10150 }, { "epoch": 77.55725190839695, "grad_norm": 1.9181137084960938, "learning_rate": 0.0001, "loss": 0.0282, "step": 10160 }, { "epoch": 77.63358778625954, "grad_norm": 1.868276834487915, "learning_rate": 0.0001, "loss": 0.029, "step": 10170 }, { "epoch": 77.70992366412214, "grad_norm": 1.9951680898666382, "learning_rate": 0.0001, "loss": 0.0288, "step": 10180 }, { "epoch": 77.78625954198473, "grad_norm": 1.8432846069335938, "learning_rate": 0.0001, "loss": 0.0278, "step": 10190 }, { "epoch": 77.86259541984732, "grad_norm": 1.9362231492996216, "learning_rate": 0.0001, "loss": 0.0278, "step": 10200 }, { "epoch": 77.93893129770993, "grad_norm": 1.6608227491378784, "learning_rate": 0.0001, "loss": 0.0282, "step": 10210 }, { "epoch": 78.01526717557252, "grad_norm": 2.2444908618927, "learning_rate": 0.0001, "loss": 0.0279, "step": 10220 }, { "epoch": 78.09160305343511, "grad_norm": 1.6329540014266968, "learning_rate": 0.0001, "loss": 0.0274, "step": 10230 }, { "epoch": 78.16793893129771, "grad_norm": 2.2004575729370117, "learning_rate": 0.0001, "loss": 0.0276, "step": 10240 }, { "epoch": 78.2442748091603, "grad_norm": 2.98291277885437, "learning_rate": 0.0001, "loss": 0.0292, "step": 10250 }, { "epoch": 78.3206106870229, "grad_norm": 2.1479852199554443, "learning_rate": 0.0001, "loss": 0.0294, "step": 10260 }, { "epoch": 78.3969465648855, "grad_norm": 2.0218281745910645, "learning_rate": 0.0001, "loss": 0.03, "step": 10270 }, { "epoch": 78.47328244274809, "grad_norm": 2.0934836864471436, "learning_rate": 0.0001, "loss": 0.0309, "step": 10280 }, { "epoch": 78.54961832061069, "grad_norm": 1.9793630838394165, "learning_rate": 0.0001, "loss": 0.0289, "step": 10290 }, { "epoch": 78.62595419847328, "grad_norm": 2.422004222869873, "learning_rate": 0.0001, "loss": 0.0298, "step": 10300 }, { "epoch": 78.70229007633588, "grad_norm": 1.9870843887329102, "learning_rate": 0.0001, "loss": 0.0296, "step": 10310 }, { "epoch": 78.77862595419847, "grad_norm": 2.4192922115325928, "learning_rate": 0.0001, "loss": 0.0291, "step": 10320 }, { "epoch": 78.85496183206106, "grad_norm": 2.2679059505462646, "learning_rate": 0.0001, "loss": 0.0306, "step": 10330 }, { "epoch": 78.93129770992367, "grad_norm": 1.687603235244751, "learning_rate": 0.0001, "loss": 0.0295, "step": 10340 }, { "epoch": 79.00763358778626, "grad_norm": 1.649583339691162, "learning_rate": 0.0001, "loss": 0.0292, "step": 10350 }, { "epoch": 79.08396946564885, "grad_norm": 1.6411957740783691, "learning_rate": 0.0001, "loss": 0.029, "step": 10360 }, { "epoch": 79.16030534351145, "grad_norm": 2.725250244140625, "learning_rate": 0.0001, "loss": 0.0303, "step": 10370 }, { "epoch": 79.23664122137404, "grad_norm": 2.185713052749634, "learning_rate": 0.0001, "loss": 0.0311, "step": 10380 }, { "epoch": 79.31297709923665, "grad_norm": 2.728951930999756, "learning_rate": 0.0001, "loss": 0.0304, "step": 10390 }, { "epoch": 79.38931297709924, "grad_norm": 1.8349037170410156, "learning_rate": 0.0001, "loss": 0.0296, "step": 10400 }, { "epoch": 79.46564885496183, "grad_norm": 2.7023067474365234, "learning_rate": 0.0001, "loss": 0.0293, "step": 10410 }, { "epoch": 79.54198473282443, "grad_norm": 2.663361072540283, "learning_rate": 0.0001, "loss": 0.0298, "step": 10420 }, { "epoch": 79.61832061068702, "grad_norm": 1.755434513092041, "learning_rate": 0.0001, "loss": 0.0296, "step": 10430 }, { "epoch": 79.69465648854961, "grad_norm": 1.8546017408370972, "learning_rate": 0.0001, "loss": 0.029, "step": 10440 }, { "epoch": 79.77099236641222, "grad_norm": 2.701925277709961, "learning_rate": 0.0001, "loss": 0.0298, "step": 10450 }, { "epoch": 79.8473282442748, "grad_norm": 2.125941753387451, "learning_rate": 0.0001, "loss": 0.0288, "step": 10460 }, { "epoch": 79.92366412213741, "grad_norm": 1.7561113834381104, "learning_rate": 0.0001, "loss": 0.0285, "step": 10470 }, { "epoch": 80.0, "grad_norm": 1.498563289642334, "learning_rate": 0.0001, "loss": 0.0291, "step": 10480 }, { "epoch": 80.07633587786259, "grad_norm": 2.284315347671509, "learning_rate": 0.0001, "loss": 0.0294, "step": 10490 }, { "epoch": 80.1526717557252, "grad_norm": 2.9022216796875, "learning_rate": 0.0001, "loss": 0.0287, "step": 10500 }, { "epoch": 80.22900763358778, "grad_norm": 2.199303150177002, "learning_rate": 0.0001, "loss": 0.0281, "step": 10510 }, { "epoch": 80.30534351145039, "grad_norm": 2.8844375610351562, "learning_rate": 0.0001, "loss": 0.0286, "step": 10520 }, { "epoch": 80.38167938931298, "grad_norm": 1.9314541816711426, "learning_rate": 0.0001, "loss": 0.0302, "step": 10530 }, { "epoch": 80.45801526717557, "grad_norm": 2.1909773349761963, "learning_rate": 0.0001, "loss": 0.0297, "step": 10540 }, { "epoch": 80.53435114503817, "grad_norm": 1.8728426694869995, "learning_rate": 0.0001, "loss": 0.0281, "step": 10550 }, { "epoch": 80.61068702290076, "grad_norm": 1.959956169128418, "learning_rate": 0.0001, "loss": 0.0278, "step": 10560 }, { "epoch": 80.68702290076335, "grad_norm": 2.021017551422119, "learning_rate": 0.0001, "loss": 0.0287, "step": 10570 }, { "epoch": 80.76335877862596, "grad_norm": 2.051701068878174, "learning_rate": 0.0001, "loss": 0.0284, "step": 10580 }, { "epoch": 80.83969465648855, "grad_norm": 2.127924919128418, "learning_rate": 0.0001, "loss": 0.0298, "step": 10590 }, { "epoch": 80.91603053435115, "grad_norm": 2.3978803157806396, "learning_rate": 0.0001, "loss": 0.029, "step": 10600 }, { "epoch": 80.99236641221374, "grad_norm": 2.2183926105499268, "learning_rate": 0.0001, "loss": 0.0291, "step": 10610 }, { "epoch": 81.06870229007633, "grad_norm": 1.9786779880523682, "learning_rate": 0.0001, "loss": 0.029, "step": 10620 }, { "epoch": 81.14503816793894, "grad_norm": 1.8764652013778687, "learning_rate": 0.0001, "loss": 0.0269, "step": 10630 }, { "epoch": 81.22137404580153, "grad_norm": 2.0697710514068604, "learning_rate": 0.0001, "loss": 0.0282, "step": 10640 }, { "epoch": 81.29770992366412, "grad_norm": 2.631648302078247, "learning_rate": 0.0001, "loss": 0.0277, "step": 10650 }, { "epoch": 81.37404580152672, "grad_norm": 1.9131652116775513, "learning_rate": 0.0001, "loss": 0.028, "step": 10660 }, { "epoch": 81.45038167938931, "grad_norm": 2.6014490127563477, "learning_rate": 0.0001, "loss": 0.0269, "step": 10670 }, { "epoch": 81.52671755725191, "grad_norm": 2.44943904876709, "learning_rate": 0.0001, "loss": 0.0274, "step": 10680 }, { "epoch": 81.6030534351145, "grad_norm": 1.901868462562561, "learning_rate": 0.0001, "loss": 0.0273, "step": 10690 }, { "epoch": 81.6793893129771, "grad_norm": 2.3217716217041016, "learning_rate": 0.0001, "loss": 0.0276, "step": 10700 }, { "epoch": 81.7557251908397, "grad_norm": 2.577629566192627, "learning_rate": 0.0001, "loss": 0.0281, "step": 10710 }, { "epoch": 81.83206106870229, "grad_norm": 2.5501110553741455, "learning_rate": 0.0001, "loss": 0.0277, "step": 10720 }, { "epoch": 81.90839694656489, "grad_norm": 2.4765372276306152, "learning_rate": 0.0001, "loss": 0.0285, "step": 10730 }, { "epoch": 81.98473282442748, "grad_norm": 1.8906701803207397, "learning_rate": 0.0001, "loss": 0.0285, "step": 10740 }, { "epoch": 82.06106870229007, "grad_norm": 2.5954701900482178, "learning_rate": 0.0001, "loss": 0.0279, "step": 10750 }, { "epoch": 82.13740458015268, "grad_norm": 2.4093728065490723, "learning_rate": 0.0001, "loss": 0.0278, "step": 10760 }, { "epoch": 82.21374045801527, "grad_norm": 2.1400680541992188, "learning_rate": 0.0001, "loss": 0.0267, "step": 10770 }, { "epoch": 82.29007633587786, "grad_norm": 2.162142515182495, "learning_rate": 0.0001, "loss": 0.0281, "step": 10780 }, { "epoch": 82.36641221374046, "grad_norm": 2.4292562007904053, "learning_rate": 0.0001, "loss": 0.0275, "step": 10790 }, { "epoch": 82.44274809160305, "grad_norm": 2.3338546752929688, "learning_rate": 0.0001, "loss": 0.0285, "step": 10800 }, { "epoch": 82.51908396946565, "grad_norm": 2.645214080810547, "learning_rate": 0.0001, "loss": 0.028, "step": 10810 }, { "epoch": 82.59541984732824, "grad_norm": 1.6632883548736572, "learning_rate": 0.0001, "loss": 0.0277, "step": 10820 }, { "epoch": 82.67175572519083, "grad_norm": 2.142306089401245, "learning_rate": 0.0001, "loss": 0.0277, "step": 10830 }, { "epoch": 82.74809160305344, "grad_norm": 2.3990318775177, "learning_rate": 0.0001, "loss": 0.0272, "step": 10840 }, { "epoch": 82.82442748091603, "grad_norm": 1.8101366758346558, "learning_rate": 0.0001, "loss": 0.027, "step": 10850 }, { "epoch": 82.90076335877862, "grad_norm": 1.89290452003479, "learning_rate": 0.0001, "loss": 0.0273, "step": 10860 }, { "epoch": 82.97709923664122, "grad_norm": 2.1134817600250244, "learning_rate": 0.0001, "loss": 0.0291, "step": 10870 }, { "epoch": 83.05343511450381, "grad_norm": 1.7673486471176147, "learning_rate": 0.0001, "loss": 0.0283, "step": 10880 }, { "epoch": 83.12977099236642, "grad_norm": 2.1532113552093506, "learning_rate": 0.0001, "loss": 0.0274, "step": 10890 }, { "epoch": 83.20610687022901, "grad_norm": 2.053982973098755, "learning_rate": 0.0001, "loss": 0.0289, "step": 10900 }, { "epoch": 83.2824427480916, "grad_norm": 1.616274118423462, "learning_rate": 0.0001, "loss": 0.0295, "step": 10910 }, { "epoch": 83.3587786259542, "grad_norm": 2.22554612159729, "learning_rate": 0.0001, "loss": 0.0286, "step": 10920 }, { "epoch": 83.43511450381679, "grad_norm": 1.7250736951828003, "learning_rate": 0.0001, "loss": 0.0281, "step": 10930 }, { "epoch": 83.5114503816794, "grad_norm": 2.167389154434204, "learning_rate": 0.0001, "loss": 0.0289, "step": 10940 }, { "epoch": 83.58778625954199, "grad_norm": 2.2665109634399414, "learning_rate": 0.0001, "loss": 0.0289, "step": 10950 }, { "epoch": 83.66412213740458, "grad_norm": 1.9995191097259521, "learning_rate": 0.0001, "loss": 0.0296, "step": 10960 }, { "epoch": 83.74045801526718, "grad_norm": 2.1778533458709717, "learning_rate": 0.0001, "loss": 0.0298, "step": 10970 }, { "epoch": 83.81679389312977, "grad_norm": 2.5118041038513184, "learning_rate": 0.0001, "loss": 0.0301, "step": 10980 }, { "epoch": 83.89312977099236, "grad_norm": 2.2061352729797363, "learning_rate": 0.0001, "loss": 0.0293, "step": 10990 }, { "epoch": 83.96946564885496, "grad_norm": 2.351837635040283, "learning_rate": 0.0001, "loss": 0.0292, "step": 11000 }, { "epoch": 84.04580152671755, "grad_norm": 1.914156198501587, "learning_rate": 0.0001, "loss": 0.0278, "step": 11010 }, { "epoch": 84.12213740458016, "grad_norm": 2.2052597999572754, "learning_rate": 0.0001, "loss": 0.0283, "step": 11020 }, { "epoch": 84.19847328244275, "grad_norm": 2.0015766620635986, "learning_rate": 0.0001, "loss": 0.0276, "step": 11030 }, { "epoch": 84.27480916030534, "grad_norm": 2.4087297916412354, "learning_rate": 0.0001, "loss": 0.0276, "step": 11040 }, { "epoch": 84.35114503816794, "grad_norm": 2.5203473567962646, "learning_rate": 0.0001, "loss": 0.0271, "step": 11050 }, { "epoch": 84.42748091603053, "grad_norm": 1.9411259889602661, "learning_rate": 0.0001, "loss": 0.0269, "step": 11060 }, { "epoch": 84.50381679389314, "grad_norm": 1.8183445930480957, "learning_rate": 0.0001, "loss": 0.0275, "step": 11070 }, { "epoch": 84.58015267175573, "grad_norm": 2.243483304977417, "learning_rate": 0.0001, "loss": 0.0276, "step": 11080 }, { "epoch": 84.65648854961832, "grad_norm": 1.7455048561096191, "learning_rate": 0.0001, "loss": 0.0295, "step": 11090 }, { "epoch": 84.73282442748092, "grad_norm": 2.23636531829834, "learning_rate": 0.0001, "loss": 0.0295, "step": 11100 }, { "epoch": 84.80916030534351, "grad_norm": 2.1698169708251953, "learning_rate": 0.0001, "loss": 0.0293, "step": 11110 }, { "epoch": 84.8854961832061, "grad_norm": 2.031522035598755, "learning_rate": 0.0001, "loss": 0.0293, "step": 11120 }, { "epoch": 84.9618320610687, "grad_norm": 2.128495931625366, "learning_rate": 0.0001, "loss": 0.0278, "step": 11130 }, { "epoch": 85.0381679389313, "grad_norm": 2.255833148956299, "learning_rate": 0.0001, "loss": 0.0295, "step": 11140 }, { "epoch": 85.1145038167939, "grad_norm": 2.0805819034576416, "learning_rate": 0.0001, "loss": 0.0287, "step": 11150 }, { "epoch": 85.19083969465649, "grad_norm": 1.857322335243225, "learning_rate": 0.0001, "loss": 0.0283, "step": 11160 }, { "epoch": 85.26717557251908, "grad_norm": 1.903228998184204, "learning_rate": 0.0001, "loss": 0.0293, "step": 11170 }, { "epoch": 85.34351145038168, "grad_norm": 2.397448778152466, "learning_rate": 0.0001, "loss": 0.0294, "step": 11180 }, { "epoch": 85.41984732824427, "grad_norm": 2.708606481552124, "learning_rate": 0.0001, "loss": 0.0289, "step": 11190 }, { "epoch": 85.49618320610686, "grad_norm": 2.332487106323242, "learning_rate": 0.0001, "loss": 0.0285, "step": 11200 }, { "epoch": 85.57251908396947, "grad_norm": 2.4110116958618164, "learning_rate": 0.0001, "loss": 0.0273, "step": 11210 }, { "epoch": 85.64885496183206, "grad_norm": 2.2653985023498535, "learning_rate": 0.0001, "loss": 0.0278, "step": 11220 }, { "epoch": 85.72519083969466, "grad_norm": 3.3893303871154785, "learning_rate": 0.0001, "loss": 0.028, "step": 11230 }, { "epoch": 85.80152671755725, "grad_norm": 1.6208544969558716, "learning_rate": 0.0001, "loss": 0.0279, "step": 11240 }, { "epoch": 85.87786259541984, "grad_norm": 1.647706389427185, "learning_rate": 0.0001, "loss": 0.0266, "step": 11250 }, { "epoch": 85.95419847328245, "grad_norm": 1.7920585870742798, "learning_rate": 0.0001, "loss": 0.0272, "step": 11260 }, { "epoch": 86.03053435114504, "grad_norm": 1.7480031251907349, "learning_rate": 0.0001, "loss": 0.0269, "step": 11270 }, { "epoch": 86.10687022900764, "grad_norm": 2.2306630611419678, "learning_rate": 0.0001, "loss": 0.0262, "step": 11280 }, { "epoch": 86.18320610687023, "grad_norm": 1.5807181596755981, "learning_rate": 0.0001, "loss": 0.0272, "step": 11290 }, { "epoch": 86.25954198473282, "grad_norm": 1.8858304023742676, "learning_rate": 0.0001, "loss": 0.0272, "step": 11300 }, { "epoch": 86.33587786259542, "grad_norm": 2.4685378074645996, "learning_rate": 0.0001, "loss": 0.028, "step": 11310 }, { "epoch": 86.41221374045801, "grad_norm": 1.978068470954895, "learning_rate": 0.0001, "loss": 0.0294, "step": 11320 }, { "epoch": 86.4885496183206, "grad_norm": 1.957485556602478, "learning_rate": 0.0001, "loss": 0.0302, "step": 11330 }, { "epoch": 86.56488549618321, "grad_norm": 2.145585298538208, "learning_rate": 0.0001, "loss": 0.0293, "step": 11340 }, { "epoch": 86.6412213740458, "grad_norm": 2.376044750213623, "learning_rate": 0.0001, "loss": 0.0309, "step": 11350 }, { "epoch": 86.7175572519084, "grad_norm": 1.8277636766433716, "learning_rate": 0.0001, "loss": 0.0305, "step": 11360 }, { "epoch": 86.79389312977099, "grad_norm": 1.928977608680725, "learning_rate": 0.0001, "loss": 0.0304, "step": 11370 }, { "epoch": 86.87022900763358, "grad_norm": 1.4652042388916016, "learning_rate": 0.0001, "loss": 0.0299, "step": 11380 }, { "epoch": 86.94656488549619, "grad_norm": 2.1937546730041504, "learning_rate": 0.0001, "loss": 0.0298, "step": 11390 }, { "epoch": 87.02290076335878, "grad_norm": 1.915789246559143, "learning_rate": 0.0001, "loss": 0.0288, "step": 11400 }, { "epoch": 87.09923664122137, "grad_norm": 1.9487838745117188, "learning_rate": 0.0001, "loss": 0.0283, "step": 11410 }, { "epoch": 87.17557251908397, "grad_norm": 1.764557123184204, "learning_rate": 0.0001, "loss": 0.0287, "step": 11420 }, { "epoch": 87.25190839694656, "grad_norm": 1.7376431226730347, "learning_rate": 0.0001, "loss": 0.0284, "step": 11430 }, { "epoch": 87.32824427480917, "grad_norm": 1.7744957208633423, "learning_rate": 0.0001, "loss": 0.0286, "step": 11440 }, { "epoch": 87.40458015267176, "grad_norm": 1.8261313438415527, "learning_rate": 0.0001, "loss": 0.0281, "step": 11450 }, { "epoch": 87.48091603053435, "grad_norm": 1.7342931032180786, "learning_rate": 0.0001, "loss": 0.028, "step": 11460 }, { "epoch": 87.55725190839695, "grad_norm": 1.9022024869918823, "learning_rate": 0.0001, "loss": 0.0297, "step": 11470 }, { "epoch": 87.63358778625954, "grad_norm": 2.201618194580078, "learning_rate": 0.0001, "loss": 0.0282, "step": 11480 }, { "epoch": 87.70992366412214, "grad_norm": 2.619689464569092, "learning_rate": 0.0001, "loss": 0.0274, "step": 11490 }, { "epoch": 87.78625954198473, "grad_norm": 2.0253217220306396, "learning_rate": 0.0001, "loss": 0.0262, "step": 11500 }, { "epoch": 87.86259541984732, "grad_norm": 1.9725637435913086, "learning_rate": 0.0001, "loss": 0.0269, "step": 11510 }, { "epoch": 87.93893129770993, "grad_norm": 1.7485506534576416, "learning_rate": 0.0001, "loss": 0.0276, "step": 11520 }, { "epoch": 88.01526717557252, "grad_norm": 2.2544643878936768, "learning_rate": 0.0001, "loss": 0.0279, "step": 11530 }, { "epoch": 88.09160305343511, "grad_norm": 2.4897453784942627, "learning_rate": 0.0001, "loss": 0.0262, "step": 11540 }, { "epoch": 88.16793893129771, "grad_norm": 1.7011996507644653, "learning_rate": 0.0001, "loss": 0.026, "step": 11550 }, { "epoch": 88.2442748091603, "grad_norm": 1.7848111391067505, "learning_rate": 0.0001, "loss": 0.0267, "step": 11560 }, { "epoch": 88.3206106870229, "grad_norm": 2.521416187286377, "learning_rate": 0.0001, "loss": 0.0274, "step": 11570 }, { "epoch": 88.3969465648855, "grad_norm": 2.1718149185180664, "learning_rate": 0.0001, "loss": 0.0273, "step": 11580 }, { "epoch": 88.47328244274809, "grad_norm": 1.716087818145752, "learning_rate": 0.0001, "loss": 0.0272, "step": 11590 }, { "epoch": 88.54961832061069, "grad_norm": 2.008735418319702, "learning_rate": 0.0001, "loss": 0.0275, "step": 11600 }, { "epoch": 88.62595419847328, "grad_norm": 2.038304328918457, "learning_rate": 0.0001, "loss": 0.0291, "step": 11610 }, { "epoch": 88.70229007633588, "grad_norm": 1.8229063749313354, "learning_rate": 0.0001, "loss": 0.0278, "step": 11620 }, { "epoch": 88.77862595419847, "grad_norm": 1.6218775510787964, "learning_rate": 0.0001, "loss": 0.0286, "step": 11630 }, { "epoch": 88.85496183206106, "grad_norm": 2.2887074947357178, "learning_rate": 0.0001, "loss": 0.0288, "step": 11640 }, { "epoch": 88.93129770992367, "grad_norm": 2.2172722816467285, "learning_rate": 0.0001, "loss": 0.0272, "step": 11650 }, { "epoch": 89.00763358778626, "grad_norm": 2.0486466884613037, "learning_rate": 0.0001, "loss": 0.0282, "step": 11660 }, { "epoch": 89.08396946564885, "grad_norm": 1.58942711353302, "learning_rate": 0.0001, "loss": 0.0267, "step": 11670 }, { "epoch": 89.16030534351145, "grad_norm": 2.7565033435821533, "learning_rate": 0.0001, "loss": 0.0263, "step": 11680 }, { "epoch": 89.23664122137404, "grad_norm": 2.4500365257263184, "learning_rate": 0.0001, "loss": 0.0284, "step": 11690 }, { "epoch": 89.31297709923665, "grad_norm": 2.4496915340423584, "learning_rate": 0.0001, "loss": 0.0277, "step": 11700 }, { "epoch": 89.38931297709924, "grad_norm": 2.007631778717041, "learning_rate": 0.0001, "loss": 0.0277, "step": 11710 }, { "epoch": 89.46564885496183, "grad_norm": 1.604849934577942, "learning_rate": 0.0001, "loss": 0.0283, "step": 11720 }, { "epoch": 89.54198473282443, "grad_norm": 1.7006183862686157, "learning_rate": 0.0001, "loss": 0.0284, "step": 11730 }, { "epoch": 89.61832061068702, "grad_norm": 1.6537704467773438, "learning_rate": 0.0001, "loss": 0.0279, "step": 11740 }, { "epoch": 89.69465648854961, "grad_norm": 2.1811344623565674, "learning_rate": 0.0001, "loss": 0.0281, "step": 11750 }, { "epoch": 89.77099236641222, "grad_norm": 1.643764853477478, "learning_rate": 0.0001, "loss": 0.0274, "step": 11760 }, { "epoch": 89.8473282442748, "grad_norm": 1.8840290307998657, "learning_rate": 0.0001, "loss": 0.0283, "step": 11770 }, { "epoch": 89.92366412213741, "grad_norm": 1.9610332250595093, "learning_rate": 0.0001, "loss": 0.0284, "step": 11780 }, { "epoch": 90.0, "grad_norm": 2.1873841285705566, "learning_rate": 0.0001, "loss": 0.0272, "step": 11790 }, { "epoch": 90.07633587786259, "grad_norm": 1.6441974639892578, "learning_rate": 0.0001, "loss": 0.0272, "step": 11800 }, { "epoch": 90.1526717557252, "grad_norm": 2.0373079776763916, "learning_rate": 0.0001, "loss": 0.027, "step": 11810 }, { "epoch": 90.22900763358778, "grad_norm": 2.1042699813842773, "learning_rate": 0.0001, "loss": 0.0275, "step": 11820 }, { "epoch": 90.30534351145039, "grad_norm": 2.212646245956421, "learning_rate": 0.0001, "loss": 0.0281, "step": 11830 }, { "epoch": 90.38167938931298, "grad_norm": 1.7019261121749878, "learning_rate": 0.0001, "loss": 0.0281, "step": 11840 }, { "epoch": 90.45801526717557, "grad_norm": 2.0297627449035645, "learning_rate": 0.0001, "loss": 0.0266, "step": 11850 }, { "epoch": 90.53435114503817, "grad_norm": 1.97756028175354, "learning_rate": 0.0001, "loss": 0.0269, "step": 11860 }, { "epoch": 90.61068702290076, "grad_norm": 2.0326085090637207, "learning_rate": 0.0001, "loss": 0.0277, "step": 11870 }, { "epoch": 90.68702290076335, "grad_norm": 2.2632765769958496, "learning_rate": 0.0001, "loss": 0.0277, "step": 11880 }, { "epoch": 90.76335877862596, "grad_norm": 2.004880666732788, "learning_rate": 0.0001, "loss": 0.027, "step": 11890 }, { "epoch": 90.83969465648855, "grad_norm": 2.2464892864227295, "learning_rate": 0.0001, "loss": 0.0273, "step": 11900 }, { "epoch": 90.91603053435115, "grad_norm": 2.583890438079834, "learning_rate": 0.0001, "loss": 0.0266, "step": 11910 }, { "epoch": 90.99236641221374, "grad_norm": 2.1703271865844727, "learning_rate": 0.0001, "loss": 0.0277, "step": 11920 }, { "epoch": 91.06870229007633, "grad_norm": 2.0116517543792725, "learning_rate": 0.0001, "loss": 0.0258, "step": 11930 }, { "epoch": 91.14503816793894, "grad_norm": 2.1539690494537354, "learning_rate": 0.0001, "loss": 0.0262, "step": 11940 }, { "epoch": 91.22137404580153, "grad_norm": 1.737656831741333, "learning_rate": 0.0001, "loss": 0.0273, "step": 11950 }, { "epoch": 91.29770992366412, "grad_norm": 1.5265249013900757, "learning_rate": 0.0001, "loss": 0.0268, "step": 11960 }, { "epoch": 91.37404580152672, "grad_norm": 1.4589312076568604, "learning_rate": 0.0001, "loss": 0.0261, "step": 11970 }, { "epoch": 91.45038167938931, "grad_norm": 2.2422075271606445, "learning_rate": 0.0001, "loss": 0.0265, "step": 11980 }, { "epoch": 91.52671755725191, "grad_norm": 1.9763215780258179, "learning_rate": 0.0001, "loss": 0.0285, "step": 11990 }, { "epoch": 91.6030534351145, "grad_norm": 2.454458236694336, "learning_rate": 0.0001, "loss": 0.0293, "step": 12000 }, { "epoch": 91.6793893129771, "grad_norm": 2.5606000423431396, "learning_rate": 0.0001, "loss": 0.0286, "step": 12010 }, { "epoch": 91.7557251908397, "grad_norm": 1.7009351253509521, "learning_rate": 0.0001, "loss": 0.0287, "step": 12020 }, { "epoch": 91.83206106870229, "grad_norm": 2.326247215270996, "learning_rate": 0.0001, "loss": 0.0279, "step": 12030 }, { "epoch": 91.90839694656489, "grad_norm": 1.9448156356811523, "learning_rate": 0.0001, "loss": 0.0275, "step": 12040 }, { "epoch": 91.98473282442748, "grad_norm": 1.3165194988250732, "learning_rate": 0.0001, "loss": 0.0283, "step": 12050 }, { "epoch": 92.06106870229007, "grad_norm": 2.20365047454834, "learning_rate": 0.0001, "loss": 0.0277, "step": 12060 }, { "epoch": 92.13740458015268, "grad_norm": 1.7925690412521362, "learning_rate": 0.0001, "loss": 0.0277, "step": 12070 }, { "epoch": 92.21374045801527, "grad_norm": 2.6208276748657227, "learning_rate": 0.0001, "loss": 0.0273, "step": 12080 }, { "epoch": 92.29007633587786, "grad_norm": 2.061520576477051, "learning_rate": 0.0001, "loss": 0.0289, "step": 12090 }, { "epoch": 92.36641221374046, "grad_norm": 1.7876814603805542, "learning_rate": 0.0001, "loss": 0.0278, "step": 12100 }, { "epoch": 92.44274809160305, "grad_norm": 2.3957793712615967, "learning_rate": 0.0001, "loss": 0.0279, "step": 12110 }, { "epoch": 92.51908396946565, "grad_norm": 2.0234436988830566, "learning_rate": 0.0001, "loss": 0.0278, "step": 12120 }, { "epoch": 92.59541984732824, "grad_norm": 1.6603822708129883, "learning_rate": 0.0001, "loss": 0.0284, "step": 12130 }, { "epoch": 92.67175572519083, "grad_norm": 1.6081066131591797, "learning_rate": 0.0001, "loss": 0.028, "step": 12140 }, { "epoch": 92.74809160305344, "grad_norm": 2.1749582290649414, "learning_rate": 0.0001, "loss": 0.0282, "step": 12150 }, { "epoch": 92.82442748091603, "grad_norm": 2.040853261947632, "learning_rate": 0.0001, "loss": 0.0275, "step": 12160 }, { "epoch": 92.90076335877862, "grad_norm": 1.8177794218063354, "learning_rate": 0.0001, "loss": 0.0272, "step": 12170 }, { "epoch": 92.97709923664122, "grad_norm": 2.4440574645996094, "learning_rate": 0.0001, "loss": 0.0273, "step": 12180 }, { "epoch": 93.05343511450381, "grad_norm": 2.207021951675415, "learning_rate": 0.0001, "loss": 0.029, "step": 12190 }, { "epoch": 93.12977099236642, "grad_norm": 2.050929546356201, "learning_rate": 0.0001, "loss": 0.0301, "step": 12200 }, { "epoch": 93.20610687022901, "grad_norm": 2.3443288803100586, "learning_rate": 0.0001, "loss": 0.028, "step": 12210 }, { "epoch": 93.2824427480916, "grad_norm": 2.2728869915008545, "learning_rate": 0.0001, "loss": 0.0271, "step": 12220 }, { "epoch": 93.3587786259542, "grad_norm": 2.062159776687622, "learning_rate": 0.0001, "loss": 0.0281, "step": 12230 }, { "epoch": 93.43511450381679, "grad_norm": 2.9072484970092773, "learning_rate": 0.0001, "loss": 0.0283, "step": 12240 }, { "epoch": 93.5114503816794, "grad_norm": 2.4082822799682617, "learning_rate": 0.0001, "loss": 0.0279, "step": 12250 }, { "epoch": 93.58778625954199, "grad_norm": 1.7157539129257202, "learning_rate": 0.0001, "loss": 0.0272, "step": 12260 }, { "epoch": 93.66412213740458, "grad_norm": 2.4433212280273438, "learning_rate": 0.0001, "loss": 0.0269, "step": 12270 }, { "epoch": 93.74045801526718, "grad_norm": 2.318206548690796, "learning_rate": 0.0001, "loss": 0.0267, "step": 12280 }, { "epoch": 93.81679389312977, "grad_norm": 2.04441237449646, "learning_rate": 0.0001, "loss": 0.0265, "step": 12290 }, { "epoch": 93.89312977099236, "grad_norm": 2.0506439208984375, "learning_rate": 0.0001, "loss": 0.0268, "step": 12300 }, { "epoch": 93.96946564885496, "grad_norm": 2.1349093914031982, "learning_rate": 0.0001, "loss": 0.0285, "step": 12310 }, { "epoch": 94.04580152671755, "grad_norm": 1.792271614074707, "learning_rate": 0.0001, "loss": 0.029, "step": 12320 }, { "epoch": 94.12213740458016, "grad_norm": 1.620333194732666, "learning_rate": 0.0001, "loss": 0.0278, "step": 12330 }, { "epoch": 94.19847328244275, "grad_norm": 1.9943668842315674, "learning_rate": 0.0001, "loss": 0.0278, "step": 12340 }, { "epoch": 94.27480916030534, "grad_norm": 1.7413588762283325, "learning_rate": 0.0001, "loss": 0.0279, "step": 12350 }, { "epoch": 94.35114503816794, "grad_norm": 2.052370548248291, "learning_rate": 0.0001, "loss": 0.0274, "step": 12360 }, { "epoch": 94.42748091603053, "grad_norm": 1.8157126903533936, "learning_rate": 0.0001, "loss": 0.0276, "step": 12370 }, { "epoch": 94.50381679389314, "grad_norm": 2.1297357082366943, "learning_rate": 0.0001, "loss": 0.0277, "step": 12380 }, { "epoch": 94.58015267175573, "grad_norm": 2.706399917602539, "learning_rate": 0.0001, "loss": 0.0268, "step": 12390 }, { "epoch": 94.65648854961832, "grad_norm": 1.9994560480117798, "learning_rate": 0.0001, "loss": 0.027, "step": 12400 }, { "epoch": 94.73282442748092, "grad_norm": 2.151804208755493, "learning_rate": 0.0001, "loss": 0.0285, "step": 12410 }, { "epoch": 94.80916030534351, "grad_norm": 2.254965305328369, "learning_rate": 0.0001, "loss": 0.0275, "step": 12420 }, { "epoch": 94.8854961832061, "grad_norm": 2.2192678451538086, "learning_rate": 0.0001, "loss": 0.0276, "step": 12430 }, { "epoch": 94.9618320610687, "grad_norm": 2.004732847213745, "learning_rate": 0.0001, "loss": 0.0281, "step": 12440 }, { "epoch": 95.0381679389313, "grad_norm": 2.084909200668335, "learning_rate": 0.0001, "loss": 0.0281, "step": 12450 }, { "epoch": 95.1145038167939, "grad_norm": 1.8144062757492065, "learning_rate": 0.0001, "loss": 0.0279, "step": 12460 }, { "epoch": 95.19083969465649, "grad_norm": 2.081183910369873, "learning_rate": 0.0001, "loss": 0.0297, "step": 12470 }, { "epoch": 95.26717557251908, "grad_norm": 1.9379130601882935, "learning_rate": 0.0001, "loss": 0.0283, "step": 12480 }, { "epoch": 95.34351145038168, "grad_norm": 1.7462767362594604, "learning_rate": 0.0001, "loss": 0.0272, "step": 12490 }, { "epoch": 95.41984732824427, "grad_norm": 1.62994384765625, "learning_rate": 0.0001, "loss": 0.027, "step": 12500 }, { "epoch": 95.49618320610686, "grad_norm": 2.5998990535736084, "learning_rate": 0.0001, "loss": 0.0275, "step": 12510 }, { "epoch": 95.57251908396947, "grad_norm": 2.269757032394409, "learning_rate": 0.0001, "loss": 0.0279, "step": 12520 }, { "epoch": 95.64885496183206, "grad_norm": 1.970633625984192, "learning_rate": 0.0001, "loss": 0.0261, "step": 12530 }, { "epoch": 95.72519083969466, "grad_norm": 1.8760147094726562, "learning_rate": 0.0001, "loss": 0.029, "step": 12540 }, { "epoch": 95.80152671755725, "grad_norm": 2.068983554840088, "learning_rate": 0.0001, "loss": 0.0267, "step": 12550 }, { "epoch": 95.87786259541984, "grad_norm": 2.1720149517059326, "learning_rate": 0.0001, "loss": 0.0265, "step": 12560 }, { "epoch": 95.95419847328245, "grad_norm": 2.3412740230560303, "learning_rate": 0.0001, "loss": 0.0268, "step": 12570 }, { "epoch": 96.03053435114504, "grad_norm": 1.7871941328048706, "learning_rate": 0.0001, "loss": 0.0277, "step": 12580 }, { "epoch": 96.10687022900764, "grad_norm": 2.1433558464050293, "learning_rate": 0.0001, "loss": 0.026, "step": 12590 }, { "epoch": 96.18320610687023, "grad_norm": 2.3395795822143555, "learning_rate": 0.0001, "loss": 0.0258, "step": 12600 }, { "epoch": 96.25954198473282, "grad_norm": 2.1531822681427, "learning_rate": 0.0001, "loss": 0.0264, "step": 12610 }, { "epoch": 96.33587786259542, "grad_norm": 1.7504849433898926, "learning_rate": 0.0001, "loss": 0.0268, "step": 12620 }, { "epoch": 96.41221374045801, "grad_norm": 1.5636529922485352, "learning_rate": 0.0001, "loss": 0.0265, "step": 12630 }, { "epoch": 96.4885496183206, "grad_norm": 1.7711431980133057, "learning_rate": 0.0001, "loss": 0.0265, "step": 12640 }, { "epoch": 96.56488549618321, "grad_norm": 1.5287121534347534, "learning_rate": 0.0001, "loss": 0.0264, "step": 12650 }, { "epoch": 96.6412213740458, "grad_norm": 2.027833938598633, "learning_rate": 0.0001, "loss": 0.0271, "step": 12660 }, { "epoch": 96.7175572519084, "grad_norm": 1.6903228759765625, "learning_rate": 0.0001, "loss": 0.0278, "step": 12670 }, { "epoch": 96.79389312977099, "grad_norm": 2.381314277648926, "learning_rate": 0.0001, "loss": 0.0261, "step": 12680 }, { "epoch": 96.87022900763358, "grad_norm": 1.7204443216323853, "learning_rate": 0.0001, "loss": 0.0253, "step": 12690 }, { "epoch": 96.94656488549619, "grad_norm": 2.2273786067962646, "learning_rate": 0.0001, "loss": 0.0269, "step": 12700 }, { "epoch": 97.02290076335878, "grad_norm": 1.4928877353668213, "learning_rate": 0.0001, "loss": 0.0264, "step": 12710 }, { "epoch": 97.09923664122137, "grad_norm": 1.9681404829025269, "learning_rate": 0.0001, "loss": 0.0266, "step": 12720 }, { "epoch": 97.17557251908397, "grad_norm": 2.339693307876587, "learning_rate": 0.0001, "loss": 0.0272, "step": 12730 }, { "epoch": 97.25190839694656, "grad_norm": 1.992285132408142, "learning_rate": 0.0001, "loss": 0.0268, "step": 12740 }, { "epoch": 97.32824427480917, "grad_norm": 2.0576207637786865, "learning_rate": 0.0001, "loss": 0.027, "step": 12750 }, { "epoch": 97.40458015267176, "grad_norm": 1.8974705934524536, "learning_rate": 0.0001, "loss": 0.0267, "step": 12760 }, { "epoch": 97.48091603053435, "grad_norm": 1.713901400566101, "learning_rate": 0.0001, "loss": 0.0273, "step": 12770 }, { "epoch": 97.55725190839695, "grad_norm": 1.8364317417144775, "learning_rate": 0.0001, "loss": 0.0271, "step": 12780 }, { "epoch": 97.63358778625954, "grad_norm": 1.4989039897918701, "learning_rate": 0.0001, "loss": 0.0264, "step": 12790 }, { "epoch": 97.70992366412214, "grad_norm": 2.4042513370513916, "learning_rate": 0.0001, "loss": 0.0266, "step": 12800 }, { "epoch": 97.78625954198473, "grad_norm": 2.03918194770813, "learning_rate": 0.0001, "loss": 0.0274, "step": 12810 }, { "epoch": 97.86259541984732, "grad_norm": 1.9604179859161377, "learning_rate": 0.0001, "loss": 0.026, "step": 12820 }, { "epoch": 97.93893129770993, "grad_norm": 1.8114110231399536, "learning_rate": 0.0001, "loss": 0.0263, "step": 12830 }, { "epoch": 98.01526717557252, "grad_norm": 1.676608920097351, "learning_rate": 0.0001, "loss": 0.0265, "step": 12840 }, { "epoch": 98.09160305343511, "grad_norm": 1.8550493717193604, "learning_rate": 0.0001, "loss": 0.0263, "step": 12850 }, { "epoch": 98.16793893129771, "grad_norm": 1.7346504926681519, "learning_rate": 0.0001, "loss": 0.0271, "step": 12860 }, { "epoch": 98.2442748091603, "grad_norm": 1.7089221477508545, "learning_rate": 0.0001, "loss": 0.0268, "step": 12870 }, { "epoch": 98.3206106870229, "grad_norm": 2.4548115730285645, "learning_rate": 0.0001, "loss": 0.0262, "step": 12880 }, { "epoch": 98.3969465648855, "grad_norm": 3.040973663330078, "learning_rate": 0.0001, "loss": 0.0256, "step": 12890 }, { "epoch": 98.47328244274809, "grad_norm": 2.492417335510254, "learning_rate": 0.0001, "loss": 0.0256, "step": 12900 }, { "epoch": 98.54961832061069, "grad_norm": 3.00152325630188, "learning_rate": 0.0001, "loss": 0.0279, "step": 12910 }, { "epoch": 98.62595419847328, "grad_norm": 2.3257269859313965, "learning_rate": 0.0001, "loss": 0.0275, "step": 12920 }, { "epoch": 98.70229007633588, "grad_norm": 2.1610448360443115, "learning_rate": 0.0001, "loss": 0.0258, "step": 12930 }, { "epoch": 98.77862595419847, "grad_norm": 1.658455491065979, "learning_rate": 0.0001, "loss": 0.0255, "step": 12940 }, { "epoch": 98.85496183206106, "grad_norm": 1.5496798753738403, "learning_rate": 0.0001, "loss": 0.0264, "step": 12950 }, { "epoch": 98.93129770992367, "grad_norm": 1.3847154378890991, "learning_rate": 0.0001, "loss": 0.0272, "step": 12960 }, { "epoch": 99.00763358778626, "grad_norm": 2.322085380554199, "learning_rate": 0.0001, "loss": 0.027, "step": 12970 }, { "epoch": 99.08396946564885, "grad_norm": 1.6285853385925293, "learning_rate": 0.0001, "loss": 0.0272, "step": 12980 }, { "epoch": 99.16030534351145, "grad_norm": 2.3184287548065186, "learning_rate": 0.0001, "loss": 0.0276, "step": 12990 }, { "epoch": 99.23664122137404, "grad_norm": 2.160177230834961, "learning_rate": 0.0001, "loss": 0.0271, "step": 13000 }, { "epoch": 99.31297709923665, "grad_norm": 1.4014972448349, "learning_rate": 0.0001, "loss": 0.0266, "step": 13010 }, { "epoch": 99.38931297709924, "grad_norm": 2.194845676422119, "learning_rate": 0.0001, "loss": 0.0279, "step": 13020 }, { "epoch": 99.46564885496183, "grad_norm": 2.1724159717559814, "learning_rate": 0.0001, "loss": 0.0271, "step": 13030 }, { "epoch": 99.54198473282443, "grad_norm": 1.9012318849563599, "learning_rate": 0.0001, "loss": 0.0265, "step": 13040 }, { "epoch": 99.61832061068702, "grad_norm": 2.6711907386779785, "learning_rate": 0.0001, "loss": 0.0285, "step": 13050 }, { "epoch": 99.69465648854961, "grad_norm": 1.8770544528961182, "learning_rate": 0.0001, "loss": 0.0283, "step": 13060 }, { "epoch": 99.77099236641222, "grad_norm": 2.315079689025879, "learning_rate": 0.0001, "loss": 0.0261, "step": 13070 }, { "epoch": 99.8473282442748, "grad_norm": 3.185803174972534, "learning_rate": 0.0001, "loss": 0.0264, "step": 13080 }, { "epoch": 99.92366412213741, "grad_norm": 2.134028673171997, "learning_rate": 0.0001, "loss": 0.0267, "step": 13090 }, { "epoch": 100.0, "grad_norm": 1.8612663745880127, "learning_rate": 0.0001, "loss": 0.027, "step": 13100 }, { "epoch": 100.07633587786259, "grad_norm": 1.9833686351776123, "learning_rate": 0.0001, "loss": 0.0264, "step": 13110 }, { "epoch": 100.1526717557252, "grad_norm": 2.3031277656555176, "learning_rate": 0.0001, "loss": 0.0261, "step": 13120 }, { "epoch": 100.22900763358778, "grad_norm": 1.873538613319397, "learning_rate": 0.0001, "loss": 0.0248, "step": 13130 }, { "epoch": 100.30534351145039, "grad_norm": 2.265451669692993, "learning_rate": 0.0001, "loss": 0.0248, "step": 13140 }, { "epoch": 100.38167938931298, "grad_norm": 1.673041820526123, "learning_rate": 0.0001, "loss": 0.0257, "step": 13150 }, { "epoch": 100.45801526717557, "grad_norm": 1.935873031616211, "learning_rate": 0.0001, "loss": 0.0266, "step": 13160 }, { "epoch": 100.53435114503817, "grad_norm": 2.3503096103668213, "learning_rate": 0.0001, "loss": 0.0253, "step": 13170 }, { "epoch": 100.61068702290076, "grad_norm": 1.8084266185760498, "learning_rate": 0.0001, "loss": 0.0266, "step": 13180 }, { "epoch": 100.68702290076335, "grad_norm": 1.4114789962768555, "learning_rate": 0.0001, "loss": 0.0273, "step": 13190 }, { "epoch": 100.76335877862596, "grad_norm": 2.127671718597412, "learning_rate": 0.0001, "loss": 0.0277, "step": 13200 }, { "epoch": 100.83969465648855, "grad_norm": 2.2973878383636475, "learning_rate": 0.0001, "loss": 0.028, "step": 13210 }, { "epoch": 100.91603053435115, "grad_norm": 2.0174946784973145, "learning_rate": 0.0001, "loss": 0.0283, "step": 13220 }, { "epoch": 100.99236641221374, "grad_norm": 1.7465158700942993, "learning_rate": 0.0001, "loss": 0.0291, "step": 13230 }, { "epoch": 101.06870229007633, "grad_norm": 1.7766789197921753, "learning_rate": 0.0001, "loss": 0.0288, "step": 13240 }, { "epoch": 101.14503816793894, "grad_norm": 1.575044870376587, "learning_rate": 0.0001, "loss": 0.0265, "step": 13250 }, { "epoch": 101.22137404580153, "grad_norm": 1.9099048376083374, "learning_rate": 0.0001, "loss": 0.0276, "step": 13260 }, { "epoch": 101.29770992366412, "grad_norm": 2.3319928646087646, "learning_rate": 0.0001, "loss": 0.0281, "step": 13270 }, { "epoch": 101.37404580152672, "grad_norm": 2.4553635120391846, "learning_rate": 0.0001, "loss": 0.027, "step": 13280 }, { "epoch": 101.45038167938931, "grad_norm": 2.060188055038452, "learning_rate": 0.0001, "loss": 0.0274, "step": 13290 }, { "epoch": 101.52671755725191, "grad_norm": 2.1323132514953613, "learning_rate": 0.0001, "loss": 0.0286, "step": 13300 }, { "epoch": 101.6030534351145, "grad_norm": 1.6371467113494873, "learning_rate": 0.0001, "loss": 0.0275, "step": 13310 }, { "epoch": 101.6793893129771, "grad_norm": 2.0572807788848877, "learning_rate": 0.0001, "loss": 0.027, "step": 13320 }, { "epoch": 101.7557251908397, "grad_norm": 1.9916040897369385, "learning_rate": 0.0001, "loss": 0.0276, "step": 13330 }, { "epoch": 101.83206106870229, "grad_norm": 1.5224597454071045, "learning_rate": 0.0001, "loss": 0.0256, "step": 13340 }, { "epoch": 101.90839694656489, "grad_norm": 1.8677860498428345, "learning_rate": 0.0001, "loss": 0.026, "step": 13350 }, { "epoch": 101.98473282442748, "grad_norm": 1.8820358514785767, "learning_rate": 0.0001, "loss": 0.0266, "step": 13360 }, { "epoch": 102.06106870229007, "grad_norm": 2.739431381225586, "learning_rate": 0.0001, "loss": 0.0269, "step": 13370 }, { "epoch": 102.13740458015268, "grad_norm": 1.828119158744812, "learning_rate": 0.0001, "loss": 0.026, "step": 13380 }, { "epoch": 102.21374045801527, "grad_norm": 2.042367696762085, "learning_rate": 0.0001, "loss": 0.0274, "step": 13390 }, { "epoch": 102.29007633587786, "grad_norm": 2.0280649662017822, "learning_rate": 0.0001, "loss": 0.0268, "step": 13400 }, { "epoch": 102.36641221374046, "grad_norm": 1.9109655618667603, "learning_rate": 0.0001, "loss": 0.028, "step": 13410 }, { "epoch": 102.44274809160305, "grad_norm": 1.9902852773666382, "learning_rate": 0.0001, "loss": 0.0273, "step": 13420 }, { "epoch": 102.51908396946565, "grad_norm": 2.0384387969970703, "learning_rate": 0.0001, "loss": 0.0262, "step": 13430 }, { "epoch": 102.59541984732824, "grad_norm": 1.9222385883331299, "learning_rate": 0.0001, "loss": 0.0262, "step": 13440 }, { "epoch": 102.67175572519083, "grad_norm": 1.3386602401733398, "learning_rate": 0.0001, "loss": 0.0269, "step": 13450 }, { "epoch": 102.74809160305344, "grad_norm": 1.528159499168396, "learning_rate": 0.0001, "loss": 0.0269, "step": 13460 }, { "epoch": 102.82442748091603, "grad_norm": 2.025310516357422, "learning_rate": 0.0001, "loss": 0.0277, "step": 13470 }, { "epoch": 102.90076335877862, "grad_norm": 1.272646188735962, "learning_rate": 0.0001, "loss": 0.027, "step": 13480 }, { "epoch": 102.97709923664122, "grad_norm": 1.8140534162521362, "learning_rate": 0.0001, "loss": 0.0269, "step": 13490 }, { "epoch": 103.05343511450381, "grad_norm": 1.674079179763794, "learning_rate": 0.0001, "loss": 0.0279, "step": 13500 }, { "epoch": 103.12977099236642, "grad_norm": 1.9771126508712769, "learning_rate": 0.0001, "loss": 0.0272, "step": 13510 }, { "epoch": 103.20610687022901, "grad_norm": 1.9426838159561157, "learning_rate": 0.0001, "loss": 0.0288, "step": 13520 }, { "epoch": 103.2824427480916, "grad_norm": 2.386338233947754, "learning_rate": 0.0001, "loss": 0.0294, "step": 13530 }, { "epoch": 103.3587786259542, "grad_norm": 1.700807809829712, "learning_rate": 0.0001, "loss": 0.028, "step": 13540 }, { "epoch": 103.43511450381679, "grad_norm": 2.140105724334717, "learning_rate": 0.0001, "loss": 0.0274, "step": 13550 }, { "epoch": 103.5114503816794, "grad_norm": 1.7366458177566528, "learning_rate": 0.0001, "loss": 0.0263, "step": 13560 }, { "epoch": 103.58778625954199, "grad_norm": 1.9462931156158447, "learning_rate": 0.0001, "loss": 0.027, "step": 13570 }, { "epoch": 103.66412213740458, "grad_norm": 1.4153809547424316, "learning_rate": 0.0001, "loss": 0.0267, "step": 13580 }, { "epoch": 103.74045801526718, "grad_norm": 2.305588483810425, "learning_rate": 0.0001, "loss": 0.0273, "step": 13590 }, { "epoch": 103.81679389312977, "grad_norm": 1.8659080266952515, "learning_rate": 0.0001, "loss": 0.029, "step": 13600 }, { "epoch": 103.89312977099236, "grad_norm": 1.947811484336853, "learning_rate": 0.0001, "loss": 0.0275, "step": 13610 }, { "epoch": 103.96946564885496, "grad_norm": 1.7041164636611938, "learning_rate": 0.0001, "loss": 0.0293, "step": 13620 }, { "epoch": 104.04580152671755, "grad_norm": 1.5779571533203125, "learning_rate": 0.0001, "loss": 0.0288, "step": 13630 }, { "epoch": 104.12213740458016, "grad_norm": 2.3642513751983643, "learning_rate": 0.0001, "loss": 0.0282, "step": 13640 }, { "epoch": 104.19847328244275, "grad_norm": 1.5471822023391724, "learning_rate": 0.0001, "loss": 0.0288, "step": 13650 }, { "epoch": 104.27480916030534, "grad_norm": 2.5675530433654785, "learning_rate": 0.0001, "loss": 0.029, "step": 13660 }, { "epoch": 104.35114503816794, "grad_norm": 1.444873332977295, "learning_rate": 0.0001, "loss": 0.0282, "step": 13670 }, { "epoch": 104.42748091603053, "grad_norm": 1.8112150430679321, "learning_rate": 0.0001, "loss": 0.0262, "step": 13680 }, { "epoch": 104.50381679389314, "grad_norm": 2.0475568771362305, "learning_rate": 0.0001, "loss": 0.0268, "step": 13690 }, { "epoch": 104.58015267175573, "grad_norm": 2.410339593887329, "learning_rate": 0.0001, "loss": 0.0263, "step": 13700 }, { "epoch": 104.65648854961832, "grad_norm": 2.342548370361328, "learning_rate": 0.0001, "loss": 0.0256, "step": 13710 }, { "epoch": 104.73282442748092, "grad_norm": 1.703364610671997, "learning_rate": 0.0001, "loss": 0.0271, "step": 13720 }, { "epoch": 104.80916030534351, "grad_norm": 1.918748378753662, "learning_rate": 0.0001, "loss": 0.0268, "step": 13730 }, { "epoch": 104.8854961832061, "grad_norm": 1.7196224927902222, "learning_rate": 0.0001, "loss": 0.027, "step": 13740 }, { "epoch": 104.9618320610687, "grad_norm": 1.5409975051879883, "learning_rate": 0.0001, "loss": 0.0262, "step": 13750 }, { "epoch": 105.0381679389313, "grad_norm": 1.603686809539795, "learning_rate": 0.0001, "loss": 0.0258, "step": 13760 }, { "epoch": 105.1145038167939, "grad_norm": 1.9122635126113892, "learning_rate": 0.0001, "loss": 0.0269, "step": 13770 }, { "epoch": 105.19083969465649, "grad_norm": 2.1836819648742676, "learning_rate": 0.0001, "loss": 0.0267, "step": 13780 }, { "epoch": 105.26717557251908, "grad_norm": 2.1935462951660156, "learning_rate": 0.0001, "loss": 0.0259, "step": 13790 }, { "epoch": 105.34351145038168, "grad_norm": 2.5028724670410156, "learning_rate": 0.0001, "loss": 0.027, "step": 13800 }, { "epoch": 105.41984732824427, "grad_norm": 1.45194673538208, "learning_rate": 0.0001, "loss": 0.0267, "step": 13810 }, { "epoch": 105.49618320610686, "grad_norm": 2.1363937854766846, "learning_rate": 0.0001, "loss": 0.0263, "step": 13820 }, { "epoch": 105.57251908396947, "grad_norm": 1.7763941287994385, "learning_rate": 0.0001, "loss": 0.0265, "step": 13830 }, { "epoch": 105.64885496183206, "grad_norm": 1.5406538248062134, "learning_rate": 0.0001, "loss": 0.0266, "step": 13840 }, { "epoch": 105.72519083969466, "grad_norm": 1.9294182062149048, "learning_rate": 0.0001, "loss": 0.0258, "step": 13850 }, { "epoch": 105.80152671755725, "grad_norm": 2.200371026992798, "learning_rate": 0.0001, "loss": 0.026, "step": 13860 }, { "epoch": 105.87786259541984, "grad_norm": 2.068359375, "learning_rate": 0.0001, "loss": 0.026, "step": 13870 }, { "epoch": 105.95419847328245, "grad_norm": 2.2413039207458496, "learning_rate": 0.0001, "loss": 0.0275, "step": 13880 }, { "epoch": 106.03053435114504, "grad_norm": 1.6548395156860352, "learning_rate": 0.0001, "loss": 0.0273, "step": 13890 }, { "epoch": 106.10687022900764, "grad_norm": 2.1615777015686035, "learning_rate": 0.0001, "loss": 0.0258, "step": 13900 }, { "epoch": 106.18320610687023, "grad_norm": 1.7091784477233887, "learning_rate": 0.0001, "loss": 0.0252, "step": 13910 }, { "epoch": 106.25954198473282, "grad_norm": 1.6703401803970337, "learning_rate": 0.0001, "loss": 0.025, "step": 13920 }, { "epoch": 106.33587786259542, "grad_norm": 2.4271137714385986, "learning_rate": 0.0001, "loss": 0.0259, "step": 13930 }, { "epoch": 106.41221374045801, "grad_norm": 1.9302237033843994, "learning_rate": 0.0001, "loss": 0.0264, "step": 13940 }, { "epoch": 106.4885496183206, "grad_norm": 1.6839969158172607, "learning_rate": 0.0001, "loss": 0.0273, "step": 13950 }, { "epoch": 106.56488549618321, "grad_norm": 1.8176937103271484, "learning_rate": 0.0001, "loss": 0.028, "step": 13960 }, { "epoch": 106.6412213740458, "grad_norm": 2.3561131954193115, "learning_rate": 0.0001, "loss": 0.0281, "step": 13970 }, { "epoch": 106.7175572519084, "grad_norm": 1.571781039237976, "learning_rate": 0.0001, "loss": 0.0287, "step": 13980 }, { "epoch": 106.79389312977099, "grad_norm": 2.0897812843322754, "learning_rate": 0.0001, "loss": 0.0284, "step": 13990 }, { "epoch": 106.87022900763358, "grad_norm": 1.6471349000930786, "learning_rate": 0.0001, "loss": 0.0287, "step": 14000 }, { "epoch": 106.94656488549619, "grad_norm": 1.9019122123718262, "learning_rate": 0.0001, "loss": 0.0276, "step": 14010 }, { "epoch": 107.02290076335878, "grad_norm": 2.1511881351470947, "learning_rate": 0.0001, "loss": 0.0267, "step": 14020 }, { "epoch": 107.09923664122137, "grad_norm": 2.299320697784424, "learning_rate": 0.0001, "loss": 0.0291, "step": 14030 }, { "epoch": 107.17557251908397, "grad_norm": 2.847635269165039, "learning_rate": 0.0001, "loss": 0.0269, "step": 14040 }, { "epoch": 107.25190839694656, "grad_norm": 2.1379647254943848, "learning_rate": 0.0001, "loss": 0.0263, "step": 14050 }, { "epoch": 107.32824427480917, "grad_norm": 1.6726840734481812, "learning_rate": 0.0001, "loss": 0.0255, "step": 14060 }, { "epoch": 107.40458015267176, "grad_norm": 3.1246273517608643, "learning_rate": 0.0001, "loss": 0.026, "step": 14070 }, { "epoch": 107.48091603053435, "grad_norm": 2.2021291255950928, "learning_rate": 0.0001, "loss": 0.0257, "step": 14080 }, { "epoch": 107.55725190839695, "grad_norm": 2.1489217281341553, "learning_rate": 0.0001, "loss": 0.0264, "step": 14090 }, { "epoch": 107.63358778625954, "grad_norm": 1.832087516784668, "learning_rate": 0.0001, "loss": 0.0267, "step": 14100 }, { "epoch": 107.70992366412214, "grad_norm": 2.3391225337982178, "learning_rate": 0.0001, "loss": 0.0268, "step": 14110 }, { "epoch": 107.78625954198473, "grad_norm": 1.6024986505508423, "learning_rate": 0.0001, "loss": 0.0256, "step": 14120 }, { "epoch": 107.86259541984732, "grad_norm": 2.0169708728790283, "learning_rate": 0.0001, "loss": 0.0257, "step": 14130 }, { "epoch": 107.93893129770993, "grad_norm": 1.675081491470337, "learning_rate": 0.0001, "loss": 0.0265, "step": 14140 }, { "epoch": 108.01526717557252, "grad_norm": 1.5355618000030518, "learning_rate": 0.0001, "loss": 0.0263, "step": 14150 }, { "epoch": 108.09160305343511, "grad_norm": 1.4880777597427368, "learning_rate": 0.0001, "loss": 0.0255, "step": 14160 }, { "epoch": 108.16793893129771, "grad_norm": 2.1000118255615234, "learning_rate": 0.0001, "loss": 0.0264, "step": 14170 }, { "epoch": 108.2442748091603, "grad_norm": 1.8350757360458374, "learning_rate": 0.0001, "loss": 0.0253, "step": 14180 }, { "epoch": 108.3206106870229, "grad_norm": 2.1062300205230713, "learning_rate": 0.0001, "loss": 0.0269, "step": 14190 }, { "epoch": 108.3969465648855, "grad_norm": 1.8613107204437256, "learning_rate": 0.0001, "loss": 0.0267, "step": 14200 }, { "epoch": 108.47328244274809, "grad_norm": 1.8723589181900024, "learning_rate": 0.0001, "loss": 0.0259, "step": 14210 }, { "epoch": 108.54961832061069, "grad_norm": 2.2070553302764893, "learning_rate": 0.0001, "loss": 0.0271, "step": 14220 }, { "epoch": 108.62595419847328, "grad_norm": 2.7128851413726807, "learning_rate": 0.0001, "loss": 0.0257, "step": 14230 }, { "epoch": 108.70229007633588, "grad_norm": 2.419053077697754, "learning_rate": 0.0001, "loss": 0.0264, "step": 14240 }, { "epoch": 108.77862595419847, "grad_norm": 1.5612834692001343, "learning_rate": 0.0001, "loss": 0.0257, "step": 14250 }, { "epoch": 108.85496183206106, "grad_norm": 1.5611684322357178, "learning_rate": 0.0001, "loss": 0.0259, "step": 14260 }, { "epoch": 108.93129770992367, "grad_norm": 1.8846875429153442, "learning_rate": 0.0001, "loss": 0.0261, "step": 14270 }, { "epoch": 109.00763358778626, "grad_norm": 1.501477837562561, "learning_rate": 0.0001, "loss": 0.0258, "step": 14280 }, { "epoch": 109.08396946564885, "grad_norm": 1.865256905555725, "learning_rate": 0.0001, "loss": 0.0257, "step": 14290 }, { "epoch": 109.16030534351145, "grad_norm": 1.4026694297790527, "learning_rate": 0.0001, "loss": 0.026, "step": 14300 }, { "epoch": 109.23664122137404, "grad_norm": 1.8338406085968018, "learning_rate": 0.0001, "loss": 0.0259, "step": 14310 }, { "epoch": 109.31297709923665, "grad_norm": 1.333419919013977, "learning_rate": 0.0001, "loss": 0.0262, "step": 14320 }, { "epoch": 109.38931297709924, "grad_norm": 1.7217533588409424, "learning_rate": 0.0001, "loss": 0.0266, "step": 14330 }, { "epoch": 109.46564885496183, "grad_norm": 2.003329038619995, "learning_rate": 0.0001, "loss": 0.0273, "step": 14340 }, { "epoch": 109.54198473282443, "grad_norm": 2.0766079425811768, "learning_rate": 0.0001, "loss": 0.0275, "step": 14350 }, { "epoch": 109.61832061068702, "grad_norm": 2.0275754928588867, "learning_rate": 0.0001, "loss": 0.0265, "step": 14360 }, { "epoch": 109.69465648854961, "grad_norm": 1.8439338207244873, "learning_rate": 0.0001, "loss": 0.0257, "step": 14370 }, { "epoch": 109.77099236641222, "grad_norm": 1.3396995067596436, "learning_rate": 0.0001, "loss": 0.0253, "step": 14380 }, { "epoch": 109.8473282442748, "grad_norm": 2.6278295516967773, "learning_rate": 0.0001, "loss": 0.0256, "step": 14390 }, { "epoch": 109.92366412213741, "grad_norm": 1.913515329360962, "learning_rate": 0.0001, "loss": 0.0263, "step": 14400 }, { "epoch": 110.0, "grad_norm": 1.8752895593643188, "learning_rate": 0.0001, "loss": 0.0264, "step": 14410 }, { "epoch": 110.07633587786259, "grad_norm": 1.771101474761963, "learning_rate": 0.0001, "loss": 0.027, "step": 14420 }, { "epoch": 110.1526717557252, "grad_norm": 2.3533873558044434, "learning_rate": 0.0001, "loss": 0.0252, "step": 14430 }, { "epoch": 110.22900763358778, "grad_norm": 2.6179544925689697, "learning_rate": 0.0001, "loss": 0.0259, "step": 14440 }, { "epoch": 110.30534351145039, "grad_norm": 2.2424890995025635, "learning_rate": 0.0001, "loss": 0.0251, "step": 14450 }, { "epoch": 110.38167938931298, "grad_norm": 2.2671687602996826, "learning_rate": 0.0001, "loss": 0.0254, "step": 14460 }, { "epoch": 110.45801526717557, "grad_norm": 1.6238926649093628, "learning_rate": 0.0001, "loss": 0.0258, "step": 14470 }, { "epoch": 110.53435114503817, "grad_norm": 2.1705501079559326, "learning_rate": 0.0001, "loss": 0.025, "step": 14480 }, { "epoch": 110.61068702290076, "grad_norm": 1.5282460451126099, "learning_rate": 0.0001, "loss": 0.0249, "step": 14490 }, { "epoch": 110.68702290076335, "grad_norm": 2.1080498695373535, "learning_rate": 0.0001, "loss": 0.0257, "step": 14500 }, { "epoch": 110.76335877862596, "grad_norm": 2.17987322807312, "learning_rate": 0.0001, "loss": 0.0257, "step": 14510 }, { "epoch": 110.83969465648855, "grad_norm": 2.0062994956970215, "learning_rate": 0.0001, "loss": 0.0268, "step": 14520 }, { "epoch": 110.91603053435115, "grad_norm": 2.5881309509277344, "learning_rate": 0.0001, "loss": 0.0253, "step": 14530 }, { "epoch": 110.99236641221374, "grad_norm": 1.7685433626174927, "learning_rate": 0.0001, "loss": 0.0254, "step": 14540 }, { "epoch": 111.06870229007633, "grad_norm": 2.0147347450256348, "learning_rate": 0.0001, "loss": 0.0249, "step": 14550 }, { "epoch": 111.14503816793894, "grad_norm": 1.6122452020645142, "learning_rate": 0.0001, "loss": 0.0268, "step": 14560 }, { "epoch": 111.22137404580153, "grad_norm": 1.6672857999801636, "learning_rate": 0.0001, "loss": 0.0245, "step": 14570 }, { "epoch": 111.29770992366412, "grad_norm": 2.3389432430267334, "learning_rate": 0.0001, "loss": 0.0256, "step": 14580 }, { "epoch": 111.37404580152672, "grad_norm": 1.5482020378112793, "learning_rate": 0.0001, "loss": 0.025, "step": 14590 }, { "epoch": 111.45038167938931, "grad_norm": 2.172473192214966, "learning_rate": 0.0001, "loss": 0.0245, "step": 14600 }, { "epoch": 111.52671755725191, "grad_norm": 1.6346776485443115, "learning_rate": 0.0001, "loss": 0.0257, "step": 14610 }, { "epoch": 111.6030534351145, "grad_norm": 1.9348939657211304, "learning_rate": 0.0001, "loss": 0.0254, "step": 14620 }, { "epoch": 111.6793893129771, "grad_norm": 2.0765647888183594, "learning_rate": 0.0001, "loss": 0.026, "step": 14630 }, { "epoch": 111.7557251908397, "grad_norm": 1.7074205875396729, "learning_rate": 0.0001, "loss": 0.0259, "step": 14640 }, { "epoch": 111.83206106870229, "grad_norm": 1.8330968618392944, "learning_rate": 0.0001, "loss": 0.0257, "step": 14650 }, { "epoch": 111.90839694656489, "grad_norm": 1.8337650299072266, "learning_rate": 0.0001, "loss": 0.026, "step": 14660 }, { "epoch": 111.98473282442748, "grad_norm": 1.375638723373413, "learning_rate": 0.0001, "loss": 0.0271, "step": 14670 }, { "epoch": 112.06106870229007, "grad_norm": 1.7264699935913086, "learning_rate": 0.0001, "loss": 0.027, "step": 14680 }, { "epoch": 112.13740458015268, "grad_norm": 2.5966849327087402, "learning_rate": 0.0001, "loss": 0.0272, "step": 14690 }, { "epoch": 112.21374045801527, "grad_norm": 2.2791976928710938, "learning_rate": 0.0001, "loss": 0.0273, "step": 14700 }, { "epoch": 112.29007633587786, "grad_norm": 2.2592978477478027, "learning_rate": 0.0001, "loss": 0.026, "step": 14710 }, { "epoch": 112.36641221374046, "grad_norm": 2.62086820602417, "learning_rate": 0.0001, "loss": 0.0266, "step": 14720 }, { "epoch": 112.44274809160305, "grad_norm": 2.071450710296631, "learning_rate": 0.0001, "loss": 0.0263, "step": 14730 }, { "epoch": 112.51908396946565, "grad_norm": 1.5072299242019653, "learning_rate": 0.0001, "loss": 0.0255, "step": 14740 }, { "epoch": 112.59541984732824, "grad_norm": 1.3815081119537354, "learning_rate": 0.0001, "loss": 0.0251, "step": 14750 }, { "epoch": 112.67175572519083, "grad_norm": 1.365954041481018, "learning_rate": 0.0001, "loss": 0.0252, "step": 14760 }, { "epoch": 112.74809160305344, "grad_norm": 2.017256021499634, "learning_rate": 0.0001, "loss": 0.0257, "step": 14770 }, { "epoch": 112.82442748091603, "grad_norm": 2.329543352127075, "learning_rate": 0.0001, "loss": 0.025, "step": 14780 }, { "epoch": 112.90076335877862, "grad_norm": 2.146838426589966, "learning_rate": 0.0001, "loss": 0.0251, "step": 14790 }, { "epoch": 112.97709923664122, "grad_norm": 2.663508415222168, "learning_rate": 0.0001, "loss": 0.0264, "step": 14800 }, { "epoch": 113.05343511450381, "grad_norm": 1.8337376117706299, "learning_rate": 0.0001, "loss": 0.0259, "step": 14810 }, { "epoch": 113.12977099236642, "grad_norm": 2.3346917629241943, "learning_rate": 0.0001, "loss": 0.0253, "step": 14820 }, { "epoch": 113.20610687022901, "grad_norm": 1.57377290725708, "learning_rate": 0.0001, "loss": 0.0241, "step": 14830 }, { "epoch": 113.2824427480916, "grad_norm": 1.8287137746810913, "learning_rate": 0.0001, "loss": 0.0252, "step": 14840 }, { "epoch": 113.3587786259542, "grad_norm": 2.025059938430786, "learning_rate": 0.0001, "loss": 0.0239, "step": 14850 }, { "epoch": 113.43511450381679, "grad_norm": 1.465946912765503, "learning_rate": 0.0001, "loss": 0.0256, "step": 14860 }, { "epoch": 113.5114503816794, "grad_norm": 1.6853965520858765, "learning_rate": 0.0001, "loss": 0.0245, "step": 14870 }, { "epoch": 113.58778625954199, "grad_norm": 1.8550448417663574, "learning_rate": 0.0001, "loss": 0.0271, "step": 14880 }, { "epoch": 113.66412213740458, "grad_norm": 2.014408826828003, "learning_rate": 0.0001, "loss": 0.0265, "step": 14890 }, { "epoch": 113.74045801526718, "grad_norm": 1.9487653970718384, "learning_rate": 0.0001, "loss": 0.0264, "step": 14900 }, { "epoch": 113.81679389312977, "grad_norm": 1.9449117183685303, "learning_rate": 0.0001, "loss": 0.0258, "step": 14910 }, { "epoch": 113.89312977099236, "grad_norm": 1.936742901802063, "learning_rate": 0.0001, "loss": 0.0255, "step": 14920 }, { "epoch": 113.96946564885496, "grad_norm": 2.0583653450012207, "learning_rate": 0.0001, "loss": 0.0257, "step": 14930 }, { "epoch": 114.04580152671755, "grad_norm": 1.9321428537368774, "learning_rate": 0.0001, "loss": 0.0261, "step": 14940 }, { "epoch": 114.12213740458016, "grad_norm": 1.6811192035675049, "learning_rate": 0.0001, "loss": 0.0259, "step": 14950 }, { "epoch": 114.19847328244275, "grad_norm": 1.9829530715942383, "learning_rate": 0.0001, "loss": 0.0254, "step": 14960 }, { "epoch": 114.27480916030534, "grad_norm": 1.683912754058838, "learning_rate": 0.0001, "loss": 0.0247, "step": 14970 }, { "epoch": 114.35114503816794, "grad_norm": 1.5881398916244507, "learning_rate": 0.0001, "loss": 0.0246, "step": 14980 }, { "epoch": 114.42748091603053, "grad_norm": 2.2326455116271973, "learning_rate": 0.0001, "loss": 0.0262, "step": 14990 }, { "epoch": 114.50381679389314, "grad_norm": 2.079730749130249, "learning_rate": 0.0001, "loss": 0.0262, "step": 15000 }, { "epoch": 114.58015267175573, "grad_norm": 2.4789414405822754, "learning_rate": 0.0001, "loss": 0.0259, "step": 15010 }, { "epoch": 114.65648854961832, "grad_norm": 2.590930700302124, "learning_rate": 0.0001, "loss": 0.0264, "step": 15020 }, { "epoch": 114.73282442748092, "grad_norm": 2.352165460586548, "learning_rate": 0.0001, "loss": 0.0252, "step": 15030 }, { "epoch": 114.80916030534351, "grad_norm": 1.8409730195999146, "learning_rate": 0.0001, "loss": 0.0254, "step": 15040 }, { "epoch": 114.8854961832061, "grad_norm": 1.688120722770691, "learning_rate": 0.0001, "loss": 0.0242, "step": 15050 }, { "epoch": 114.9618320610687, "grad_norm": 2.0241076946258545, "learning_rate": 0.0001, "loss": 0.0254, "step": 15060 }, { "epoch": 115.0381679389313, "grad_norm": 2.6924779415130615, "learning_rate": 0.0001, "loss": 0.0269, "step": 15070 }, { "epoch": 115.1145038167939, "grad_norm": 1.8466442823410034, "learning_rate": 0.0001, "loss": 0.0269, "step": 15080 }, { "epoch": 115.19083969465649, "grad_norm": 2.006572961807251, "learning_rate": 0.0001, "loss": 0.026, "step": 15090 }, { "epoch": 115.26717557251908, "grad_norm": 3.0930116176605225, "learning_rate": 0.0001, "loss": 0.0259, "step": 15100 }, { "epoch": 115.34351145038168, "grad_norm": 2.6286890506744385, "learning_rate": 0.0001, "loss": 0.0265, "step": 15110 }, { "epoch": 115.41984732824427, "grad_norm": 1.8086265325546265, "learning_rate": 0.0001, "loss": 0.0244, "step": 15120 }, { "epoch": 115.49618320610686, "grad_norm": 2.329252004623413, "learning_rate": 0.0001, "loss": 0.0258, "step": 15130 }, { "epoch": 115.57251908396947, "grad_norm": 2.005523681640625, "learning_rate": 0.0001, "loss": 0.0255, "step": 15140 }, { "epoch": 115.64885496183206, "grad_norm": 1.884843349456787, "learning_rate": 0.0001, "loss": 0.0254, "step": 15150 }, { "epoch": 115.72519083969466, "grad_norm": 1.3556348085403442, "learning_rate": 0.0001, "loss": 0.0261, "step": 15160 }, { "epoch": 115.80152671755725, "grad_norm": 1.4958380460739136, "learning_rate": 0.0001, "loss": 0.0246, "step": 15170 }, { "epoch": 115.87786259541984, "grad_norm": 1.888242244720459, "learning_rate": 0.0001, "loss": 0.0258, "step": 15180 }, { "epoch": 115.95419847328245, "grad_norm": 2.1639037132263184, "learning_rate": 0.0001, "loss": 0.0268, "step": 15190 }, { "epoch": 116.03053435114504, "grad_norm": 1.5670878887176514, "learning_rate": 0.0001, "loss": 0.0252, "step": 15200 }, { "epoch": 116.10687022900764, "grad_norm": 1.6756706237792969, "learning_rate": 0.0001, "loss": 0.0252, "step": 15210 }, { "epoch": 116.18320610687023, "grad_norm": 1.525733470916748, "learning_rate": 0.0001, "loss": 0.0258, "step": 15220 }, { "epoch": 116.25954198473282, "grad_norm": 2.0264508724212646, "learning_rate": 0.0001, "loss": 0.0249, "step": 15230 }, { "epoch": 116.33587786259542, "grad_norm": 1.8771530389785767, "learning_rate": 0.0001, "loss": 0.0254, "step": 15240 }, { "epoch": 116.41221374045801, "grad_norm": 1.8668025732040405, "learning_rate": 0.0001, "loss": 0.0252, "step": 15250 }, { "epoch": 116.4885496183206, "grad_norm": 2.365960121154785, "learning_rate": 0.0001, "loss": 0.0249, "step": 15260 }, { "epoch": 116.56488549618321, "grad_norm": 1.709107756614685, "learning_rate": 0.0001, "loss": 0.0252, "step": 15270 }, { "epoch": 116.6412213740458, "grad_norm": 1.9496164321899414, "learning_rate": 0.0001, "loss": 0.0254, "step": 15280 }, { "epoch": 116.7175572519084, "grad_norm": 2.4106786251068115, "learning_rate": 0.0001, "loss": 0.0261, "step": 15290 }, { "epoch": 116.79389312977099, "grad_norm": 1.8348307609558105, "learning_rate": 0.0001, "loss": 0.0262, "step": 15300 }, { "epoch": 116.87022900763358, "grad_norm": 1.9022836685180664, "learning_rate": 0.0001, "loss": 0.0269, "step": 15310 }, { "epoch": 116.94656488549619, "grad_norm": 2.411651849746704, "learning_rate": 0.0001, "loss": 0.0262, "step": 15320 }, { "epoch": 117.02290076335878, "grad_norm": 1.4580509662628174, "learning_rate": 0.0001, "loss": 0.0249, "step": 15330 }, { "epoch": 117.09923664122137, "grad_norm": 1.9735075235366821, "learning_rate": 0.0001, "loss": 0.0251, "step": 15340 }, { "epoch": 117.17557251908397, "grad_norm": 1.7291923761367798, "learning_rate": 0.0001, "loss": 0.0242, "step": 15350 }, { "epoch": 117.25190839694656, "grad_norm": 1.4364008903503418, "learning_rate": 0.0001, "loss": 0.0245, "step": 15360 }, { "epoch": 117.32824427480917, "grad_norm": 2.0841617584228516, "learning_rate": 0.0001, "loss": 0.0235, "step": 15370 }, { "epoch": 117.40458015267176, "grad_norm": 1.4696444272994995, "learning_rate": 0.0001, "loss": 0.024, "step": 15380 }, { "epoch": 117.48091603053435, "grad_norm": 2.1728620529174805, "learning_rate": 0.0001, "loss": 0.0261, "step": 15390 }, { "epoch": 117.55725190839695, "grad_norm": 1.8555303812026978, "learning_rate": 0.0001, "loss": 0.0256, "step": 15400 }, { "epoch": 117.63358778625954, "grad_norm": 1.4966517686843872, "learning_rate": 0.0001, "loss": 0.0253, "step": 15410 }, { "epoch": 117.70992366412214, "grad_norm": 2.3192079067230225, "learning_rate": 0.0001, "loss": 0.0258, "step": 15420 }, { "epoch": 117.78625954198473, "grad_norm": 1.7937722206115723, "learning_rate": 0.0001, "loss": 0.0253, "step": 15430 }, { "epoch": 117.86259541984732, "grad_norm": 1.5581895112991333, "learning_rate": 0.0001, "loss": 0.0252, "step": 15440 }, { "epoch": 117.93893129770993, "grad_norm": 1.5753506422042847, "learning_rate": 0.0001, "loss": 0.0254, "step": 15450 }, { "epoch": 118.01526717557252, "grad_norm": 1.4206081628799438, "learning_rate": 0.0001, "loss": 0.0255, "step": 15460 }, { "epoch": 118.09160305343511, "grad_norm": 2.1512374877929688, "learning_rate": 0.0001, "loss": 0.0258, "step": 15470 }, { "epoch": 118.16793893129771, "grad_norm": 2.0737228393554688, "learning_rate": 0.0001, "loss": 0.0273, "step": 15480 }, { "epoch": 118.2442748091603, "grad_norm": 1.9602850675582886, "learning_rate": 0.0001, "loss": 0.0261, "step": 15490 }, { "epoch": 118.3206106870229, "grad_norm": 2.0509796142578125, "learning_rate": 0.0001, "loss": 0.0266, "step": 15500 }, { "epoch": 118.3969465648855, "grad_norm": 1.9820879697799683, "learning_rate": 0.0001, "loss": 0.0255, "step": 15510 }, { "epoch": 118.47328244274809, "grad_norm": 1.86448335647583, "learning_rate": 0.0001, "loss": 0.0244, "step": 15520 }, { "epoch": 118.54961832061069, "grad_norm": 1.9419654607772827, "learning_rate": 0.0001, "loss": 0.0247, "step": 15530 }, { "epoch": 118.62595419847328, "grad_norm": 2.47489333152771, "learning_rate": 0.0001, "loss": 0.0254, "step": 15540 }, { "epoch": 118.70229007633588, "grad_norm": 1.7001042366027832, "learning_rate": 0.0001, "loss": 0.024, "step": 15550 }, { "epoch": 118.77862595419847, "grad_norm": 1.4148337841033936, "learning_rate": 0.0001, "loss": 0.0246, "step": 15560 }, { "epoch": 118.85496183206106, "grad_norm": 2.7960731983184814, "learning_rate": 0.0001, "loss": 0.0253, "step": 15570 }, { "epoch": 118.93129770992367, "grad_norm": 1.7280665636062622, "learning_rate": 0.0001, "loss": 0.025, "step": 15580 }, { "epoch": 119.00763358778626, "grad_norm": 1.6151024103164673, "learning_rate": 0.0001, "loss": 0.0242, "step": 15590 }, { "epoch": 119.08396946564885, "grad_norm": 1.2589495182037354, "learning_rate": 0.0001, "loss": 0.023, "step": 15600 }, { "epoch": 119.16030534351145, "grad_norm": 2.3939781188964844, "learning_rate": 0.0001, "loss": 0.0237, "step": 15610 }, { "epoch": 119.23664122137404, "grad_norm": 1.746652364730835, "learning_rate": 0.0001, "loss": 0.0246, "step": 15620 }, { "epoch": 119.31297709923665, "grad_norm": 2.4038126468658447, "learning_rate": 0.0001, "loss": 0.0248, "step": 15630 }, { "epoch": 119.38931297709924, "grad_norm": 1.746677279472351, "learning_rate": 0.0001, "loss": 0.0251, "step": 15640 }, { "epoch": 119.46564885496183, "grad_norm": 1.5188875198364258, "learning_rate": 0.0001, "loss": 0.0253, "step": 15650 }, { "epoch": 119.54198473282443, "grad_norm": 1.8718074560165405, "learning_rate": 0.0001, "loss": 0.0244, "step": 15660 }, { "epoch": 119.61832061068702, "grad_norm": 1.8563330173492432, "learning_rate": 0.0001, "loss": 0.0237, "step": 15670 }, { "epoch": 119.69465648854961, "grad_norm": 1.5635520219802856, "learning_rate": 0.0001, "loss": 0.0246, "step": 15680 }, { "epoch": 119.77099236641222, "grad_norm": 2.0395240783691406, "learning_rate": 0.0001, "loss": 0.0253, "step": 15690 }, { "epoch": 119.8473282442748, "grad_norm": 1.8876255750656128, "learning_rate": 0.0001, "loss": 0.0248, "step": 15700 }, { "epoch": 119.92366412213741, "grad_norm": 1.6750431060791016, "learning_rate": 0.0001, "loss": 0.0261, "step": 15710 }, { "epoch": 120.0, "grad_norm": 1.1838289499282837, "learning_rate": 0.0001, "loss": 0.0251, "step": 15720 }, { "epoch": 120.07633587786259, "grad_norm": 1.637638807296753, "learning_rate": 0.0001, "loss": 0.0252, "step": 15730 }, { "epoch": 120.1526717557252, "grad_norm": 1.3162132501602173, "learning_rate": 0.0001, "loss": 0.0262, "step": 15740 }, { "epoch": 120.22900763358778, "grad_norm": 1.3929551839828491, "learning_rate": 0.0001, "loss": 0.0251, "step": 15750 }, { "epoch": 120.30534351145039, "grad_norm": 2.7941837310791016, "learning_rate": 0.0001, "loss": 0.0262, "step": 15760 }, { "epoch": 120.38167938931298, "grad_norm": 1.6508995294570923, "learning_rate": 0.0001, "loss": 0.0254, "step": 15770 }, { "epoch": 120.45801526717557, "grad_norm": 1.6067787408828735, "learning_rate": 0.0001, "loss": 0.0258, "step": 15780 }, { "epoch": 120.53435114503817, "grad_norm": 2.128896474838257, "learning_rate": 0.0001, "loss": 0.0244, "step": 15790 }, { "epoch": 120.61068702290076, "grad_norm": 1.7550603151321411, "learning_rate": 0.0001, "loss": 0.0248, "step": 15800 }, { "epoch": 120.68702290076335, "grad_norm": 2.0526845455169678, "learning_rate": 0.0001, "loss": 0.0257, "step": 15810 }, { "epoch": 120.76335877862596, "grad_norm": 1.750895380973816, "learning_rate": 0.0001, "loss": 0.0243, "step": 15820 }, { "epoch": 120.83969465648855, "grad_norm": 1.295495629310608, "learning_rate": 0.0001, "loss": 0.0236, "step": 15830 }, { "epoch": 120.91603053435115, "grad_norm": 2.3013792037963867, "learning_rate": 0.0001, "loss": 0.0241, "step": 15840 }, { "epoch": 120.99236641221374, "grad_norm": 2.035681962966919, "learning_rate": 0.0001, "loss": 0.0241, "step": 15850 }, { "epoch": 121.06870229007633, "grad_norm": 1.697551965713501, "learning_rate": 0.0001, "loss": 0.0244, "step": 15860 }, { "epoch": 121.14503816793894, "grad_norm": 1.3557212352752686, "learning_rate": 0.0001, "loss": 0.024, "step": 15870 }, { "epoch": 121.22137404580153, "grad_norm": 1.8172125816345215, "learning_rate": 0.0001, "loss": 0.0244, "step": 15880 }, { "epoch": 121.29770992366412, "grad_norm": 2.312432050704956, "learning_rate": 0.0001, "loss": 0.0244, "step": 15890 }, { "epoch": 121.37404580152672, "grad_norm": 2.3072333335876465, "learning_rate": 0.0001, "loss": 0.0245, "step": 15900 }, { "epoch": 121.45038167938931, "grad_norm": 1.7157551050186157, "learning_rate": 0.0001, "loss": 0.0244, "step": 15910 }, { "epoch": 121.52671755725191, "grad_norm": 1.7888585329055786, "learning_rate": 0.0001, "loss": 0.0241, "step": 15920 }, { "epoch": 121.6030534351145, "grad_norm": 1.8260431289672852, "learning_rate": 0.0001, "loss": 0.0241, "step": 15930 }, { "epoch": 121.6793893129771, "grad_norm": 1.778313159942627, "learning_rate": 0.0001, "loss": 0.0242, "step": 15940 }, { "epoch": 121.7557251908397, "grad_norm": 2.2923953533172607, "learning_rate": 0.0001, "loss": 0.0247, "step": 15950 }, { "epoch": 121.83206106870229, "grad_norm": 2.256901264190674, "learning_rate": 0.0001, "loss": 0.0251, "step": 15960 }, { "epoch": 121.90839694656489, "grad_norm": 1.604381799697876, "learning_rate": 0.0001, "loss": 0.0246, "step": 15970 }, { "epoch": 121.98473282442748, "grad_norm": 1.6402384042739868, "learning_rate": 0.0001, "loss": 0.0237, "step": 15980 }, { "epoch": 122.06106870229007, "grad_norm": 1.7560867071151733, "learning_rate": 0.0001, "loss": 0.0241, "step": 15990 }, { "epoch": 122.13740458015268, "grad_norm": 2.2440099716186523, "learning_rate": 0.0001, "loss": 0.0239, "step": 16000 }, { "epoch": 122.21374045801527, "grad_norm": 1.3417513370513916, "learning_rate": 0.0001, "loss": 0.0253, "step": 16010 }, { "epoch": 122.29007633587786, "grad_norm": 1.4212861061096191, "learning_rate": 0.0001, "loss": 0.0259, "step": 16020 }, { "epoch": 122.36641221374046, "grad_norm": 2.1146910190582275, "learning_rate": 0.0001, "loss": 0.0249, "step": 16030 }, { "epoch": 122.44274809160305, "grad_norm": 1.5228506326675415, "learning_rate": 0.0001, "loss": 0.0246, "step": 16040 }, { "epoch": 122.51908396946565, "grad_norm": 1.3759709596633911, "learning_rate": 0.0001, "loss": 0.0251, "step": 16050 }, { "epoch": 122.59541984732824, "grad_norm": 1.8848570585250854, "learning_rate": 0.0001, "loss": 0.0265, "step": 16060 }, { "epoch": 122.67175572519083, "grad_norm": 1.8568482398986816, "learning_rate": 0.0001, "loss": 0.0274, "step": 16070 }, { "epoch": 122.74809160305344, "grad_norm": 1.5361436605453491, "learning_rate": 0.0001, "loss": 0.0246, "step": 16080 }, { "epoch": 122.82442748091603, "grad_norm": 1.4472092390060425, "learning_rate": 0.0001, "loss": 0.0239, "step": 16090 }, { "epoch": 122.90076335877862, "grad_norm": 1.7086504697799683, "learning_rate": 0.0001, "loss": 0.0247, "step": 16100 }, { "epoch": 122.97709923664122, "grad_norm": 1.7791690826416016, "learning_rate": 0.0001, "loss": 0.0244, "step": 16110 }, { "epoch": 123.05343511450381, "grad_norm": 1.703456997871399, "learning_rate": 0.0001, "loss": 0.0254, "step": 16120 }, { "epoch": 123.12977099236642, "grad_norm": 2.4206230640411377, "learning_rate": 0.0001, "loss": 0.0253, "step": 16130 }, { "epoch": 123.20610687022901, "grad_norm": 2.1517505645751953, "learning_rate": 0.0001, "loss": 0.0253, "step": 16140 }, { "epoch": 123.2824427480916, "grad_norm": 2.8271121978759766, "learning_rate": 0.0001, "loss": 0.0245, "step": 16150 }, { "epoch": 123.3587786259542, "grad_norm": 2.9381327629089355, "learning_rate": 0.0001, "loss": 0.0248, "step": 16160 }, { "epoch": 123.43511450381679, "grad_norm": 2.3510942459106445, "learning_rate": 0.0001, "loss": 0.0245, "step": 16170 }, { "epoch": 123.5114503816794, "grad_norm": 2.3057491779327393, "learning_rate": 0.0001, "loss": 0.0242, "step": 16180 }, { "epoch": 123.58778625954199, "grad_norm": 1.5303407907485962, "learning_rate": 0.0001, "loss": 0.0238, "step": 16190 }, { "epoch": 123.66412213740458, "grad_norm": 2.1387438774108887, "learning_rate": 0.0001, "loss": 0.0241, "step": 16200 }, { "epoch": 123.74045801526718, "grad_norm": 1.9674426317214966, "learning_rate": 0.0001, "loss": 0.0239, "step": 16210 }, { "epoch": 123.81679389312977, "grad_norm": 1.9130791425704956, "learning_rate": 0.0001, "loss": 0.0245, "step": 16220 }, { "epoch": 123.89312977099236, "grad_norm": 1.379544734954834, "learning_rate": 0.0001, "loss": 0.0244, "step": 16230 }, { "epoch": 123.96946564885496, "grad_norm": 1.9312430620193481, "learning_rate": 0.0001, "loss": 0.0237, "step": 16240 }, { "epoch": 124.04580152671755, "grad_norm": 1.4378705024719238, "learning_rate": 0.0001, "loss": 0.0239, "step": 16250 }, { "epoch": 124.12213740458016, "grad_norm": 1.9389605522155762, "learning_rate": 0.0001, "loss": 0.0244, "step": 16260 }, { "epoch": 124.19847328244275, "grad_norm": 1.6096863746643066, "learning_rate": 0.0001, "loss": 0.0236, "step": 16270 }, { "epoch": 124.27480916030534, "grad_norm": 1.7904376983642578, "learning_rate": 0.0001, "loss": 0.0235, "step": 16280 }, { "epoch": 124.35114503816794, "grad_norm": 2.1865174770355225, "learning_rate": 0.0001, "loss": 0.0237, "step": 16290 }, { "epoch": 124.42748091603053, "grad_norm": 1.3436956405639648, "learning_rate": 0.0001, "loss": 0.0232, "step": 16300 }, { "epoch": 124.50381679389314, "grad_norm": 1.9371331930160522, "learning_rate": 0.0001, "loss": 0.0244, "step": 16310 }, { "epoch": 124.58015267175573, "grad_norm": 1.7309271097183228, "learning_rate": 0.0001, "loss": 0.0248, "step": 16320 }, { "epoch": 124.65648854961832, "grad_norm": 1.862630009651184, "learning_rate": 0.0001, "loss": 0.0239, "step": 16330 }, { "epoch": 124.73282442748092, "grad_norm": 1.8628296852111816, "learning_rate": 0.0001, "loss": 0.0247, "step": 16340 }, { "epoch": 124.80916030534351, "grad_norm": 2.2309763431549072, "learning_rate": 0.0001, "loss": 0.0256, "step": 16350 }, { "epoch": 124.8854961832061, "grad_norm": 2.0379345417022705, "learning_rate": 0.0001, "loss": 0.0268, "step": 16360 }, { "epoch": 124.9618320610687, "grad_norm": 2.584515333175659, "learning_rate": 0.0001, "loss": 0.0252, "step": 16370 }, { "epoch": 125.0381679389313, "grad_norm": 2.2119827270507812, "learning_rate": 0.0001, "loss": 0.0254, "step": 16380 }, { "epoch": 125.1145038167939, "grad_norm": 2.1146557331085205, "learning_rate": 0.0001, "loss": 0.0249, "step": 16390 }, { "epoch": 125.19083969465649, "grad_norm": 2.0483484268188477, "learning_rate": 0.0001, "loss": 0.0238, "step": 16400 }, { "epoch": 125.26717557251908, "grad_norm": 2.2982728481292725, "learning_rate": 0.0001, "loss": 0.0246, "step": 16410 }, { "epoch": 125.34351145038168, "grad_norm": 2.016889810562134, "learning_rate": 0.0001, "loss": 0.0249, "step": 16420 }, { "epoch": 125.41984732824427, "grad_norm": 2.534792900085449, "learning_rate": 0.0001, "loss": 0.0241, "step": 16430 }, { "epoch": 125.49618320610686, "grad_norm": 1.92035973072052, "learning_rate": 0.0001, "loss": 0.0245, "step": 16440 }, { "epoch": 125.57251908396947, "grad_norm": 1.8099064826965332, "learning_rate": 0.0001, "loss": 0.0239, "step": 16450 }, { "epoch": 125.64885496183206, "grad_norm": 2.320101499557495, "learning_rate": 0.0001, "loss": 0.024, "step": 16460 }, { "epoch": 125.72519083969466, "grad_norm": 2.423621654510498, "learning_rate": 0.0001, "loss": 0.0239, "step": 16470 }, { "epoch": 125.80152671755725, "grad_norm": 1.8318268060684204, "learning_rate": 0.0001, "loss": 0.0229, "step": 16480 }, { "epoch": 125.87786259541984, "grad_norm": 2.7308120727539062, "learning_rate": 0.0001, "loss": 0.0239, "step": 16490 }, { "epoch": 125.95419847328245, "grad_norm": 1.7165706157684326, "learning_rate": 0.0001, "loss": 0.0218, "step": 16500 }, { "epoch": 126.03053435114504, "grad_norm": 2.23111891746521, "learning_rate": 0.0001, "loss": 0.023, "step": 16510 }, { "epoch": 126.10687022900764, "grad_norm": 2.194178342819214, "learning_rate": 0.0001, "loss": 0.0228, "step": 16520 }, { "epoch": 126.18320610687023, "grad_norm": 2.3519949913024902, "learning_rate": 0.0001, "loss": 0.0225, "step": 16530 }, { "epoch": 126.25954198473282, "grad_norm": 1.8070989847183228, "learning_rate": 0.0001, "loss": 0.0232, "step": 16540 }, { "epoch": 126.33587786259542, "grad_norm": 1.9677554368972778, "learning_rate": 0.0001, "loss": 0.0231, "step": 16550 }, { "epoch": 126.41221374045801, "grad_norm": 2.798417568206787, "learning_rate": 0.0001, "loss": 0.0226, "step": 16560 }, { "epoch": 126.4885496183206, "grad_norm": 2.096165180206299, "learning_rate": 0.0001, "loss": 0.0238, "step": 16570 }, { "epoch": 126.56488549618321, "grad_norm": 2.0896694660186768, "learning_rate": 0.0001, "loss": 0.023, "step": 16580 }, { "epoch": 126.6412213740458, "grad_norm": 2.3884434700012207, "learning_rate": 0.0001, "loss": 0.023, "step": 16590 }, { "epoch": 126.7175572519084, "grad_norm": 1.8235700130462646, "learning_rate": 0.0001, "loss": 0.0236, "step": 16600 }, { "epoch": 126.79389312977099, "grad_norm": 2.095262050628662, "learning_rate": 0.0001, "loss": 0.0232, "step": 16610 }, { "epoch": 126.87022900763358, "grad_norm": 1.713771939277649, "learning_rate": 0.0001, "loss": 0.0237, "step": 16620 }, { "epoch": 126.94656488549619, "grad_norm": 1.9034526348114014, "learning_rate": 0.0001, "loss": 0.0232, "step": 16630 }, { "epoch": 127.02290076335878, "grad_norm": 2.1984546184539795, "learning_rate": 0.0001, "loss": 0.0245, "step": 16640 }, { "epoch": 127.09923664122137, "grad_norm": 2.635342597961426, "learning_rate": 0.0001, "loss": 0.0238, "step": 16650 }, { "epoch": 127.17557251908397, "grad_norm": 2.1449527740478516, "learning_rate": 0.0001, "loss": 0.0234, "step": 16660 }, { "epoch": 127.25190839694656, "grad_norm": 1.8335840702056885, "learning_rate": 0.0001, "loss": 0.0231, "step": 16670 }, { "epoch": 127.32824427480917, "grad_norm": 1.7487876415252686, "learning_rate": 0.0001, "loss": 0.0236, "step": 16680 }, { "epoch": 127.40458015267176, "grad_norm": 1.8571327924728394, "learning_rate": 0.0001, "loss": 0.0232, "step": 16690 }, { "epoch": 127.48091603053435, "grad_norm": 1.5652072429656982, "learning_rate": 0.0001, "loss": 0.0242, "step": 16700 }, { "epoch": 127.55725190839695, "grad_norm": 1.8138909339904785, "learning_rate": 0.0001, "loss": 0.0243, "step": 16710 }, { "epoch": 127.63358778625954, "grad_norm": 2.3368799686431885, "learning_rate": 0.0001, "loss": 0.0243, "step": 16720 }, { "epoch": 127.70992366412214, "grad_norm": 2.3406167030334473, "learning_rate": 0.0001, "loss": 0.0239, "step": 16730 }, { "epoch": 127.78625954198473, "grad_norm": 1.7942403554916382, "learning_rate": 0.0001, "loss": 0.025, "step": 16740 }, { "epoch": 127.86259541984732, "grad_norm": 2.771207094192505, "learning_rate": 0.0001, "loss": 0.0242, "step": 16750 }, { "epoch": 127.93893129770993, "grad_norm": 1.6801270246505737, "learning_rate": 0.0001, "loss": 0.0241, "step": 16760 }, { "epoch": 128.01526717557252, "grad_norm": 1.6370314359664917, "learning_rate": 0.0001, "loss": 0.0233, "step": 16770 }, { "epoch": 128.0916030534351, "grad_norm": 1.6873974800109863, "learning_rate": 0.0001, "loss": 0.0231, "step": 16780 }, { "epoch": 128.1679389312977, "grad_norm": 2.436624526977539, "learning_rate": 0.0001, "loss": 0.023, "step": 16790 }, { "epoch": 128.24427480916032, "grad_norm": 2.789461374282837, "learning_rate": 0.0001, "loss": 0.0252, "step": 16800 }, { "epoch": 128.3206106870229, "grad_norm": 2.59100341796875, "learning_rate": 0.0001, "loss": 0.0243, "step": 16810 }, { "epoch": 128.3969465648855, "grad_norm": 1.3165935277938843, "learning_rate": 0.0001, "loss": 0.0237, "step": 16820 }, { "epoch": 128.4732824427481, "grad_norm": 1.9049394130706787, "learning_rate": 0.0001, "loss": 0.0237, "step": 16830 }, { "epoch": 128.54961832061068, "grad_norm": 1.6251397132873535, "learning_rate": 0.0001, "loss": 0.0238, "step": 16840 }, { "epoch": 128.6259541984733, "grad_norm": 2.102013111114502, "learning_rate": 0.0001, "loss": 0.0238, "step": 16850 }, { "epoch": 128.70229007633588, "grad_norm": 1.7725673913955688, "learning_rate": 0.0001, "loss": 0.0236, "step": 16860 }, { "epoch": 128.77862595419847, "grad_norm": 2.0830681324005127, "learning_rate": 0.0001, "loss": 0.0246, "step": 16870 }, { "epoch": 128.85496183206106, "grad_norm": 2.3509674072265625, "learning_rate": 0.0001, "loss": 0.0254, "step": 16880 }, { "epoch": 128.93129770992365, "grad_norm": 1.8996829986572266, "learning_rate": 0.0001, "loss": 0.0251, "step": 16890 }, { "epoch": 129.00763358778627, "grad_norm": 2.4166884422302246, "learning_rate": 0.0001, "loss": 0.0246, "step": 16900 }, { "epoch": 129.08396946564886, "grad_norm": 2.275219678878784, "learning_rate": 0.0001, "loss": 0.0251, "step": 16910 }, { "epoch": 129.16030534351145, "grad_norm": 1.9694429636001587, "learning_rate": 0.0001, "loss": 0.0245, "step": 16920 }, { "epoch": 129.23664122137404, "grad_norm": 2.083876371383667, "learning_rate": 0.0001, "loss": 0.0251, "step": 16930 }, { "epoch": 129.31297709923663, "grad_norm": 2.1330416202545166, "learning_rate": 0.0001, "loss": 0.0254, "step": 16940 }, { "epoch": 129.38931297709922, "grad_norm": 2.171612024307251, "learning_rate": 0.0001, "loss": 0.0246, "step": 16950 }, { "epoch": 129.46564885496184, "grad_norm": 1.9905825853347778, "learning_rate": 0.0001, "loss": 0.0244, "step": 16960 }, { "epoch": 129.54198473282443, "grad_norm": 2.5902140140533447, "learning_rate": 0.0001, "loss": 0.0243, "step": 16970 }, { "epoch": 129.61832061068702, "grad_norm": 2.9320695400238037, "learning_rate": 0.0001, "loss": 0.0248, "step": 16980 }, { "epoch": 129.6946564885496, "grad_norm": 3.1223666667938232, "learning_rate": 0.0001, "loss": 0.0241, "step": 16990 }, { "epoch": 129.7709923664122, "grad_norm": 1.484989047050476, "learning_rate": 0.0001, "loss": 0.023, "step": 17000 }, { "epoch": 129.84732824427482, "grad_norm": 2.4951813220977783, "learning_rate": 0.0001, "loss": 0.0238, "step": 17010 }, { "epoch": 129.9236641221374, "grad_norm": 1.5903056859970093, "learning_rate": 0.0001, "loss": 0.0233, "step": 17020 }, { "epoch": 130.0, "grad_norm": 2.065276861190796, "learning_rate": 0.0001, "loss": 0.0244, "step": 17030 }, { "epoch": 130.0763358778626, "grad_norm": 1.9295934438705444, "learning_rate": 0.0001, "loss": 0.0235, "step": 17040 }, { "epoch": 130.15267175572518, "grad_norm": 1.8086984157562256, "learning_rate": 0.0001, "loss": 0.0233, "step": 17050 }, { "epoch": 130.2290076335878, "grad_norm": 2.5908076763153076, "learning_rate": 0.0001, "loss": 0.024, "step": 17060 }, { "epoch": 130.3053435114504, "grad_norm": 1.9548143148422241, "learning_rate": 0.0001, "loss": 0.0241, "step": 17070 }, { "epoch": 130.38167938931298, "grad_norm": 1.7045865058898926, "learning_rate": 0.0001, "loss": 0.0236, "step": 17080 }, { "epoch": 130.45801526717557, "grad_norm": 2.5864062309265137, "learning_rate": 0.0001, "loss": 0.024, "step": 17090 }, { "epoch": 130.53435114503816, "grad_norm": 2.2013890743255615, "learning_rate": 0.0001, "loss": 0.0244, "step": 17100 }, { "epoch": 130.61068702290078, "grad_norm": 1.695995569229126, "learning_rate": 0.0001, "loss": 0.0254, "step": 17110 }, { "epoch": 130.68702290076337, "grad_norm": 1.823116660118103, "learning_rate": 0.0001, "loss": 0.0246, "step": 17120 }, { "epoch": 130.76335877862596, "grad_norm": 1.2603741884231567, "learning_rate": 0.0001, "loss": 0.0238, "step": 17130 }, { "epoch": 130.83969465648855, "grad_norm": 1.7719241380691528, "learning_rate": 0.0001, "loss": 0.0243, "step": 17140 }, { "epoch": 130.91603053435114, "grad_norm": 1.829355239868164, "learning_rate": 0.0001, "loss": 0.0249, "step": 17150 }, { "epoch": 130.99236641221373, "grad_norm": 2.0080044269561768, "learning_rate": 0.0001, "loss": 0.0251, "step": 17160 }, { "epoch": 131.06870229007635, "grad_norm": 1.8260619640350342, "learning_rate": 0.0001, "loss": 0.0246, "step": 17170 }, { "epoch": 131.14503816793894, "grad_norm": 1.9120230674743652, "learning_rate": 0.0001, "loss": 0.0254, "step": 17180 }, { "epoch": 131.22137404580153, "grad_norm": 1.7523702383041382, "learning_rate": 0.0001, "loss": 0.0254, "step": 17190 }, { "epoch": 131.29770992366412, "grad_norm": 1.9870518445968628, "learning_rate": 0.0001, "loss": 0.0265, "step": 17200 }, { "epoch": 131.3740458015267, "grad_norm": 1.526145100593567, "learning_rate": 0.0001, "loss": 0.026, "step": 17210 }, { "epoch": 131.45038167938932, "grad_norm": 2.649848222732544, "learning_rate": 0.0001, "loss": 0.0248, "step": 17220 }, { "epoch": 131.5267175572519, "grad_norm": 2.544611692428589, "learning_rate": 0.0001, "loss": 0.0258, "step": 17230 }, { "epoch": 131.6030534351145, "grad_norm": 1.7605948448181152, "learning_rate": 0.0001, "loss": 0.0258, "step": 17240 }, { "epoch": 131.6793893129771, "grad_norm": 2.0810811519622803, "learning_rate": 0.0001, "loss": 0.0248, "step": 17250 }, { "epoch": 131.75572519083968, "grad_norm": 1.4834784269332886, "learning_rate": 0.0001, "loss": 0.0246, "step": 17260 }, { "epoch": 131.8320610687023, "grad_norm": 2.631695508956909, "learning_rate": 0.0001, "loss": 0.0243, "step": 17270 }, { "epoch": 131.9083969465649, "grad_norm": 2.364500045776367, "learning_rate": 0.0001, "loss": 0.0241, "step": 17280 }, { "epoch": 131.98473282442748, "grad_norm": 1.5901628732681274, "learning_rate": 0.0001, "loss": 0.0243, "step": 17290 }, { "epoch": 132.06106870229007, "grad_norm": 1.991389513015747, "learning_rate": 0.0001, "loss": 0.0239, "step": 17300 }, { "epoch": 132.13740458015266, "grad_norm": 2.2480506896972656, "learning_rate": 0.0001, "loss": 0.0239, "step": 17310 }, { "epoch": 132.21374045801528, "grad_norm": 1.7686328887939453, "learning_rate": 0.0001, "loss": 0.0235, "step": 17320 }, { "epoch": 132.29007633587787, "grad_norm": 1.5501338243484497, "learning_rate": 0.0001, "loss": 0.0238, "step": 17330 }, { "epoch": 132.36641221374046, "grad_norm": 1.8876698017120361, "learning_rate": 0.0001, "loss": 0.0232, "step": 17340 }, { "epoch": 132.44274809160305, "grad_norm": 1.4958032369613647, "learning_rate": 0.0001, "loss": 0.0247, "step": 17350 }, { "epoch": 132.51908396946564, "grad_norm": 1.585284948348999, "learning_rate": 0.0001, "loss": 0.0247, "step": 17360 }, { "epoch": 132.59541984732823, "grad_norm": 1.8021756410598755, "learning_rate": 0.0001, "loss": 0.0244, "step": 17370 }, { "epoch": 132.67175572519085, "grad_norm": 2.165189743041992, "learning_rate": 0.0001, "loss": 0.0252, "step": 17380 }, { "epoch": 132.74809160305344, "grad_norm": 1.8125700950622559, "learning_rate": 0.0001, "loss": 0.0244, "step": 17390 }, { "epoch": 132.82442748091603, "grad_norm": 2.3665685653686523, "learning_rate": 0.0001, "loss": 0.0241, "step": 17400 }, { "epoch": 132.90076335877862, "grad_norm": 1.7300716638565063, "learning_rate": 0.0001, "loss": 0.024, "step": 17410 }, { "epoch": 132.9770992366412, "grad_norm": 1.4007598161697388, "learning_rate": 0.0001, "loss": 0.0236, "step": 17420 }, { "epoch": 133.05343511450383, "grad_norm": 1.8414175510406494, "learning_rate": 0.0001, "loss": 0.0236, "step": 17430 }, { "epoch": 133.12977099236642, "grad_norm": 2.072113037109375, "learning_rate": 0.0001, "loss": 0.0233, "step": 17440 }, { "epoch": 133.206106870229, "grad_norm": 2.293640613555908, "learning_rate": 0.0001, "loss": 0.0236, "step": 17450 }, { "epoch": 133.2824427480916, "grad_norm": 1.919061303138733, "learning_rate": 0.0001, "loss": 0.0237, "step": 17460 }, { "epoch": 133.3587786259542, "grad_norm": 1.7010351419448853, "learning_rate": 0.0001, "loss": 0.024, "step": 17470 }, { "epoch": 133.4351145038168, "grad_norm": 1.7262523174285889, "learning_rate": 0.0001, "loss": 0.0241, "step": 17480 }, { "epoch": 133.5114503816794, "grad_norm": 1.272637128829956, "learning_rate": 0.0001, "loss": 0.0243, "step": 17490 }, { "epoch": 133.58778625954199, "grad_norm": 1.584294319152832, "learning_rate": 0.0001, "loss": 0.0251, "step": 17500 }, { "epoch": 133.66412213740458, "grad_norm": 1.6738758087158203, "learning_rate": 0.0001, "loss": 0.0253, "step": 17510 }, { "epoch": 133.74045801526717, "grad_norm": 1.7234468460083008, "learning_rate": 0.0001, "loss": 0.0257, "step": 17520 }, { "epoch": 133.81679389312978, "grad_norm": 1.6638462543487549, "learning_rate": 0.0001, "loss": 0.0251, "step": 17530 }, { "epoch": 133.89312977099237, "grad_norm": 2.389390468597412, "learning_rate": 0.0001, "loss": 0.0249, "step": 17540 }, { "epoch": 133.96946564885496, "grad_norm": 2.2606136798858643, "learning_rate": 0.0001, "loss": 0.0255, "step": 17550 }, { "epoch": 134.04580152671755, "grad_norm": 1.6278581619262695, "learning_rate": 0.0001, "loss": 0.0242, "step": 17560 }, { "epoch": 134.12213740458014, "grad_norm": 2.1357390880584717, "learning_rate": 0.0001, "loss": 0.0252, "step": 17570 }, { "epoch": 134.19847328244273, "grad_norm": 1.7148653268814087, "learning_rate": 0.0001, "loss": 0.0254, "step": 17580 }, { "epoch": 134.27480916030535, "grad_norm": 1.7333632707595825, "learning_rate": 0.0001, "loss": 0.0248, "step": 17590 }, { "epoch": 134.35114503816794, "grad_norm": 1.7834539413452148, "learning_rate": 0.0001, "loss": 0.0246, "step": 17600 }, { "epoch": 134.42748091603053, "grad_norm": 1.8663445711135864, "learning_rate": 0.0001, "loss": 0.0239, "step": 17610 }, { "epoch": 134.50381679389312, "grad_norm": 1.8901255130767822, "learning_rate": 0.0001, "loss": 0.0243, "step": 17620 }, { "epoch": 134.5801526717557, "grad_norm": 2.0247488021850586, "learning_rate": 0.0001, "loss": 0.0241, "step": 17630 }, { "epoch": 134.65648854961833, "grad_norm": 1.8260968923568726, "learning_rate": 0.0001, "loss": 0.0246, "step": 17640 }, { "epoch": 134.73282442748092, "grad_norm": 1.5347256660461426, "learning_rate": 0.0001, "loss": 0.0247, "step": 17650 }, { "epoch": 134.8091603053435, "grad_norm": 2.090719223022461, "learning_rate": 0.0001, "loss": 0.0256, "step": 17660 }, { "epoch": 134.8854961832061, "grad_norm": 1.607559084892273, "learning_rate": 0.0001, "loss": 0.0256, "step": 17670 }, { "epoch": 134.9618320610687, "grad_norm": 1.590576171875, "learning_rate": 0.0001, "loss": 0.0259, "step": 17680 }, { "epoch": 135.0381679389313, "grad_norm": 1.8677852153778076, "learning_rate": 0.0001, "loss": 0.0255, "step": 17690 }, { "epoch": 135.1145038167939, "grad_norm": 2.371096611022949, "learning_rate": 0.0001, "loss": 0.0276, "step": 17700 }, { "epoch": 135.1908396946565, "grad_norm": 2.40842342376709, "learning_rate": 0.0001, "loss": 0.0287, "step": 17710 }, { "epoch": 135.26717557251908, "grad_norm": 1.9158827066421509, "learning_rate": 0.0001, "loss": 0.0276, "step": 17720 }, { "epoch": 135.34351145038167, "grad_norm": 1.8305095434188843, "learning_rate": 0.0001, "loss": 0.0265, "step": 17730 }, { "epoch": 135.4198473282443, "grad_norm": 1.6170059442520142, "learning_rate": 0.0001, "loss": 0.0263, "step": 17740 }, { "epoch": 135.49618320610688, "grad_norm": 1.7662345170974731, "learning_rate": 0.0001, "loss": 0.0255, "step": 17750 }, { "epoch": 135.57251908396947, "grad_norm": 1.624845027923584, "learning_rate": 0.0001, "loss": 0.025, "step": 17760 }, { "epoch": 135.64885496183206, "grad_norm": 1.8932509422302246, "learning_rate": 0.0001, "loss": 0.0253, "step": 17770 }, { "epoch": 135.72519083969465, "grad_norm": 1.7705029249191284, "learning_rate": 0.0001, "loss": 0.0255, "step": 17780 }, { "epoch": 135.80152671755727, "grad_norm": 1.4806798696517944, "learning_rate": 0.0001, "loss": 0.0257, "step": 17790 }, { "epoch": 135.87786259541986, "grad_norm": 1.3202948570251465, "learning_rate": 0.0001, "loss": 0.0245, "step": 17800 }, { "epoch": 135.95419847328245, "grad_norm": 1.7669066190719604, "learning_rate": 0.0001, "loss": 0.0256, "step": 17810 }, { "epoch": 136.03053435114504, "grad_norm": 1.6365677118301392, "learning_rate": 0.0001, "loss": 0.0264, "step": 17820 }, { "epoch": 136.10687022900763, "grad_norm": 1.6235955953598022, "learning_rate": 0.0001, "loss": 0.026, "step": 17830 }, { "epoch": 136.18320610687022, "grad_norm": 2.0492846965789795, "learning_rate": 0.0001, "loss": 0.0261, "step": 17840 }, { "epoch": 136.25954198473283, "grad_norm": 1.6765235662460327, "learning_rate": 0.0001, "loss": 0.0257, "step": 17850 }, { "epoch": 136.33587786259542, "grad_norm": 1.84732186794281, "learning_rate": 0.0001, "loss": 0.0262, "step": 17860 }, { "epoch": 136.41221374045801, "grad_norm": 1.4112446308135986, "learning_rate": 0.0001, "loss": 0.0261, "step": 17870 }, { "epoch": 136.4885496183206, "grad_norm": 2.3396568298339844, "learning_rate": 0.0001, "loss": 0.0255, "step": 17880 }, { "epoch": 136.5648854961832, "grad_norm": 2.145880937576294, "learning_rate": 0.0001, "loss": 0.0266, "step": 17890 }, { "epoch": 136.6412213740458, "grad_norm": 1.7020949125289917, "learning_rate": 0.0001, "loss": 0.0267, "step": 17900 }, { "epoch": 136.7175572519084, "grad_norm": 2.1749041080474854, "learning_rate": 0.0001, "loss": 0.0263, "step": 17910 }, { "epoch": 136.793893129771, "grad_norm": 2.5275936126708984, "learning_rate": 0.0001, "loss": 0.0245, "step": 17920 }, { "epoch": 136.87022900763358, "grad_norm": 1.486377477645874, "learning_rate": 0.0001, "loss": 0.0253, "step": 17930 }, { "epoch": 136.94656488549617, "grad_norm": 2.346461057662964, "learning_rate": 0.0001, "loss": 0.0252, "step": 17940 }, { "epoch": 137.0229007633588, "grad_norm": 1.5139367580413818, "learning_rate": 0.0001, "loss": 0.0244, "step": 17950 }, { "epoch": 137.09923664122138, "grad_norm": 1.8141592741012573, "learning_rate": 0.0001, "loss": 0.0236, "step": 17960 }, { "epoch": 137.17557251908397, "grad_norm": 2.1064908504486084, "learning_rate": 0.0001, "loss": 0.0238, "step": 17970 }, { "epoch": 137.25190839694656, "grad_norm": 1.8296928405761719, "learning_rate": 0.0001, "loss": 0.0245, "step": 17980 }, { "epoch": 137.32824427480915, "grad_norm": 1.5814824104309082, "learning_rate": 0.0001, "loss": 0.0237, "step": 17990 }, { "epoch": 137.40458015267177, "grad_norm": 1.6516484022140503, "learning_rate": 0.0001, "loss": 0.0232, "step": 18000 }, { "epoch": 137.48091603053436, "grad_norm": 1.9787766933441162, "learning_rate": 0.0001, "loss": 0.0233, "step": 18010 }, { "epoch": 137.55725190839695, "grad_norm": 2.011317253112793, "learning_rate": 0.0001, "loss": 0.0236, "step": 18020 }, { "epoch": 137.63358778625954, "grad_norm": 2.0006279945373535, "learning_rate": 0.0001, "loss": 0.0236, "step": 18030 }, { "epoch": 137.70992366412213, "grad_norm": 1.3485209941864014, "learning_rate": 0.0001, "loss": 0.0237, "step": 18040 }, { "epoch": 137.78625954198472, "grad_norm": 1.2767022848129272, "learning_rate": 0.0001, "loss": 0.0236, "step": 18050 }, { "epoch": 137.86259541984734, "grad_norm": 1.8262702226638794, "learning_rate": 0.0001, "loss": 0.0235, "step": 18060 }, { "epoch": 137.93893129770993, "grad_norm": 1.9936009645462036, "learning_rate": 0.0001, "loss": 0.0242, "step": 18070 }, { "epoch": 138.01526717557252, "grad_norm": 2.0218558311462402, "learning_rate": 0.0001, "loss": 0.0228, "step": 18080 }, { "epoch": 138.0916030534351, "grad_norm": 1.7554450035095215, "learning_rate": 0.0001, "loss": 0.0236, "step": 18090 }, { "epoch": 138.1679389312977, "grad_norm": 2.0641117095947266, "learning_rate": 0.0001, "loss": 0.0232, "step": 18100 }, { "epoch": 138.24427480916032, "grad_norm": 2.1536149978637695, "learning_rate": 0.0001, "loss": 0.0239, "step": 18110 }, { "epoch": 138.3206106870229, "grad_norm": 1.785131573677063, "learning_rate": 0.0001, "loss": 0.0235, "step": 18120 }, { "epoch": 138.3969465648855, "grad_norm": 2.4867613315582275, "learning_rate": 0.0001, "loss": 0.0241, "step": 18130 }, { "epoch": 138.4732824427481, "grad_norm": 2.1554479598999023, "learning_rate": 0.0001, "loss": 0.0227, "step": 18140 }, { "epoch": 138.54961832061068, "grad_norm": 2.1281261444091797, "learning_rate": 0.0001, "loss": 0.0229, "step": 18150 }, { "epoch": 138.6259541984733, "grad_norm": 2.0595333576202393, "learning_rate": 0.0001, "loss": 0.0231, "step": 18160 }, { "epoch": 138.70229007633588, "grad_norm": 1.8964797258377075, "learning_rate": 0.0001, "loss": 0.0244, "step": 18170 }, { "epoch": 138.77862595419847, "grad_norm": 1.81574285030365, "learning_rate": 0.0001, "loss": 0.0228, "step": 18180 }, { "epoch": 138.85496183206106, "grad_norm": 2.481729030609131, "learning_rate": 0.0001, "loss": 0.0229, "step": 18190 }, { "epoch": 138.93129770992365, "grad_norm": 1.9384708404541016, "learning_rate": 0.0001, "loss": 0.0227, "step": 18200 }, { "epoch": 139.00763358778627, "grad_norm": 1.975895643234253, "learning_rate": 0.0001, "loss": 0.0232, "step": 18210 }, { "epoch": 139.08396946564886, "grad_norm": 1.6916577816009521, "learning_rate": 0.0001, "loss": 0.0232, "step": 18220 }, { "epoch": 139.16030534351145, "grad_norm": 1.7056564092636108, "learning_rate": 0.0001, "loss": 0.0228, "step": 18230 }, { "epoch": 139.23664122137404, "grad_norm": 1.4677714109420776, "learning_rate": 0.0001, "loss": 0.0233, "step": 18240 }, { "epoch": 139.31297709923663, "grad_norm": 1.4529156684875488, "learning_rate": 0.0001, "loss": 0.0221, "step": 18250 }, { "epoch": 139.38931297709922, "grad_norm": 1.6169939041137695, "learning_rate": 0.0001, "loss": 0.0216, "step": 18260 }, { "epoch": 139.46564885496184, "grad_norm": 2.151128053665161, "learning_rate": 0.0001, "loss": 0.0226, "step": 18270 }, { "epoch": 139.54198473282443, "grad_norm": 2.070307731628418, "learning_rate": 0.0001, "loss": 0.0237, "step": 18280 }, { "epoch": 139.61832061068702, "grad_norm": 1.41827392578125, "learning_rate": 0.0001, "loss": 0.0229, "step": 18290 }, { "epoch": 139.6946564885496, "grad_norm": 1.7656997442245483, "learning_rate": 0.0001, "loss": 0.0226, "step": 18300 }, { "epoch": 139.7709923664122, "grad_norm": 2.19321870803833, "learning_rate": 0.0001, "loss": 0.0226, "step": 18310 }, { "epoch": 139.84732824427482, "grad_norm": 2.2900772094726562, "learning_rate": 0.0001, "loss": 0.0231, "step": 18320 }, { "epoch": 139.9236641221374, "grad_norm": 2.404597520828247, "learning_rate": 0.0001, "loss": 0.0229, "step": 18330 }, { "epoch": 140.0, "grad_norm": 1.861312985420227, "learning_rate": 0.0001, "loss": 0.0226, "step": 18340 }, { "epoch": 140.0763358778626, "grad_norm": 1.6206108331680298, "learning_rate": 0.0001, "loss": 0.0224, "step": 18350 }, { "epoch": 140.15267175572518, "grad_norm": 1.569796085357666, "learning_rate": 0.0001, "loss": 0.0228, "step": 18360 }, { "epoch": 140.2290076335878, "grad_norm": 2.4651591777801514, "learning_rate": 0.0001, "loss": 0.0218, "step": 18370 }, { "epoch": 140.3053435114504, "grad_norm": 2.3144779205322266, "learning_rate": 0.0001, "loss": 0.0226, "step": 18380 }, { "epoch": 140.38167938931298, "grad_norm": 1.748473048210144, "learning_rate": 0.0001, "loss": 0.023, "step": 18390 }, { "epoch": 140.45801526717557, "grad_norm": 1.8416175842285156, "learning_rate": 0.0001, "loss": 0.0225, "step": 18400 }, { "epoch": 140.53435114503816, "grad_norm": 1.7939825057983398, "learning_rate": 0.0001, "loss": 0.0217, "step": 18410 }, { "epoch": 140.61068702290078, "grad_norm": 1.251869559288025, "learning_rate": 0.0001, "loss": 0.022, "step": 18420 }, { "epoch": 140.68702290076337, "grad_norm": 1.7626636028289795, "learning_rate": 0.0001, "loss": 0.023, "step": 18430 }, { "epoch": 140.76335877862596, "grad_norm": 1.7751753330230713, "learning_rate": 0.0001, "loss": 0.0235, "step": 18440 }, { "epoch": 140.83969465648855, "grad_norm": 1.9806745052337646, "learning_rate": 0.0001, "loss": 0.0256, "step": 18450 }, { "epoch": 140.91603053435114, "grad_norm": 1.4961674213409424, "learning_rate": 0.0001, "loss": 0.0261, "step": 18460 }, { "epoch": 140.99236641221373, "grad_norm": 1.3152166604995728, "learning_rate": 0.0001, "loss": 0.0256, "step": 18470 }, { "epoch": 141.06870229007635, "grad_norm": 2.225908041000366, "learning_rate": 0.0001, "loss": 0.0238, "step": 18480 }, { "epoch": 141.14503816793894, "grad_norm": 1.5874313116073608, "learning_rate": 0.0001, "loss": 0.0251, "step": 18490 }, { "epoch": 141.22137404580153, "grad_norm": 2.2353591918945312, "learning_rate": 0.0001, "loss": 0.0247, "step": 18500 }, { "epoch": 141.29770992366412, "grad_norm": 1.7588859796524048, "learning_rate": 0.0001, "loss": 0.0251, "step": 18510 }, { "epoch": 141.3740458015267, "grad_norm": 1.6024439334869385, "learning_rate": 0.0001, "loss": 0.0248, "step": 18520 }, { "epoch": 141.45038167938932, "grad_norm": 1.6579993963241577, "learning_rate": 0.0001, "loss": 0.0236, "step": 18530 }, { "epoch": 141.5267175572519, "grad_norm": 1.7040019035339355, "learning_rate": 0.0001, "loss": 0.0239, "step": 18540 }, { "epoch": 141.6030534351145, "grad_norm": 2.2158310413360596, "learning_rate": 0.0001, "loss": 0.0248, "step": 18550 }, { "epoch": 141.6793893129771, "grad_norm": 1.6036994457244873, "learning_rate": 0.0001, "loss": 0.0247, "step": 18560 }, { "epoch": 141.75572519083968, "grad_norm": 2.465074300765991, "learning_rate": 0.0001, "loss": 0.0256, "step": 18570 }, { "epoch": 141.8320610687023, "grad_norm": 1.531445026397705, "learning_rate": 0.0001, "loss": 0.0236, "step": 18580 }, { "epoch": 141.9083969465649, "grad_norm": 2.5781092643737793, "learning_rate": 0.0001, "loss": 0.0227, "step": 18590 }, { "epoch": 141.98473282442748, "grad_norm": 2.609844207763672, "learning_rate": 0.0001, "loss": 0.0227, "step": 18600 }, { "epoch": 142.06106870229007, "grad_norm": 2.6921119689941406, "learning_rate": 0.0001, "loss": 0.0234, "step": 18610 }, { "epoch": 142.13740458015266, "grad_norm": 2.042236566543579, "learning_rate": 0.0001, "loss": 0.0233, "step": 18620 }, { "epoch": 142.21374045801528, "grad_norm": 2.372300386428833, "learning_rate": 0.0001, "loss": 0.0233, "step": 18630 }, { "epoch": 142.29007633587787, "grad_norm": 1.5678398609161377, "learning_rate": 0.0001, "loss": 0.024, "step": 18640 }, { "epoch": 142.36641221374046, "grad_norm": 1.4929572343826294, "learning_rate": 0.0001, "loss": 0.0239, "step": 18650 }, { "epoch": 142.44274809160305, "grad_norm": 1.4599945545196533, "learning_rate": 0.0001, "loss": 0.0245, "step": 18660 }, { "epoch": 142.51908396946564, "grad_norm": 1.8987780809402466, "learning_rate": 0.0001, "loss": 0.0256, "step": 18670 }, { "epoch": 142.59541984732823, "grad_norm": 1.9025704860687256, "learning_rate": 0.0001, "loss": 0.0256, "step": 18680 }, { "epoch": 142.67175572519085, "grad_norm": 1.4561814069747925, "learning_rate": 0.0001, "loss": 0.0242, "step": 18690 }, { "epoch": 142.74809160305344, "grad_norm": 1.54768967628479, "learning_rate": 0.0001, "loss": 0.0238, "step": 18700 }, { "epoch": 142.82442748091603, "grad_norm": 1.6330435276031494, "learning_rate": 0.0001, "loss": 0.0231, "step": 18710 }, { "epoch": 142.90076335877862, "grad_norm": 1.7944693565368652, "learning_rate": 0.0001, "loss": 0.0241, "step": 18720 }, { "epoch": 142.9770992366412, "grad_norm": 1.617865800857544, "learning_rate": 0.0001, "loss": 0.0238, "step": 18730 }, { "epoch": 143.05343511450383, "grad_norm": 1.6488045454025269, "learning_rate": 0.0001, "loss": 0.0248, "step": 18740 }, { "epoch": 143.12977099236642, "grad_norm": 1.973853588104248, "learning_rate": 0.0001, "loss": 0.0242, "step": 18750 }, { "epoch": 143.206106870229, "grad_norm": 1.874089241027832, "learning_rate": 0.0001, "loss": 0.0241, "step": 18760 }, { "epoch": 143.2824427480916, "grad_norm": 1.6558207273483276, "learning_rate": 0.0001, "loss": 0.0235, "step": 18770 }, { "epoch": 143.3587786259542, "grad_norm": 2.090569019317627, "learning_rate": 0.0001, "loss": 0.0237, "step": 18780 }, { "epoch": 143.4351145038168, "grad_norm": 1.7200157642364502, "learning_rate": 0.0001, "loss": 0.0234, "step": 18790 }, { "epoch": 143.5114503816794, "grad_norm": 1.8201185464859009, "learning_rate": 0.0001, "loss": 0.0239, "step": 18800 }, { "epoch": 143.58778625954199, "grad_norm": 1.6051697731018066, "learning_rate": 0.0001, "loss": 0.0232, "step": 18810 }, { "epoch": 143.66412213740458, "grad_norm": 1.7153327465057373, "learning_rate": 0.0001, "loss": 0.0238, "step": 18820 }, { "epoch": 143.74045801526717, "grad_norm": 2.321765422821045, "learning_rate": 0.0001, "loss": 0.0237, "step": 18830 }, { "epoch": 143.81679389312978, "grad_norm": 2.1289901733398438, "learning_rate": 0.0001, "loss": 0.0234, "step": 18840 }, { "epoch": 143.89312977099237, "grad_norm": 2.5311765670776367, "learning_rate": 0.0001, "loss": 0.0232, "step": 18850 }, { "epoch": 143.96946564885496, "grad_norm": 1.8864034414291382, "learning_rate": 0.0001, "loss": 0.0229, "step": 18860 }, { "epoch": 144.04580152671755, "grad_norm": 2.0097994804382324, "learning_rate": 0.0001, "loss": 0.0229, "step": 18870 }, { "epoch": 144.12213740458014, "grad_norm": 1.4302304983139038, "learning_rate": 0.0001, "loss": 0.0225, "step": 18880 }, { "epoch": 144.19847328244273, "grad_norm": 1.5591930150985718, "learning_rate": 0.0001, "loss": 0.0226, "step": 18890 }, { "epoch": 144.27480916030535, "grad_norm": 1.2349088191986084, "learning_rate": 0.0001, "loss": 0.0221, "step": 18900 }, { "epoch": 144.35114503816794, "grad_norm": 1.490890383720398, "learning_rate": 0.0001, "loss": 0.023, "step": 18910 }, { "epoch": 144.42748091603053, "grad_norm": 1.5644499063491821, "learning_rate": 0.0001, "loss": 0.0238, "step": 18920 }, { "epoch": 144.50381679389312, "grad_norm": 2.713182210922241, "learning_rate": 0.0001, "loss": 0.0248, "step": 18930 }, { "epoch": 144.5801526717557, "grad_norm": 2.4140241146087646, "learning_rate": 0.0001, "loss": 0.0237, "step": 18940 }, { "epoch": 144.65648854961833, "grad_norm": 2.142439126968384, "learning_rate": 0.0001, "loss": 0.0236, "step": 18950 }, { "epoch": 144.73282442748092, "grad_norm": 2.095541000366211, "learning_rate": 0.0001, "loss": 0.0229, "step": 18960 }, { "epoch": 144.8091603053435, "grad_norm": 2.766317129135132, "learning_rate": 0.0001, "loss": 0.0218, "step": 18970 }, { "epoch": 144.8854961832061, "grad_norm": 2.5355727672576904, "learning_rate": 0.0001, "loss": 0.0222, "step": 18980 }, { "epoch": 144.9618320610687, "grad_norm": 2.280672788619995, "learning_rate": 0.0001, "loss": 0.0227, "step": 18990 }, { "epoch": 145.0381679389313, "grad_norm": 2.8176050186157227, "learning_rate": 0.0001, "loss": 0.0221, "step": 19000 }, { "epoch": 145.1145038167939, "grad_norm": 1.4921622276306152, "learning_rate": 0.0001, "loss": 0.0224, "step": 19010 }, { "epoch": 145.1908396946565, "grad_norm": 1.6795213222503662, "learning_rate": 0.0001, "loss": 0.0221, "step": 19020 }, { "epoch": 145.26717557251908, "grad_norm": 1.7908555269241333, "learning_rate": 0.0001, "loss": 0.0224, "step": 19030 }, { "epoch": 145.34351145038167, "grad_norm": 1.4827083349227905, "learning_rate": 0.0001, "loss": 0.0224, "step": 19040 }, { "epoch": 145.4198473282443, "grad_norm": 1.8576819896697998, "learning_rate": 0.0001, "loss": 0.0223, "step": 19050 }, { "epoch": 145.49618320610688, "grad_norm": 1.5228997468948364, "learning_rate": 0.0001, "loss": 0.0226, "step": 19060 }, { "epoch": 145.57251908396947, "grad_norm": 1.5266523361206055, "learning_rate": 0.0001, "loss": 0.0229, "step": 19070 }, { "epoch": 145.64885496183206, "grad_norm": 1.584536075592041, "learning_rate": 0.0001, "loss": 0.0219, "step": 19080 }, { "epoch": 145.72519083969465, "grad_norm": 1.4066451787948608, "learning_rate": 0.0001, "loss": 0.0229, "step": 19090 }, { "epoch": 145.80152671755727, "grad_norm": 1.8911315202713013, "learning_rate": 0.0001, "loss": 0.0235, "step": 19100 }, { "epoch": 145.87786259541986, "grad_norm": 1.8923910856246948, "learning_rate": 0.0001, "loss": 0.024, "step": 19110 }, { "epoch": 145.95419847328245, "grad_norm": 1.7185736894607544, "learning_rate": 0.0001, "loss": 0.0238, "step": 19120 }, { "epoch": 146.03053435114504, "grad_norm": 2.319606065750122, "learning_rate": 0.0001, "loss": 0.0233, "step": 19130 }, { "epoch": 146.10687022900763, "grad_norm": 1.5207267999649048, "learning_rate": 0.0001, "loss": 0.0237, "step": 19140 }, { "epoch": 146.18320610687022, "grad_norm": 1.44071364402771, "learning_rate": 0.0001, "loss": 0.0234, "step": 19150 }, { "epoch": 146.25954198473283, "grad_norm": 1.5966678857803345, "learning_rate": 0.0001, "loss": 0.0228, "step": 19160 }, { "epoch": 146.33587786259542, "grad_norm": 1.7550181150436401, "learning_rate": 0.0001, "loss": 0.0234, "step": 19170 }, { "epoch": 146.41221374045801, "grad_norm": 2.190492630004883, "learning_rate": 0.0001, "loss": 0.0232, "step": 19180 }, { "epoch": 146.4885496183206, "grad_norm": 2.3966526985168457, "learning_rate": 0.0001, "loss": 0.0235, "step": 19190 }, { "epoch": 146.5648854961832, "grad_norm": 2.014775276184082, "learning_rate": 0.0001, "loss": 0.0224, "step": 19200 }, { "epoch": 146.6412213740458, "grad_norm": 2.3381433486938477, "learning_rate": 0.0001, "loss": 0.0235, "step": 19210 }, { "epoch": 146.7175572519084, "grad_norm": 1.4027516841888428, "learning_rate": 0.0001, "loss": 0.0227, "step": 19220 }, { "epoch": 146.793893129771, "grad_norm": 1.4419931173324585, "learning_rate": 0.0001, "loss": 0.0225, "step": 19230 }, { "epoch": 146.87022900763358, "grad_norm": 1.6612884998321533, "learning_rate": 0.0001, "loss": 0.0232, "step": 19240 }, { "epoch": 146.94656488549617, "grad_norm": 2.634620189666748, "learning_rate": 0.0001, "loss": 0.0233, "step": 19250 }, { "epoch": 147.0229007633588, "grad_norm": 2.019786834716797, "learning_rate": 0.0001, "loss": 0.0231, "step": 19260 }, { "epoch": 147.09923664122138, "grad_norm": 1.9124294519424438, "learning_rate": 0.0001, "loss": 0.0231, "step": 19270 }, { "epoch": 147.17557251908397, "grad_norm": 1.583674669265747, "learning_rate": 0.0001, "loss": 0.0237, "step": 19280 }, { "epoch": 147.25190839694656, "grad_norm": 2.7291452884674072, "learning_rate": 0.0001, "loss": 0.024, "step": 19290 }, { "epoch": 147.32824427480915, "grad_norm": 1.8250149488449097, "learning_rate": 0.0001, "loss": 0.0224, "step": 19300 }, { "epoch": 147.40458015267177, "grad_norm": 1.7534140348434448, "learning_rate": 0.0001, "loss": 0.0226, "step": 19310 }, { "epoch": 147.48091603053436, "grad_norm": 1.769402265548706, "learning_rate": 0.0001, "loss": 0.0231, "step": 19320 }, { "epoch": 147.55725190839695, "grad_norm": 1.4815322160720825, "learning_rate": 0.0001, "loss": 0.0225, "step": 19330 }, { "epoch": 147.63358778625954, "grad_norm": 2.247523307800293, "learning_rate": 0.0001, "loss": 0.0222, "step": 19340 }, { "epoch": 147.70992366412213, "grad_norm": 1.6464420557022095, "learning_rate": 0.0001, "loss": 0.0226, "step": 19350 }, { "epoch": 147.78625954198472, "grad_norm": 1.5817058086395264, "learning_rate": 0.0001, "loss": 0.0231, "step": 19360 }, { "epoch": 147.86259541984734, "grad_norm": 1.8207142353057861, "learning_rate": 0.0001, "loss": 0.0224, "step": 19370 }, { "epoch": 147.93893129770993, "grad_norm": 2.1463236808776855, "learning_rate": 0.0001, "loss": 0.0232, "step": 19380 }, { "epoch": 148.01526717557252, "grad_norm": 2.0860211849212646, "learning_rate": 0.0001, "loss": 0.0223, "step": 19390 }, { "epoch": 148.0916030534351, "grad_norm": 1.7384068965911865, "learning_rate": 0.0001, "loss": 0.0233, "step": 19400 }, { "epoch": 148.1679389312977, "grad_norm": 1.804158329963684, "learning_rate": 0.0001, "loss": 0.0227, "step": 19410 }, { "epoch": 148.24427480916032, "grad_norm": 1.8705806732177734, "learning_rate": 0.0001, "loss": 0.0232, "step": 19420 }, { "epoch": 148.3206106870229, "grad_norm": 1.7086182832717896, "learning_rate": 0.0001, "loss": 0.0228, "step": 19430 }, { "epoch": 148.3969465648855, "grad_norm": 1.4622129201889038, "learning_rate": 0.0001, "loss": 0.0219, "step": 19440 }, { "epoch": 148.4732824427481, "grad_norm": 1.8405702114105225, "learning_rate": 0.0001, "loss": 0.0223, "step": 19450 }, { "epoch": 148.54961832061068, "grad_norm": 1.5342963933944702, "learning_rate": 0.0001, "loss": 0.0234, "step": 19460 }, { "epoch": 148.6259541984733, "grad_norm": 1.7189654111862183, "learning_rate": 0.0001, "loss": 0.0229, "step": 19470 }, { "epoch": 148.70229007633588, "grad_norm": 2.134758949279785, "learning_rate": 0.0001, "loss": 0.0236, "step": 19480 }, { "epoch": 148.77862595419847, "grad_norm": 2.195810317993164, "learning_rate": 0.0001, "loss": 0.0228, "step": 19490 }, { "epoch": 148.85496183206106, "grad_norm": 1.589704155921936, "learning_rate": 0.0001, "loss": 0.0224, "step": 19500 }, { "epoch": 148.93129770992365, "grad_norm": 1.4758087396621704, "learning_rate": 0.0001, "loss": 0.022, "step": 19510 }, { "epoch": 149.00763358778627, "grad_norm": 2.2490713596343994, "learning_rate": 0.0001, "loss": 0.0237, "step": 19520 }, { "epoch": 149.08396946564886, "grad_norm": 1.9370534420013428, "learning_rate": 0.0001, "loss": 0.022, "step": 19530 }, { "epoch": 149.16030534351145, "grad_norm": 2.062602996826172, "learning_rate": 0.0001, "loss": 0.0222, "step": 19540 }, { "epoch": 149.23664122137404, "grad_norm": 2.3263800144195557, "learning_rate": 0.0001, "loss": 0.0232, "step": 19550 }, { "epoch": 149.31297709923663, "grad_norm": 2.0333786010742188, "learning_rate": 0.0001, "loss": 0.0232, "step": 19560 }, { "epoch": 149.38931297709922, "grad_norm": 1.6431976556777954, "learning_rate": 0.0001, "loss": 0.0231, "step": 19570 }, { "epoch": 149.46564885496184, "grad_norm": 1.6877412796020508, "learning_rate": 0.0001, "loss": 0.023, "step": 19580 }, { "epoch": 149.54198473282443, "grad_norm": 1.9111511707305908, "learning_rate": 0.0001, "loss": 0.0234, "step": 19590 }, { "epoch": 149.61832061068702, "grad_norm": 1.923061728477478, "learning_rate": 0.0001, "loss": 0.0239, "step": 19600 }, { "epoch": 149.6946564885496, "grad_norm": 2.1813807487487793, "learning_rate": 0.0001, "loss": 0.0239, "step": 19610 }, { "epoch": 149.7709923664122, "grad_norm": 1.7091639041900635, "learning_rate": 0.0001, "loss": 0.0244, "step": 19620 }, { "epoch": 149.84732824427482, "grad_norm": 2.0543103218078613, "learning_rate": 0.0001, "loss": 0.0236, "step": 19630 }, { "epoch": 149.9236641221374, "grad_norm": 1.8037424087524414, "learning_rate": 0.0001, "loss": 0.0237, "step": 19640 }, { "epoch": 150.0, "grad_norm": 2.0387673377990723, "learning_rate": 0.0001, "loss": 0.0225, "step": 19650 }, { "epoch": 150.0763358778626, "grad_norm": 2.0109243392944336, "learning_rate": 0.0001, "loss": 0.0232, "step": 19660 }, { "epoch": 150.15267175572518, "grad_norm": 2.4625983238220215, "learning_rate": 0.0001, "loss": 0.0228, "step": 19670 }, { "epoch": 150.2290076335878, "grad_norm": 1.5297874212265015, "learning_rate": 0.0001, "loss": 0.0219, "step": 19680 }, { "epoch": 150.3053435114504, "grad_norm": 2.23103404045105, "learning_rate": 0.0001, "loss": 0.0212, "step": 19690 }, { "epoch": 150.38167938931298, "grad_norm": 1.5405712127685547, "learning_rate": 0.0001, "loss": 0.0224, "step": 19700 }, { "epoch": 150.45801526717557, "grad_norm": 1.3857311010360718, "learning_rate": 0.0001, "loss": 0.0222, "step": 19710 }, { "epoch": 150.53435114503816, "grad_norm": 1.9597268104553223, "learning_rate": 0.0001, "loss": 0.0229, "step": 19720 }, { "epoch": 150.61068702290078, "grad_norm": 1.9449282884597778, "learning_rate": 0.0001, "loss": 0.0228, "step": 19730 }, { "epoch": 150.68702290076337, "grad_norm": 2.206061840057373, "learning_rate": 0.0001, "loss": 0.0229, "step": 19740 }, { "epoch": 150.76335877862596, "grad_norm": 1.9930037260055542, "learning_rate": 0.0001, "loss": 0.023, "step": 19750 }, { "epoch": 150.83969465648855, "grad_norm": 1.4824227094650269, "learning_rate": 0.0001, "loss": 0.0221, "step": 19760 }, { "epoch": 150.91603053435114, "grad_norm": 1.5950818061828613, "learning_rate": 0.0001, "loss": 0.0224, "step": 19770 }, { "epoch": 150.99236641221373, "grad_norm": 1.991788625717163, "learning_rate": 0.0001, "loss": 0.0218, "step": 19780 }, { "epoch": 151.06870229007635, "grad_norm": 1.7095696926116943, "learning_rate": 0.0001, "loss": 0.0232, "step": 19790 }, { "epoch": 151.14503816793894, "grad_norm": 1.3337736129760742, "learning_rate": 0.0001, "loss": 0.0226, "step": 19800 }, { "epoch": 151.22137404580153, "grad_norm": 2.1689517498016357, "learning_rate": 0.0001, "loss": 0.024, "step": 19810 }, { "epoch": 151.29770992366412, "grad_norm": 1.7319650650024414, "learning_rate": 0.0001, "loss": 0.0238, "step": 19820 }, { "epoch": 151.3740458015267, "grad_norm": 1.8656560182571411, "learning_rate": 0.0001, "loss": 0.0232, "step": 19830 }, { "epoch": 151.45038167938932, "grad_norm": 1.8544877767562866, "learning_rate": 0.0001, "loss": 0.0225, "step": 19840 }, { "epoch": 151.5267175572519, "grad_norm": 2.464477300643921, "learning_rate": 0.0001, "loss": 0.0235, "step": 19850 }, { "epoch": 151.6030534351145, "grad_norm": 2.3663899898529053, "learning_rate": 0.0001, "loss": 0.0229, "step": 19860 }, { "epoch": 151.6793893129771, "grad_norm": 2.028512477874756, "learning_rate": 0.0001, "loss": 0.0228, "step": 19870 }, { "epoch": 151.75572519083968, "grad_norm": 1.5044915676116943, "learning_rate": 0.0001, "loss": 0.0228, "step": 19880 }, { "epoch": 151.8320610687023, "grad_norm": 2.248030424118042, "learning_rate": 0.0001, "loss": 0.0228, "step": 19890 }, { "epoch": 151.9083969465649, "grad_norm": 2.7113351821899414, "learning_rate": 0.0001, "loss": 0.0219, "step": 19900 }, { "epoch": 151.98473282442748, "grad_norm": 1.9502358436584473, "learning_rate": 0.0001, "loss": 0.0236, "step": 19910 }, { "epoch": 152.06106870229007, "grad_norm": 1.5177921056747437, "learning_rate": 0.0001, "loss": 0.0216, "step": 19920 }, { "epoch": 152.13740458015266, "grad_norm": 1.3092845678329468, "learning_rate": 0.0001, "loss": 0.0217, "step": 19930 }, { "epoch": 152.21374045801528, "grad_norm": 1.7648773193359375, "learning_rate": 0.0001, "loss": 0.022, "step": 19940 }, { "epoch": 152.29007633587787, "grad_norm": 1.5612967014312744, "learning_rate": 0.0001, "loss": 0.0228, "step": 19950 }, { "epoch": 152.36641221374046, "grad_norm": 1.858564853668213, "learning_rate": 0.0001, "loss": 0.023, "step": 19960 }, { "epoch": 152.44274809160305, "grad_norm": 1.4415937662124634, "learning_rate": 0.0001, "loss": 0.0223, "step": 19970 }, { "epoch": 152.51908396946564, "grad_norm": 1.1106112003326416, "learning_rate": 0.0001, "loss": 0.0218, "step": 19980 }, { "epoch": 152.59541984732823, "grad_norm": 1.3994282484054565, "learning_rate": 0.0001, "loss": 0.023, "step": 19990 }, { "epoch": 152.67175572519085, "grad_norm": 2.033216714859009, "learning_rate": 0.0001, "loss": 0.0241, "step": 20000 }, { "epoch": 152.74809160305344, "grad_norm": 2.6634275913238525, "learning_rate": 0.0001, "loss": 0.0239, "step": 20010 }, { "epoch": 152.82442748091603, "grad_norm": 2.095069169998169, "learning_rate": 0.0001, "loss": 0.0234, "step": 20020 }, { "epoch": 152.90076335877862, "grad_norm": 3.5900886058807373, "learning_rate": 0.0001, "loss": 0.023, "step": 20030 }, { "epoch": 152.9770992366412, "grad_norm": 2.7166759967803955, "learning_rate": 0.0001, "loss": 0.0228, "step": 20040 }, { "epoch": 153.05343511450383, "grad_norm": 1.9750175476074219, "learning_rate": 0.0001, "loss": 0.022, "step": 20050 }, { "epoch": 153.12977099236642, "grad_norm": 1.8120908737182617, "learning_rate": 0.0001, "loss": 0.021, "step": 20060 }, { "epoch": 153.206106870229, "grad_norm": 1.7338941097259521, "learning_rate": 0.0001, "loss": 0.0212, "step": 20070 }, { "epoch": 153.2824427480916, "grad_norm": 2.0505688190460205, "learning_rate": 0.0001, "loss": 0.0216, "step": 20080 }, { "epoch": 153.3587786259542, "grad_norm": 1.9430211782455444, "learning_rate": 0.0001, "loss": 0.0227, "step": 20090 }, { "epoch": 153.4351145038168, "grad_norm": 2.024275779724121, "learning_rate": 0.0001, "loss": 0.0243, "step": 20100 }, { "epoch": 153.5114503816794, "grad_norm": 3.195266008377075, "learning_rate": 0.0001, "loss": 0.0233, "step": 20110 }, { "epoch": 153.58778625954199, "grad_norm": 3.5222599506378174, "learning_rate": 0.0001, "loss": 0.0254, "step": 20120 }, { "epoch": 153.66412213740458, "grad_norm": 2.4723639488220215, "learning_rate": 0.0001, "loss": 0.0237, "step": 20130 }, { "epoch": 153.74045801526717, "grad_norm": 3.374467134475708, "learning_rate": 0.0001, "loss": 0.0228, "step": 20140 }, { "epoch": 153.81679389312978, "grad_norm": 2.180366277694702, "learning_rate": 0.0001, "loss": 0.0222, "step": 20150 }, { "epoch": 153.89312977099237, "grad_norm": 2.339445114135742, "learning_rate": 0.0001, "loss": 0.0211, "step": 20160 }, { "epoch": 153.96946564885496, "grad_norm": 2.1860435009002686, "learning_rate": 0.0001, "loss": 0.0214, "step": 20170 }, { "epoch": 154.04580152671755, "grad_norm": 2.1641323566436768, "learning_rate": 0.0001, "loss": 0.0205, "step": 20180 }, { "epoch": 154.12213740458014, "grad_norm": 1.9945554733276367, "learning_rate": 0.0001, "loss": 0.0203, "step": 20190 }, { "epoch": 154.19847328244273, "grad_norm": 2.02128267288208, "learning_rate": 0.0001, "loss": 0.0217, "step": 20200 }, { "epoch": 154.27480916030535, "grad_norm": 1.706535816192627, "learning_rate": 0.0001, "loss": 0.021, "step": 20210 }, { "epoch": 154.35114503816794, "grad_norm": 1.7006375789642334, "learning_rate": 0.0001, "loss": 0.0214, "step": 20220 }, { "epoch": 154.42748091603053, "grad_norm": 1.3947381973266602, "learning_rate": 0.0001, "loss": 0.0215, "step": 20230 }, { "epoch": 154.50381679389312, "grad_norm": 1.7772727012634277, "learning_rate": 0.0001, "loss": 0.0218, "step": 20240 }, { "epoch": 154.5801526717557, "grad_norm": 1.6285043954849243, "learning_rate": 0.0001, "loss": 0.0221, "step": 20250 }, { "epoch": 154.65648854961833, "grad_norm": 1.5370588302612305, "learning_rate": 0.0001, "loss": 0.0225, "step": 20260 }, { "epoch": 154.73282442748092, "grad_norm": 1.2778910398483276, "learning_rate": 0.0001, "loss": 0.0218, "step": 20270 }, { "epoch": 154.8091603053435, "grad_norm": 1.7086045742034912, "learning_rate": 0.0001, "loss": 0.0221, "step": 20280 }, { "epoch": 154.8854961832061, "grad_norm": 2.4366657733917236, "learning_rate": 0.0001, "loss": 0.0226, "step": 20290 }, { "epoch": 154.9618320610687, "grad_norm": 1.679747223854065, "learning_rate": 0.0001, "loss": 0.0224, "step": 20300 }, { "epoch": 155.0381679389313, "grad_norm": 1.9735270738601685, "learning_rate": 0.0001, "loss": 0.0236, "step": 20310 }, { "epoch": 155.1145038167939, "grad_norm": 2.4519248008728027, "learning_rate": 0.0001, "loss": 0.0222, "step": 20320 }, { "epoch": 155.1908396946565, "grad_norm": 1.7672796249389648, "learning_rate": 0.0001, "loss": 0.0225, "step": 20330 }, { "epoch": 155.26717557251908, "grad_norm": 1.96281099319458, "learning_rate": 0.0001, "loss": 0.0218, "step": 20340 }, { "epoch": 155.34351145038167, "grad_norm": 1.212877631187439, "learning_rate": 0.0001, "loss": 0.0224, "step": 20350 }, { "epoch": 155.4198473282443, "grad_norm": 1.6876603364944458, "learning_rate": 0.0001, "loss": 0.0223, "step": 20360 }, { "epoch": 155.49618320610688, "grad_norm": 1.442894458770752, "learning_rate": 0.0001, "loss": 0.022, "step": 20370 }, { "epoch": 155.57251908396947, "grad_norm": 1.5333373546600342, "learning_rate": 0.0001, "loss": 0.0217, "step": 20380 }, { "epoch": 155.64885496183206, "grad_norm": 1.9774178266525269, "learning_rate": 0.0001, "loss": 0.023, "step": 20390 }, { "epoch": 155.72519083969465, "grad_norm": 1.2164281606674194, "learning_rate": 0.0001, "loss": 0.0235, "step": 20400 }, { "epoch": 155.80152671755727, "grad_norm": 1.8258938789367676, "learning_rate": 0.0001, "loss": 0.0232, "step": 20410 }, { "epoch": 155.87786259541986, "grad_norm": 1.9329631328582764, "learning_rate": 0.0001, "loss": 0.0231, "step": 20420 }, { "epoch": 155.95419847328245, "grad_norm": 2.1637556552886963, "learning_rate": 0.0001, "loss": 0.0228, "step": 20430 }, { "epoch": 156.03053435114504, "grad_norm": 1.5575392246246338, "learning_rate": 0.0001, "loss": 0.0227, "step": 20440 }, { "epoch": 156.10687022900763, "grad_norm": 1.7726606130599976, "learning_rate": 0.0001, "loss": 0.023, "step": 20450 }, { "epoch": 156.18320610687022, "grad_norm": 2.6433327198028564, "learning_rate": 0.0001, "loss": 0.0218, "step": 20460 }, { "epoch": 156.25954198473283, "grad_norm": 1.5608073472976685, "learning_rate": 0.0001, "loss": 0.0225, "step": 20470 }, { "epoch": 156.33587786259542, "grad_norm": 1.9678473472595215, "learning_rate": 0.0001, "loss": 0.0237, "step": 20480 }, { "epoch": 156.41221374045801, "grad_norm": 1.7237086296081543, "learning_rate": 0.0001, "loss": 0.0221, "step": 20490 }, { "epoch": 156.4885496183206, "grad_norm": 2.132045269012451, "learning_rate": 0.0001, "loss": 0.022, "step": 20500 }, { "epoch": 156.5648854961832, "grad_norm": 1.2970499992370605, "learning_rate": 0.0001, "loss": 0.0213, "step": 20510 }, { "epoch": 156.6412213740458, "grad_norm": 2.190141439437866, "learning_rate": 0.0001, "loss": 0.0219, "step": 20520 }, { "epoch": 156.7175572519084, "grad_norm": 1.4638142585754395, "learning_rate": 0.0001, "loss": 0.0226, "step": 20530 }, { "epoch": 156.793893129771, "grad_norm": 1.337630271911621, "learning_rate": 0.0001, "loss": 0.0225, "step": 20540 }, { "epoch": 156.87022900763358, "grad_norm": 1.3416345119476318, "learning_rate": 0.0001, "loss": 0.0219, "step": 20550 }, { "epoch": 156.94656488549617, "grad_norm": 1.737494707107544, "learning_rate": 0.0001, "loss": 0.0225, "step": 20560 }, { "epoch": 157.0229007633588, "grad_norm": 1.9312658309936523, "learning_rate": 0.0001, "loss": 0.0226, "step": 20570 }, { "epoch": 157.09923664122138, "grad_norm": 1.723759412765503, "learning_rate": 0.0001, "loss": 0.022, "step": 20580 }, { "epoch": 157.17557251908397, "grad_norm": 1.160834789276123, "learning_rate": 0.0001, "loss": 0.0222, "step": 20590 }, { "epoch": 157.25190839694656, "grad_norm": 2.2235686779022217, "learning_rate": 0.0001, "loss": 0.0223, "step": 20600 }, { "epoch": 157.32824427480915, "grad_norm": 1.553209662437439, "learning_rate": 0.0001, "loss": 0.0232, "step": 20610 }, { "epoch": 157.40458015267177, "grad_norm": 1.9157912731170654, "learning_rate": 0.0001, "loss": 0.0232, "step": 20620 }, { "epoch": 157.48091603053436, "grad_norm": 2.324876070022583, "learning_rate": 0.0001, "loss": 0.0231, "step": 20630 }, { "epoch": 157.55725190839695, "grad_norm": 1.5198075771331787, "learning_rate": 0.0001, "loss": 0.0225, "step": 20640 }, { "epoch": 157.63358778625954, "grad_norm": 2.1136093139648438, "learning_rate": 0.0001, "loss": 0.0226, "step": 20650 }, { "epoch": 157.70992366412213, "grad_norm": 1.7197810411453247, "learning_rate": 0.0001, "loss": 0.0225, "step": 20660 }, { "epoch": 157.78625954198472, "grad_norm": 1.5631693601608276, "learning_rate": 0.0001, "loss": 0.0232, "step": 20670 }, { "epoch": 157.86259541984734, "grad_norm": 1.235395908355713, "learning_rate": 0.0001, "loss": 0.0224, "step": 20680 }, { "epoch": 157.93893129770993, "grad_norm": 2.150519609451294, "learning_rate": 0.0001, "loss": 0.0237, "step": 20690 }, { "epoch": 158.01526717557252, "grad_norm": 1.9886951446533203, "learning_rate": 0.0001, "loss": 0.0244, "step": 20700 }, { "epoch": 158.0916030534351, "grad_norm": 1.9335685968399048, "learning_rate": 0.0001, "loss": 0.024, "step": 20710 }, { "epoch": 158.1679389312977, "grad_norm": 1.418335199356079, "learning_rate": 0.0001, "loss": 0.0238, "step": 20720 }, { "epoch": 158.24427480916032, "grad_norm": 1.5999822616577148, "learning_rate": 0.0001, "loss": 0.0243, "step": 20730 }, { "epoch": 158.3206106870229, "grad_norm": 1.7888798713684082, "learning_rate": 0.0001, "loss": 0.0245, "step": 20740 }, { "epoch": 158.3969465648855, "grad_norm": 1.560165524482727, "learning_rate": 0.0001, "loss": 0.0251, "step": 20750 }, { "epoch": 158.4732824427481, "grad_norm": 1.668843150138855, "learning_rate": 0.0001, "loss": 0.024, "step": 20760 }, { "epoch": 158.54961832061068, "grad_norm": 1.5955833196640015, "learning_rate": 0.0001, "loss": 0.0248, "step": 20770 }, { "epoch": 158.6259541984733, "grad_norm": 1.3566563129425049, "learning_rate": 0.0001, "loss": 0.0239, "step": 20780 }, { "epoch": 158.70229007633588, "grad_norm": 1.5992969274520874, "learning_rate": 0.0001, "loss": 0.0222, "step": 20790 }, { "epoch": 158.77862595419847, "grad_norm": 1.3938175439834595, "learning_rate": 0.0001, "loss": 0.0242, "step": 20800 }, { "epoch": 158.85496183206106, "grad_norm": 1.641372561454773, "learning_rate": 0.0001, "loss": 0.0232, "step": 20810 }, { "epoch": 158.93129770992365, "grad_norm": 2.4115989208221436, "learning_rate": 0.0001, "loss": 0.0246, "step": 20820 }, { "epoch": 159.00763358778627, "grad_norm": 2.2648093700408936, "learning_rate": 0.0001, "loss": 0.0243, "step": 20830 }, { "epoch": 159.08396946564886, "grad_norm": 2.4227991104125977, "learning_rate": 0.0001, "loss": 0.0234, "step": 20840 }, { "epoch": 159.16030534351145, "grad_norm": 1.5228561162948608, "learning_rate": 0.0001, "loss": 0.0219, "step": 20850 }, { "epoch": 159.23664122137404, "grad_norm": 2.162625551223755, "learning_rate": 0.0001, "loss": 0.0231, "step": 20860 }, { "epoch": 159.31297709923663, "grad_norm": 2.173746347427368, "learning_rate": 0.0001, "loss": 0.0235, "step": 20870 }, { "epoch": 159.38931297709922, "grad_norm": 1.4672915935516357, "learning_rate": 0.0001, "loss": 0.0219, "step": 20880 }, { "epoch": 159.46564885496184, "grad_norm": 2.5573930740356445, "learning_rate": 0.0001, "loss": 0.0226, "step": 20890 }, { "epoch": 159.54198473282443, "grad_norm": 1.8001632690429688, "learning_rate": 0.0001, "loss": 0.0228, "step": 20900 }, { "epoch": 159.61832061068702, "grad_norm": 2.841303825378418, "learning_rate": 0.0001, "loss": 0.0215, "step": 20910 }, { "epoch": 159.6946564885496, "grad_norm": 2.897516965866089, "learning_rate": 0.0001, "loss": 0.023, "step": 20920 }, { "epoch": 159.7709923664122, "grad_norm": 1.9642887115478516, "learning_rate": 0.0001, "loss": 0.0226, "step": 20930 }, { "epoch": 159.84732824427482, "grad_norm": 2.1953418254852295, "learning_rate": 0.0001, "loss": 0.0228, "step": 20940 }, { "epoch": 159.9236641221374, "grad_norm": 2.060640335083008, "learning_rate": 0.0001, "loss": 0.0226, "step": 20950 }, { "epoch": 160.0, "grad_norm": 1.8610405921936035, "learning_rate": 0.0001, "loss": 0.0224, "step": 20960 }, { "epoch": 160.0763358778626, "grad_norm": 2.650118827819824, "learning_rate": 0.0001, "loss": 0.0226, "step": 20970 }, { "epoch": 160.15267175572518, "grad_norm": 1.8298012018203735, "learning_rate": 0.0001, "loss": 0.0218, "step": 20980 }, { "epoch": 160.2290076335878, "grad_norm": 1.741580843925476, "learning_rate": 0.0001, "loss": 0.0217, "step": 20990 }, { "epoch": 160.3053435114504, "grad_norm": 2.2207155227661133, "learning_rate": 0.0001, "loss": 0.0219, "step": 21000 }, { "epoch": 160.38167938931298, "grad_norm": 1.9793070554733276, "learning_rate": 0.0001, "loss": 0.0218, "step": 21010 }, { "epoch": 160.45801526717557, "grad_norm": 1.9044231176376343, "learning_rate": 0.0001, "loss": 0.0216, "step": 21020 }, { "epoch": 160.53435114503816, "grad_norm": 2.1002867221832275, "learning_rate": 0.0001, "loss": 0.023, "step": 21030 }, { "epoch": 160.61068702290078, "grad_norm": 2.071359157562256, "learning_rate": 0.0001, "loss": 0.0229, "step": 21040 }, { "epoch": 160.68702290076337, "grad_norm": 2.687701463699341, "learning_rate": 0.0001, "loss": 0.0228, "step": 21050 }, { "epoch": 160.76335877862596, "grad_norm": 2.027771472930908, "learning_rate": 0.0001, "loss": 0.0228, "step": 21060 }, { "epoch": 160.83969465648855, "grad_norm": 1.5135585069656372, "learning_rate": 0.0001, "loss": 0.0226, "step": 21070 }, { "epoch": 160.91603053435114, "grad_norm": 1.5295920372009277, "learning_rate": 0.0001, "loss": 0.0213, "step": 21080 }, { "epoch": 160.99236641221373, "grad_norm": 1.556797742843628, "learning_rate": 0.0001, "loss": 0.0218, "step": 21090 }, { "epoch": 161.06870229007635, "grad_norm": 1.6905088424682617, "learning_rate": 0.0001, "loss": 0.0217, "step": 21100 }, { "epoch": 161.14503816793894, "grad_norm": 2.696230411529541, "learning_rate": 0.0001, "loss": 0.0239, "step": 21110 }, { "epoch": 161.22137404580153, "grad_norm": 2.189103364944458, "learning_rate": 0.0001, "loss": 0.0234, "step": 21120 }, { "epoch": 161.29770992366412, "grad_norm": 2.0637717247009277, "learning_rate": 0.0001, "loss": 0.0236, "step": 21130 }, { "epoch": 161.3740458015267, "grad_norm": 1.6538993120193481, "learning_rate": 0.0001, "loss": 0.0232, "step": 21140 }, { "epoch": 161.45038167938932, "grad_norm": 1.8302137851715088, "learning_rate": 0.0001, "loss": 0.0218, "step": 21150 }, { "epoch": 161.5267175572519, "grad_norm": 1.5508488416671753, "learning_rate": 0.0001, "loss": 0.0225, "step": 21160 }, { "epoch": 161.6030534351145, "grad_norm": 2.2469322681427, "learning_rate": 0.0001, "loss": 0.0214, "step": 21170 }, { "epoch": 161.6793893129771, "grad_norm": 1.9213944673538208, "learning_rate": 0.0001, "loss": 0.0228, "step": 21180 }, { "epoch": 161.75572519083968, "grad_norm": 1.8453326225280762, "learning_rate": 0.0001, "loss": 0.0232, "step": 21190 }, { "epoch": 161.8320610687023, "grad_norm": 2.3942835330963135, "learning_rate": 0.0001, "loss": 0.0236, "step": 21200 }, { "epoch": 161.9083969465649, "grad_norm": 1.207602858543396, "learning_rate": 0.0001, "loss": 0.0231, "step": 21210 }, { "epoch": 161.98473282442748, "grad_norm": 1.5184985399246216, "learning_rate": 0.0001, "loss": 0.0228, "step": 21220 }, { "epoch": 162.06106870229007, "grad_norm": 1.9011670351028442, "learning_rate": 0.0001, "loss": 0.0233, "step": 21230 }, { "epoch": 162.13740458015266, "grad_norm": 1.4306609630584717, "learning_rate": 0.0001, "loss": 0.0246, "step": 21240 }, { "epoch": 162.21374045801528, "grad_norm": 2.1581804752349854, "learning_rate": 0.0001, "loss": 0.0244, "step": 21250 }, { "epoch": 162.29007633587787, "grad_norm": 1.6539276838302612, "learning_rate": 0.0001, "loss": 0.0251, "step": 21260 }, { "epoch": 162.36641221374046, "grad_norm": 2.347217082977295, "learning_rate": 0.0001, "loss": 0.0249, "step": 21270 }, { "epoch": 162.44274809160305, "grad_norm": 1.461394190788269, "learning_rate": 0.0001, "loss": 0.0244, "step": 21280 }, { "epoch": 162.51908396946564, "grad_norm": 1.8420989513397217, "learning_rate": 0.0001, "loss": 0.023, "step": 21290 }, { "epoch": 162.59541984732823, "grad_norm": 1.7283047437667847, "learning_rate": 0.0001, "loss": 0.0225, "step": 21300 }, { "epoch": 162.67175572519085, "grad_norm": 1.3291535377502441, "learning_rate": 0.0001, "loss": 0.023, "step": 21310 }, { "epoch": 162.74809160305344, "grad_norm": 1.7306084632873535, "learning_rate": 0.0001, "loss": 0.0232, "step": 21320 }, { "epoch": 162.82442748091603, "grad_norm": 1.669945478439331, "learning_rate": 0.0001, "loss": 0.0238, "step": 21330 }, { "epoch": 162.90076335877862, "grad_norm": 1.6542117595672607, "learning_rate": 0.0001, "loss": 0.0237, "step": 21340 }, { "epoch": 162.9770992366412, "grad_norm": 2.3019251823425293, "learning_rate": 0.0001, "loss": 0.0233, "step": 21350 }, { "epoch": 163.05343511450383, "grad_norm": 1.320403814315796, "learning_rate": 0.0001, "loss": 0.0225, "step": 21360 }, { "epoch": 163.12977099236642, "grad_norm": 1.927878737449646, "learning_rate": 0.0001, "loss": 0.0227, "step": 21370 }, { "epoch": 163.206106870229, "grad_norm": 1.9642177820205688, "learning_rate": 0.0001, "loss": 0.0228, "step": 21380 }, { "epoch": 163.2824427480916, "grad_norm": 2.2045435905456543, "learning_rate": 0.0001, "loss": 0.0231, "step": 21390 }, { "epoch": 163.3587786259542, "grad_norm": 1.5938889980316162, "learning_rate": 0.0001, "loss": 0.0229, "step": 21400 }, { "epoch": 163.4351145038168, "grad_norm": 2.288153886795044, "learning_rate": 0.0001, "loss": 0.0235, "step": 21410 }, { "epoch": 163.5114503816794, "grad_norm": 1.4675546884536743, "learning_rate": 0.0001, "loss": 0.023, "step": 21420 }, { "epoch": 163.58778625954199, "grad_norm": 1.5411453247070312, "learning_rate": 0.0001, "loss": 0.0225, "step": 21430 }, { "epoch": 163.66412213740458, "grad_norm": 2.329625129699707, "learning_rate": 0.0001, "loss": 0.023, "step": 21440 }, { "epoch": 163.74045801526717, "grad_norm": 2.125457286834717, "learning_rate": 0.0001, "loss": 0.0229, "step": 21450 }, { "epoch": 163.81679389312978, "grad_norm": 1.6455180644989014, "learning_rate": 0.0001, "loss": 0.0224, "step": 21460 }, { "epoch": 163.89312977099237, "grad_norm": 1.6706815958023071, "learning_rate": 0.0001, "loss": 0.0216, "step": 21470 }, { "epoch": 163.96946564885496, "grad_norm": 1.7077192068099976, "learning_rate": 0.0001, "loss": 0.0216, "step": 21480 }, { "epoch": 164.04580152671755, "grad_norm": 2.2540478706359863, "learning_rate": 0.0001, "loss": 0.0221, "step": 21490 }, { "epoch": 164.12213740458014, "grad_norm": 1.8492074012756348, "learning_rate": 0.0001, "loss": 0.023, "step": 21500 }, { "epoch": 164.19847328244273, "grad_norm": 1.9196491241455078, "learning_rate": 0.0001, "loss": 0.0226, "step": 21510 }, { "epoch": 164.27480916030535, "grad_norm": 1.5825941562652588, "learning_rate": 0.0001, "loss": 0.0216, "step": 21520 }, { "epoch": 164.35114503816794, "grad_norm": 1.800052285194397, "learning_rate": 0.0001, "loss": 0.0225, "step": 21530 }, { "epoch": 164.42748091603053, "grad_norm": 2.7707035541534424, "learning_rate": 0.0001, "loss": 0.0231, "step": 21540 }, { "epoch": 164.50381679389312, "grad_norm": 1.4967827796936035, "learning_rate": 0.0001, "loss": 0.0228, "step": 21550 }, { "epoch": 164.5801526717557, "grad_norm": 1.3447129726409912, "learning_rate": 0.0001, "loss": 0.0222, "step": 21560 }, { "epoch": 164.65648854961833, "grad_norm": 1.311567783355713, "learning_rate": 0.0001, "loss": 0.0215, "step": 21570 }, { "epoch": 164.73282442748092, "grad_norm": 1.8684263229370117, "learning_rate": 0.0001, "loss": 0.022, "step": 21580 }, { "epoch": 164.8091603053435, "grad_norm": 2.0928337574005127, "learning_rate": 0.0001, "loss": 0.0212, "step": 21590 }, { "epoch": 164.8854961832061, "grad_norm": 1.3690582513809204, "learning_rate": 0.0001, "loss": 0.0211, "step": 21600 }, { "epoch": 164.9618320610687, "grad_norm": 1.315799593925476, "learning_rate": 0.0001, "loss": 0.0228, "step": 21610 }, { "epoch": 165.0381679389313, "grad_norm": 2.06595778465271, "learning_rate": 0.0001, "loss": 0.0222, "step": 21620 }, { "epoch": 165.1145038167939, "grad_norm": 1.92926824092865, "learning_rate": 0.0001, "loss": 0.0228, "step": 21630 }, { "epoch": 165.1908396946565, "grad_norm": 1.5361056327819824, "learning_rate": 0.0001, "loss": 0.0218, "step": 21640 }, { "epoch": 165.26717557251908, "grad_norm": 2.3015835285186768, "learning_rate": 0.0001, "loss": 0.0239, "step": 21650 }, { "epoch": 165.34351145038167, "grad_norm": 2.166611909866333, "learning_rate": 0.0001, "loss": 0.0219, "step": 21660 }, { "epoch": 165.4198473282443, "grad_norm": 2.1090946197509766, "learning_rate": 0.0001, "loss": 0.0213, "step": 21670 }, { "epoch": 165.49618320610688, "grad_norm": 2.156492233276367, "learning_rate": 0.0001, "loss": 0.0222, "step": 21680 }, { "epoch": 165.57251908396947, "grad_norm": 1.6741328239440918, "learning_rate": 0.0001, "loss": 0.0219, "step": 21690 }, { "epoch": 165.64885496183206, "grad_norm": 1.5692726373672485, "learning_rate": 0.0001, "loss": 0.0215, "step": 21700 }, { "epoch": 165.72519083969465, "grad_norm": 2.173657178878784, "learning_rate": 0.0001, "loss": 0.0212, "step": 21710 }, { "epoch": 165.80152671755727, "grad_norm": 1.5870939493179321, "learning_rate": 0.0001, "loss": 0.0221, "step": 21720 }, { "epoch": 165.87786259541986, "grad_norm": 1.7792344093322754, "learning_rate": 0.0001, "loss": 0.0223, "step": 21730 }, { "epoch": 165.95419847328245, "grad_norm": 2.4655873775482178, "learning_rate": 0.0001, "loss": 0.0219, "step": 21740 }, { "epoch": 166.03053435114504, "grad_norm": 1.9168108701705933, "learning_rate": 0.0001, "loss": 0.0221, "step": 21750 }, { "epoch": 166.10687022900763, "grad_norm": 1.2115076780319214, "learning_rate": 0.0001, "loss": 0.0215, "step": 21760 }, { "epoch": 166.18320610687022, "grad_norm": 1.3329217433929443, "learning_rate": 0.0001, "loss": 0.0223, "step": 21770 }, { "epoch": 166.25954198473283, "grad_norm": 1.372704029083252, "learning_rate": 0.0001, "loss": 0.023, "step": 21780 }, { "epoch": 166.33587786259542, "grad_norm": 1.3280328512191772, "learning_rate": 0.0001, "loss": 0.0233, "step": 21790 }, { "epoch": 166.41221374045801, "grad_norm": 2.0022058486938477, "learning_rate": 0.0001, "loss": 0.023, "step": 21800 }, { "epoch": 166.4885496183206, "grad_norm": 1.5718485116958618, "learning_rate": 0.0001, "loss": 0.0227, "step": 21810 }, { "epoch": 166.5648854961832, "grad_norm": 2.4222145080566406, "learning_rate": 0.0001, "loss": 0.0221, "step": 21820 }, { "epoch": 166.6412213740458, "grad_norm": 1.5047003030776978, "learning_rate": 0.0001, "loss": 0.0228, "step": 21830 }, { "epoch": 166.7175572519084, "grad_norm": 1.364322304725647, "learning_rate": 0.0001, "loss": 0.0216, "step": 21840 }, { "epoch": 166.793893129771, "grad_norm": 1.4773656129837036, "learning_rate": 0.0001, "loss": 0.0223, "step": 21850 }, { "epoch": 166.87022900763358, "grad_norm": 1.9001561403274536, "learning_rate": 0.0001, "loss": 0.0217, "step": 21860 }, { "epoch": 166.94656488549617, "grad_norm": 2.9062881469726562, "learning_rate": 0.0001, "loss": 0.0217, "step": 21870 }, { "epoch": 167.0229007633588, "grad_norm": 1.6220593452453613, "learning_rate": 0.0001, "loss": 0.0212, "step": 21880 }, { "epoch": 167.09923664122138, "grad_norm": 1.8843563795089722, "learning_rate": 0.0001, "loss": 0.0207, "step": 21890 }, { "epoch": 167.17557251908397, "grad_norm": 1.7807495594024658, "learning_rate": 0.0001, "loss": 0.0206, "step": 21900 }, { "epoch": 167.25190839694656, "grad_norm": 1.3884310722351074, "learning_rate": 0.0001, "loss": 0.0209, "step": 21910 }, { "epoch": 167.32824427480915, "grad_norm": 1.570243000984192, "learning_rate": 0.0001, "loss": 0.0205, "step": 21920 }, { "epoch": 167.40458015267177, "grad_norm": 2.030144453048706, "learning_rate": 0.0001, "loss": 0.0212, "step": 21930 }, { "epoch": 167.48091603053436, "grad_norm": 2.218545913696289, "learning_rate": 0.0001, "loss": 0.0212, "step": 21940 }, { "epoch": 167.55725190839695, "grad_norm": 2.422698736190796, "learning_rate": 0.0001, "loss": 0.0213, "step": 21950 }, { "epoch": 167.63358778625954, "grad_norm": 2.0288004875183105, "learning_rate": 0.0001, "loss": 0.021, "step": 21960 }, { "epoch": 167.70992366412213, "grad_norm": 2.1546318531036377, "learning_rate": 0.0001, "loss": 0.0214, "step": 21970 }, { "epoch": 167.78625954198472, "grad_norm": 1.7608442306518555, "learning_rate": 0.0001, "loss": 0.0218, "step": 21980 }, { "epoch": 167.86259541984734, "grad_norm": 1.7497482299804688, "learning_rate": 0.0001, "loss": 0.0228, "step": 21990 }, { "epoch": 167.93893129770993, "grad_norm": 1.9544154405593872, "learning_rate": 0.0001, "loss": 0.0222, "step": 22000 }, { "epoch": 168.01526717557252, "grad_norm": 1.921897292137146, "learning_rate": 0.0001, "loss": 0.0218, "step": 22010 }, { "epoch": 168.0916030534351, "grad_norm": 1.7276482582092285, "learning_rate": 0.0001, "loss": 0.0229, "step": 22020 }, { "epoch": 168.1679389312977, "grad_norm": 1.7362651824951172, "learning_rate": 0.0001, "loss": 0.0214, "step": 22030 }, { "epoch": 168.24427480916032, "grad_norm": 2.168593168258667, "learning_rate": 0.0001, "loss": 0.0221, "step": 22040 }, { "epoch": 168.3206106870229, "grad_norm": 1.5920201539993286, "learning_rate": 0.0001, "loss": 0.0229, "step": 22050 }, { "epoch": 168.3969465648855, "grad_norm": 1.8681761026382446, "learning_rate": 0.0001, "loss": 0.022, "step": 22060 }, { "epoch": 168.4732824427481, "grad_norm": 2.371202230453491, "learning_rate": 0.0001, "loss": 0.0226, "step": 22070 }, { "epoch": 168.54961832061068, "grad_norm": 1.5926083326339722, "learning_rate": 0.0001, "loss": 0.0224, "step": 22080 }, { "epoch": 168.6259541984733, "grad_norm": 1.6500896215438843, "learning_rate": 0.0001, "loss": 0.0217, "step": 22090 }, { "epoch": 168.70229007633588, "grad_norm": 1.7714884281158447, "learning_rate": 0.0001, "loss": 0.0217, "step": 22100 }, { "epoch": 168.77862595419847, "grad_norm": 1.963346242904663, "learning_rate": 0.0001, "loss": 0.0225, "step": 22110 }, { "epoch": 168.85496183206106, "grad_norm": 2.3053863048553467, "learning_rate": 0.0001, "loss": 0.0217, "step": 22120 }, { "epoch": 168.93129770992365, "grad_norm": 1.207205057144165, "learning_rate": 0.0001, "loss": 0.0209, "step": 22130 }, { "epoch": 169.00763358778627, "grad_norm": 1.6502596139907837, "learning_rate": 0.0001, "loss": 0.021, "step": 22140 }, { "epoch": 169.08396946564886, "grad_norm": 1.086119294166565, "learning_rate": 0.0001, "loss": 0.0217, "step": 22150 }, { "epoch": 169.16030534351145, "grad_norm": 2.876680374145508, "learning_rate": 0.0001, "loss": 0.0232, "step": 22160 }, { "epoch": 169.23664122137404, "grad_norm": 2.4049222469329834, "learning_rate": 0.0001, "loss": 0.0229, "step": 22170 }, { "epoch": 169.31297709923663, "grad_norm": 1.9603906869888306, "learning_rate": 0.0001, "loss": 0.0223, "step": 22180 }, { "epoch": 169.38931297709922, "grad_norm": 2.1025452613830566, "learning_rate": 0.0001, "loss": 0.0222, "step": 22190 }, { "epoch": 169.46564885496184, "grad_norm": 2.2270915508270264, "learning_rate": 0.0001, "loss": 0.0219, "step": 22200 }, { "epoch": 169.54198473282443, "grad_norm": 1.7365612983703613, "learning_rate": 0.0001, "loss": 0.0216, "step": 22210 }, { "epoch": 169.61832061068702, "grad_norm": 1.9957469701766968, "learning_rate": 0.0001, "loss": 0.0222, "step": 22220 }, { "epoch": 169.6946564885496, "grad_norm": 1.9072980880737305, "learning_rate": 0.0001, "loss": 0.0217, "step": 22230 }, { "epoch": 169.7709923664122, "grad_norm": 1.8167704343795776, "learning_rate": 0.0001, "loss": 0.0214, "step": 22240 }, { "epoch": 169.84732824427482, "grad_norm": 1.4379886388778687, "learning_rate": 0.0001, "loss": 0.0214, "step": 22250 }, { "epoch": 169.9236641221374, "grad_norm": 1.4165899753570557, "learning_rate": 0.0001, "loss": 0.021, "step": 22260 }, { "epoch": 170.0, "grad_norm": 1.9446052312850952, "learning_rate": 0.0001, "loss": 0.0206, "step": 22270 }, { "epoch": 170.0763358778626, "grad_norm": 2.2885234355926514, "learning_rate": 0.0001, "loss": 0.0212, "step": 22280 }, { "epoch": 170.15267175572518, "grad_norm": 1.9917221069335938, "learning_rate": 0.0001, "loss": 0.0209, "step": 22290 }, { "epoch": 170.2290076335878, "grad_norm": 1.358998417854309, "learning_rate": 0.0001, "loss": 0.0207, "step": 22300 }, { "epoch": 170.3053435114504, "grad_norm": 2.1055665016174316, "learning_rate": 0.0001, "loss": 0.0203, "step": 22310 }, { "epoch": 170.38167938931298, "grad_norm": 2.5007755756378174, "learning_rate": 0.0001, "loss": 0.0206, "step": 22320 }, { "epoch": 170.45801526717557, "grad_norm": 2.099022388458252, "learning_rate": 0.0001, "loss": 0.0206, "step": 22330 }, { "epoch": 170.53435114503816, "grad_norm": 1.6562551259994507, "learning_rate": 0.0001, "loss": 0.0211, "step": 22340 }, { "epoch": 170.61068702290078, "grad_norm": 1.9748826026916504, "learning_rate": 0.0001, "loss": 0.0211, "step": 22350 }, { "epoch": 170.68702290076337, "grad_norm": 1.4840904474258423, "learning_rate": 0.0001, "loss": 0.0212, "step": 22360 }, { "epoch": 170.76335877862596, "grad_norm": 1.5095648765563965, "learning_rate": 0.0001, "loss": 0.0219, "step": 22370 }, { "epoch": 170.83969465648855, "grad_norm": 1.8259320259094238, "learning_rate": 0.0001, "loss": 0.0215, "step": 22380 }, { "epoch": 170.91603053435114, "grad_norm": 1.529832124710083, "learning_rate": 0.0001, "loss": 0.0227, "step": 22390 }, { "epoch": 170.99236641221373, "grad_norm": 1.500813364982605, "learning_rate": 0.0001, "loss": 0.023, "step": 22400 }, { "epoch": 171.06870229007635, "grad_norm": 2.0732688903808594, "learning_rate": 0.0001, "loss": 0.0227, "step": 22410 }, { "epoch": 171.14503816793894, "grad_norm": 1.6453624963760376, "learning_rate": 0.0001, "loss": 0.0216, "step": 22420 }, { "epoch": 171.22137404580153, "grad_norm": 1.9134691953659058, "learning_rate": 0.0001, "loss": 0.0228, "step": 22430 }, { "epoch": 171.29770992366412, "grad_norm": 1.6582732200622559, "learning_rate": 0.0001, "loss": 0.022, "step": 22440 }, { "epoch": 171.3740458015267, "grad_norm": 1.4566854238510132, "learning_rate": 0.0001, "loss": 0.022, "step": 22450 }, { "epoch": 171.45038167938932, "grad_norm": 1.5370690822601318, "learning_rate": 0.0001, "loss": 0.0223, "step": 22460 }, { "epoch": 171.5267175572519, "grad_norm": 1.944685459136963, "learning_rate": 0.0001, "loss": 0.0238, "step": 22470 }, { "epoch": 171.6030534351145, "grad_norm": 1.1829198598861694, "learning_rate": 0.0001, "loss": 0.0245, "step": 22480 }, { "epoch": 171.6793893129771, "grad_norm": 1.6947907209396362, "learning_rate": 0.0001, "loss": 0.0234, "step": 22490 }, { "epoch": 171.75572519083968, "grad_norm": 1.7128280401229858, "learning_rate": 0.0001, "loss": 0.0229, "step": 22500 }, { "epoch": 171.8320610687023, "grad_norm": 1.4393528699874878, "learning_rate": 0.0001, "loss": 0.0224, "step": 22510 }, { "epoch": 171.9083969465649, "grad_norm": 1.6767370700836182, "learning_rate": 0.0001, "loss": 0.0226, "step": 22520 }, { "epoch": 171.98473282442748, "grad_norm": 1.5434006452560425, "learning_rate": 0.0001, "loss": 0.0216, "step": 22530 }, { "epoch": 172.06106870229007, "grad_norm": 2.381965160369873, "learning_rate": 0.0001, "loss": 0.0219, "step": 22540 }, { "epoch": 172.13740458015266, "grad_norm": 2.3163888454437256, "learning_rate": 0.0001, "loss": 0.0229, "step": 22550 }, { "epoch": 172.21374045801528, "grad_norm": 1.9249721765518188, "learning_rate": 0.0001, "loss": 0.0228, "step": 22560 }, { "epoch": 172.29007633587787, "grad_norm": 1.3112709522247314, "learning_rate": 0.0001, "loss": 0.0225, "step": 22570 }, { "epoch": 172.36641221374046, "grad_norm": 1.425394058227539, "learning_rate": 0.0001, "loss": 0.0223, "step": 22580 }, { "epoch": 172.44274809160305, "grad_norm": 1.7495349645614624, "learning_rate": 0.0001, "loss": 0.0219, "step": 22590 }, { "epoch": 172.51908396946564, "grad_norm": 1.574785590171814, "learning_rate": 0.0001, "loss": 0.022, "step": 22600 }, { "epoch": 172.59541984732823, "grad_norm": 1.7773194313049316, "learning_rate": 0.0001, "loss": 0.0206, "step": 22610 }, { "epoch": 172.67175572519085, "grad_norm": 1.9789533615112305, "learning_rate": 0.0001, "loss": 0.0212, "step": 22620 }, { "epoch": 172.74809160305344, "grad_norm": 1.7124643325805664, "learning_rate": 0.0001, "loss": 0.0223, "step": 22630 }, { "epoch": 172.82442748091603, "grad_norm": 2.2491819858551025, "learning_rate": 0.0001, "loss": 0.0229, "step": 22640 }, { "epoch": 172.90076335877862, "grad_norm": 1.4172966480255127, "learning_rate": 0.0001, "loss": 0.0232, "step": 22650 }, { "epoch": 172.9770992366412, "grad_norm": 2.89172101020813, "learning_rate": 0.0001, "loss": 0.0241, "step": 22660 }, { "epoch": 173.05343511450383, "grad_norm": 1.8342483043670654, "learning_rate": 0.0001, "loss": 0.0232, "step": 22670 }, { "epoch": 173.12977099236642, "grad_norm": 1.2697004079818726, "learning_rate": 0.0001, "loss": 0.022, "step": 22680 }, { "epoch": 173.206106870229, "grad_norm": 1.4993033409118652, "learning_rate": 0.0001, "loss": 0.0217, "step": 22690 }, { "epoch": 173.2824427480916, "grad_norm": 1.7246832847595215, "learning_rate": 0.0001, "loss": 0.0221, "step": 22700 }, { "epoch": 173.3587786259542, "grad_norm": 2.1477181911468506, "learning_rate": 0.0001, "loss": 0.0213, "step": 22710 }, { "epoch": 173.4351145038168, "grad_norm": 1.7059417963027954, "learning_rate": 0.0001, "loss": 0.0219, "step": 22720 }, { "epoch": 173.5114503816794, "grad_norm": 1.3060420751571655, "learning_rate": 0.0001, "loss": 0.0209, "step": 22730 }, { "epoch": 173.58778625954199, "grad_norm": 1.4111950397491455, "learning_rate": 0.0001, "loss": 0.0211, "step": 22740 }, { "epoch": 173.66412213740458, "grad_norm": 1.8502399921417236, "learning_rate": 0.0001, "loss": 0.0214, "step": 22750 }, { "epoch": 173.74045801526717, "grad_norm": 1.7360258102416992, "learning_rate": 0.0001, "loss": 0.0229, "step": 22760 }, { "epoch": 173.81679389312978, "grad_norm": 1.7936232089996338, "learning_rate": 0.0001, "loss": 0.0231, "step": 22770 }, { "epoch": 173.89312977099237, "grad_norm": 1.9499050378799438, "learning_rate": 0.0001, "loss": 0.0231, "step": 22780 }, { "epoch": 173.96946564885496, "grad_norm": 1.4462395906448364, "learning_rate": 0.0001, "loss": 0.0227, "step": 22790 }, { "epoch": 174.04580152671755, "grad_norm": 2.602766990661621, "learning_rate": 0.0001, "loss": 0.0225, "step": 22800 }, { "epoch": 174.12213740458014, "grad_norm": 1.725661277770996, "learning_rate": 0.0001, "loss": 0.0233, "step": 22810 }, { "epoch": 174.19847328244273, "grad_norm": 1.9464209079742432, "learning_rate": 0.0001, "loss": 0.0216, "step": 22820 }, { "epoch": 174.27480916030535, "grad_norm": 1.415336012840271, "learning_rate": 0.0001, "loss": 0.022, "step": 22830 }, { "epoch": 174.35114503816794, "grad_norm": 1.9603220224380493, "learning_rate": 0.0001, "loss": 0.0227, "step": 22840 }, { "epoch": 174.42748091603053, "grad_norm": 1.6594618558883667, "learning_rate": 0.0001, "loss": 0.0241, "step": 22850 }, { "epoch": 174.50381679389312, "grad_norm": 1.7286323308944702, "learning_rate": 0.0001, "loss": 0.0247, "step": 22860 }, { "epoch": 174.5801526717557, "grad_norm": 1.671622395515442, "learning_rate": 0.0001, "loss": 0.0233, "step": 22870 }, { "epoch": 174.65648854961833, "grad_norm": 1.7323306798934937, "learning_rate": 0.0001, "loss": 0.0222, "step": 22880 }, { "epoch": 174.73282442748092, "grad_norm": 1.2820415496826172, "learning_rate": 0.0001, "loss": 0.0229, "step": 22890 }, { "epoch": 174.8091603053435, "grad_norm": 2.170872688293457, "learning_rate": 0.0001, "loss": 0.0225, "step": 22900 }, { "epoch": 174.8854961832061, "grad_norm": 1.876339316368103, "learning_rate": 0.0001, "loss": 0.0232, "step": 22910 }, { "epoch": 174.9618320610687, "grad_norm": 1.6856611967086792, "learning_rate": 0.0001, "loss": 0.0228, "step": 22920 }, { "epoch": 175.0381679389313, "grad_norm": 1.6870908737182617, "learning_rate": 0.0001, "loss": 0.0223, "step": 22930 }, { "epoch": 175.1145038167939, "grad_norm": 2.2220895290374756, "learning_rate": 0.0001, "loss": 0.0223, "step": 22940 }, { "epoch": 175.1908396946565, "grad_norm": 1.3819546699523926, "learning_rate": 0.0001, "loss": 0.0225, "step": 22950 }, { "epoch": 175.26717557251908, "grad_norm": 1.7581202983856201, "learning_rate": 0.0001, "loss": 0.0209, "step": 22960 }, { "epoch": 175.34351145038167, "grad_norm": 1.5827678442001343, "learning_rate": 0.0001, "loss": 0.0219, "step": 22970 }, { "epoch": 175.4198473282443, "grad_norm": 1.8515571355819702, "learning_rate": 0.0001, "loss": 0.0208, "step": 22980 }, { "epoch": 175.49618320610688, "grad_norm": 1.6494163274765015, "learning_rate": 0.0001, "loss": 0.0214, "step": 22990 }, { "epoch": 175.57251908396947, "grad_norm": 1.9911894798278809, "learning_rate": 0.0001, "loss": 0.0218, "step": 23000 }, { "epoch": 175.64885496183206, "grad_norm": 1.599664330482483, "learning_rate": 0.0001, "loss": 0.0222, "step": 23010 }, { "epoch": 175.72519083969465, "grad_norm": 1.9557398557662964, "learning_rate": 0.0001, "loss": 0.022, "step": 23020 }, { "epoch": 175.80152671755727, "grad_norm": 1.5706526041030884, "learning_rate": 0.0001, "loss": 0.022, "step": 23030 }, { "epoch": 175.87786259541986, "grad_norm": 1.5998494625091553, "learning_rate": 0.0001, "loss": 0.0224, "step": 23040 }, { "epoch": 175.95419847328245, "grad_norm": 1.2812048196792603, "learning_rate": 0.0001, "loss": 0.0221, "step": 23050 }, { "epoch": 176.03053435114504, "grad_norm": 1.667405366897583, "learning_rate": 0.0001, "loss": 0.0216, "step": 23060 }, { "epoch": 176.10687022900763, "grad_norm": 1.4794055223464966, "learning_rate": 0.0001, "loss": 0.0228, "step": 23070 }, { "epoch": 176.18320610687022, "grad_norm": 2.2206647396087646, "learning_rate": 0.0001, "loss": 0.0229, "step": 23080 }, { "epoch": 176.25954198473283, "grad_norm": 2.435009002685547, "learning_rate": 0.0001, "loss": 0.0228, "step": 23090 }, { "epoch": 176.33587786259542, "grad_norm": 1.7252695560455322, "learning_rate": 0.0001, "loss": 0.0228, "step": 23100 }, { "epoch": 176.41221374045801, "grad_norm": 2.182196855545044, "learning_rate": 0.0001, "loss": 0.0217, "step": 23110 }, { "epoch": 176.4885496183206, "grad_norm": 2.2744691371917725, "learning_rate": 0.0001, "loss": 0.0205, "step": 23120 }, { "epoch": 176.5648854961832, "grad_norm": 1.227611780166626, "learning_rate": 0.0001, "loss": 0.021, "step": 23130 }, { "epoch": 176.6412213740458, "grad_norm": 1.0991308689117432, "learning_rate": 0.0001, "loss": 0.0214, "step": 23140 }, { "epoch": 176.7175572519084, "grad_norm": 2.7247660160064697, "learning_rate": 0.0001, "loss": 0.0207, "step": 23150 }, { "epoch": 176.793893129771, "grad_norm": 2.1144115924835205, "learning_rate": 0.0001, "loss": 0.0217, "step": 23160 }, { "epoch": 176.87022900763358, "grad_norm": 1.5654641389846802, "learning_rate": 0.0001, "loss": 0.0224, "step": 23170 }, { "epoch": 176.94656488549617, "grad_norm": 2.0250115394592285, "learning_rate": 0.0001, "loss": 0.0217, "step": 23180 }, { "epoch": 177.0229007633588, "grad_norm": 1.2246999740600586, "learning_rate": 0.0001, "loss": 0.0209, "step": 23190 }, { "epoch": 177.09923664122138, "grad_norm": 1.5260474681854248, "learning_rate": 0.0001, "loss": 0.0206, "step": 23200 }, { "epoch": 177.17557251908397, "grad_norm": 1.5454713106155396, "learning_rate": 0.0001, "loss": 0.0208, "step": 23210 }, { "epoch": 177.25190839694656, "grad_norm": 1.3704653978347778, "learning_rate": 0.0001, "loss": 0.0211, "step": 23220 }, { "epoch": 177.32824427480915, "grad_norm": 2.506273031234741, "learning_rate": 0.0001, "loss": 0.0209, "step": 23230 }, { "epoch": 177.40458015267177, "grad_norm": 1.6447663307189941, "learning_rate": 0.0001, "loss": 0.021, "step": 23240 }, { "epoch": 177.48091603053436, "grad_norm": 1.934885025024414, "learning_rate": 0.0001, "loss": 0.021, "step": 23250 }, { "epoch": 177.55725190839695, "grad_norm": 1.445265531539917, "learning_rate": 0.0001, "loss": 0.0217, "step": 23260 }, { "epoch": 177.63358778625954, "grad_norm": 1.461861491203308, "learning_rate": 0.0001, "loss": 0.0215, "step": 23270 }, { "epoch": 177.70992366412213, "grad_norm": 2.2309372425079346, "learning_rate": 0.0001, "loss": 0.0212, "step": 23280 }, { "epoch": 177.78625954198472, "grad_norm": 1.6086236238479614, "learning_rate": 0.0001, "loss": 0.0215, "step": 23290 }, { "epoch": 177.86259541984734, "grad_norm": 2.035792589187622, "learning_rate": 0.0001, "loss": 0.0209, "step": 23300 }, { "epoch": 177.93893129770993, "grad_norm": 1.5424104928970337, "learning_rate": 0.0001, "loss": 0.0215, "step": 23310 }, { "epoch": 178.01526717557252, "grad_norm": 1.1691855192184448, "learning_rate": 0.0001, "loss": 0.0209, "step": 23320 }, { "epoch": 178.0916030534351, "grad_norm": 1.4711965322494507, "learning_rate": 0.0001, "loss": 0.0207, "step": 23330 }, { "epoch": 178.1679389312977, "grad_norm": 1.3122197389602661, "learning_rate": 0.0001, "loss": 0.0209, "step": 23340 }, { "epoch": 178.24427480916032, "grad_norm": 1.998096227645874, "learning_rate": 0.0001, "loss": 0.0217, "step": 23350 }, { "epoch": 178.3206106870229, "grad_norm": 1.6266369819641113, "learning_rate": 0.0001, "loss": 0.0218, "step": 23360 }, { "epoch": 178.3969465648855, "grad_norm": 1.7069343328475952, "learning_rate": 0.0001, "loss": 0.0222, "step": 23370 }, { "epoch": 178.4732824427481, "grad_norm": 1.9495598077774048, "learning_rate": 0.0001, "loss": 0.0231, "step": 23380 }, { "epoch": 178.54961832061068, "grad_norm": 1.9329026937484741, "learning_rate": 0.0001, "loss": 0.022, "step": 23390 }, { "epoch": 178.6259541984733, "grad_norm": 1.8020412921905518, "learning_rate": 0.0001, "loss": 0.0223, "step": 23400 }, { "epoch": 178.70229007633588, "grad_norm": 2.0395169258117676, "learning_rate": 0.0001, "loss": 0.0218, "step": 23410 }, { "epoch": 178.77862595419847, "grad_norm": 2.1409263610839844, "learning_rate": 0.0001, "loss": 0.0207, "step": 23420 }, { "epoch": 178.85496183206106, "grad_norm": 1.718335509300232, "learning_rate": 0.0001, "loss": 0.0211, "step": 23430 }, { "epoch": 178.93129770992365, "grad_norm": 1.9731471538543701, "learning_rate": 0.0001, "loss": 0.0219, "step": 23440 }, { "epoch": 179.00763358778627, "grad_norm": 1.4988259077072144, "learning_rate": 0.0001, "loss": 0.0224, "step": 23450 }, { "epoch": 179.08396946564886, "grad_norm": 1.362685203552246, "learning_rate": 0.0001, "loss": 0.022, "step": 23460 }, { "epoch": 179.16030534351145, "grad_norm": 2.079068422317505, "learning_rate": 0.0001, "loss": 0.0229, "step": 23470 }, { "epoch": 179.23664122137404, "grad_norm": 1.4700963497161865, "learning_rate": 0.0001, "loss": 0.0227, "step": 23480 }, { "epoch": 179.31297709923663, "grad_norm": 2.0773797035217285, "learning_rate": 0.0001, "loss": 0.0217, "step": 23490 }, { "epoch": 179.38931297709922, "grad_norm": 1.7251348495483398, "learning_rate": 0.0001, "loss": 0.0227, "step": 23500 }, { "epoch": 179.46564885496184, "grad_norm": 2.8634581565856934, "learning_rate": 0.0001, "loss": 0.0216, "step": 23510 }, { "epoch": 179.54198473282443, "grad_norm": 1.637608528137207, "learning_rate": 0.0001, "loss": 0.0211, "step": 23520 }, { "epoch": 179.61832061068702, "grad_norm": 1.93489408493042, "learning_rate": 0.0001, "loss": 0.0209, "step": 23530 }, { "epoch": 179.6946564885496, "grad_norm": 1.8865607976913452, "learning_rate": 0.0001, "loss": 0.0215, "step": 23540 }, { "epoch": 179.7709923664122, "grad_norm": 1.4447447061538696, "learning_rate": 0.0001, "loss": 0.0222, "step": 23550 }, { "epoch": 179.84732824427482, "grad_norm": 2.55912709236145, "learning_rate": 0.0001, "loss": 0.0224, "step": 23560 }, { "epoch": 179.9236641221374, "grad_norm": 1.8555114269256592, "learning_rate": 0.0001, "loss": 0.0218, "step": 23570 }, { "epoch": 180.0, "grad_norm": 1.7340373992919922, "learning_rate": 0.0001, "loss": 0.0215, "step": 23580 }, { "epoch": 180.0763358778626, "grad_norm": 2.1897952556610107, "learning_rate": 0.0001, "loss": 0.0209, "step": 23590 }, { "epoch": 180.15267175572518, "grad_norm": 2.108966588973999, "learning_rate": 0.0001, "loss": 0.0207, "step": 23600 }, { "epoch": 180.2290076335878, "grad_norm": 2.078275442123413, "learning_rate": 0.0001, "loss": 0.0223, "step": 23610 }, { "epoch": 180.3053435114504, "grad_norm": 2.6599504947662354, "learning_rate": 0.0001, "loss": 0.0214, "step": 23620 }, { "epoch": 180.38167938931298, "grad_norm": 2.6282622814178467, "learning_rate": 0.0001, "loss": 0.0206, "step": 23630 }, { "epoch": 180.45801526717557, "grad_norm": 2.3378348350524902, "learning_rate": 0.0001, "loss": 0.0209, "step": 23640 }, { "epoch": 180.53435114503816, "grad_norm": 1.4437836408615112, "learning_rate": 0.0001, "loss": 0.0194, "step": 23650 }, { "epoch": 180.61068702290078, "grad_norm": 1.3158305883407593, "learning_rate": 0.0001, "loss": 0.0203, "step": 23660 }, { "epoch": 180.68702290076337, "grad_norm": 1.554937720298767, "learning_rate": 0.0001, "loss": 0.0208, "step": 23670 }, { "epoch": 180.76335877862596, "grad_norm": 1.4327739477157593, "learning_rate": 0.0001, "loss": 0.0208, "step": 23680 }, { "epoch": 180.83969465648855, "grad_norm": 2.2751147747039795, "learning_rate": 0.0001, "loss": 0.0207, "step": 23690 }, { "epoch": 180.91603053435114, "grad_norm": 2.290031909942627, "learning_rate": 0.0001, "loss": 0.0218, "step": 23700 }, { "epoch": 180.99236641221373, "grad_norm": 2.28201961517334, "learning_rate": 0.0001, "loss": 0.022, "step": 23710 }, { "epoch": 181.06870229007635, "grad_norm": 1.825775384902954, "learning_rate": 0.0001, "loss": 0.0211, "step": 23720 }, { "epoch": 181.14503816793894, "grad_norm": 1.7590025663375854, "learning_rate": 0.0001, "loss": 0.021, "step": 23730 }, { "epoch": 181.22137404580153, "grad_norm": 1.2263292074203491, "learning_rate": 0.0001, "loss": 0.0205, "step": 23740 }, { "epoch": 181.29770992366412, "grad_norm": 1.3882137537002563, "learning_rate": 0.0001, "loss": 0.0213, "step": 23750 }, { "epoch": 181.3740458015267, "grad_norm": 1.7285140752792358, "learning_rate": 0.0001, "loss": 0.0211, "step": 23760 }, { "epoch": 181.45038167938932, "grad_norm": 1.2065813541412354, "learning_rate": 0.0001, "loss": 0.0214, "step": 23770 }, { "epoch": 181.5267175572519, "grad_norm": 1.363983392715454, "learning_rate": 0.0001, "loss": 0.0216, "step": 23780 }, { "epoch": 181.6030534351145, "grad_norm": 1.289522409439087, "learning_rate": 0.0001, "loss": 0.022, "step": 23790 }, { "epoch": 181.6793893129771, "grad_norm": 1.254873275756836, "learning_rate": 0.0001, "loss": 0.0229, "step": 23800 }, { "epoch": 181.75572519083968, "grad_norm": 1.754231572151184, "learning_rate": 0.0001, "loss": 0.0214, "step": 23810 }, { "epoch": 181.8320610687023, "grad_norm": 1.6247626543045044, "learning_rate": 0.0001, "loss": 0.0219, "step": 23820 }, { "epoch": 181.9083969465649, "grad_norm": 1.6458961963653564, "learning_rate": 0.0001, "loss": 0.0219, "step": 23830 }, { "epoch": 181.98473282442748, "grad_norm": 1.4291177988052368, "learning_rate": 0.0001, "loss": 0.021, "step": 23840 }, { "epoch": 182.06106870229007, "grad_norm": 1.4339404106140137, "learning_rate": 0.0001, "loss": 0.022, "step": 23850 }, { "epoch": 182.13740458015266, "grad_norm": 1.625072717666626, "learning_rate": 0.0001, "loss": 0.0227, "step": 23860 }, { "epoch": 182.21374045801528, "grad_norm": 1.7074978351593018, "learning_rate": 0.0001, "loss": 0.0238, "step": 23870 }, { "epoch": 182.29007633587787, "grad_norm": 1.8346022367477417, "learning_rate": 0.0001, "loss": 0.0244, "step": 23880 }, { "epoch": 182.36641221374046, "grad_norm": 1.7996799945831299, "learning_rate": 0.0001, "loss": 0.0234, "step": 23890 }, { "epoch": 182.44274809160305, "grad_norm": 1.5501830577850342, "learning_rate": 0.0001, "loss": 0.0227, "step": 23900 }, { "epoch": 182.51908396946564, "grad_norm": 1.4164342880249023, "learning_rate": 0.0001, "loss": 0.0225, "step": 23910 }, { "epoch": 182.59541984732823, "grad_norm": 1.707708477973938, "learning_rate": 0.0001, "loss": 0.0216, "step": 23920 }, { "epoch": 182.67175572519085, "grad_norm": 1.5814367532730103, "learning_rate": 0.0001, "loss": 0.0218, "step": 23930 }, { "epoch": 182.74809160305344, "grad_norm": 0.991327702999115, "learning_rate": 0.0001, "loss": 0.0208, "step": 23940 }, { "epoch": 182.82442748091603, "grad_norm": 1.4550538063049316, "learning_rate": 0.0001, "loss": 0.0213, "step": 23950 }, { "epoch": 182.90076335877862, "grad_norm": 1.621055245399475, "learning_rate": 0.0001, "loss": 0.0206, "step": 23960 }, { "epoch": 182.9770992366412, "grad_norm": 2.0230774879455566, "learning_rate": 0.0001, "loss": 0.0212, "step": 23970 }, { "epoch": 183.05343511450383, "grad_norm": 1.9626582860946655, "learning_rate": 0.0001, "loss": 0.0214, "step": 23980 }, { "epoch": 183.12977099236642, "grad_norm": 2.1521873474121094, "learning_rate": 0.0001, "loss": 0.0213, "step": 23990 }, { "epoch": 183.206106870229, "grad_norm": 1.5870939493179321, "learning_rate": 0.0001, "loss": 0.0218, "step": 24000 }, { "epoch": 183.2824427480916, "grad_norm": 1.4056739807128906, "learning_rate": 0.0001, "loss": 0.0208, "step": 24010 }, { "epoch": 183.3587786259542, "grad_norm": 1.3566814661026, "learning_rate": 0.0001, "loss": 0.0212, "step": 24020 }, { "epoch": 183.4351145038168, "grad_norm": 1.5375303030014038, "learning_rate": 0.0001, "loss": 0.0218, "step": 24030 }, { "epoch": 183.5114503816794, "grad_norm": 1.4487076997756958, "learning_rate": 0.0001, "loss": 0.0212, "step": 24040 }, { "epoch": 183.58778625954199, "grad_norm": 1.6572365760803223, "learning_rate": 0.0001, "loss": 0.0226, "step": 24050 }, { "epoch": 183.66412213740458, "grad_norm": 1.662877082824707, "learning_rate": 0.0001, "loss": 0.0218, "step": 24060 }, { "epoch": 183.74045801526717, "grad_norm": 1.8005154132843018, "learning_rate": 0.0001, "loss": 0.0223, "step": 24070 }, { "epoch": 183.81679389312978, "grad_norm": 1.6186962127685547, "learning_rate": 0.0001, "loss": 0.0225, "step": 24080 }, { "epoch": 183.89312977099237, "grad_norm": 1.987959384918213, "learning_rate": 0.0001, "loss": 0.0222, "step": 24090 }, { "epoch": 183.96946564885496, "grad_norm": 1.678130865097046, "learning_rate": 0.0001, "loss": 0.0212, "step": 24100 }, { "epoch": 184.04580152671755, "grad_norm": 1.8585387468338013, "learning_rate": 0.0001, "loss": 0.021, "step": 24110 }, { "epoch": 184.12213740458014, "grad_norm": 2.800595760345459, "learning_rate": 0.0001, "loss": 0.022, "step": 24120 }, { "epoch": 184.19847328244273, "grad_norm": 2.0730392932891846, "learning_rate": 0.0001, "loss": 0.0214, "step": 24130 }, { "epoch": 184.27480916030535, "grad_norm": 2.462068796157837, "learning_rate": 0.0001, "loss": 0.0207, "step": 24140 }, { "epoch": 184.35114503816794, "grad_norm": 1.6177124977111816, "learning_rate": 0.0001, "loss": 0.0211, "step": 24150 }, { "epoch": 184.42748091603053, "grad_norm": 2.919520854949951, "learning_rate": 0.0001, "loss": 0.0207, "step": 24160 }, { "epoch": 184.50381679389312, "grad_norm": 2.103503465652466, "learning_rate": 0.0001, "loss": 0.0207, "step": 24170 }, { "epoch": 184.5801526717557, "grad_norm": 1.8151739835739136, "learning_rate": 0.0001, "loss": 0.0204, "step": 24180 }, { "epoch": 184.65648854961833, "grad_norm": 2.707106113433838, "learning_rate": 0.0001, "loss": 0.0207, "step": 24190 }, { "epoch": 184.73282442748092, "grad_norm": 1.8534914255142212, "learning_rate": 0.0001, "loss": 0.0203, "step": 24200 }, { "epoch": 184.8091603053435, "grad_norm": 2.129941701889038, "learning_rate": 0.0001, "loss": 0.0203, "step": 24210 }, { "epoch": 184.8854961832061, "grad_norm": 1.6789530515670776, "learning_rate": 0.0001, "loss": 0.0204, "step": 24220 }, { "epoch": 184.9618320610687, "grad_norm": 2.2978951930999756, "learning_rate": 0.0001, "loss": 0.0208, "step": 24230 }, { "epoch": 185.0381679389313, "grad_norm": 1.4639513492584229, "learning_rate": 0.0001, "loss": 0.0206, "step": 24240 }, { "epoch": 185.1145038167939, "grad_norm": 2.227553606033325, "learning_rate": 0.0001, "loss": 0.0211, "step": 24250 }, { "epoch": 185.1908396946565, "grad_norm": 2.034788131713867, "learning_rate": 0.0001, "loss": 0.0209, "step": 24260 }, { "epoch": 185.26717557251908, "grad_norm": 2.187082052230835, "learning_rate": 0.0001, "loss": 0.0211, "step": 24270 }, { "epoch": 185.34351145038167, "grad_norm": 2.472611427307129, "learning_rate": 0.0001, "loss": 0.0214, "step": 24280 }, { "epoch": 185.4198473282443, "grad_norm": 2.05346941947937, "learning_rate": 0.0001, "loss": 0.0206, "step": 24290 }, { "epoch": 185.49618320610688, "grad_norm": 1.5020560026168823, "learning_rate": 0.0001, "loss": 0.0198, "step": 24300 }, { "epoch": 185.57251908396947, "grad_norm": 1.895699143409729, "learning_rate": 0.0001, "loss": 0.0199, "step": 24310 }, { "epoch": 185.64885496183206, "grad_norm": 1.9496965408325195, "learning_rate": 0.0001, "loss": 0.02, "step": 24320 }, { "epoch": 185.72519083969465, "grad_norm": 1.7097110748291016, "learning_rate": 0.0001, "loss": 0.0199, "step": 24330 }, { "epoch": 185.80152671755727, "grad_norm": 1.3029290437698364, "learning_rate": 0.0001, "loss": 0.0194, "step": 24340 }, { "epoch": 185.87786259541986, "grad_norm": 1.6116409301757812, "learning_rate": 0.0001, "loss": 0.0207, "step": 24350 }, { "epoch": 185.95419847328245, "grad_norm": 1.9740408658981323, "learning_rate": 0.0001, "loss": 0.0203, "step": 24360 }, { "epoch": 186.03053435114504, "grad_norm": 1.556488037109375, "learning_rate": 0.0001, "loss": 0.0209, "step": 24370 }, { "epoch": 186.10687022900763, "grad_norm": 1.4059616327285767, "learning_rate": 0.0001, "loss": 0.0215, "step": 24380 }, { "epoch": 186.18320610687022, "grad_norm": 1.8327434062957764, "learning_rate": 0.0001, "loss": 0.0222, "step": 24390 }, { "epoch": 186.25954198473283, "grad_norm": 2.3722164630889893, "learning_rate": 0.0001, "loss": 0.0231, "step": 24400 }, { "epoch": 186.33587786259542, "grad_norm": 2.0672130584716797, "learning_rate": 0.0001, "loss": 0.0218, "step": 24410 }, { "epoch": 186.41221374045801, "grad_norm": 2.1230273246765137, "learning_rate": 0.0001, "loss": 0.0209, "step": 24420 }, { "epoch": 186.4885496183206, "grad_norm": 1.9044562578201294, "learning_rate": 0.0001, "loss": 0.0206, "step": 24430 }, { "epoch": 186.5648854961832, "grad_norm": 1.579028606414795, "learning_rate": 0.0001, "loss": 0.0201, "step": 24440 }, { "epoch": 186.6412213740458, "grad_norm": 2.0431718826293945, "learning_rate": 0.0001, "loss": 0.0199, "step": 24450 }, { "epoch": 186.7175572519084, "grad_norm": 1.76737642288208, "learning_rate": 0.0001, "loss": 0.0202, "step": 24460 }, { "epoch": 186.793893129771, "grad_norm": 1.4224921464920044, "learning_rate": 0.0001, "loss": 0.0195, "step": 24470 }, { "epoch": 186.87022900763358, "grad_norm": 1.326766014099121, "learning_rate": 0.0001, "loss": 0.0195, "step": 24480 }, { "epoch": 186.94656488549617, "grad_norm": 1.6101688146591187, "learning_rate": 0.0001, "loss": 0.02, "step": 24490 }, { "epoch": 187.0229007633588, "grad_norm": 1.6738961935043335, "learning_rate": 0.0001, "loss": 0.0198, "step": 24500 }, { "epoch": 187.09923664122138, "grad_norm": 1.1907485723495483, "learning_rate": 0.0001, "loss": 0.0205, "step": 24510 }, { "epoch": 187.17557251908397, "grad_norm": 1.8487001657485962, "learning_rate": 0.0001, "loss": 0.0207, "step": 24520 }, { "epoch": 187.25190839694656, "grad_norm": 1.3636102676391602, "learning_rate": 0.0001, "loss": 0.021, "step": 24530 }, { "epoch": 187.32824427480915, "grad_norm": 1.7319059371948242, "learning_rate": 0.0001, "loss": 0.0213, "step": 24540 }, { "epoch": 187.40458015267177, "grad_norm": 2.5530214309692383, "learning_rate": 0.0001, "loss": 0.0206, "step": 24550 }, { "epoch": 187.48091603053436, "grad_norm": 2.6330156326293945, "learning_rate": 0.0001, "loss": 0.0209, "step": 24560 }, { "epoch": 187.55725190839695, "grad_norm": 2.077301263809204, "learning_rate": 0.0001, "loss": 0.0206, "step": 24570 }, { "epoch": 187.63358778625954, "grad_norm": 2.1170239448547363, "learning_rate": 0.0001, "loss": 0.021, "step": 24580 }, { "epoch": 187.70992366412213, "grad_norm": 2.1051957607269287, "learning_rate": 0.0001, "loss": 0.0206, "step": 24590 }, { "epoch": 187.78625954198472, "grad_norm": 2.341588020324707, "learning_rate": 0.0001, "loss": 0.0209, "step": 24600 }, { "epoch": 187.86259541984734, "grad_norm": 1.614212155342102, "learning_rate": 0.0001, "loss": 0.021, "step": 24610 }, { "epoch": 187.93893129770993, "grad_norm": 2.516505479812622, "learning_rate": 0.0001, "loss": 0.0209, "step": 24620 }, { "epoch": 188.01526717557252, "grad_norm": 2.0920915603637695, "learning_rate": 0.0001, "loss": 0.0211, "step": 24630 }, { "epoch": 188.0916030534351, "grad_norm": 2.0410637855529785, "learning_rate": 0.0001, "loss": 0.0203, "step": 24640 }, { "epoch": 188.1679389312977, "grad_norm": 1.686797022819519, "learning_rate": 0.0001, "loss": 0.0202, "step": 24650 }, { "epoch": 188.24427480916032, "grad_norm": 1.5657457113265991, "learning_rate": 0.0001, "loss": 0.0204, "step": 24660 }, { "epoch": 188.3206106870229, "grad_norm": 1.560410737991333, "learning_rate": 0.0001, "loss": 0.0205, "step": 24670 }, { "epoch": 188.3969465648855, "grad_norm": 1.5390305519104004, "learning_rate": 0.0001, "loss": 0.0208, "step": 24680 }, { "epoch": 188.4732824427481, "grad_norm": 1.6408581733703613, "learning_rate": 0.0001, "loss": 0.0207, "step": 24690 }, { "epoch": 188.54961832061068, "grad_norm": 1.7326037883758545, "learning_rate": 0.0001, "loss": 0.0213, "step": 24700 }, { "epoch": 188.6259541984733, "grad_norm": 1.9865597486495972, "learning_rate": 0.0001, "loss": 0.0223, "step": 24710 }, { "epoch": 188.70229007633588, "grad_norm": 2.563878297805786, "learning_rate": 0.0001, "loss": 0.0218, "step": 24720 }, { "epoch": 188.77862595419847, "grad_norm": 1.9122099876403809, "learning_rate": 0.0001, "loss": 0.0212, "step": 24730 }, { "epoch": 188.85496183206106, "grad_norm": 1.5954242944717407, "learning_rate": 0.0001, "loss": 0.0211, "step": 24740 }, { "epoch": 188.93129770992365, "grad_norm": 1.6656742095947266, "learning_rate": 0.0001, "loss": 0.0208, "step": 24750 }, { "epoch": 189.00763358778627, "grad_norm": 1.8426915407180786, "learning_rate": 0.0001, "loss": 0.0211, "step": 24760 }, { "epoch": 189.08396946564886, "grad_norm": 1.5240758657455444, "learning_rate": 0.0001, "loss": 0.021, "step": 24770 }, { "epoch": 189.16030534351145, "grad_norm": 2.47446346282959, "learning_rate": 0.0001, "loss": 0.021, "step": 24780 }, { "epoch": 189.23664122137404, "grad_norm": 2.4581313133239746, "learning_rate": 0.0001, "loss": 0.0214, "step": 24790 }, { "epoch": 189.31297709923663, "grad_norm": 1.7017483711242676, "learning_rate": 0.0001, "loss": 0.0209, "step": 24800 }, { "epoch": 189.38931297709922, "grad_norm": 1.8851197957992554, "learning_rate": 0.0001, "loss": 0.0204, "step": 24810 }, { "epoch": 189.46564885496184, "grad_norm": 1.4101073741912842, "learning_rate": 0.0001, "loss": 0.0192, "step": 24820 }, { "epoch": 189.54198473282443, "grad_norm": 1.4338527917861938, "learning_rate": 0.0001, "loss": 0.0199, "step": 24830 }, { "epoch": 189.61832061068702, "grad_norm": 1.7569819688796997, "learning_rate": 0.0001, "loss": 0.021, "step": 24840 }, { "epoch": 189.6946564885496, "grad_norm": 1.742276906967163, "learning_rate": 0.0001, "loss": 0.0201, "step": 24850 }, { "epoch": 189.7709923664122, "grad_norm": 2.006625175476074, "learning_rate": 0.0001, "loss": 0.0214, "step": 24860 }, { "epoch": 189.84732824427482, "grad_norm": 1.8266856670379639, "learning_rate": 0.0001, "loss": 0.0209, "step": 24870 }, { "epoch": 189.9236641221374, "grad_norm": 1.8758834600448608, "learning_rate": 0.0001, "loss": 0.0205, "step": 24880 }, { "epoch": 190.0, "grad_norm": 1.9195538759231567, "learning_rate": 0.0001, "loss": 0.0208, "step": 24890 }, { "epoch": 190.0763358778626, "grad_norm": 1.6389490365982056, "learning_rate": 0.0001, "loss": 0.0202, "step": 24900 }, { "epoch": 190.15267175572518, "grad_norm": 1.7357898950576782, "learning_rate": 0.0001, "loss": 0.0203, "step": 24910 }, { "epoch": 190.2290076335878, "grad_norm": 1.4256343841552734, "learning_rate": 0.0001, "loss": 0.0205, "step": 24920 }, { "epoch": 190.3053435114504, "grad_norm": 1.1464303731918335, "learning_rate": 0.0001, "loss": 0.0208, "step": 24930 }, { "epoch": 190.38167938931298, "grad_norm": 2.0040061473846436, "learning_rate": 0.0001, "loss": 0.0213, "step": 24940 }, { "epoch": 190.45801526717557, "grad_norm": 1.5865610837936401, "learning_rate": 0.0001, "loss": 0.0203, "step": 24950 }, { "epoch": 190.53435114503816, "grad_norm": 1.9267361164093018, "learning_rate": 0.0001, "loss": 0.0214, "step": 24960 }, { "epoch": 190.61068702290078, "grad_norm": 2.169950008392334, "learning_rate": 0.0001, "loss": 0.0209, "step": 24970 }, { "epoch": 190.68702290076337, "grad_norm": 1.487210750579834, "learning_rate": 0.0001, "loss": 0.0209, "step": 24980 }, { "epoch": 190.76335877862596, "grad_norm": 1.936001181602478, "learning_rate": 0.0001, "loss": 0.0205, "step": 24990 }, { "epoch": 190.83969465648855, "grad_norm": 2.012246608734131, "learning_rate": 0.0001, "loss": 0.0219, "step": 25000 }, { "epoch": 190.91603053435114, "grad_norm": 1.6860244274139404, "learning_rate": 0.0001, "loss": 0.021, "step": 25010 }, { "epoch": 190.99236641221373, "grad_norm": 1.7539801597595215, "learning_rate": 0.0001, "loss": 0.0204, "step": 25020 }, { "epoch": 191.06870229007635, "grad_norm": 1.6664352416992188, "learning_rate": 0.0001, "loss": 0.0206, "step": 25030 }, { "epoch": 191.14503816793894, "grad_norm": 2.0140435695648193, "learning_rate": 0.0001, "loss": 0.0212, "step": 25040 }, { "epoch": 191.22137404580153, "grad_norm": 1.685986042022705, "learning_rate": 0.0001, "loss": 0.0212, "step": 25050 }, { "epoch": 191.29770992366412, "grad_norm": 2.2997403144836426, "learning_rate": 0.0001, "loss": 0.0212, "step": 25060 }, { "epoch": 191.3740458015267, "grad_norm": 2.013061285018921, "learning_rate": 0.0001, "loss": 0.021, "step": 25070 }, { "epoch": 191.45038167938932, "grad_norm": 1.6849156618118286, "learning_rate": 0.0001, "loss": 0.021, "step": 25080 }, { "epoch": 191.5267175572519, "grad_norm": 1.7962619066238403, "learning_rate": 0.0001, "loss": 0.0212, "step": 25090 }, { "epoch": 191.6030534351145, "grad_norm": 1.5315865278244019, "learning_rate": 0.0001, "loss": 0.0214, "step": 25100 }, { "epoch": 191.6793893129771, "grad_norm": 1.5507243871688843, "learning_rate": 0.0001, "loss": 0.0206, "step": 25110 }, { "epoch": 191.75572519083968, "grad_norm": 1.5645508766174316, "learning_rate": 0.0001, "loss": 0.0202, "step": 25120 }, { "epoch": 191.8320610687023, "grad_norm": 1.5418697595596313, "learning_rate": 0.0001, "loss": 0.0206, "step": 25130 }, { "epoch": 191.9083969465649, "grad_norm": 1.83246910572052, "learning_rate": 0.0001, "loss": 0.021, "step": 25140 }, { "epoch": 191.98473282442748, "grad_norm": 1.6846472024917603, "learning_rate": 0.0001, "loss": 0.0218, "step": 25150 }, { "epoch": 192.06106870229007, "grad_norm": 1.7745792865753174, "learning_rate": 0.0001, "loss": 0.0219, "step": 25160 }, { "epoch": 192.13740458015266, "grad_norm": 1.8291758298873901, "learning_rate": 0.0001, "loss": 0.0212, "step": 25170 }, { "epoch": 192.21374045801528, "grad_norm": 2.0109879970550537, "learning_rate": 0.0001, "loss": 0.021, "step": 25180 }, { "epoch": 192.29007633587787, "grad_norm": 2.0082485675811768, "learning_rate": 0.0001, "loss": 0.0211, "step": 25190 }, { "epoch": 192.36641221374046, "grad_norm": 1.7560397386550903, "learning_rate": 0.0001, "loss": 0.0204, "step": 25200 }, { "epoch": 192.44274809160305, "grad_norm": 1.622100591659546, "learning_rate": 0.0001, "loss": 0.0199, "step": 25210 }, { "epoch": 192.51908396946564, "grad_norm": 1.4455454349517822, "learning_rate": 0.0001, "loss": 0.0212, "step": 25220 }, { "epoch": 192.59541984732823, "grad_norm": 1.6919325590133667, "learning_rate": 0.0001, "loss": 0.021, "step": 25230 }, { "epoch": 192.67175572519085, "grad_norm": 1.7698153257369995, "learning_rate": 0.0001, "loss": 0.0205, "step": 25240 }, { "epoch": 192.74809160305344, "grad_norm": 1.8345613479614258, "learning_rate": 0.0001, "loss": 0.0204, "step": 25250 }, { "epoch": 192.82442748091603, "grad_norm": 2.0554630756378174, "learning_rate": 0.0001, "loss": 0.0204, "step": 25260 }, { "epoch": 192.90076335877862, "grad_norm": 1.207247257232666, "learning_rate": 0.0001, "loss": 0.0195, "step": 25270 }, { "epoch": 192.9770992366412, "grad_norm": 1.9996036291122437, "learning_rate": 0.0001, "loss": 0.02, "step": 25280 }, { "epoch": 193.05343511450383, "grad_norm": 1.3551374673843384, "learning_rate": 0.0001, "loss": 0.02, "step": 25290 }, { "epoch": 193.12977099236642, "grad_norm": 1.1612240076065063, "learning_rate": 0.0001, "loss": 0.0206, "step": 25300 }, { "epoch": 193.206106870229, "grad_norm": 1.5771969556808472, "learning_rate": 0.0001, "loss": 0.0207, "step": 25310 }, { "epoch": 193.2824427480916, "grad_norm": 1.3401286602020264, "learning_rate": 0.0001, "loss": 0.0201, "step": 25320 }, { "epoch": 193.3587786259542, "grad_norm": 1.8073445558547974, "learning_rate": 0.0001, "loss": 0.0198, "step": 25330 }, { "epoch": 193.4351145038168, "grad_norm": 1.5679103136062622, "learning_rate": 0.0001, "loss": 0.0211, "step": 25340 }, { "epoch": 193.5114503816794, "grad_norm": 1.6002110242843628, "learning_rate": 0.0001, "loss": 0.0206, "step": 25350 }, { "epoch": 193.58778625954199, "grad_norm": 2.028174877166748, "learning_rate": 0.0001, "loss": 0.0201, "step": 25360 }, { "epoch": 193.66412213740458, "grad_norm": 1.377624750137329, "learning_rate": 0.0001, "loss": 0.0202, "step": 25370 }, { "epoch": 193.74045801526717, "grad_norm": 2.047410726547241, "learning_rate": 0.0001, "loss": 0.0209, "step": 25380 }, { "epoch": 193.81679389312978, "grad_norm": 1.5137287378311157, "learning_rate": 0.0001, "loss": 0.0205, "step": 25390 }, { "epoch": 193.89312977099237, "grad_norm": 2.174386739730835, "learning_rate": 0.0001, "loss": 0.0203, "step": 25400 }, { "epoch": 193.96946564885496, "grad_norm": 1.7005951404571533, "learning_rate": 0.0001, "loss": 0.0203, "step": 25410 }, { "epoch": 194.04580152671755, "grad_norm": 1.681911587715149, "learning_rate": 0.0001, "loss": 0.0192, "step": 25420 }, { "epoch": 194.12213740458014, "grad_norm": 1.9381603002548218, "learning_rate": 0.0001, "loss": 0.0204, "step": 25430 }, { "epoch": 194.19847328244273, "grad_norm": 1.8146319389343262, "learning_rate": 0.0001, "loss": 0.0204, "step": 25440 }, { "epoch": 194.27480916030535, "grad_norm": 1.3676148653030396, "learning_rate": 0.0001, "loss": 0.0205, "step": 25450 }, { "epoch": 194.35114503816794, "grad_norm": 1.580313801765442, "learning_rate": 0.0001, "loss": 0.0207, "step": 25460 }, { "epoch": 194.42748091603053, "grad_norm": 1.796258807182312, "learning_rate": 0.0001, "loss": 0.0209, "step": 25470 }, { "epoch": 194.50381679389312, "grad_norm": 1.5504627227783203, "learning_rate": 0.0001, "loss": 0.0213, "step": 25480 }, { "epoch": 194.5801526717557, "grad_norm": 1.4488433599472046, "learning_rate": 0.0001, "loss": 0.0217, "step": 25490 }, { "epoch": 194.65648854961833, "grad_norm": 1.4247689247131348, "learning_rate": 0.0001, "loss": 0.0211, "step": 25500 }, { "epoch": 194.73282442748092, "grad_norm": 1.7369853258132935, "learning_rate": 0.0001, "loss": 0.0215, "step": 25510 }, { "epoch": 194.8091603053435, "grad_norm": 2.550147771835327, "learning_rate": 0.0001, "loss": 0.0212, "step": 25520 }, { "epoch": 194.8854961832061, "grad_norm": 1.7774994373321533, "learning_rate": 0.0001, "loss": 0.0215, "step": 25530 }, { "epoch": 194.9618320610687, "grad_norm": 1.8196895122528076, "learning_rate": 0.0001, "loss": 0.0218, "step": 25540 }, { "epoch": 195.0381679389313, "grad_norm": 1.6165357828140259, "learning_rate": 0.0001, "loss": 0.0211, "step": 25550 }, { "epoch": 195.1145038167939, "grad_norm": 1.4221878051757812, "learning_rate": 0.0001, "loss": 0.0214, "step": 25560 }, { "epoch": 195.1908396946565, "grad_norm": 1.2976670265197754, "learning_rate": 0.0001, "loss": 0.0219, "step": 25570 }, { "epoch": 195.26717557251908, "grad_norm": 2.2212655544281006, "learning_rate": 0.0001, "loss": 0.0202, "step": 25580 }, { "epoch": 195.34351145038167, "grad_norm": 1.5515273809432983, "learning_rate": 0.0001, "loss": 0.0205, "step": 25590 }, { "epoch": 195.4198473282443, "grad_norm": 1.6764265298843384, "learning_rate": 0.0001, "loss": 0.0223, "step": 25600 }, { "epoch": 195.49618320610688, "grad_norm": 1.7600173950195312, "learning_rate": 0.0001, "loss": 0.0229, "step": 25610 }, { "epoch": 195.57251908396947, "grad_norm": 1.5097618103027344, "learning_rate": 0.0001, "loss": 0.0231, "step": 25620 }, { "epoch": 195.64885496183206, "grad_norm": 2.06498122215271, "learning_rate": 0.0001, "loss": 0.0222, "step": 25630 }, { "epoch": 195.72519083969465, "grad_norm": 2.4085283279418945, "learning_rate": 0.0001, "loss": 0.0223, "step": 25640 }, { "epoch": 195.80152671755727, "grad_norm": 1.88239586353302, "learning_rate": 0.0001, "loss": 0.0221, "step": 25650 }, { "epoch": 195.87786259541986, "grad_norm": 1.5626360177993774, "learning_rate": 0.0001, "loss": 0.022, "step": 25660 }, { "epoch": 195.95419847328245, "grad_norm": 1.9201804399490356, "learning_rate": 0.0001, "loss": 0.0209, "step": 25670 }, { "epoch": 196.03053435114504, "grad_norm": 1.4151155948638916, "learning_rate": 0.0001, "loss": 0.0203, "step": 25680 }, { "epoch": 196.10687022900763, "grad_norm": 2.089506149291992, "learning_rate": 0.0001, "loss": 0.0213, "step": 25690 }, { "epoch": 196.18320610687022, "grad_norm": 1.5939885377883911, "learning_rate": 0.0001, "loss": 0.0205, "step": 25700 }, { "epoch": 196.25954198473283, "grad_norm": 1.898643970489502, "learning_rate": 0.0001, "loss": 0.0203, "step": 25710 }, { "epoch": 196.33587786259542, "grad_norm": 2.1500778198242188, "learning_rate": 0.0001, "loss": 0.0203, "step": 25720 }, { "epoch": 196.41221374045801, "grad_norm": 1.9085943698883057, "learning_rate": 0.0001, "loss": 0.0203, "step": 25730 }, { "epoch": 196.4885496183206, "grad_norm": 2.607948064804077, "learning_rate": 0.0001, "loss": 0.0202, "step": 25740 }, { "epoch": 196.5648854961832, "grad_norm": 1.820949912071228, "learning_rate": 0.0001, "loss": 0.0204, "step": 25750 }, { "epoch": 196.6412213740458, "grad_norm": 1.6572890281677246, "learning_rate": 0.0001, "loss": 0.0191, "step": 25760 }, { "epoch": 196.7175572519084, "grad_norm": 2.128269910812378, "learning_rate": 0.0001, "loss": 0.0203, "step": 25770 }, { "epoch": 196.793893129771, "grad_norm": 1.5117156505584717, "learning_rate": 0.0001, "loss": 0.0206, "step": 25780 }, { "epoch": 196.87022900763358, "grad_norm": 1.9089518785476685, "learning_rate": 0.0001, "loss": 0.0203, "step": 25790 }, { "epoch": 196.94656488549617, "grad_norm": 1.5729318857192993, "learning_rate": 0.0001, "loss": 0.0218, "step": 25800 }, { "epoch": 197.0229007633588, "grad_norm": 1.4225314855575562, "learning_rate": 0.0001, "loss": 0.0207, "step": 25810 }, { "epoch": 197.09923664122138, "grad_norm": 1.952863097190857, "learning_rate": 0.0001, "loss": 0.0216, "step": 25820 }, { "epoch": 197.17557251908397, "grad_norm": 1.9769736528396606, "learning_rate": 0.0001, "loss": 0.0217, "step": 25830 }, { "epoch": 197.25190839694656, "grad_norm": 1.2563809156417847, "learning_rate": 0.0001, "loss": 0.0218, "step": 25840 }, { "epoch": 197.32824427480915, "grad_norm": 1.8597203493118286, "learning_rate": 0.0001, "loss": 0.0205, "step": 25850 }, { "epoch": 197.40458015267177, "grad_norm": 1.3368136882781982, "learning_rate": 0.0001, "loss": 0.0215, "step": 25860 }, { "epoch": 197.48091603053436, "grad_norm": 1.3501331806182861, "learning_rate": 0.0001, "loss": 0.0212, "step": 25870 }, { "epoch": 197.55725190839695, "grad_norm": 1.5942165851593018, "learning_rate": 0.0001, "loss": 0.0215, "step": 25880 }, { "epoch": 197.63358778625954, "grad_norm": 1.2879201173782349, "learning_rate": 0.0001, "loss": 0.0223, "step": 25890 }, { "epoch": 197.70992366412213, "grad_norm": 1.4393723011016846, "learning_rate": 0.0001, "loss": 0.023, "step": 25900 }, { "epoch": 197.78625954198472, "grad_norm": 1.791897177696228, "learning_rate": 0.0001, "loss": 0.0238, "step": 25910 }, { "epoch": 197.86259541984734, "grad_norm": 1.1778590679168701, "learning_rate": 0.0001, "loss": 0.0218, "step": 25920 }, { "epoch": 197.93893129770993, "grad_norm": 1.344442367553711, "learning_rate": 0.0001, "loss": 0.0213, "step": 25930 }, { "epoch": 198.01526717557252, "grad_norm": 1.4362016916275024, "learning_rate": 0.0001, "loss": 0.0211, "step": 25940 }, { "epoch": 198.0916030534351, "grad_norm": 1.273647427558899, "learning_rate": 0.0001, "loss": 0.0207, "step": 25950 }, { "epoch": 198.1679389312977, "grad_norm": 1.3756766319274902, "learning_rate": 0.0001, "loss": 0.0208, "step": 25960 }, { "epoch": 198.24427480916032, "grad_norm": 2.4455935955047607, "learning_rate": 0.0001, "loss": 0.0205, "step": 25970 }, { "epoch": 198.3206106870229, "grad_norm": 1.862897515296936, "learning_rate": 0.0001, "loss": 0.0207, "step": 25980 }, { "epoch": 198.3969465648855, "grad_norm": 2.104339122772217, "learning_rate": 0.0001, "loss": 0.0214, "step": 25990 }, { "epoch": 198.4732824427481, "grad_norm": 1.591981291770935, "learning_rate": 0.0001, "loss": 0.0211, "step": 26000 }, { "epoch": 198.54961832061068, "grad_norm": 1.748861312866211, "learning_rate": 0.0001, "loss": 0.021, "step": 26010 }, { "epoch": 198.6259541984733, "grad_norm": 1.9624388217926025, "learning_rate": 0.0001, "loss": 0.0206, "step": 26020 }, { "epoch": 198.70229007633588, "grad_norm": 1.8516641855239868, "learning_rate": 0.0001, "loss": 0.0218, "step": 26030 }, { "epoch": 198.77862595419847, "grad_norm": 1.784555435180664, "learning_rate": 0.0001, "loss": 0.0216, "step": 26040 }, { "epoch": 198.85496183206106, "grad_norm": 1.2281900644302368, "learning_rate": 0.0001, "loss": 0.0213, "step": 26050 }, { "epoch": 198.93129770992365, "grad_norm": 1.7220531702041626, "learning_rate": 0.0001, "loss": 0.0204, "step": 26060 }, { "epoch": 199.00763358778627, "grad_norm": 1.7228924036026, "learning_rate": 0.0001, "loss": 0.0207, "step": 26070 }, { "epoch": 199.08396946564886, "grad_norm": 1.4724980592727661, "learning_rate": 0.0001, "loss": 0.0202, "step": 26080 }, { "epoch": 199.16030534351145, "grad_norm": 1.5389896631240845, "learning_rate": 0.0001, "loss": 0.0206, "step": 26090 }, { "epoch": 199.23664122137404, "grad_norm": 1.3764731884002686, "learning_rate": 0.0001, "loss": 0.0204, "step": 26100 }, { "epoch": 199.31297709923663, "grad_norm": 1.8386181592941284, "learning_rate": 0.0001, "loss": 0.0203, "step": 26110 }, { "epoch": 199.38931297709922, "grad_norm": 1.9860702753067017, "learning_rate": 0.0001, "loss": 0.0215, "step": 26120 }, { "epoch": 199.46564885496184, "grad_norm": 2.3243515491485596, "learning_rate": 0.0001, "loss": 0.0215, "step": 26130 }, { "epoch": 199.54198473282443, "grad_norm": 2.5021979808807373, "learning_rate": 0.0001, "loss": 0.0215, "step": 26140 }, { "epoch": 199.61832061068702, "grad_norm": 1.7285618782043457, "learning_rate": 0.0001, "loss": 0.0219, "step": 26150 }, { "epoch": 199.6946564885496, "grad_norm": 1.8069965839385986, "learning_rate": 0.0001, "loss": 0.0218, "step": 26160 }, { "epoch": 199.7709923664122, "grad_norm": 2.117262125015259, "learning_rate": 0.0001, "loss": 0.0217, "step": 26170 }, { "epoch": 199.84732824427482, "grad_norm": 1.4245035648345947, "learning_rate": 0.0001, "loss": 0.0211, "step": 26180 }, { "epoch": 199.9236641221374, "grad_norm": 1.5899196863174438, "learning_rate": 0.0001, "loss": 0.02, "step": 26190 }, { "epoch": 200.0, "grad_norm": 1.827039361000061, "learning_rate": 0.0001, "loss": 0.0201, "step": 26200 }, { "epoch": 200.0763358778626, "grad_norm": 1.4052727222442627, "learning_rate": 0.0001, "loss": 0.0209, "step": 26210 }, { "epoch": 200.15267175572518, "grad_norm": 1.5331180095672607, "learning_rate": 0.0001, "loss": 0.0217, "step": 26220 }, { "epoch": 200.2290076335878, "grad_norm": 1.9471148252487183, "learning_rate": 0.0001, "loss": 0.0215, "step": 26230 }, { "epoch": 200.3053435114504, "grad_norm": 2.5415072441101074, "learning_rate": 0.0001, "loss": 0.0214, "step": 26240 }, { "epoch": 200.38167938931298, "grad_norm": 2.3740501403808594, "learning_rate": 0.0001, "loss": 0.021, "step": 26250 }, { "epoch": 200.45801526717557, "grad_norm": 1.9660924673080444, "learning_rate": 0.0001, "loss": 0.0202, "step": 26260 }, { "epoch": 200.53435114503816, "grad_norm": 2.5231664180755615, "learning_rate": 0.0001, "loss": 0.0206, "step": 26270 }, { "epoch": 200.61068702290078, "grad_norm": 1.5139880180358887, "learning_rate": 0.0001, "loss": 0.0202, "step": 26280 }, { "epoch": 200.68702290076337, "grad_norm": 1.5237690210342407, "learning_rate": 0.0001, "loss": 0.0196, "step": 26290 }, { "epoch": 200.76335877862596, "grad_norm": 1.4430533647537231, "learning_rate": 0.0001, "loss": 0.0199, "step": 26300 }, { "epoch": 200.83969465648855, "grad_norm": 1.7605549097061157, "learning_rate": 0.0001, "loss": 0.0204, "step": 26310 }, { "epoch": 200.91603053435114, "grad_norm": 1.6951361894607544, "learning_rate": 0.0001, "loss": 0.0216, "step": 26320 }, { "epoch": 200.99236641221373, "grad_norm": 2.0131945610046387, "learning_rate": 0.0001, "loss": 0.0219, "step": 26330 }, { "epoch": 201.06870229007635, "grad_norm": 1.4570903778076172, "learning_rate": 0.0001, "loss": 0.0206, "step": 26340 }, { "epoch": 201.14503816793894, "grad_norm": 1.58893620967865, "learning_rate": 0.0001, "loss": 0.0216, "step": 26350 }, { "epoch": 201.22137404580153, "grad_norm": 1.8565753698349, "learning_rate": 0.0001, "loss": 0.0213, "step": 26360 }, { "epoch": 201.29770992366412, "grad_norm": 2.8051223754882812, "learning_rate": 0.0001, "loss": 0.0221, "step": 26370 }, { "epoch": 201.3740458015267, "grad_norm": 2.968824863433838, "learning_rate": 0.0001, "loss": 0.0211, "step": 26380 }, { "epoch": 201.45038167938932, "grad_norm": 3.2743008136749268, "learning_rate": 0.0001, "loss": 0.0205, "step": 26390 }, { "epoch": 201.5267175572519, "grad_norm": 2.040377140045166, "learning_rate": 0.0001, "loss": 0.0208, "step": 26400 }, { "epoch": 201.6030534351145, "grad_norm": 2.7658770084381104, "learning_rate": 0.0001, "loss": 0.0203, "step": 26410 }, { "epoch": 201.6793893129771, "grad_norm": 2.439661979675293, "learning_rate": 0.0001, "loss": 0.0197, "step": 26420 }, { "epoch": 201.75572519083968, "grad_norm": 1.5621082782745361, "learning_rate": 0.0001, "loss": 0.0196, "step": 26430 }, { "epoch": 201.8320610687023, "grad_norm": 1.810871958732605, "learning_rate": 0.0001, "loss": 0.0193, "step": 26440 }, { "epoch": 201.9083969465649, "grad_norm": 2.2744202613830566, "learning_rate": 0.0001, "loss": 0.0199, "step": 26450 }, { "epoch": 201.98473282442748, "grad_norm": 2.0999209880828857, "learning_rate": 0.0001, "loss": 0.0195, "step": 26460 }, { "epoch": 202.06106870229007, "grad_norm": 2.7262158393859863, "learning_rate": 0.0001, "loss": 0.0189, "step": 26470 }, { "epoch": 202.13740458015266, "grad_norm": 2.030538320541382, "learning_rate": 0.0001, "loss": 0.019, "step": 26480 }, { "epoch": 202.21374045801528, "grad_norm": 1.4374977350234985, "learning_rate": 0.0001, "loss": 0.0189, "step": 26490 }, { "epoch": 202.29007633587787, "grad_norm": 1.7394020557403564, "learning_rate": 0.0001, "loss": 0.0186, "step": 26500 }, { "epoch": 202.36641221374046, "grad_norm": 1.954837679862976, "learning_rate": 0.0001, "loss": 0.0193, "step": 26510 }, { "epoch": 202.44274809160305, "grad_norm": 1.2735214233398438, "learning_rate": 0.0001, "loss": 0.0186, "step": 26520 }, { "epoch": 202.51908396946564, "grad_norm": 1.5060038566589355, "learning_rate": 0.0001, "loss": 0.0195, "step": 26530 }, { "epoch": 202.59541984732823, "grad_norm": 1.8587881326675415, "learning_rate": 0.0001, "loss": 0.0202, "step": 26540 }, { "epoch": 202.67175572519085, "grad_norm": 1.3965219259262085, "learning_rate": 0.0001, "loss": 0.0197, "step": 26550 }, { "epoch": 202.74809160305344, "grad_norm": 1.5911325216293335, "learning_rate": 0.0001, "loss": 0.0197, "step": 26560 }, { "epoch": 202.82442748091603, "grad_norm": 2.04811692237854, "learning_rate": 0.0001, "loss": 0.0212, "step": 26570 }, { "epoch": 202.90076335877862, "grad_norm": 1.6008570194244385, "learning_rate": 0.0001, "loss": 0.02, "step": 26580 }, { "epoch": 202.9770992366412, "grad_norm": 2.025291681289673, "learning_rate": 0.0001, "loss": 0.0199, "step": 26590 }, { "epoch": 203.05343511450383, "grad_norm": 1.7323241233825684, "learning_rate": 0.0001, "loss": 0.0213, "step": 26600 }, { "epoch": 203.12977099236642, "grad_norm": 1.4775371551513672, "learning_rate": 0.0001, "loss": 0.0203, "step": 26610 }, { "epoch": 203.206106870229, "grad_norm": 1.5055314302444458, "learning_rate": 0.0001, "loss": 0.0201, "step": 26620 }, { "epoch": 203.2824427480916, "grad_norm": 2.007391929626465, "learning_rate": 0.0001, "loss": 0.0203, "step": 26630 }, { "epoch": 203.3587786259542, "grad_norm": 1.3941255807876587, "learning_rate": 0.0001, "loss": 0.0203, "step": 26640 }, { "epoch": 203.4351145038168, "grad_norm": 1.9527591466903687, "learning_rate": 0.0001, "loss": 0.0206, "step": 26650 }, { "epoch": 203.5114503816794, "grad_norm": 1.9040284156799316, "learning_rate": 0.0001, "loss": 0.0199, "step": 26660 }, { "epoch": 203.58778625954199, "grad_norm": 1.5050666332244873, "learning_rate": 0.0001, "loss": 0.0203, "step": 26670 }, { "epoch": 203.66412213740458, "grad_norm": 1.3774352073669434, "learning_rate": 0.0001, "loss": 0.0205, "step": 26680 }, { "epoch": 203.74045801526717, "grad_norm": 1.4383190870285034, "learning_rate": 0.0001, "loss": 0.0202, "step": 26690 }, { "epoch": 203.81679389312978, "grad_norm": 1.7433987855911255, "learning_rate": 0.0001, "loss": 0.0201, "step": 26700 }, { "epoch": 203.89312977099237, "grad_norm": 2.039682149887085, "learning_rate": 0.0001, "loss": 0.0209, "step": 26710 }, { "epoch": 203.96946564885496, "grad_norm": 2.5937283039093018, "learning_rate": 0.0001, "loss": 0.0199, "step": 26720 }, { "epoch": 204.04580152671755, "grad_norm": 1.8101080656051636, "learning_rate": 0.0001, "loss": 0.0199, "step": 26730 }, { "epoch": 204.12213740458014, "grad_norm": 1.9091938734054565, "learning_rate": 0.0001, "loss": 0.02, "step": 26740 }, { "epoch": 204.19847328244273, "grad_norm": 1.5184134244918823, "learning_rate": 0.0001, "loss": 0.0201, "step": 26750 }, { "epoch": 204.27480916030535, "grad_norm": 1.5610036849975586, "learning_rate": 0.0001, "loss": 0.019, "step": 26760 }, { "epoch": 204.35114503816794, "grad_norm": 1.5335311889648438, "learning_rate": 0.0001, "loss": 0.0194, "step": 26770 }, { "epoch": 204.42748091603053, "grad_norm": 1.027923583984375, "learning_rate": 0.0001, "loss": 0.0203, "step": 26780 }, { "epoch": 204.50381679389312, "grad_norm": 1.2624523639678955, "learning_rate": 0.0001, "loss": 0.021, "step": 26790 }, { "epoch": 204.5801526717557, "grad_norm": 1.6056592464447021, "learning_rate": 0.0001, "loss": 0.0199, "step": 26800 }, { "epoch": 204.65648854961833, "grad_norm": 1.6135568618774414, "learning_rate": 0.0001, "loss": 0.0215, "step": 26810 }, { "epoch": 204.73282442748092, "grad_norm": 2.390652656555176, "learning_rate": 0.0001, "loss": 0.021, "step": 26820 }, { "epoch": 204.8091603053435, "grad_norm": 1.831555962562561, "learning_rate": 0.0001, "loss": 0.0211, "step": 26830 }, { "epoch": 204.8854961832061, "grad_norm": 2.2205379009246826, "learning_rate": 0.0001, "loss": 0.0204, "step": 26840 }, { "epoch": 204.9618320610687, "grad_norm": 1.942641019821167, "learning_rate": 0.0001, "loss": 0.0202, "step": 26850 }, { "epoch": 205.0381679389313, "grad_norm": 2.216310501098633, "learning_rate": 0.0001, "loss": 0.0203, "step": 26860 }, { "epoch": 205.1145038167939, "grad_norm": 1.6680171489715576, "learning_rate": 0.0001, "loss": 0.0204, "step": 26870 }, { "epoch": 205.1908396946565, "grad_norm": 1.7223719358444214, "learning_rate": 0.0001, "loss": 0.0201, "step": 26880 }, { "epoch": 205.26717557251908, "grad_norm": 2.5038950443267822, "learning_rate": 0.0001, "loss": 0.0209, "step": 26890 }, { "epoch": 205.34351145038167, "grad_norm": 2.8057918548583984, "learning_rate": 0.0001, "loss": 0.0206, "step": 26900 }, { "epoch": 205.4198473282443, "grad_norm": 1.6448322534561157, "learning_rate": 0.0001, "loss": 0.0203, "step": 26910 }, { "epoch": 205.49618320610688, "grad_norm": 2.0137784481048584, "learning_rate": 0.0001, "loss": 0.0198, "step": 26920 }, { "epoch": 205.57251908396947, "grad_norm": 2.208261013031006, "learning_rate": 0.0001, "loss": 0.0208, "step": 26930 }, { "epoch": 205.64885496183206, "grad_norm": 1.5348082780838013, "learning_rate": 0.0001, "loss": 0.0199, "step": 26940 }, { "epoch": 205.72519083969465, "grad_norm": 1.3385905027389526, "learning_rate": 0.0001, "loss": 0.0205, "step": 26950 }, { "epoch": 205.80152671755727, "grad_norm": 1.8609826564788818, "learning_rate": 0.0001, "loss": 0.0204, "step": 26960 }, { "epoch": 205.87786259541986, "grad_norm": 1.1026450395584106, "learning_rate": 0.0001, "loss": 0.0198, "step": 26970 }, { "epoch": 205.95419847328245, "grad_norm": 1.9429023265838623, "learning_rate": 0.0001, "loss": 0.0199, "step": 26980 }, { "epoch": 206.03053435114504, "grad_norm": 1.4814035892486572, "learning_rate": 0.0001, "loss": 0.02, "step": 26990 }, { "epoch": 206.10687022900763, "grad_norm": 2.4459147453308105, "learning_rate": 0.0001, "loss": 0.02, "step": 27000 }, { "epoch": 206.18320610687022, "grad_norm": 1.6097416877746582, "learning_rate": 0.0001, "loss": 0.0204, "step": 27010 }, { "epoch": 206.25954198473283, "grad_norm": 1.6661375761032104, "learning_rate": 0.0001, "loss": 0.021, "step": 27020 }, { "epoch": 206.33587786259542, "grad_norm": 2.353231906890869, "learning_rate": 0.0001, "loss": 0.0208, "step": 27030 }, { "epoch": 206.41221374045801, "grad_norm": 2.0205042362213135, "learning_rate": 0.0001, "loss": 0.0206, "step": 27040 }, { "epoch": 206.4885496183206, "grad_norm": 1.8184890747070312, "learning_rate": 0.0001, "loss": 0.0207, "step": 27050 }, { "epoch": 206.5648854961832, "grad_norm": 2.4536590576171875, "learning_rate": 0.0001, "loss": 0.0208, "step": 27060 }, { "epoch": 206.6412213740458, "grad_norm": 1.3955814838409424, "learning_rate": 0.0001, "loss": 0.0204, "step": 27070 }, { "epoch": 206.7175572519084, "grad_norm": 2.1149237155914307, "learning_rate": 0.0001, "loss": 0.0205, "step": 27080 }, { "epoch": 206.793893129771, "grad_norm": 2.007566452026367, "learning_rate": 0.0001, "loss": 0.02, "step": 27090 }, { "epoch": 206.87022900763358, "grad_norm": 1.5441783666610718, "learning_rate": 0.0001, "loss": 0.02, "step": 27100 }, { "epoch": 206.94656488549617, "grad_norm": 2.099058151245117, "learning_rate": 0.0001, "loss": 0.0198, "step": 27110 }, { "epoch": 207.0229007633588, "grad_norm": 1.4849094152450562, "learning_rate": 0.0001, "loss": 0.0199, "step": 27120 }, { "epoch": 207.09923664122138, "grad_norm": 2.7093236446380615, "learning_rate": 0.0001, "loss": 0.0202, "step": 27130 }, { "epoch": 207.17557251908397, "grad_norm": 1.6180262565612793, "learning_rate": 0.0001, "loss": 0.0201, "step": 27140 }, { "epoch": 207.25190839694656, "grad_norm": 1.8052557706832886, "learning_rate": 0.0001, "loss": 0.0205, "step": 27150 }, { "epoch": 207.32824427480915, "grad_norm": 1.8942919969558716, "learning_rate": 0.0001, "loss": 0.0205, "step": 27160 }, { "epoch": 207.40458015267177, "grad_norm": 1.3810951709747314, "learning_rate": 0.0001, "loss": 0.0211, "step": 27170 }, { "epoch": 207.48091603053436, "grad_norm": 1.5881009101867676, "learning_rate": 0.0001, "loss": 0.0206, "step": 27180 }, { "epoch": 207.55725190839695, "grad_norm": 1.664138674736023, "learning_rate": 0.0001, "loss": 0.0213, "step": 27190 }, { "epoch": 207.63358778625954, "grad_norm": 1.3936665058135986, "learning_rate": 0.0001, "loss": 0.0202, "step": 27200 }, { "epoch": 207.70992366412213, "grad_norm": 1.674176812171936, "learning_rate": 0.0001, "loss": 0.0208, "step": 27210 }, { "epoch": 207.78625954198472, "grad_norm": 1.1129387617111206, "learning_rate": 0.0001, "loss": 0.0205, "step": 27220 }, { "epoch": 207.86259541984734, "grad_norm": 1.5804208517074585, "learning_rate": 0.0001, "loss": 0.0201, "step": 27230 }, { "epoch": 207.93893129770993, "grad_norm": 1.6368402242660522, "learning_rate": 0.0001, "loss": 0.0216, "step": 27240 }, { "epoch": 208.01526717557252, "grad_norm": 2.2811367511749268, "learning_rate": 0.0001, "loss": 0.0209, "step": 27250 }, { "epoch": 208.0916030534351, "grad_norm": 2.004406213760376, "learning_rate": 0.0001, "loss": 0.0208, "step": 27260 }, { "epoch": 208.1679389312977, "grad_norm": 2.462174892425537, "learning_rate": 0.0001, "loss": 0.0207, "step": 27270 }, { "epoch": 208.24427480916032, "grad_norm": 1.910258173942566, "learning_rate": 0.0001, "loss": 0.0211, "step": 27280 }, { "epoch": 208.3206106870229, "grad_norm": 1.8702048063278198, "learning_rate": 0.0001, "loss": 0.0203, "step": 27290 }, { "epoch": 208.3969465648855, "grad_norm": 2.6437692642211914, "learning_rate": 0.0001, "loss": 0.0204, "step": 27300 }, { "epoch": 208.4732824427481, "grad_norm": 1.6656861305236816, "learning_rate": 0.0001, "loss": 0.0198, "step": 27310 }, { "epoch": 208.54961832061068, "grad_norm": 1.814435362815857, "learning_rate": 0.0001, "loss": 0.0198, "step": 27320 }, { "epoch": 208.6259541984733, "grad_norm": 1.8766393661499023, "learning_rate": 0.0001, "loss": 0.02, "step": 27330 }, { "epoch": 208.70229007633588, "grad_norm": 1.673209309577942, "learning_rate": 0.0001, "loss": 0.0194, "step": 27340 }, { "epoch": 208.77862595419847, "grad_norm": 2.2662484645843506, "learning_rate": 0.0001, "loss": 0.0194, "step": 27350 }, { "epoch": 208.85496183206106, "grad_norm": 1.9312435388565063, "learning_rate": 0.0001, "loss": 0.0202, "step": 27360 }, { "epoch": 208.93129770992365, "grad_norm": 1.6189842224121094, "learning_rate": 0.0001, "loss": 0.0207, "step": 27370 }, { "epoch": 209.00763358778627, "grad_norm": 1.3744523525238037, "learning_rate": 0.0001, "loss": 0.0207, "step": 27380 }, { "epoch": 209.08396946564886, "grad_norm": 1.8684215545654297, "learning_rate": 0.0001, "loss": 0.0208, "step": 27390 }, { "epoch": 209.16030534351145, "grad_norm": 2.1375551223754883, "learning_rate": 0.0001, "loss": 0.0211, "step": 27400 }, { "epoch": 209.23664122137404, "grad_norm": 2.1450648307800293, "learning_rate": 0.0001, "loss": 0.0206, "step": 27410 }, { "epoch": 209.31297709923663, "grad_norm": 2.5473690032958984, "learning_rate": 0.0001, "loss": 0.0205, "step": 27420 }, { "epoch": 209.38931297709922, "grad_norm": 3.035773992538452, "learning_rate": 0.0001, "loss": 0.02, "step": 27430 }, { "epoch": 209.46564885496184, "grad_norm": 1.864399790763855, "learning_rate": 0.0001, "loss": 0.02, "step": 27440 }, { "epoch": 209.54198473282443, "grad_norm": 1.7866169214248657, "learning_rate": 0.0001, "loss": 0.0193, "step": 27450 }, { "epoch": 209.61832061068702, "grad_norm": 1.8969860076904297, "learning_rate": 0.0001, "loss": 0.0193, "step": 27460 }, { "epoch": 209.6946564885496, "grad_norm": 2.2415926456451416, "learning_rate": 0.0001, "loss": 0.0193, "step": 27470 }, { "epoch": 209.7709923664122, "grad_norm": 1.5537514686584473, "learning_rate": 0.0001, "loss": 0.0199, "step": 27480 }, { "epoch": 209.84732824427482, "grad_norm": 1.8727991580963135, "learning_rate": 0.0001, "loss": 0.0195, "step": 27490 }, { "epoch": 209.9236641221374, "grad_norm": 1.4338096380233765, "learning_rate": 0.0001, "loss": 0.0209, "step": 27500 }, { "epoch": 210.0, "grad_norm": 2.0573959350585938, "learning_rate": 0.0001, "loss": 0.0202, "step": 27510 }, { "epoch": 210.0763358778626, "grad_norm": 2.0980446338653564, "learning_rate": 0.0001, "loss": 0.0216, "step": 27520 }, { "epoch": 210.15267175572518, "grad_norm": 2.2707955837249756, "learning_rate": 0.0001, "loss": 0.021, "step": 27530 }, { "epoch": 210.2290076335878, "grad_norm": 1.931077480316162, "learning_rate": 0.0001, "loss": 0.0207, "step": 27540 }, { "epoch": 210.3053435114504, "grad_norm": 1.688913345336914, "learning_rate": 0.0001, "loss": 0.0203, "step": 27550 }, { "epoch": 210.38167938931298, "grad_norm": 1.597499132156372, "learning_rate": 0.0001, "loss": 0.0204, "step": 27560 }, { "epoch": 210.45801526717557, "grad_norm": 1.6438912153244019, "learning_rate": 0.0001, "loss": 0.0196, "step": 27570 }, { "epoch": 210.53435114503816, "grad_norm": 1.4914565086364746, "learning_rate": 0.0001, "loss": 0.0205, "step": 27580 }, { "epoch": 210.61068702290078, "grad_norm": 1.351320505142212, "learning_rate": 0.0001, "loss": 0.0201, "step": 27590 }, { "epoch": 210.68702290076337, "grad_norm": 1.5868878364562988, "learning_rate": 0.0001, "loss": 0.021, "step": 27600 }, { "epoch": 210.76335877862596, "grad_norm": 1.6748393774032593, "learning_rate": 0.0001, "loss": 0.0211, "step": 27610 }, { "epoch": 210.83969465648855, "grad_norm": 1.5330474376678467, "learning_rate": 0.0001, "loss": 0.0212, "step": 27620 }, { "epoch": 210.91603053435114, "grad_norm": 1.6041748523712158, "learning_rate": 0.0001, "loss": 0.0212, "step": 27630 }, { "epoch": 210.99236641221373, "grad_norm": 2.4768457412719727, "learning_rate": 0.0001, "loss": 0.021, "step": 27640 }, { "epoch": 211.06870229007635, "grad_norm": 2.4593119621276855, "learning_rate": 0.0001, "loss": 0.0211, "step": 27650 }, { "epoch": 211.14503816793894, "grad_norm": 1.9183717966079712, "learning_rate": 0.0001, "loss": 0.0206, "step": 27660 }, { "epoch": 211.22137404580153, "grad_norm": 2.341362953186035, "learning_rate": 0.0001, "loss": 0.0215, "step": 27670 }, { "epoch": 211.29770992366412, "grad_norm": 2.0096004009246826, "learning_rate": 0.0001, "loss": 0.0205, "step": 27680 }, { "epoch": 211.3740458015267, "grad_norm": 1.899532437324524, "learning_rate": 0.0001, "loss": 0.02, "step": 27690 }, { "epoch": 211.45038167938932, "grad_norm": 1.6768296957015991, "learning_rate": 0.0001, "loss": 0.0205, "step": 27700 }, { "epoch": 211.5267175572519, "grad_norm": 1.4656563997268677, "learning_rate": 0.0001, "loss": 0.02, "step": 27710 }, { "epoch": 211.6030534351145, "grad_norm": 2.306624174118042, "learning_rate": 0.0001, "loss": 0.0195, "step": 27720 }, { "epoch": 211.6793893129771, "grad_norm": 1.681687355041504, "learning_rate": 0.0001, "loss": 0.0188, "step": 27730 }, { "epoch": 211.75572519083968, "grad_norm": 2.0045135021209717, "learning_rate": 0.0001, "loss": 0.0197, "step": 27740 }, { "epoch": 211.8320610687023, "grad_norm": 1.8094873428344727, "learning_rate": 0.0001, "loss": 0.0197, "step": 27750 }, { "epoch": 211.9083969465649, "grad_norm": 1.9790680408477783, "learning_rate": 0.0001, "loss": 0.019, "step": 27760 }, { "epoch": 211.98473282442748, "grad_norm": 1.6372815370559692, "learning_rate": 0.0001, "loss": 0.0198, "step": 27770 }, { "epoch": 212.06106870229007, "grad_norm": 1.5920032262802124, "learning_rate": 0.0001, "loss": 0.0191, "step": 27780 }, { "epoch": 212.13740458015266, "grad_norm": 1.5952187776565552, "learning_rate": 0.0001, "loss": 0.02, "step": 27790 }, { "epoch": 212.21374045801528, "grad_norm": 2.477078914642334, "learning_rate": 0.0001, "loss": 0.0197, "step": 27800 }, { "epoch": 212.29007633587787, "grad_norm": 1.9837394952774048, "learning_rate": 0.0001, "loss": 0.0202, "step": 27810 }, { "epoch": 212.36641221374046, "grad_norm": 1.9152506589889526, "learning_rate": 0.0001, "loss": 0.0201, "step": 27820 }, { "epoch": 212.44274809160305, "grad_norm": 1.5122016668319702, "learning_rate": 0.0001, "loss": 0.0196, "step": 27830 }, { "epoch": 212.51908396946564, "grad_norm": 1.7879258394241333, "learning_rate": 0.0001, "loss": 0.0204, "step": 27840 }, { "epoch": 212.59541984732823, "grad_norm": 1.58941650390625, "learning_rate": 0.0001, "loss": 0.0204, "step": 27850 }, { "epoch": 212.67175572519085, "grad_norm": 2.588716506958008, "learning_rate": 0.0001, "loss": 0.0199, "step": 27860 }, { "epoch": 212.74809160305344, "grad_norm": 1.7323440313339233, "learning_rate": 0.0001, "loss": 0.0203, "step": 27870 }, { "epoch": 212.82442748091603, "grad_norm": 1.3111238479614258, "learning_rate": 0.0001, "loss": 0.019, "step": 27880 }, { "epoch": 212.90076335877862, "grad_norm": 1.963696837425232, "learning_rate": 0.0001, "loss": 0.0204, "step": 27890 }, { "epoch": 212.9770992366412, "grad_norm": 1.7117180824279785, "learning_rate": 0.0001, "loss": 0.0204, "step": 27900 }, { "epoch": 213.05343511450383, "grad_norm": 2.2912437915802, "learning_rate": 0.0001, "loss": 0.0201, "step": 27910 }, { "epoch": 213.12977099236642, "grad_norm": 1.4395489692687988, "learning_rate": 0.0001, "loss": 0.0205, "step": 27920 }, { "epoch": 213.206106870229, "grad_norm": 2.0851316452026367, "learning_rate": 0.0001, "loss": 0.0208, "step": 27930 }, { "epoch": 213.2824427480916, "grad_norm": 1.8116750717163086, "learning_rate": 0.0001, "loss": 0.0215, "step": 27940 }, { "epoch": 213.3587786259542, "grad_norm": 1.747558355331421, "learning_rate": 0.0001, "loss": 0.0209, "step": 27950 }, { "epoch": 213.4351145038168, "grad_norm": 1.3816044330596924, "learning_rate": 0.0001, "loss": 0.02, "step": 27960 }, { "epoch": 213.5114503816794, "grad_norm": 1.652869701385498, "learning_rate": 0.0001, "loss": 0.021, "step": 27970 }, { "epoch": 213.58778625954199, "grad_norm": 2.17498517036438, "learning_rate": 0.0001, "loss": 0.0208, "step": 27980 }, { "epoch": 213.66412213740458, "grad_norm": 1.6759119033813477, "learning_rate": 0.0001, "loss": 0.0202, "step": 27990 }, { "epoch": 213.74045801526717, "grad_norm": 1.8141099214553833, "learning_rate": 0.0001, "loss": 0.02, "step": 28000 }, { "epoch": 213.81679389312978, "grad_norm": 1.516993761062622, "learning_rate": 0.0001, "loss": 0.021, "step": 28010 }, { "epoch": 213.89312977099237, "grad_norm": 1.9303538799285889, "learning_rate": 0.0001, "loss": 0.0207, "step": 28020 }, { "epoch": 213.96946564885496, "grad_norm": 1.1123241186141968, "learning_rate": 0.0001, "loss": 0.02, "step": 28030 }, { "epoch": 214.04580152671755, "grad_norm": 1.820546269416809, "learning_rate": 0.0001, "loss": 0.0201, "step": 28040 }, { "epoch": 214.12213740458014, "grad_norm": 1.8463058471679688, "learning_rate": 0.0001, "loss": 0.0202, "step": 28050 }, { "epoch": 214.19847328244273, "grad_norm": 1.694929599761963, "learning_rate": 0.0001, "loss": 0.0196, "step": 28060 }, { "epoch": 214.27480916030535, "grad_norm": 1.514792799949646, "learning_rate": 0.0001, "loss": 0.0194, "step": 28070 }, { "epoch": 214.35114503816794, "grad_norm": 1.9629075527191162, "learning_rate": 0.0001, "loss": 0.0204, "step": 28080 }, { "epoch": 214.42748091603053, "grad_norm": 1.9067411422729492, "learning_rate": 0.0001, "loss": 0.0205, "step": 28090 }, { "epoch": 214.50381679389312, "grad_norm": 2.059335231781006, "learning_rate": 0.0001, "loss": 0.0202, "step": 28100 }, { "epoch": 214.5801526717557, "grad_norm": 1.611331582069397, "learning_rate": 0.0001, "loss": 0.0199, "step": 28110 }, { "epoch": 214.65648854961833, "grad_norm": 2.794905424118042, "learning_rate": 0.0001, "loss": 0.0196, "step": 28120 }, { "epoch": 214.73282442748092, "grad_norm": 2.1255276203155518, "learning_rate": 0.0001, "loss": 0.0189, "step": 28130 }, { "epoch": 214.8091603053435, "grad_norm": 2.4690215587615967, "learning_rate": 0.0001, "loss": 0.0191, "step": 28140 }, { "epoch": 214.8854961832061, "grad_norm": 2.5606839656829834, "learning_rate": 0.0001, "loss": 0.0189, "step": 28150 }, { "epoch": 214.9618320610687, "grad_norm": 1.978377103805542, "learning_rate": 0.0001, "loss": 0.0189, "step": 28160 }, { "epoch": 215.0381679389313, "grad_norm": 1.4231321811676025, "learning_rate": 0.0001, "loss": 0.0187, "step": 28170 }, { "epoch": 215.1145038167939, "grad_norm": 1.6771881580352783, "learning_rate": 0.0001, "loss": 0.0186, "step": 28180 }, { "epoch": 215.1908396946565, "grad_norm": 2.4929535388946533, "learning_rate": 0.0001, "loss": 0.0187, "step": 28190 }, { "epoch": 215.26717557251908, "grad_norm": 2.1066317558288574, "learning_rate": 0.0001, "loss": 0.0191, "step": 28200 }, { "epoch": 215.34351145038167, "grad_norm": 1.3996052742004395, "learning_rate": 0.0001, "loss": 0.0189, "step": 28210 }, { "epoch": 215.4198473282443, "grad_norm": 1.5581459999084473, "learning_rate": 0.0001, "loss": 0.0192, "step": 28220 }, { "epoch": 215.49618320610688, "grad_norm": 1.813713550567627, "learning_rate": 0.0001, "loss": 0.0192, "step": 28230 }, { "epoch": 215.57251908396947, "grad_norm": 1.8921396732330322, "learning_rate": 0.0001, "loss": 0.019, "step": 28240 }, { "epoch": 215.64885496183206, "grad_norm": 1.2824344635009766, "learning_rate": 0.0001, "loss": 0.0198, "step": 28250 }, { "epoch": 215.72519083969465, "grad_norm": 1.722132921218872, "learning_rate": 0.0001, "loss": 0.0202, "step": 28260 }, { "epoch": 215.80152671755727, "grad_norm": 1.6188160181045532, "learning_rate": 0.0001, "loss": 0.0203, "step": 28270 }, { "epoch": 215.87786259541986, "grad_norm": 1.9405170679092407, "learning_rate": 0.0001, "loss": 0.0205, "step": 28280 }, { "epoch": 215.95419847328245, "grad_norm": 2.347334623336792, "learning_rate": 0.0001, "loss": 0.02, "step": 28290 }, { "epoch": 216.03053435114504, "grad_norm": 1.240739345550537, "learning_rate": 0.0001, "loss": 0.0201, "step": 28300 }, { "epoch": 216.10687022900763, "grad_norm": 1.9451292753219604, "learning_rate": 0.0001, "loss": 0.0194, "step": 28310 }, { "epoch": 216.18320610687022, "grad_norm": 2.229991912841797, "learning_rate": 0.0001, "loss": 0.019, "step": 28320 }, { "epoch": 216.25954198473283, "grad_norm": 2.4645748138427734, "learning_rate": 0.0001, "loss": 0.0198, "step": 28330 }, { "epoch": 216.33587786259542, "grad_norm": 1.5712730884552002, "learning_rate": 0.0001, "loss": 0.0195, "step": 28340 }, { "epoch": 216.41221374045801, "grad_norm": 1.4513264894485474, "learning_rate": 0.0001, "loss": 0.02, "step": 28350 }, { "epoch": 216.4885496183206, "grad_norm": 2.2770092487335205, "learning_rate": 0.0001, "loss": 0.0198, "step": 28360 }, { "epoch": 216.5648854961832, "grad_norm": 1.7679721117019653, "learning_rate": 0.0001, "loss": 0.0196, "step": 28370 }, { "epoch": 216.6412213740458, "grad_norm": 1.8558305501937866, "learning_rate": 0.0001, "loss": 0.0197, "step": 28380 }, { "epoch": 216.7175572519084, "grad_norm": 1.4815568923950195, "learning_rate": 0.0001, "loss": 0.0194, "step": 28390 }, { "epoch": 216.793893129771, "grad_norm": 1.1153957843780518, "learning_rate": 0.0001, "loss": 0.0191, "step": 28400 }, { "epoch": 216.87022900763358, "grad_norm": 1.5491752624511719, "learning_rate": 0.0001, "loss": 0.0198, "step": 28410 }, { "epoch": 216.94656488549617, "grad_norm": 1.3681238889694214, "learning_rate": 0.0001, "loss": 0.0203, "step": 28420 }, { "epoch": 217.0229007633588, "grad_norm": 1.324243426322937, "learning_rate": 0.0001, "loss": 0.0208, "step": 28430 }, { "epoch": 217.09923664122138, "grad_norm": 1.559936285018921, "learning_rate": 0.0001, "loss": 0.0209, "step": 28440 }, { "epoch": 217.17557251908397, "grad_norm": 2.412653684616089, "learning_rate": 0.0001, "loss": 0.0213, "step": 28450 }, { "epoch": 217.25190839694656, "grad_norm": 1.3397325277328491, "learning_rate": 0.0001, "loss": 0.0212, "step": 28460 }, { "epoch": 217.32824427480915, "grad_norm": 2.0694284439086914, "learning_rate": 0.0001, "loss": 0.0205, "step": 28470 }, { "epoch": 217.40458015267177, "grad_norm": 1.3615003824234009, "learning_rate": 0.0001, "loss": 0.0208, "step": 28480 }, { "epoch": 217.48091603053436, "grad_norm": 1.739808201789856, "learning_rate": 0.0001, "loss": 0.0204, "step": 28490 }, { "epoch": 217.55725190839695, "grad_norm": 1.4641845226287842, "learning_rate": 0.0001, "loss": 0.0204, "step": 28500 }, { "epoch": 217.63358778625954, "grad_norm": 1.5120306015014648, "learning_rate": 0.0001, "loss": 0.0208, "step": 28510 }, { "epoch": 217.70992366412213, "grad_norm": 1.5747963190078735, "learning_rate": 0.0001, "loss": 0.0203, "step": 28520 }, { "epoch": 217.78625954198472, "grad_norm": 1.6080116033554077, "learning_rate": 0.0001, "loss": 0.0212, "step": 28530 }, { "epoch": 217.86259541984734, "grad_norm": 1.5958623886108398, "learning_rate": 0.0001, "loss": 0.0212, "step": 28540 }, { "epoch": 217.93893129770993, "grad_norm": 1.3535460233688354, "learning_rate": 0.0001, "loss": 0.0203, "step": 28550 }, { "epoch": 218.01526717557252, "grad_norm": 2.142479658126831, "learning_rate": 0.0001, "loss": 0.0212, "step": 28560 }, { "epoch": 218.0916030534351, "grad_norm": 1.582614779472351, "learning_rate": 0.0001, "loss": 0.021, "step": 28570 }, { "epoch": 218.1679389312977, "grad_norm": 1.8747440576553345, "learning_rate": 0.0001, "loss": 0.0201, "step": 28580 }, { "epoch": 218.24427480916032, "grad_norm": 1.6657357215881348, "learning_rate": 0.0001, "loss": 0.021, "step": 28590 }, { "epoch": 218.3206106870229, "grad_norm": 1.3278779983520508, "learning_rate": 0.0001, "loss": 0.021, "step": 28600 }, { "epoch": 218.3969465648855, "grad_norm": 1.5700058937072754, "learning_rate": 0.0001, "loss": 0.0206, "step": 28610 }, { "epoch": 218.4732824427481, "grad_norm": 2.127697467803955, "learning_rate": 0.0001, "loss": 0.0213, "step": 28620 }, { "epoch": 218.54961832061068, "grad_norm": 1.688131332397461, "learning_rate": 0.0001, "loss": 0.0205, "step": 28630 }, { "epoch": 218.6259541984733, "grad_norm": 2.255859851837158, "learning_rate": 0.0001, "loss": 0.0211, "step": 28640 }, { "epoch": 218.70229007633588, "grad_norm": 2.6652491092681885, "learning_rate": 0.0001, "loss": 0.0211, "step": 28650 }, { "epoch": 218.77862595419847, "grad_norm": 2.0566415786743164, "learning_rate": 0.0001, "loss": 0.0208, "step": 28660 }, { "epoch": 218.85496183206106, "grad_norm": 2.1104767322540283, "learning_rate": 0.0001, "loss": 0.02, "step": 28670 }, { "epoch": 218.93129770992365, "grad_norm": 1.6617987155914307, "learning_rate": 0.0001, "loss": 0.0198, "step": 28680 }, { "epoch": 219.00763358778627, "grad_norm": 2.0542428493499756, "learning_rate": 0.0001, "loss": 0.0203, "step": 28690 }, { "epoch": 219.08396946564886, "grad_norm": 1.545452356338501, "learning_rate": 0.0001, "loss": 0.0194, "step": 28700 }, { "epoch": 219.16030534351145, "grad_norm": 1.436671495437622, "learning_rate": 0.0001, "loss": 0.0196, "step": 28710 }, { "epoch": 219.23664122137404, "grad_norm": 1.4000232219696045, "learning_rate": 0.0001, "loss": 0.019, "step": 28720 }, { "epoch": 219.31297709923663, "grad_norm": 1.7459250688552856, "learning_rate": 0.0001, "loss": 0.0198, "step": 28730 }, { "epoch": 219.38931297709922, "grad_norm": 1.7112761735916138, "learning_rate": 0.0001, "loss": 0.0193, "step": 28740 }, { "epoch": 219.46564885496184, "grad_norm": 1.6293737888336182, "learning_rate": 0.0001, "loss": 0.0193, "step": 28750 }, { "epoch": 219.54198473282443, "grad_norm": 2.15759015083313, "learning_rate": 0.0001, "loss": 0.0191, "step": 28760 }, { "epoch": 219.61832061068702, "grad_norm": 2.468843460083008, "learning_rate": 0.0001, "loss": 0.0192, "step": 28770 }, { "epoch": 219.6946564885496, "grad_norm": 2.1061410903930664, "learning_rate": 0.0001, "loss": 0.019, "step": 28780 }, { "epoch": 219.7709923664122, "grad_norm": 1.7314035892486572, "learning_rate": 0.0001, "loss": 0.0187, "step": 28790 }, { "epoch": 219.84732824427482, "grad_norm": 1.3484599590301514, "learning_rate": 0.0001, "loss": 0.0187, "step": 28800 }, { "epoch": 219.9236641221374, "grad_norm": 2.020348072052002, "learning_rate": 0.0001, "loss": 0.0193, "step": 28810 }, { "epoch": 220.0, "grad_norm": 1.6040616035461426, "learning_rate": 0.0001, "loss": 0.019, "step": 28820 }, { "epoch": 220.0763358778626, "grad_norm": 1.445194959640503, "learning_rate": 0.0001, "loss": 0.019, "step": 28830 }, { "epoch": 220.15267175572518, "grad_norm": 1.6593080759048462, "learning_rate": 0.0001, "loss": 0.0192, "step": 28840 }, { "epoch": 220.2290076335878, "grad_norm": 2.3975141048431396, "learning_rate": 0.0001, "loss": 0.0198, "step": 28850 }, { "epoch": 220.3053435114504, "grad_norm": 1.4975179433822632, "learning_rate": 0.0001, "loss": 0.0191, "step": 28860 }, { "epoch": 220.38167938931298, "grad_norm": 1.3744868040084839, "learning_rate": 0.0001, "loss": 0.0191, "step": 28870 }, { "epoch": 220.45801526717557, "grad_norm": 1.6490219831466675, "learning_rate": 0.0001, "loss": 0.0196, "step": 28880 }, { "epoch": 220.53435114503816, "grad_norm": 1.7780630588531494, "learning_rate": 0.0001, "loss": 0.0201, "step": 28890 }, { "epoch": 220.61068702290078, "grad_norm": 1.6373153924942017, "learning_rate": 0.0001, "loss": 0.0203, "step": 28900 }, { "epoch": 220.68702290076337, "grad_norm": 1.8755545616149902, "learning_rate": 0.0001, "loss": 0.0194, "step": 28910 }, { "epoch": 220.76335877862596, "grad_norm": 1.9904776811599731, "learning_rate": 0.0001, "loss": 0.0199, "step": 28920 }, { "epoch": 220.83969465648855, "grad_norm": 1.5554630756378174, "learning_rate": 0.0001, "loss": 0.0194, "step": 28930 }, { "epoch": 220.91603053435114, "grad_norm": 1.5496746301651, "learning_rate": 0.0001, "loss": 0.0197, "step": 28940 }, { "epoch": 220.99236641221373, "grad_norm": 2.098242998123169, "learning_rate": 0.0001, "loss": 0.0209, "step": 28950 }, { "epoch": 221.06870229007635, "grad_norm": 1.5969104766845703, "learning_rate": 0.0001, "loss": 0.02, "step": 28960 }, { "epoch": 221.14503816793894, "grad_norm": 1.2332878112792969, "learning_rate": 0.0001, "loss": 0.0208, "step": 28970 }, { "epoch": 221.22137404580153, "grad_norm": 2.0147318840026855, "learning_rate": 0.0001, "loss": 0.0202, "step": 28980 }, { "epoch": 221.29770992366412, "grad_norm": 1.5716989040374756, "learning_rate": 0.0001, "loss": 0.0197, "step": 28990 }, { "epoch": 221.3740458015267, "grad_norm": 1.8190624713897705, "learning_rate": 0.0001, "loss": 0.0207, "step": 29000 }, { "epoch": 221.45038167938932, "grad_norm": 1.5354857444763184, "learning_rate": 0.0001, "loss": 0.0209, "step": 29010 }, { "epoch": 221.5267175572519, "grad_norm": 2.5572292804718018, "learning_rate": 0.0001, "loss": 0.0213, "step": 29020 }, { "epoch": 221.6030534351145, "grad_norm": 1.2282094955444336, "learning_rate": 0.0001, "loss": 0.0205, "step": 29030 }, { "epoch": 221.6793893129771, "grad_norm": 1.4324383735656738, "learning_rate": 0.0001, "loss": 0.0205, "step": 29040 }, { "epoch": 221.75572519083968, "grad_norm": 2.144275188446045, "learning_rate": 0.0001, "loss": 0.0208, "step": 29050 }, { "epoch": 221.8320610687023, "grad_norm": 1.4130526781082153, "learning_rate": 0.0001, "loss": 0.0202, "step": 29060 }, { "epoch": 221.9083969465649, "grad_norm": 1.5283437967300415, "learning_rate": 0.0001, "loss": 0.0208, "step": 29070 }, { "epoch": 221.98473282442748, "grad_norm": 1.1189489364624023, "learning_rate": 0.0001, "loss": 0.0202, "step": 29080 }, { "epoch": 222.06106870229007, "grad_norm": 1.7385103702545166, "learning_rate": 0.0001, "loss": 0.0197, "step": 29090 }, { "epoch": 222.13740458015266, "grad_norm": 1.5122185945510864, "learning_rate": 0.0001, "loss": 0.0195, "step": 29100 }, { "epoch": 222.21374045801528, "grad_norm": 1.5466212034225464, "learning_rate": 0.0001, "loss": 0.0195, "step": 29110 }, { "epoch": 222.29007633587787, "grad_norm": 1.852888584136963, "learning_rate": 0.0001, "loss": 0.02, "step": 29120 }, { "epoch": 222.36641221374046, "grad_norm": 1.5056545734405518, "learning_rate": 0.0001, "loss": 0.0202, "step": 29130 }, { "epoch": 222.44274809160305, "grad_norm": 2.5993728637695312, "learning_rate": 0.0001, "loss": 0.0205, "step": 29140 }, { "epoch": 222.51908396946564, "grad_norm": 1.8644473552703857, "learning_rate": 0.0001, "loss": 0.0202, "step": 29150 }, { "epoch": 222.59541984732823, "grad_norm": 1.4358985424041748, "learning_rate": 0.0001, "loss": 0.0195, "step": 29160 }, { "epoch": 222.67175572519085, "grad_norm": 1.6311460733413696, "learning_rate": 0.0001, "loss": 0.0198, "step": 29170 }, { "epoch": 222.74809160305344, "grad_norm": 1.231732964515686, "learning_rate": 0.0001, "loss": 0.0206, "step": 29180 }, { "epoch": 222.82442748091603, "grad_norm": 2.1773524284362793, "learning_rate": 0.0001, "loss": 0.0201, "step": 29190 }, { "epoch": 222.90076335877862, "grad_norm": 2.727030038833618, "learning_rate": 0.0001, "loss": 0.0201, "step": 29200 }, { "epoch": 222.9770992366412, "grad_norm": 1.4476256370544434, "learning_rate": 0.0001, "loss": 0.0204, "step": 29210 }, { "epoch": 223.05343511450383, "grad_norm": 1.6426969766616821, "learning_rate": 0.0001, "loss": 0.0198, "step": 29220 }, { "epoch": 223.12977099236642, "grad_norm": 1.6916193962097168, "learning_rate": 0.0001, "loss": 0.0208, "step": 29230 }, { "epoch": 223.206106870229, "grad_norm": 2.4865825176239014, "learning_rate": 0.0001, "loss": 0.0203, "step": 29240 }, { "epoch": 223.2824427480916, "grad_norm": 1.8812756538391113, "learning_rate": 0.0001, "loss": 0.0202, "step": 29250 }, { "epoch": 223.3587786259542, "grad_norm": 1.5301048755645752, "learning_rate": 0.0001, "loss": 0.0195, "step": 29260 }, { "epoch": 223.4351145038168, "grad_norm": 2.2134146690368652, "learning_rate": 0.0001, "loss": 0.0191, "step": 29270 }, { "epoch": 223.5114503816794, "grad_norm": 1.815268874168396, "learning_rate": 0.0001, "loss": 0.0192, "step": 29280 }, { "epoch": 223.58778625954199, "grad_norm": 1.6932717561721802, "learning_rate": 0.0001, "loss": 0.0189, "step": 29290 }, { "epoch": 223.66412213740458, "grad_norm": 1.8051141500473022, "learning_rate": 0.0001, "loss": 0.0194, "step": 29300 }, { "epoch": 223.74045801526717, "grad_norm": 1.6701586246490479, "learning_rate": 0.0001, "loss": 0.0193, "step": 29310 }, { "epoch": 223.81679389312978, "grad_norm": 1.7425373792648315, "learning_rate": 0.0001, "loss": 0.0191, "step": 29320 }, { "epoch": 223.89312977099237, "grad_norm": 1.7045090198516846, "learning_rate": 0.0001, "loss": 0.0191, "step": 29330 }, { "epoch": 223.96946564885496, "grad_norm": 2.1696650981903076, "learning_rate": 0.0001, "loss": 0.0198, "step": 29340 }, { "epoch": 224.04580152671755, "grad_norm": 2.160710096359253, "learning_rate": 0.0001, "loss": 0.0194, "step": 29350 }, { "epoch": 224.12213740458014, "grad_norm": 1.5531034469604492, "learning_rate": 0.0001, "loss": 0.0189, "step": 29360 }, { "epoch": 224.19847328244273, "grad_norm": 1.576005220413208, "learning_rate": 0.0001, "loss": 0.0196, "step": 29370 }, { "epoch": 224.27480916030535, "grad_norm": 1.3345789909362793, "learning_rate": 0.0001, "loss": 0.0191, "step": 29380 }, { "epoch": 224.35114503816794, "grad_norm": 2.0492944717407227, "learning_rate": 0.0001, "loss": 0.0198, "step": 29390 }, { "epoch": 224.42748091603053, "grad_norm": 1.498372197151184, "learning_rate": 0.0001, "loss": 0.0196, "step": 29400 }, { "epoch": 224.50381679389312, "grad_norm": 1.9966559410095215, "learning_rate": 0.0001, "loss": 0.0201, "step": 29410 }, { "epoch": 224.5801526717557, "grad_norm": 2.0624799728393555, "learning_rate": 0.0001, "loss": 0.0196, "step": 29420 }, { "epoch": 224.65648854961833, "grad_norm": 1.797250509262085, "learning_rate": 0.0001, "loss": 0.0194, "step": 29430 }, { "epoch": 224.73282442748092, "grad_norm": 1.675264596939087, "learning_rate": 0.0001, "loss": 0.0201, "step": 29440 }, { "epoch": 224.8091603053435, "grad_norm": 2.4206621646881104, "learning_rate": 0.0001, "loss": 0.0199, "step": 29450 }, { "epoch": 224.8854961832061, "grad_norm": 2.2179152965545654, "learning_rate": 0.0001, "loss": 0.0203, "step": 29460 }, { "epoch": 224.9618320610687, "grad_norm": 1.8060888051986694, "learning_rate": 0.0001, "loss": 0.0207, "step": 29470 }, { "epoch": 225.0381679389313, "grad_norm": 1.758692741394043, "learning_rate": 0.0001, "loss": 0.0202, "step": 29480 }, { "epoch": 225.1145038167939, "grad_norm": 1.5889911651611328, "learning_rate": 0.0001, "loss": 0.02, "step": 29490 }, { "epoch": 225.1908396946565, "grad_norm": 2.1339170932769775, "learning_rate": 0.0001, "loss": 0.0199, "step": 29500 }, { "epoch": 225.26717557251908, "grad_norm": 1.4491697549819946, "learning_rate": 0.0001, "loss": 0.0199, "step": 29510 }, { "epoch": 225.34351145038167, "grad_norm": 1.6307644844055176, "learning_rate": 0.0001, "loss": 0.0196, "step": 29520 }, { "epoch": 225.4198473282443, "grad_norm": 1.4546895027160645, "learning_rate": 0.0001, "loss": 0.0191, "step": 29530 }, { "epoch": 225.49618320610688, "grad_norm": 1.9260454177856445, "learning_rate": 0.0001, "loss": 0.019, "step": 29540 }, { "epoch": 225.57251908396947, "grad_norm": 1.62917959690094, "learning_rate": 0.0001, "loss": 0.0193, "step": 29550 }, { "epoch": 225.64885496183206, "grad_norm": 2.3202507495880127, "learning_rate": 0.0001, "loss": 0.0195, "step": 29560 }, { "epoch": 225.72519083969465, "grad_norm": 1.4705801010131836, "learning_rate": 0.0001, "loss": 0.019, "step": 29570 }, { "epoch": 225.80152671755727, "grad_norm": 2.179410934448242, "learning_rate": 0.0001, "loss": 0.0188, "step": 29580 }, { "epoch": 225.87786259541986, "grad_norm": 1.7950325012207031, "learning_rate": 0.0001, "loss": 0.02, "step": 29590 }, { "epoch": 225.95419847328245, "grad_norm": 1.2794673442840576, "learning_rate": 0.0001, "loss": 0.0186, "step": 29600 }, { "epoch": 226.03053435114504, "grad_norm": 1.4904955625534058, "learning_rate": 0.0001, "loss": 0.0191, "step": 29610 }, { "epoch": 226.10687022900763, "grad_norm": 2.1621596813201904, "learning_rate": 0.0001, "loss": 0.0194, "step": 29620 }, { "epoch": 226.18320610687022, "grad_norm": 1.612284541130066, "learning_rate": 0.0001, "loss": 0.0194, "step": 29630 }, { "epoch": 226.25954198473283, "grad_norm": 2.3074147701263428, "learning_rate": 0.0001, "loss": 0.0193, "step": 29640 }, { "epoch": 226.33587786259542, "grad_norm": 1.771267294883728, "learning_rate": 0.0001, "loss": 0.0193, "step": 29650 }, { "epoch": 226.41221374045801, "grad_norm": 2.1699681282043457, "learning_rate": 0.0001, "loss": 0.0192, "step": 29660 }, { "epoch": 226.4885496183206, "grad_norm": 1.4951281547546387, "learning_rate": 0.0001, "loss": 0.0186, "step": 29670 }, { "epoch": 226.5648854961832, "grad_norm": 1.5192015171051025, "learning_rate": 0.0001, "loss": 0.019, "step": 29680 }, { "epoch": 226.6412213740458, "grad_norm": 1.749383568763733, "learning_rate": 0.0001, "loss": 0.0195, "step": 29690 }, { "epoch": 226.7175572519084, "grad_norm": 1.6344712972640991, "learning_rate": 0.0001, "loss": 0.0185, "step": 29700 }, { "epoch": 226.793893129771, "grad_norm": 1.7094981670379639, "learning_rate": 0.0001, "loss": 0.0186, "step": 29710 }, { "epoch": 226.87022900763358, "grad_norm": 1.6329665184020996, "learning_rate": 0.0001, "loss": 0.0189, "step": 29720 }, { "epoch": 226.94656488549617, "grad_norm": 2.2806806564331055, "learning_rate": 0.0001, "loss": 0.0195, "step": 29730 }, { "epoch": 227.0229007633588, "grad_norm": 1.9027892351150513, "learning_rate": 0.0001, "loss": 0.0189, "step": 29740 }, { "epoch": 227.09923664122138, "grad_norm": 1.2402926683425903, "learning_rate": 0.0001, "loss": 0.0196, "step": 29750 }, { "epoch": 227.17557251908397, "grad_norm": 1.845613956451416, "learning_rate": 0.0001, "loss": 0.0188, "step": 29760 }, { "epoch": 227.25190839694656, "grad_norm": 1.5191941261291504, "learning_rate": 0.0001, "loss": 0.0191, "step": 29770 }, { "epoch": 227.32824427480915, "grad_norm": 1.969158411026001, "learning_rate": 0.0001, "loss": 0.0186, "step": 29780 }, { "epoch": 227.40458015267177, "grad_norm": 1.129394769668579, "learning_rate": 0.0001, "loss": 0.0188, "step": 29790 }, { "epoch": 227.48091603053436, "grad_norm": 1.412137746810913, "learning_rate": 0.0001, "loss": 0.0186, "step": 29800 }, { "epoch": 227.55725190839695, "grad_norm": 1.7291431427001953, "learning_rate": 0.0001, "loss": 0.0194, "step": 29810 }, { "epoch": 227.63358778625954, "grad_norm": 1.5055334568023682, "learning_rate": 0.0001, "loss": 0.02, "step": 29820 }, { "epoch": 227.70992366412213, "grad_norm": 1.4898438453674316, "learning_rate": 0.0001, "loss": 0.0203, "step": 29830 }, { "epoch": 227.78625954198472, "grad_norm": 1.4075088500976562, "learning_rate": 0.0001, "loss": 0.0203, "step": 29840 }, { "epoch": 227.86259541984734, "grad_norm": 1.7032837867736816, "learning_rate": 0.0001, "loss": 0.0198, "step": 29850 }, { "epoch": 227.93893129770993, "grad_norm": 1.408571720123291, "learning_rate": 0.0001, "loss": 0.0206, "step": 29860 }, { "epoch": 228.01526717557252, "grad_norm": 1.3365085124969482, "learning_rate": 0.0001, "loss": 0.0212, "step": 29870 }, { "epoch": 228.0916030534351, "grad_norm": 1.2977961301803589, "learning_rate": 0.0001, "loss": 0.0205, "step": 29880 }, { "epoch": 228.1679389312977, "grad_norm": 1.9620128870010376, "learning_rate": 0.0001, "loss": 0.019, "step": 29890 }, { "epoch": 228.24427480916032, "grad_norm": 1.5918060541152954, "learning_rate": 0.0001, "loss": 0.0191, "step": 29900 }, { "epoch": 228.3206106870229, "grad_norm": 1.631442904472351, "learning_rate": 0.0001, "loss": 0.0196, "step": 29910 }, { "epoch": 228.3969465648855, "grad_norm": 1.5648412704467773, "learning_rate": 0.0001, "loss": 0.0198, "step": 29920 }, { "epoch": 228.4732824427481, "grad_norm": 1.5613449811935425, "learning_rate": 0.0001, "loss": 0.0195, "step": 29930 }, { "epoch": 228.54961832061068, "grad_norm": 1.5085835456848145, "learning_rate": 0.0001, "loss": 0.0192, "step": 29940 }, { "epoch": 228.6259541984733, "grad_norm": 2.1215932369232178, "learning_rate": 0.0001, "loss": 0.0196, "step": 29950 }, { "epoch": 228.70229007633588, "grad_norm": 1.8856043815612793, "learning_rate": 0.0001, "loss": 0.0197, "step": 29960 }, { "epoch": 228.77862595419847, "grad_norm": 2.028282403945923, "learning_rate": 0.0001, "loss": 0.0195, "step": 29970 }, { "epoch": 228.85496183206106, "grad_norm": 1.8524106740951538, "learning_rate": 0.0001, "loss": 0.0191, "step": 29980 }, { "epoch": 228.93129770992365, "grad_norm": 1.6796132326126099, "learning_rate": 0.0001, "loss": 0.0193, "step": 29990 }, { "epoch": 229.00763358778627, "grad_norm": 1.6787506341934204, "learning_rate": 0.0001, "loss": 0.0195, "step": 30000 }, { "epoch": 229.00763358778627, "step": 30000, "total_flos": 0.0, "train_loss": 0.021504405591388543, "train_runtime": 9723.7824, "train_samples_per_second": 197.454, "train_steps_per_second": 3.085 } ], "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 }