{ "best_metric": null, "best_model_checkpoint": null, "epoch": 26.851644663235625, "eval_steps": 500, "global_step": 60000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.004475274110539271, "grad_norm": 4.44952917098999, "learning_rate": 3.3333333333333335e-07, "loss": 1.3919, "step": 10 }, { "epoch": 0.008950548221078542, "grad_norm": 3.579458713531494, "learning_rate": 6.666666666666667e-07, "loss": 1.3904, "step": 20 }, { "epoch": 0.013425822331617812, "grad_norm": 5.267451763153076, "learning_rate": 1.0000000000000002e-06, "loss": 1.3869, "step": 30 }, { "epoch": 0.017901096442157084, "grad_norm": 3.191934823989868, "learning_rate": 1.3333333333333334e-06, "loss": 1.3831, "step": 40 }, { "epoch": 0.022376370552696354, "grad_norm": 3.0674550533294678, "learning_rate": 1.6666666666666667e-06, "loss": 1.373, "step": 50 }, { "epoch": 0.026851644663235624, "grad_norm": 4.113929748535156, "learning_rate": 2.0000000000000003e-06, "loss": 1.3744, "step": 60 }, { "epoch": 0.031326918773774894, "grad_norm": 5.452544212341309, "learning_rate": 2.3333333333333336e-06, "loss": 1.367, "step": 70 }, { "epoch": 0.03580219288431417, "grad_norm": 5.044178009033203, "learning_rate": 2.666666666666667e-06, "loss": 1.3567, "step": 80 }, { "epoch": 0.040277466994853434, "grad_norm": 5.948019504547119, "learning_rate": 3e-06, "loss": 1.3444, "step": 90 }, { "epoch": 0.04475274110539271, "grad_norm": 3.4619510173797607, "learning_rate": 3.3333333333333333e-06, "loss": 1.3399, "step": 100 }, { "epoch": 0.049228015215931974, "grad_norm": 11.397383689880371, "learning_rate": 3.666666666666667e-06, "loss": 1.3287, "step": 110 }, { "epoch": 0.05370328932647125, "grad_norm": 3.402496576309204, "learning_rate": 4.000000000000001e-06, "loss": 1.3195, "step": 120 }, { "epoch": 0.058178563437010514, "grad_norm": 6.357070446014404, "learning_rate": 4.333333333333334e-06, "loss": 1.3084, "step": 130 }, { "epoch": 0.06265383754754979, "grad_norm": 3.6552987098693848, "learning_rate": 4.666666666666667e-06, "loss": 1.2956, "step": 140 }, { "epoch": 0.06712911165808906, "grad_norm": 8.76594352722168, "learning_rate": 5e-06, "loss": 1.2873, "step": 150 }, { "epoch": 0.07160438576862833, "grad_norm": 5.0727434158325195, "learning_rate": 5.333333333333334e-06, "loss": 1.277, "step": 160 }, { "epoch": 0.0760796598791676, "grad_norm": 4.647884845733643, "learning_rate": 5.666666666666667e-06, "loss": 1.2726, "step": 170 }, { "epoch": 0.08055493398970687, "grad_norm": 4.086989879608154, "learning_rate": 6e-06, "loss": 1.2812, "step": 180 }, { "epoch": 0.08503020810024614, "grad_norm": 3.3524842262268066, "learning_rate": 6.333333333333334e-06, "loss": 1.2544, "step": 190 }, { "epoch": 0.08950548221078541, "grad_norm": 10.989836692810059, "learning_rate": 6.666666666666667e-06, "loss": 1.2667, "step": 200 }, { "epoch": 0.09398075632132469, "grad_norm": 8.459324836730957, "learning_rate": 7.000000000000001e-06, "loss": 1.2593, "step": 210 }, { "epoch": 0.09845603043186395, "grad_norm": 7.355454921722412, "learning_rate": 7.333333333333334e-06, "loss": 1.2529, "step": 220 }, { "epoch": 0.10293130454240322, "grad_norm": 4.223244667053223, "learning_rate": 7.666666666666667e-06, "loss": 1.2434, "step": 230 }, { "epoch": 0.1074065786529425, "grad_norm": 29.191478729248047, "learning_rate": 8.000000000000001e-06, "loss": 1.249, "step": 240 }, { "epoch": 0.11188185276348177, "grad_norm": 5.550793647766113, "learning_rate": 8.333333333333334e-06, "loss": 1.2577, "step": 250 }, { "epoch": 0.11635712687402103, "grad_norm": 15.218239784240723, "learning_rate": 8.666666666666668e-06, "loss": 1.2443, "step": 260 }, { "epoch": 0.1208324009845603, "grad_norm": 8.252185821533203, "learning_rate": 9e-06, "loss": 1.2494, "step": 270 }, { "epoch": 0.12530767509509957, "grad_norm": 9.197221755981445, "learning_rate": 9.333333333333334e-06, "loss": 1.2401, "step": 280 }, { "epoch": 0.12978294920563885, "grad_norm": 3.049924373626709, "learning_rate": 9.666666666666667e-06, "loss": 1.2421, "step": 290 }, { "epoch": 0.13425822331617812, "grad_norm": 41.47608947753906, "learning_rate": 1e-05, "loss": 1.2513, "step": 300 }, { "epoch": 0.1387334974267174, "grad_norm": 4.9081926345825195, "learning_rate": 1.0333333333333333e-05, "loss": 1.2506, "step": 310 }, { "epoch": 0.14320877153725667, "grad_norm": 8.71500301361084, "learning_rate": 1.0666666666666667e-05, "loss": 1.2513, "step": 320 }, { "epoch": 0.14768404564779591, "grad_norm": 11.26369571685791, "learning_rate": 1.1000000000000001e-05, "loss": 1.2523, "step": 330 }, { "epoch": 0.1521593197583352, "grad_norm": 5.4679856300354, "learning_rate": 1.1333333333333334e-05, "loss": 1.2411, "step": 340 }, { "epoch": 0.15663459386887446, "grad_norm": 8.881339073181152, "learning_rate": 1.1666666666666668e-05, "loss": 1.2405, "step": 350 }, { "epoch": 0.16110986797941373, "grad_norm": 9.603055000305176, "learning_rate": 1.2e-05, "loss": 1.2355, "step": 360 }, { "epoch": 0.165585142089953, "grad_norm": 5.888444900512695, "learning_rate": 1.2333333333333334e-05, "loss": 1.2372, "step": 370 }, { "epoch": 0.17006041620049228, "grad_norm": 4.384479522705078, "learning_rate": 1.2666666666666668e-05, "loss": 1.2251, "step": 380 }, { "epoch": 0.17453569031103155, "grad_norm": 5.680290222167969, "learning_rate": 1.3000000000000001e-05, "loss": 1.2261, "step": 390 }, { "epoch": 0.17901096442157083, "grad_norm": 5.963222503662109, "learning_rate": 1.3333333333333333e-05, "loss": 1.2233, "step": 400 }, { "epoch": 0.1834862385321101, "grad_norm": 4.027234077453613, "learning_rate": 1.3666666666666666e-05, "loss": 1.2237, "step": 410 }, { "epoch": 0.18796151264264938, "grad_norm": 9.110264778137207, "learning_rate": 1.4000000000000001e-05, "loss": 1.218, "step": 420 }, { "epoch": 0.19243678675318862, "grad_norm": 4.758606910705566, "learning_rate": 1.4333333333333334e-05, "loss": 1.1999, "step": 430 }, { "epoch": 0.1969120608637279, "grad_norm": 2.807481050491333, "learning_rate": 1.4666666666666668e-05, "loss": 1.1987, "step": 440 }, { "epoch": 0.20138733497426717, "grad_norm": 2.7931525707244873, "learning_rate": 1.5e-05, "loss": 1.1959, "step": 450 }, { "epoch": 0.20586260908480644, "grad_norm": 2.2306315898895264, "learning_rate": 1.5333333333333334e-05, "loss": 1.1941, "step": 460 }, { "epoch": 0.21033788319534572, "grad_norm": 6.576282024383545, "learning_rate": 1.5666666666666667e-05, "loss": 1.1772, "step": 470 }, { "epoch": 0.214813157305885, "grad_norm": 13.208002090454102, "learning_rate": 1.6000000000000003e-05, "loss": 1.1664, "step": 480 }, { "epoch": 0.21928843141642426, "grad_norm": 1.9115084409713745, "learning_rate": 1.6333333333333335e-05, "loss": 1.1535, "step": 490 }, { "epoch": 0.22376370552696354, "grad_norm": 1.3418084383010864, "learning_rate": 1.6666666666666667e-05, "loss": 1.1472, "step": 500 }, { "epoch": 0.2282389796375028, "grad_norm": 0.852054238319397, "learning_rate": 1.7000000000000003e-05, "loss": 1.1295, "step": 510 }, { "epoch": 0.23271425374804205, "grad_norm": 0.8802962899208069, "learning_rate": 1.7333333333333336e-05, "loss": 1.1277, "step": 520 }, { "epoch": 0.23718952785858133, "grad_norm": 0.6697993874549866, "learning_rate": 1.7666666666666668e-05, "loss": 1.1163, "step": 530 }, { "epoch": 0.2416648019691206, "grad_norm": 0.613279402256012, "learning_rate": 1.8e-05, "loss": 1.1082, "step": 540 }, { "epoch": 0.24614007607965988, "grad_norm": 0.6393995881080627, "learning_rate": 1.8333333333333333e-05, "loss": 1.0997, "step": 550 }, { "epoch": 0.25061535019019915, "grad_norm": 0.4291166663169861, "learning_rate": 1.866666666666667e-05, "loss": 1.0944, "step": 560 }, { "epoch": 0.2550906243007384, "grad_norm": 0.7893696427345276, "learning_rate": 1.9e-05, "loss": 1.0911, "step": 570 }, { "epoch": 0.2595658984112777, "grad_norm": 0.5890133380889893, "learning_rate": 1.9333333333333333e-05, "loss": 1.0878, "step": 580 }, { "epoch": 0.26404117252181697, "grad_norm": 0.5559952855110168, "learning_rate": 1.9666666666666666e-05, "loss": 1.0826, "step": 590 }, { "epoch": 0.26851644663235624, "grad_norm": 0.5822196006774902, "learning_rate": 2e-05, "loss": 1.0849, "step": 600 }, { "epoch": 0.2729917207428955, "grad_norm": 0.6737690567970276, "learning_rate": 2.0333333333333334e-05, "loss": 1.0821, "step": 610 }, { "epoch": 0.2774669948534348, "grad_norm": 0.5887278318405151, "learning_rate": 2.0666666666666666e-05, "loss": 1.0794, "step": 620 }, { "epoch": 0.28194226896397406, "grad_norm": 0.5727159380912781, "learning_rate": 2.1e-05, "loss": 1.073, "step": 630 }, { "epoch": 0.28641754307451334, "grad_norm": 0.5685022473335266, "learning_rate": 2.1333333333333335e-05, "loss": 1.0737, "step": 640 }, { "epoch": 0.2908928171850526, "grad_norm": 0.7646859288215637, "learning_rate": 2.1666666666666667e-05, "loss": 1.0715, "step": 650 }, { "epoch": 0.29536809129559183, "grad_norm": 0.6545053124427795, "learning_rate": 2.2000000000000003e-05, "loss": 1.0703, "step": 660 }, { "epoch": 0.2998433654061311, "grad_norm": 0.505450963973999, "learning_rate": 2.2333333333333335e-05, "loss": 1.0715, "step": 670 }, { "epoch": 0.3043186395166704, "grad_norm": 0.6096774339675903, "learning_rate": 2.2666666666666668e-05, "loss": 1.066, "step": 680 }, { "epoch": 0.30879391362720965, "grad_norm": 0.6588497757911682, "learning_rate": 2.3000000000000003e-05, "loss": 1.0628, "step": 690 }, { "epoch": 0.3132691877377489, "grad_norm": 0.9111886024475098, "learning_rate": 2.3333333333333336e-05, "loss": 1.0614, "step": 700 }, { "epoch": 0.3177444618482882, "grad_norm": 0.665205180644989, "learning_rate": 2.3666666666666668e-05, "loss": 1.064, "step": 710 }, { "epoch": 0.32221973595882747, "grad_norm": 0.6015230417251587, "learning_rate": 2.4e-05, "loss": 1.0635, "step": 720 }, { "epoch": 0.32669501006936674, "grad_norm": 0.6945334076881409, "learning_rate": 2.4333333333333336e-05, "loss": 1.061, "step": 730 }, { "epoch": 0.331170284179906, "grad_norm": 0.689781665802002, "learning_rate": 2.466666666666667e-05, "loss": 1.0644, "step": 740 }, { "epoch": 0.3356455582904453, "grad_norm": 0.43018200993537903, "learning_rate": 2.5e-05, "loss": 1.0676, "step": 750 }, { "epoch": 0.34012083240098456, "grad_norm": 0.3710629343986511, "learning_rate": 2.5333333333333337e-05, "loss": 1.0567, "step": 760 }, { "epoch": 0.34459610651152384, "grad_norm": 0.34247830510139465, "learning_rate": 2.5666666666666666e-05, "loss": 1.0538, "step": 770 }, { "epoch": 0.3490713806220631, "grad_norm": 0.39325398206710815, "learning_rate": 2.6000000000000002e-05, "loss": 1.0546, "step": 780 }, { "epoch": 0.3535466547326024, "grad_norm": 0.37366417050361633, "learning_rate": 2.633333333333333e-05, "loss": 1.0588, "step": 790 }, { "epoch": 0.35802192884314166, "grad_norm": 0.6455649137496948, "learning_rate": 2.6666666666666667e-05, "loss": 1.0573, "step": 800 }, { "epoch": 0.36249720295368093, "grad_norm": 0.6332912445068359, "learning_rate": 2.7000000000000002e-05, "loss": 1.0528, "step": 810 }, { "epoch": 0.3669724770642202, "grad_norm": 0.49358344078063965, "learning_rate": 2.733333333333333e-05, "loss": 1.0533, "step": 820 }, { "epoch": 0.3714477511747595, "grad_norm": 0.5301892161369324, "learning_rate": 2.7666666666666667e-05, "loss": 1.0507, "step": 830 }, { "epoch": 0.37592302528529875, "grad_norm": 0.4063838720321655, "learning_rate": 2.8000000000000003e-05, "loss": 1.0529, "step": 840 }, { "epoch": 0.38039829939583797, "grad_norm": 0.45295289158821106, "learning_rate": 2.8333333333333335e-05, "loss": 1.0487, "step": 850 }, { "epoch": 0.38487357350637724, "grad_norm": 0.5931718349456787, "learning_rate": 2.8666666666666668e-05, "loss": 1.0555, "step": 860 }, { "epoch": 0.3893488476169165, "grad_norm": 0.7279201745986938, "learning_rate": 2.9e-05, "loss": 1.0434, "step": 870 }, { "epoch": 0.3938241217274558, "grad_norm": 0.48830801248550415, "learning_rate": 2.9333333333333336e-05, "loss": 1.0449, "step": 880 }, { "epoch": 0.39829939583799506, "grad_norm": 0.5698959231376648, "learning_rate": 2.9666666666666672e-05, "loss": 1.0395, "step": 890 }, { "epoch": 0.40277466994853434, "grad_norm": 0.5753142833709717, "learning_rate": 3e-05, "loss": 1.0326, "step": 900 }, { "epoch": 0.4072499440590736, "grad_norm": 0.42692816257476807, "learning_rate": 3.0333333333333337e-05, "loss": 1.0188, "step": 910 }, { "epoch": 0.4117252181696129, "grad_norm": 0.661515474319458, "learning_rate": 3.066666666666667e-05, "loss": 0.9952, "step": 920 }, { "epoch": 0.41620049228015216, "grad_norm": 0.6225526332855225, "learning_rate": 3.1e-05, "loss": 0.9722, "step": 930 }, { "epoch": 0.42067576639069143, "grad_norm": 0.8899055123329163, "learning_rate": 3.1333333333333334e-05, "loss": 0.9279, "step": 940 }, { "epoch": 0.4251510405012307, "grad_norm": 0.9131832122802734, "learning_rate": 3.1666666666666666e-05, "loss": 0.8738, "step": 950 }, { "epoch": 0.42962631461177, "grad_norm": 0.8791340589523315, "learning_rate": 3.2000000000000005e-05, "loss": 0.825, "step": 960 }, { "epoch": 0.43410158872230925, "grad_norm": 0.8399432897567749, "learning_rate": 3.233333333333333e-05, "loss": 0.7689, "step": 970 }, { "epoch": 0.4385768628328485, "grad_norm": 0.9326962828636169, "learning_rate": 3.266666666666667e-05, "loss": 0.7106, "step": 980 }, { "epoch": 0.4430521369433878, "grad_norm": 0.764723002910614, "learning_rate": 3.3e-05, "loss": 0.66, "step": 990 }, { "epoch": 0.44752741105392707, "grad_norm": 0.8456273078918457, "learning_rate": 3.3333333333333335e-05, "loss": 0.6042, "step": 1000 }, { "epoch": 0.45200268516446634, "grad_norm": 0.7793407440185547, "learning_rate": 3.366666666666667e-05, "loss": 0.5498, "step": 1010 }, { "epoch": 0.4564779592750056, "grad_norm": 0.8557698726654053, "learning_rate": 3.4000000000000007e-05, "loss": 0.4974, "step": 1020 }, { "epoch": 0.4609532333855449, "grad_norm": 0.9878665208816528, "learning_rate": 3.433333333333333e-05, "loss": 0.4735, "step": 1030 }, { "epoch": 0.4654285074960841, "grad_norm": 0.7022286653518677, "learning_rate": 3.466666666666667e-05, "loss": 0.422, "step": 1040 }, { "epoch": 0.4699037816066234, "grad_norm": 0.9560638070106506, "learning_rate": 3.5e-05, "loss": 0.4127, "step": 1050 }, { "epoch": 0.47437905571716266, "grad_norm": 0.8896686434745789, "learning_rate": 3.5333333333333336e-05, "loss": 0.3772, "step": 1060 }, { "epoch": 0.47885432982770193, "grad_norm": 0.907202959060669, "learning_rate": 3.566666666666667e-05, "loss": 0.3678, "step": 1070 }, { "epoch": 0.4833296039382412, "grad_norm": 0.7822281122207642, "learning_rate": 3.6e-05, "loss": 0.3272, "step": 1080 }, { "epoch": 0.4878048780487805, "grad_norm": 0.8363612294197083, "learning_rate": 3.633333333333333e-05, "loss": 0.3042, "step": 1090 }, { "epoch": 0.49228015215931975, "grad_norm": 0.7413959503173828, "learning_rate": 3.6666666666666666e-05, "loss": 0.2941, "step": 1100 }, { "epoch": 0.496755426269859, "grad_norm": 0.7976481914520264, "learning_rate": 3.7e-05, "loss": 0.2814, "step": 1110 }, { "epoch": 0.5012307003803983, "grad_norm": 0.8663702607154846, "learning_rate": 3.733333333333334e-05, "loss": 0.2626, "step": 1120 }, { "epoch": 0.5057059744909376, "grad_norm": 0.8075986504554749, "learning_rate": 3.766666666666667e-05, "loss": 0.2528, "step": 1130 }, { "epoch": 0.5101812486014768, "grad_norm": 1.0972553491592407, "learning_rate": 3.8e-05, "loss": 0.2324, "step": 1140 }, { "epoch": 0.5146565227120161, "grad_norm": 0.8539525270462036, "learning_rate": 3.8333333333333334e-05, "loss": 0.2241, "step": 1150 }, { "epoch": 0.5191317968225554, "grad_norm": 0.7775368690490723, "learning_rate": 3.866666666666667e-05, "loss": 0.2145, "step": 1160 }, { "epoch": 0.5236070709330947, "grad_norm": 1.0671647787094116, "learning_rate": 3.9000000000000006e-05, "loss": 0.2046, "step": 1170 }, { "epoch": 0.5280823450436339, "grad_norm": 0.8560193777084351, "learning_rate": 3.933333333333333e-05, "loss": 0.2081, "step": 1180 }, { "epoch": 0.5325576191541732, "grad_norm": 0.864108145236969, "learning_rate": 3.966666666666667e-05, "loss": 0.2019, "step": 1190 }, { "epoch": 0.5370328932647125, "grad_norm": 0.8566147089004517, "learning_rate": 4e-05, "loss": 0.1967, "step": 1200 }, { "epoch": 0.5415081673752518, "grad_norm": 0.8582598567008972, "learning_rate": 4.0333333333333336e-05, "loss": 0.1993, "step": 1210 }, { "epoch": 0.545983441485791, "grad_norm": 0.9574819803237915, "learning_rate": 4.066666666666667e-05, "loss": 0.1889, "step": 1220 }, { "epoch": 0.5504587155963303, "grad_norm": 1.1248778104782104, "learning_rate": 4.1e-05, "loss": 0.1868, "step": 1230 }, { "epoch": 0.5549339897068696, "grad_norm": 0.7771704196929932, "learning_rate": 4.133333333333333e-05, "loss": 0.186, "step": 1240 }, { "epoch": 0.5594092638174089, "grad_norm": 0.6983978152275085, "learning_rate": 4.166666666666667e-05, "loss": 0.1876, "step": 1250 }, { "epoch": 0.5638845379279481, "grad_norm": 0.8146538734436035, "learning_rate": 4.2e-05, "loss": 0.1842, "step": 1260 }, { "epoch": 0.5683598120384874, "grad_norm": 0.7569879293441772, "learning_rate": 4.233333333333334e-05, "loss": 0.1776, "step": 1270 }, { "epoch": 0.5728350861490267, "grad_norm": 0.8377714157104492, "learning_rate": 4.266666666666667e-05, "loss": 0.1892, "step": 1280 }, { "epoch": 0.577310360259566, "grad_norm": 0.7476964592933655, "learning_rate": 4.3e-05, "loss": 0.1736, "step": 1290 }, { "epoch": 0.5817856343701052, "grad_norm": 0.7043957710266113, "learning_rate": 4.3333333333333334e-05, "loss": 0.1684, "step": 1300 }, { "epoch": 0.5862609084806445, "grad_norm": 0.7796573638916016, "learning_rate": 4.3666666666666666e-05, "loss": 0.1668, "step": 1310 }, { "epoch": 0.5907361825911837, "grad_norm": 0.9181379675865173, "learning_rate": 4.4000000000000006e-05, "loss": 0.1713, "step": 1320 }, { "epoch": 0.5952114567017229, "grad_norm": 0.7918029427528381, "learning_rate": 4.433333333333334e-05, "loss": 0.1742, "step": 1330 }, { "epoch": 0.5996867308122622, "grad_norm": 0.7342538237571716, "learning_rate": 4.466666666666667e-05, "loss": 0.1698, "step": 1340 }, { "epoch": 0.6041620049228015, "grad_norm": 0.6402199864387512, "learning_rate": 4.5e-05, "loss": 0.1714, "step": 1350 }, { "epoch": 0.6086372790333408, "grad_norm": 0.7858932018280029, "learning_rate": 4.5333333333333335e-05, "loss": 0.1656, "step": 1360 }, { "epoch": 0.61311255314388, "grad_norm": 0.7971068024635315, "learning_rate": 4.566666666666667e-05, "loss": 0.1647, "step": 1370 }, { "epoch": 0.6175878272544193, "grad_norm": 0.7253530621528625, "learning_rate": 4.600000000000001e-05, "loss": 0.1568, "step": 1380 }, { "epoch": 0.6220631013649586, "grad_norm": 0.7274602055549622, "learning_rate": 4.633333333333333e-05, "loss": 0.1621, "step": 1390 }, { "epoch": 0.6265383754754978, "grad_norm": 0.7389922142028809, "learning_rate": 4.666666666666667e-05, "loss": 0.1636, "step": 1400 }, { "epoch": 0.6310136495860371, "grad_norm": 0.7553339600563049, "learning_rate": 4.7e-05, "loss": 0.1643, "step": 1410 }, { "epoch": 0.6354889236965764, "grad_norm": 0.7003543376922607, "learning_rate": 4.7333333333333336e-05, "loss": 0.1571, "step": 1420 }, { "epoch": 0.6399641978071157, "grad_norm": 0.7014659643173218, "learning_rate": 4.766666666666667e-05, "loss": 0.1679, "step": 1430 }, { "epoch": 0.6444394719176549, "grad_norm": 0.7665868401527405, "learning_rate": 4.8e-05, "loss": 0.1594, "step": 1440 }, { "epoch": 0.6489147460281942, "grad_norm": 0.693705141544342, "learning_rate": 4.8333333333333334e-05, "loss": 0.1576, "step": 1450 }, { "epoch": 0.6533900201387335, "grad_norm": 0.7653425931930542, "learning_rate": 4.866666666666667e-05, "loss": 0.1599, "step": 1460 }, { "epoch": 0.6578652942492728, "grad_norm": 0.7983621954917908, "learning_rate": 4.9e-05, "loss": 0.1615, "step": 1470 }, { "epoch": 0.662340568359812, "grad_norm": 0.6714649796485901, "learning_rate": 4.933333333333334e-05, "loss": 0.1587, "step": 1480 }, { "epoch": 0.6668158424703513, "grad_norm": 0.6793619990348816, "learning_rate": 4.966666666666667e-05, "loss": 0.1513, "step": 1490 }, { "epoch": 0.6712911165808906, "grad_norm": 0.758057713508606, "learning_rate": 5e-05, "loss": 0.1514, "step": 1500 }, { "epoch": 0.6757663906914299, "grad_norm": 0.8115227222442627, "learning_rate": 5.0333333333333335e-05, "loss": 0.1515, "step": 1510 }, { "epoch": 0.6802416648019691, "grad_norm": 0.6773813366889954, "learning_rate": 5.0666666666666674e-05, "loss": 0.1535, "step": 1520 }, { "epoch": 0.6847169389125084, "grad_norm": 0.695502519607544, "learning_rate": 5.1000000000000006e-05, "loss": 0.143, "step": 1530 }, { "epoch": 0.6891922130230477, "grad_norm": 0.7313320636749268, "learning_rate": 5.133333333333333e-05, "loss": 0.1479, "step": 1540 }, { "epoch": 0.693667487133587, "grad_norm": 0.7531288862228394, "learning_rate": 5.166666666666667e-05, "loss": 0.1581, "step": 1550 }, { "epoch": 0.6981427612441262, "grad_norm": 0.6932622790336609, "learning_rate": 5.2000000000000004e-05, "loss": 0.1481, "step": 1560 }, { "epoch": 0.7026180353546655, "grad_norm": 0.7276437878608704, "learning_rate": 5.2333333333333336e-05, "loss": 0.1541, "step": 1570 }, { "epoch": 0.7070933094652048, "grad_norm": 0.7125598192214966, "learning_rate": 5.266666666666666e-05, "loss": 0.1464, "step": 1580 }, { "epoch": 0.711568583575744, "grad_norm": 0.7061651349067688, "learning_rate": 5.300000000000001e-05, "loss": 0.1454, "step": 1590 }, { "epoch": 0.7160438576862833, "grad_norm": 0.6304736137390137, "learning_rate": 5.333333333333333e-05, "loss": 0.1466, "step": 1600 }, { "epoch": 0.7205191317968226, "grad_norm": 0.6544821262359619, "learning_rate": 5.3666666666666666e-05, "loss": 0.1447, "step": 1610 }, { "epoch": 0.7249944059073619, "grad_norm": 0.7029533982276917, "learning_rate": 5.4000000000000005e-05, "loss": 0.1405, "step": 1620 }, { "epoch": 0.7294696800179011, "grad_norm": 0.7970440983772278, "learning_rate": 5.433333333333334e-05, "loss": 0.1507, "step": 1630 }, { "epoch": 0.7339449541284404, "grad_norm": 0.6224236488342285, "learning_rate": 5.466666666666666e-05, "loss": 0.1423, "step": 1640 }, { "epoch": 0.7384202282389797, "grad_norm": 0.6939723491668701, "learning_rate": 5.500000000000001e-05, "loss": 0.1512, "step": 1650 }, { "epoch": 0.742895502349519, "grad_norm": 0.6903528571128845, "learning_rate": 5.5333333333333334e-05, "loss": 0.1425, "step": 1660 }, { "epoch": 0.7473707764600582, "grad_norm": 0.7100644111633301, "learning_rate": 5.566666666666667e-05, "loss": 0.1438, "step": 1670 }, { "epoch": 0.7518460505705975, "grad_norm": 0.7564582824707031, "learning_rate": 5.6000000000000006e-05, "loss": 0.1427, "step": 1680 }, { "epoch": 0.7563213246811367, "grad_norm": 0.5775176882743835, "learning_rate": 5.633333333333334e-05, "loss": 0.145, "step": 1690 }, { "epoch": 0.7607965987916759, "grad_norm": 0.8500620722770691, "learning_rate": 5.666666666666667e-05, "loss": 0.1457, "step": 1700 }, { "epoch": 0.7652718729022152, "grad_norm": 0.6986157298088074, "learning_rate": 5.6999999999999996e-05, "loss": 0.1315, "step": 1710 }, { "epoch": 0.7697471470127545, "grad_norm": 0.7648106217384338, "learning_rate": 5.7333333333333336e-05, "loss": 0.148, "step": 1720 }, { "epoch": 0.7742224211232938, "grad_norm": 0.7905831336975098, "learning_rate": 5.766666666666667e-05, "loss": 0.1459, "step": 1730 }, { "epoch": 0.778697695233833, "grad_norm": 0.7604420781135559, "learning_rate": 5.8e-05, "loss": 0.1403, "step": 1740 }, { "epoch": 0.7831729693443723, "grad_norm": 0.6655674576759338, "learning_rate": 5.833333333333334e-05, "loss": 0.1336, "step": 1750 }, { "epoch": 0.7876482434549116, "grad_norm": 0.7892205715179443, "learning_rate": 5.866666666666667e-05, "loss": 0.1367, "step": 1760 }, { "epoch": 0.7921235175654509, "grad_norm": 0.6175403594970703, "learning_rate": 5.9e-05, "loss": 0.1365, "step": 1770 }, { "epoch": 0.7965987916759901, "grad_norm": 0.6012178659439087, "learning_rate": 5.9333333333333343e-05, "loss": 0.1387, "step": 1780 }, { "epoch": 0.8010740657865294, "grad_norm": 0.6702210903167725, "learning_rate": 5.966666666666667e-05, "loss": 0.1377, "step": 1790 }, { "epoch": 0.8055493398970687, "grad_norm": 0.6882554292678833, "learning_rate": 6e-05, "loss": 0.1409, "step": 1800 }, { "epoch": 0.810024614007608, "grad_norm": 0.6881408095359802, "learning_rate": 6.033333333333334e-05, "loss": 0.1372, "step": 1810 }, { "epoch": 0.8144998881181472, "grad_norm": 0.6163994669914246, "learning_rate": 6.066666666666667e-05, "loss": 0.1341, "step": 1820 }, { "epoch": 0.8189751622286865, "grad_norm": 0.5717507004737854, "learning_rate": 6.1e-05, "loss": 0.1372, "step": 1830 }, { "epoch": 0.8234504363392258, "grad_norm": 0.6699105501174927, "learning_rate": 6.133333333333334e-05, "loss": 0.1329, "step": 1840 }, { "epoch": 0.827925710449765, "grad_norm": 0.5753948092460632, "learning_rate": 6.166666666666667e-05, "loss": 0.1377, "step": 1850 }, { "epoch": 0.8324009845603043, "grad_norm": 0.5978674292564392, "learning_rate": 6.2e-05, "loss": 0.1313, "step": 1860 }, { "epoch": 0.8368762586708436, "grad_norm": 0.7257053256034851, "learning_rate": 6.233333333333334e-05, "loss": 0.1301, "step": 1870 }, { "epoch": 0.8413515327813829, "grad_norm": 0.5644158720970154, "learning_rate": 6.266666666666667e-05, "loss": 0.1351, "step": 1880 }, { "epoch": 0.8458268068919221, "grad_norm": 0.5729243755340576, "learning_rate": 6.3e-05, "loss": 0.133, "step": 1890 }, { "epoch": 0.8503020810024614, "grad_norm": 0.7270042300224304, "learning_rate": 6.333333333333333e-05, "loss": 0.1389, "step": 1900 }, { "epoch": 0.8547773551130007, "grad_norm": 0.6678065657615662, "learning_rate": 6.366666666666668e-05, "loss": 0.1305, "step": 1910 }, { "epoch": 0.85925262922354, "grad_norm": 0.6274886131286621, "learning_rate": 6.400000000000001e-05, "loss": 0.1318, "step": 1920 }, { "epoch": 0.8637279033340792, "grad_norm": 0.5678911805152893, "learning_rate": 6.433333333333333e-05, "loss": 0.1313, "step": 1930 }, { "epoch": 0.8682031774446185, "grad_norm": 0.7149289846420288, "learning_rate": 6.466666666666666e-05, "loss": 0.1334, "step": 1940 }, { "epoch": 0.8726784515551578, "grad_norm": 0.6517453789710999, "learning_rate": 6.500000000000001e-05, "loss": 0.1277, "step": 1950 }, { "epoch": 0.877153725665697, "grad_norm": 0.6262755393981934, "learning_rate": 6.533333333333334e-05, "loss": 0.1393, "step": 1960 }, { "epoch": 0.8816289997762363, "grad_norm": 0.6474819779396057, "learning_rate": 6.566666666666666e-05, "loss": 0.1304, "step": 1970 }, { "epoch": 0.8861042738867756, "grad_norm": 0.6068485975265503, "learning_rate": 6.6e-05, "loss": 0.1337, "step": 1980 }, { "epoch": 0.8905795479973149, "grad_norm": 0.6437543630599976, "learning_rate": 6.633333333333334e-05, "loss": 0.1356, "step": 1990 }, { "epoch": 0.8950548221078541, "grad_norm": 0.7113824486732483, "learning_rate": 6.666666666666667e-05, "loss": 0.1367, "step": 2000 }, { "epoch": 0.8995300962183934, "grad_norm": 0.6371535062789917, "learning_rate": 6.7e-05, "loss": 0.1352, "step": 2010 }, { "epoch": 0.9040053703289327, "grad_norm": 0.6540469527244568, "learning_rate": 6.733333333333333e-05, "loss": 0.1236, "step": 2020 }, { "epoch": 0.908480644439472, "grad_norm": 0.5861204862594604, "learning_rate": 6.766666666666667e-05, "loss": 0.1234, "step": 2030 }, { "epoch": 0.9129559185500112, "grad_norm": 0.6951718926429749, "learning_rate": 6.800000000000001e-05, "loss": 0.1273, "step": 2040 }, { "epoch": 0.9174311926605505, "grad_norm": 0.522785484790802, "learning_rate": 6.833333333333333e-05, "loss": 0.133, "step": 2050 }, { "epoch": 0.9219064667710898, "grad_norm": 0.6765419244766235, "learning_rate": 6.866666666666666e-05, "loss": 0.1308, "step": 2060 }, { "epoch": 0.926381740881629, "grad_norm": 0.7043836116790771, "learning_rate": 6.9e-05, "loss": 0.1293, "step": 2070 }, { "epoch": 0.9308570149921682, "grad_norm": 0.620452344417572, "learning_rate": 6.933333333333334e-05, "loss": 0.1297, "step": 2080 }, { "epoch": 0.9353322891027075, "grad_norm": 0.5202000141143799, "learning_rate": 6.966666666666668e-05, "loss": 0.1266, "step": 2090 }, { "epoch": 0.9398075632132468, "grad_norm": 0.594566285610199, "learning_rate": 7e-05, "loss": 0.1183, "step": 2100 }, { "epoch": 0.944282837323786, "grad_norm": 0.6313052773475647, "learning_rate": 7.033333333333334e-05, "loss": 0.1202, "step": 2110 }, { "epoch": 0.9487581114343253, "grad_norm": 0.5528560280799866, "learning_rate": 7.066666666666667e-05, "loss": 0.1195, "step": 2120 }, { "epoch": 0.9532333855448646, "grad_norm": 0.5841560959815979, "learning_rate": 7.1e-05, "loss": 0.1254, "step": 2130 }, { "epoch": 0.9577086596554039, "grad_norm": 0.544403612613678, "learning_rate": 7.133333333333334e-05, "loss": 0.1251, "step": 2140 }, { "epoch": 0.9621839337659431, "grad_norm": 0.5796578526496887, "learning_rate": 7.166666666666667e-05, "loss": 0.1243, "step": 2150 }, { "epoch": 0.9666592078764824, "grad_norm": 0.5786649584770203, "learning_rate": 7.2e-05, "loss": 0.1233, "step": 2160 }, { "epoch": 0.9711344819870217, "grad_norm": 0.6096514463424683, "learning_rate": 7.233333333333335e-05, "loss": 0.1233, "step": 2170 }, { "epoch": 0.975609756097561, "grad_norm": 0.5841403007507324, "learning_rate": 7.266666666666667e-05, "loss": 0.1204, "step": 2180 }, { "epoch": 0.9800850302081002, "grad_norm": 0.609112560749054, "learning_rate": 7.3e-05, "loss": 0.1258, "step": 2190 }, { "epoch": 0.9845603043186395, "grad_norm": 0.6360578536987305, "learning_rate": 7.333333333333333e-05, "loss": 0.1262, "step": 2200 }, { "epoch": 0.9890355784291788, "grad_norm": 0.5733288526535034, "learning_rate": 7.366666666666668e-05, "loss": 0.1277, "step": 2210 }, { "epoch": 0.993510852539718, "grad_norm": 0.5313320755958557, "learning_rate": 7.4e-05, "loss": 0.1211, "step": 2220 }, { "epoch": 0.9979861266502573, "grad_norm": 0.6055431962013245, "learning_rate": 7.433333333333333e-05, "loss": 0.1212, "step": 2230 }, { "epoch": 1.0024614007607966, "grad_norm": 0.5910041332244873, "learning_rate": 7.466666666666667e-05, "loss": 0.1177, "step": 2240 }, { "epoch": 1.006936674871336, "grad_norm": 0.5385250449180603, "learning_rate": 7.500000000000001e-05, "loss": 0.1215, "step": 2250 }, { "epoch": 1.0114119489818751, "grad_norm": 0.5837197303771973, "learning_rate": 7.533333333333334e-05, "loss": 0.1144, "step": 2260 }, { "epoch": 1.0158872230924143, "grad_norm": 0.5825377702713013, "learning_rate": 7.566666666666667e-05, "loss": 0.1167, "step": 2270 }, { "epoch": 1.0203624972029537, "grad_norm": 0.514037549495697, "learning_rate": 7.6e-05, "loss": 0.1204, "step": 2280 }, { "epoch": 1.0248377713134929, "grad_norm": 0.4897834360599518, "learning_rate": 7.633333333333334e-05, "loss": 0.1164, "step": 2290 }, { "epoch": 1.0293130454240322, "grad_norm": 0.5620017647743225, "learning_rate": 7.666666666666667e-05, "loss": 0.1261, "step": 2300 }, { "epoch": 1.0337883195345714, "grad_norm": 0.5106384754180908, "learning_rate": 7.7e-05, "loss": 0.1196, "step": 2310 }, { "epoch": 1.0382635936451108, "grad_norm": 0.5749973058700562, "learning_rate": 7.733333333333333e-05, "loss": 0.1178, "step": 2320 }, { "epoch": 1.04273886775565, "grad_norm": 0.5566390752792358, "learning_rate": 7.766666666666667e-05, "loss": 0.1153, "step": 2330 }, { "epoch": 1.0472141418661893, "grad_norm": 0.5946740508079529, "learning_rate": 7.800000000000001e-05, "loss": 0.1242, "step": 2340 }, { "epoch": 1.0516894159767285, "grad_norm": 0.5788619518280029, "learning_rate": 7.833333333333333e-05, "loss": 0.1265, "step": 2350 }, { "epoch": 1.0561646900872679, "grad_norm": 0.5738166570663452, "learning_rate": 7.866666666666666e-05, "loss": 0.1242, "step": 2360 }, { "epoch": 1.060639964197807, "grad_norm": 0.5274977684020996, "learning_rate": 7.900000000000001e-05, "loss": 0.1218, "step": 2370 }, { "epoch": 1.0651152383083464, "grad_norm": 0.5607765316963196, "learning_rate": 7.933333333333334e-05, "loss": 0.1203, "step": 2380 }, { "epoch": 1.0695905124188856, "grad_norm": 0.5642150640487671, "learning_rate": 7.966666666666666e-05, "loss": 0.125, "step": 2390 }, { "epoch": 1.074065786529425, "grad_norm": 0.4726608097553253, "learning_rate": 8e-05, "loss": 0.1178, "step": 2400 }, { "epoch": 1.0785410606399641, "grad_norm": 0.5550168752670288, "learning_rate": 8.033333333333334e-05, "loss": 0.1191, "step": 2410 }, { "epoch": 1.0830163347505035, "grad_norm": 0.6237925291061401, "learning_rate": 8.066666666666667e-05, "loss": 0.1194, "step": 2420 }, { "epoch": 1.0874916088610427, "grad_norm": 0.4940697252750397, "learning_rate": 8.1e-05, "loss": 0.1134, "step": 2430 }, { "epoch": 1.091966882971582, "grad_norm": 0.5143089294433594, "learning_rate": 8.133333333333334e-05, "loss": 0.1156, "step": 2440 }, { "epoch": 1.0964421570821212, "grad_norm": 0.49556687474250793, "learning_rate": 8.166666666666667e-05, "loss": 0.1128, "step": 2450 }, { "epoch": 1.1009174311926606, "grad_norm": 0.5184047818183899, "learning_rate": 8.2e-05, "loss": 0.1187, "step": 2460 }, { "epoch": 1.1053927053031998, "grad_norm": 0.4919315278530121, "learning_rate": 8.233333333333333e-05, "loss": 0.1181, "step": 2470 }, { "epoch": 1.1098679794137392, "grad_norm": 0.507210373878479, "learning_rate": 8.266666666666667e-05, "loss": 0.1182, "step": 2480 }, { "epoch": 1.1143432535242783, "grad_norm": 0.6839494109153748, "learning_rate": 8.3e-05, "loss": 0.1217, "step": 2490 }, { "epoch": 1.1188185276348177, "grad_norm": 0.5790488123893738, "learning_rate": 8.333333333333334e-05, "loss": 0.1178, "step": 2500 }, { "epoch": 1.1232938017453569, "grad_norm": 0.5511919856071472, "learning_rate": 8.366666666666668e-05, "loss": 0.1209, "step": 2510 }, { "epoch": 1.1277690758558963, "grad_norm": 0.5629762411117554, "learning_rate": 8.4e-05, "loss": 0.116, "step": 2520 }, { "epoch": 1.1322443499664354, "grad_norm": 0.5521126985549927, "learning_rate": 8.433333333333334e-05, "loss": 0.1147, "step": 2530 }, { "epoch": 1.1367196240769748, "grad_norm": 0.568841278553009, "learning_rate": 8.466666666666667e-05, "loss": 0.113, "step": 2540 }, { "epoch": 1.141194898187514, "grad_norm": 0.49081435799598694, "learning_rate": 8.5e-05, "loss": 0.1149, "step": 2550 }, { "epoch": 1.1456701722980533, "grad_norm": 0.5653166174888611, "learning_rate": 8.533333333333334e-05, "loss": 0.1127, "step": 2560 }, { "epoch": 1.1501454464085925, "grad_norm": 0.5538638234138489, "learning_rate": 8.566666666666667e-05, "loss": 0.117, "step": 2570 }, { "epoch": 1.154620720519132, "grad_norm": 0.5171223282814026, "learning_rate": 8.6e-05, "loss": 0.1158, "step": 2580 }, { "epoch": 1.159095994629671, "grad_norm": 0.5910136103630066, "learning_rate": 8.633333333333334e-05, "loss": 0.1088, "step": 2590 }, { "epoch": 1.1635712687402102, "grad_norm": 0.533953070640564, "learning_rate": 8.666666666666667e-05, "loss": 0.1143, "step": 2600 }, { "epoch": 1.1680465428507496, "grad_norm": 0.4594457149505615, "learning_rate": 8.7e-05, "loss": 0.1092, "step": 2610 }, { "epoch": 1.172521816961289, "grad_norm": 0.4782252013683319, "learning_rate": 8.733333333333333e-05, "loss": 0.1092, "step": 2620 }, { "epoch": 1.1769970910718281, "grad_norm": 0.4285776913166046, "learning_rate": 8.766666666666668e-05, "loss": 0.1129, "step": 2630 }, { "epoch": 1.1814723651823673, "grad_norm": 0.5105476975440979, "learning_rate": 8.800000000000001e-05, "loss": 0.1119, "step": 2640 }, { "epoch": 1.1859476392929067, "grad_norm": 0.47147491574287415, "learning_rate": 8.833333333333333e-05, "loss": 0.111, "step": 2650 }, { "epoch": 1.190422913403446, "grad_norm": 0.5246534943580627, "learning_rate": 8.866666666666668e-05, "loss": 0.1122, "step": 2660 }, { "epoch": 1.1948981875139852, "grad_norm": 0.49215176701545715, "learning_rate": 8.900000000000001e-05, "loss": 0.1097, "step": 2670 }, { "epoch": 1.1993734616245244, "grad_norm": 0.4793473780155182, "learning_rate": 8.933333333333334e-05, "loss": 0.1111, "step": 2680 }, { "epoch": 1.2038487357350638, "grad_norm": 0.45086634159088135, "learning_rate": 8.966666666666666e-05, "loss": 0.1117, "step": 2690 }, { "epoch": 1.208324009845603, "grad_norm": 0.43648120760917664, "learning_rate": 9e-05, "loss": 0.1124, "step": 2700 }, { "epoch": 1.2127992839561423, "grad_norm": 0.4713621437549591, "learning_rate": 9.033333333333334e-05, "loss": 0.1141, "step": 2710 }, { "epoch": 1.2172745580666815, "grad_norm": 0.4679372012615204, "learning_rate": 9.066666666666667e-05, "loss": 0.1078, "step": 2720 }, { "epoch": 1.2217498321772209, "grad_norm": 0.5537595748901367, "learning_rate": 9.1e-05, "loss": 0.1084, "step": 2730 }, { "epoch": 1.22622510628776, "grad_norm": 0.4873969554901123, "learning_rate": 9.133333333333334e-05, "loss": 0.1087, "step": 2740 }, { "epoch": 1.2307003803982994, "grad_norm": 0.48900479078292847, "learning_rate": 9.166666666666667e-05, "loss": 0.1148, "step": 2750 }, { "epoch": 1.2351756545088386, "grad_norm": 0.5581976175308228, "learning_rate": 9.200000000000001e-05, "loss": 0.1107, "step": 2760 }, { "epoch": 1.239650928619378, "grad_norm": 0.5638328790664673, "learning_rate": 9.233333333333333e-05, "loss": 0.1073, "step": 2770 }, { "epoch": 1.2441262027299171, "grad_norm": 0.5085706114768982, "learning_rate": 9.266666666666666e-05, "loss": 0.1162, "step": 2780 }, { "epoch": 1.2486014768404565, "grad_norm": 0.4759654104709625, "learning_rate": 9.300000000000001e-05, "loss": 0.1147, "step": 2790 }, { "epoch": 1.2530767509509957, "grad_norm": 0.47653117775917053, "learning_rate": 9.333333333333334e-05, "loss": 0.1148, "step": 2800 }, { "epoch": 1.257552025061535, "grad_norm": 0.4472392797470093, "learning_rate": 9.366666666666668e-05, "loss": 0.1146, "step": 2810 }, { "epoch": 1.2620272991720742, "grad_norm": 0.4451843202114105, "learning_rate": 9.4e-05, "loss": 0.1097, "step": 2820 }, { "epoch": 1.2665025732826136, "grad_norm": 0.44669222831726074, "learning_rate": 9.433333333333334e-05, "loss": 0.1101, "step": 2830 }, { "epoch": 1.2709778473931528, "grad_norm": 0.46961072087287903, "learning_rate": 9.466666666666667e-05, "loss": 0.1116, "step": 2840 }, { "epoch": 1.2754531215036922, "grad_norm": 0.42727145552635193, "learning_rate": 9.5e-05, "loss": 0.1085, "step": 2850 }, { "epoch": 1.2799283956142313, "grad_norm": 0.4085581600666046, "learning_rate": 9.533333333333334e-05, "loss": 0.1095, "step": 2860 }, { "epoch": 1.2844036697247707, "grad_norm": 0.48853516578674316, "learning_rate": 9.566666666666667e-05, "loss": 0.115, "step": 2870 }, { "epoch": 1.2888789438353099, "grad_norm": 0.4088858664035797, "learning_rate": 9.6e-05, "loss": 0.1069, "step": 2880 }, { "epoch": 1.2933542179458493, "grad_norm": 0.5364995002746582, "learning_rate": 9.633333333333335e-05, "loss": 0.1115, "step": 2890 }, { "epoch": 1.2978294920563884, "grad_norm": 0.5326026082038879, "learning_rate": 9.666666666666667e-05, "loss": 0.1078, "step": 2900 }, { "epoch": 1.3023047661669278, "grad_norm": 0.5500506162643433, "learning_rate": 9.7e-05, "loss": 0.1056, "step": 2910 }, { "epoch": 1.306780040277467, "grad_norm": 0.49915164709091187, "learning_rate": 9.733333333333335e-05, "loss": 0.102, "step": 2920 }, { "epoch": 1.3112553143880064, "grad_norm": 0.5130475163459778, "learning_rate": 9.766666666666668e-05, "loss": 0.1094, "step": 2930 }, { "epoch": 1.3157305884985455, "grad_norm": 0.4941202700138092, "learning_rate": 9.8e-05, "loss": 0.1046, "step": 2940 }, { "epoch": 1.3202058626090847, "grad_norm": 0.5666459798812866, "learning_rate": 9.833333333333333e-05, "loss": 0.106, "step": 2950 }, { "epoch": 1.324681136719624, "grad_norm": 0.4268014430999756, "learning_rate": 9.866666666666668e-05, "loss": 0.1079, "step": 2960 }, { "epoch": 1.3291564108301634, "grad_norm": 0.43363118171691895, "learning_rate": 9.900000000000001e-05, "loss": 0.1072, "step": 2970 }, { "epoch": 1.3336316849407026, "grad_norm": 0.47528865933418274, "learning_rate": 9.933333333333334e-05, "loss": 0.1071, "step": 2980 }, { "epoch": 1.3381069590512418, "grad_norm": 0.48455917835235596, "learning_rate": 9.966666666666667e-05, "loss": 0.1058, "step": 2990 }, { "epoch": 1.3425822331617812, "grad_norm": 0.5155789852142334, "learning_rate": 0.0001, "loss": 0.1074, "step": 3000 }, { "epoch": 1.3470575072723205, "grad_norm": 0.4875307083129883, "learning_rate": 9.99999924056601e-05, "loss": 0.1079, "step": 3010 }, { "epoch": 1.3515327813828597, "grad_norm": 0.4895997941493988, "learning_rate": 9.999996962264266e-05, "loss": 0.1105, "step": 3020 }, { "epoch": 1.3560080554933989, "grad_norm": 0.5641862750053406, "learning_rate": 9.999993165095463e-05, "loss": 0.1031, "step": 3030 }, { "epoch": 1.3604833296039383, "grad_norm": 0.4951086938381195, "learning_rate": 9.999987849060753e-05, "loss": 0.107, "step": 3040 }, { "epoch": 1.3649586037144776, "grad_norm": 0.49325281381607056, "learning_rate": 9.999981014161752e-05, "loss": 0.1023, "step": 3050 }, { "epoch": 1.3694338778250168, "grad_norm": 0.38637715578079224, "learning_rate": 9.999972660400536e-05, "loss": 0.1066, "step": 3060 }, { "epoch": 1.373909151935556, "grad_norm": 0.48560211062431335, "learning_rate": 9.999962787779642e-05, "loss": 0.1059, "step": 3070 }, { "epoch": 1.3783844260460953, "grad_norm": 0.4483901858329773, "learning_rate": 9.999951396302069e-05, "loss": 0.1018, "step": 3080 }, { "epoch": 1.3828597001566347, "grad_norm": 0.41429996490478516, "learning_rate": 9.999938485971279e-05, "loss": 0.0992, "step": 3090 }, { "epoch": 1.387334974267174, "grad_norm": 0.5105229020118713, "learning_rate": 9.999924056791192e-05, "loss": 0.1032, "step": 3100 }, { "epoch": 1.391810248377713, "grad_norm": 0.45511287450790405, "learning_rate": 9.999908108766191e-05, "loss": 0.1089, "step": 3110 }, { "epoch": 1.3962855224882524, "grad_norm": 0.3848251700401306, "learning_rate": 9.999890641901125e-05, "loss": 0.1041, "step": 3120 }, { "epoch": 1.4007607965987916, "grad_norm": 0.4416135549545288, "learning_rate": 9.999871656201292e-05, "loss": 0.1019, "step": 3130 }, { "epoch": 1.405236070709331, "grad_norm": 0.4897851049900055, "learning_rate": 9.999851151672466e-05, "loss": 0.0959, "step": 3140 }, { "epoch": 1.4097113448198701, "grad_norm": 0.41581323742866516, "learning_rate": 9.999829128320874e-05, "loss": 0.1029, "step": 3150 }, { "epoch": 1.4141866189304095, "grad_norm": 0.5126276016235352, "learning_rate": 9.999805586153205e-05, "loss": 0.1005, "step": 3160 }, { "epoch": 1.4186618930409487, "grad_norm": 0.42555391788482666, "learning_rate": 9.99978052517661e-05, "loss": 0.1011, "step": 3170 }, { "epoch": 1.423137167151488, "grad_norm": 0.39005616307258606, "learning_rate": 9.999753945398704e-05, "loss": 0.1086, "step": 3180 }, { "epoch": 1.4276124412620272, "grad_norm": 0.4278729557991028, "learning_rate": 9.999725846827562e-05, "loss": 0.1025, "step": 3190 }, { "epoch": 1.4320877153725666, "grad_norm": 0.4404760003089905, "learning_rate": 9.999696229471716e-05, "loss": 0.1022, "step": 3200 }, { "epoch": 1.4365629894831058, "grad_norm": 0.4938987195491791, "learning_rate": 9.999665093340165e-05, "loss": 0.1104, "step": 3210 }, { "epoch": 1.4410382635936452, "grad_norm": 0.45319247245788574, "learning_rate": 9.999632438442367e-05, "loss": 0.1011, "step": 3220 }, { "epoch": 1.4455135377041843, "grad_norm": 0.4730934798717499, "learning_rate": 9.999598264788241e-05, "loss": 0.1042, "step": 3230 }, { "epoch": 1.4499888118147237, "grad_norm": 0.501909613609314, "learning_rate": 9.99956257238817e-05, "loss": 0.0998, "step": 3240 }, { "epoch": 1.4544640859252629, "grad_norm": 0.47599083185195923, "learning_rate": 9.999525361252996e-05, "loss": 0.1052, "step": 3250 }, { "epoch": 1.4589393600358023, "grad_norm": 0.4442799687385559, "learning_rate": 9.999486631394021e-05, "loss": 0.102, "step": 3260 }, { "epoch": 1.4634146341463414, "grad_norm": 0.4084952473640442, "learning_rate": 9.999446382823013e-05, "loss": 0.0959, "step": 3270 }, { "epoch": 1.4678899082568808, "grad_norm": 0.46767422556877136, "learning_rate": 9.999404615552194e-05, "loss": 0.1008, "step": 3280 }, { "epoch": 1.47236518236742, "grad_norm": 0.41369113326072693, "learning_rate": 9.999361329594254e-05, "loss": 0.0948, "step": 3290 }, { "epoch": 1.4768404564779591, "grad_norm": 0.48125502467155457, "learning_rate": 9.999316524962345e-05, "loss": 0.1027, "step": 3300 }, { "epoch": 1.4813157305884985, "grad_norm": 0.4092821776866913, "learning_rate": 9.999270201670074e-05, "loss": 0.104, "step": 3310 }, { "epoch": 1.485791004699038, "grad_norm": 0.4498721957206726, "learning_rate": 9.999222359731514e-05, "loss": 0.1085, "step": 3320 }, { "epoch": 1.490266278809577, "grad_norm": 0.4059138298034668, "learning_rate": 9.999172999161198e-05, "loss": 0.103, "step": 3330 }, { "epoch": 1.4947415529201162, "grad_norm": 0.4086931645870209, "learning_rate": 9.999122119974121e-05, "loss": 0.1037, "step": 3340 }, { "epoch": 1.4992168270306556, "grad_norm": 0.3797138035297394, "learning_rate": 9.999069722185737e-05, "loss": 0.0957, "step": 3350 }, { "epoch": 1.503692101141195, "grad_norm": 0.5023220181465149, "learning_rate": 9.999015805811965e-05, "loss": 0.1056, "step": 3360 }, { "epoch": 1.5081673752517342, "grad_norm": 0.4977303743362427, "learning_rate": 9.998960370869181e-05, "loss": 0.1021, "step": 3370 }, { "epoch": 1.5126426493622733, "grad_norm": 0.4326496720314026, "learning_rate": 9.998903417374228e-05, "loss": 0.105, "step": 3380 }, { "epoch": 1.5171179234728127, "grad_norm": 0.3843882977962494, "learning_rate": 9.998844945344405e-05, "loss": 0.0999, "step": 3390 }, { "epoch": 1.521593197583352, "grad_norm": 0.43838825821876526, "learning_rate": 9.998784954797474e-05, "loss": 0.0975, "step": 3400 }, { "epoch": 1.5260684716938913, "grad_norm": 0.4529236853122711, "learning_rate": 9.998723445751658e-05, "loss": 0.102, "step": 3410 }, { "epoch": 1.5305437458044304, "grad_norm": 0.4923490285873413, "learning_rate": 9.998660418225645e-05, "loss": 0.0975, "step": 3420 }, { "epoch": 1.5350190199149698, "grad_norm": 0.5249075889587402, "learning_rate": 9.998595872238577e-05, "loss": 0.102, "step": 3430 }, { "epoch": 1.5394942940255092, "grad_norm": 0.4152672588825226, "learning_rate": 9.998529807810064e-05, "loss": 0.1007, "step": 3440 }, { "epoch": 1.5439695681360484, "grad_norm": 0.44197019934654236, "learning_rate": 9.998462224960175e-05, "loss": 0.0987, "step": 3450 }, { "epoch": 1.5484448422465875, "grad_norm": 0.5058916211128235, "learning_rate": 9.998393123709438e-05, "loss": 0.0994, "step": 3460 }, { "epoch": 1.552920116357127, "grad_norm": 0.41246268153190613, "learning_rate": 9.998322504078843e-05, "loss": 0.0936, "step": 3470 }, { "epoch": 1.5573953904676663, "grad_norm": 0.4820858836174011, "learning_rate": 9.998250366089848e-05, "loss": 0.0997, "step": 3480 }, { "epoch": 1.5618706645782054, "grad_norm": 0.4364597797393799, "learning_rate": 9.99817670976436e-05, "loss": 0.0988, "step": 3490 }, { "epoch": 1.5663459386887446, "grad_norm": 0.4721068739891052, "learning_rate": 9.998101535124758e-05, "loss": 0.0965, "step": 3500 }, { "epoch": 1.570821212799284, "grad_norm": 0.45089778304100037, "learning_rate": 9.998024842193876e-05, "loss": 0.098, "step": 3510 }, { "epoch": 1.5752964869098234, "grad_norm": 0.3806210458278656, "learning_rate": 9.997946630995013e-05, "loss": 0.0975, "step": 3520 }, { "epoch": 1.5797717610203625, "grad_norm": 0.3983749747276306, "learning_rate": 9.997866901551926e-05, "loss": 0.0931, "step": 3530 }, { "epoch": 1.5842470351309017, "grad_norm": 0.3988895118236542, "learning_rate": 9.997785653888835e-05, "loss": 0.0985, "step": 3540 }, { "epoch": 1.588722309241441, "grad_norm": 0.4235253930091858, "learning_rate": 9.997702888030423e-05, "loss": 0.0957, "step": 3550 }, { "epoch": 1.5931975833519805, "grad_norm": 0.42720550298690796, "learning_rate": 9.997618604001829e-05, "loss": 0.0957, "step": 3560 }, { "epoch": 1.5976728574625194, "grad_norm": 0.48270803689956665, "learning_rate": 9.997532801828658e-05, "loss": 0.0949, "step": 3570 }, { "epoch": 1.6021481315730588, "grad_norm": 0.3508908748626709, "learning_rate": 9.997445481536973e-05, "loss": 0.0919, "step": 3580 }, { "epoch": 1.6066234056835982, "grad_norm": 0.42955511808395386, "learning_rate": 9.997356643153303e-05, "loss": 0.1015, "step": 3590 }, { "epoch": 1.6110986797941373, "grad_norm": 0.39853668212890625, "learning_rate": 9.997266286704631e-05, "loss": 0.0948, "step": 3600 }, { "epoch": 1.6155739539046765, "grad_norm": 0.4788673222064972, "learning_rate": 9.997174412218406e-05, "loss": 0.0918, "step": 3610 }, { "epoch": 1.620049228015216, "grad_norm": 0.42267143726348877, "learning_rate": 9.997081019722537e-05, "loss": 0.0992, "step": 3620 }, { "epoch": 1.6245245021257553, "grad_norm": 0.38731813430786133, "learning_rate": 9.996986109245395e-05, "loss": 0.0921, "step": 3630 }, { "epoch": 1.6289997762362944, "grad_norm": 0.420601487159729, "learning_rate": 9.99688968081581e-05, "loss": 0.0992, "step": 3640 }, { "epoch": 1.6334750503468336, "grad_norm": 0.3665914833545685, "learning_rate": 9.996791734463077e-05, "loss": 0.0952, "step": 3650 }, { "epoch": 1.637950324457373, "grad_norm": 0.4015063941478729, "learning_rate": 9.996692270216947e-05, "loss": 0.0986, "step": 3660 }, { "epoch": 1.6424255985679124, "grad_norm": 0.3970155715942383, "learning_rate": 9.996591288107635e-05, "loss": 0.0942, "step": 3670 }, { "epoch": 1.6469008726784515, "grad_norm": 0.4373382329940796, "learning_rate": 9.996488788165816e-05, "loss": 0.0952, "step": 3680 }, { "epoch": 1.6513761467889907, "grad_norm": 0.40770816802978516, "learning_rate": 9.996384770422629e-05, "loss": 0.1004, "step": 3690 }, { "epoch": 1.65585142089953, "grad_norm": 0.34220781922340393, "learning_rate": 9.996279234909671e-05, "loss": 0.098, "step": 3700 }, { "epoch": 1.6603266950100695, "grad_norm": 0.36061933636665344, "learning_rate": 9.996172181659e-05, "loss": 0.0976, "step": 3710 }, { "epoch": 1.6648019691206086, "grad_norm": 0.3926573693752289, "learning_rate": 9.996063610703137e-05, "loss": 0.0943, "step": 3720 }, { "epoch": 1.6692772432311478, "grad_norm": 0.4306560158729553, "learning_rate": 9.995953522075061e-05, "loss": 0.0926, "step": 3730 }, { "epoch": 1.6737525173416872, "grad_norm": 0.3631528913974762, "learning_rate": 9.995841915808218e-05, "loss": 0.0933, "step": 3740 }, { "epoch": 1.6782277914522266, "grad_norm": 0.5099853277206421, "learning_rate": 9.995728791936504e-05, "loss": 0.0911, "step": 3750 }, { "epoch": 1.6827030655627657, "grad_norm": 0.43600550293922424, "learning_rate": 9.995614150494293e-05, "loss": 0.0892, "step": 3760 }, { "epoch": 1.6871783396733049, "grad_norm": 0.4157733917236328, "learning_rate": 9.995497991516401e-05, "loss": 0.0903, "step": 3770 }, { "epoch": 1.6916536137838443, "grad_norm": 0.4562300741672516, "learning_rate": 9.995380315038119e-05, "loss": 0.0988, "step": 3780 }, { "epoch": 1.6961288878943837, "grad_norm": 0.4017011523246765, "learning_rate": 9.995261121095194e-05, "loss": 0.0905, "step": 3790 }, { "epoch": 1.7006041620049228, "grad_norm": 0.40697386860847473, "learning_rate": 9.99514040972383e-05, "loss": 0.0915, "step": 3800 }, { "epoch": 1.705079436115462, "grad_norm": 0.43616846203804016, "learning_rate": 9.9950181809607e-05, "loss": 0.0885, "step": 3810 }, { "epoch": 1.7095547102260014, "grad_norm": 0.4465291202068329, "learning_rate": 9.994894434842932e-05, "loss": 0.0975, "step": 3820 }, { "epoch": 1.7140299843365407, "grad_norm": 0.3840556740760803, "learning_rate": 9.994769171408118e-05, "loss": 0.0998, "step": 3830 }, { "epoch": 1.71850525844708, "grad_norm": 0.38856860995292664, "learning_rate": 9.994642390694308e-05, "loss": 0.0928, "step": 3840 }, { "epoch": 1.722980532557619, "grad_norm": 0.3750779926776886, "learning_rate": 9.994514092740015e-05, "loss": 0.0941, "step": 3850 }, { "epoch": 1.7274558066681585, "grad_norm": 0.34980079531669617, "learning_rate": 9.994384277584214e-05, "loss": 0.0923, "step": 3860 }, { "epoch": 1.7319310807786978, "grad_norm": 0.43672624230384827, "learning_rate": 9.99425294526634e-05, "loss": 0.0912, "step": 3870 }, { "epoch": 1.736406354889237, "grad_norm": 0.438382625579834, "learning_rate": 9.994120095826285e-05, "loss": 0.0906, "step": 3880 }, { "epoch": 1.7408816289997762, "grad_norm": 0.3888714015483856, "learning_rate": 9.993985729304408e-05, "loss": 0.0933, "step": 3890 }, { "epoch": 1.7453569031103155, "grad_norm": 0.33190596103668213, "learning_rate": 9.993849845741524e-05, "loss": 0.092, "step": 3900 }, { "epoch": 1.749832177220855, "grad_norm": 0.36195722222328186, "learning_rate": 9.993712445178913e-05, "loss": 0.0876, "step": 3910 }, { "epoch": 1.754307451331394, "grad_norm": 0.347540944814682, "learning_rate": 9.99357352765831e-05, "loss": 0.0915, "step": 3920 }, { "epoch": 1.7587827254419333, "grad_norm": 0.3235618472099304, "learning_rate": 9.99343309322192e-05, "loss": 0.0895, "step": 3930 }, { "epoch": 1.7632579995524726, "grad_norm": 0.3163175582885742, "learning_rate": 9.9932911419124e-05, "loss": 0.0943, "step": 3940 }, { "epoch": 1.7677332736630118, "grad_norm": 0.3927968144416809, "learning_rate": 9.99314767377287e-05, "loss": 0.0956, "step": 3950 }, { "epoch": 1.772208547773551, "grad_norm": 0.37205854058265686, "learning_rate": 9.993002688846913e-05, "loss": 0.094, "step": 3960 }, { "epoch": 1.7766838218840904, "grad_norm": 0.36051449179649353, "learning_rate": 9.992856187178572e-05, "loss": 0.0896, "step": 3970 }, { "epoch": 1.7811590959946297, "grad_norm": 0.4134378135204315, "learning_rate": 9.99270816881235e-05, "loss": 0.094, "step": 3980 }, { "epoch": 1.785634370105169, "grad_norm": 0.33943161368370056, "learning_rate": 9.992558633793212e-05, "loss": 0.0981, "step": 3990 }, { "epoch": 1.790109644215708, "grad_norm": 0.39966100454330444, "learning_rate": 9.992407582166581e-05, "loss": 0.0912, "step": 4000 }, { "epoch": 1.7945849183262474, "grad_norm": 0.3985198736190796, "learning_rate": 9.992255013978344e-05, "loss": 0.0841, "step": 4010 }, { "epoch": 1.7990601924367868, "grad_norm": 0.3571254014968872, "learning_rate": 9.992100929274846e-05, "loss": 0.0919, "step": 4020 }, { "epoch": 1.803535466547326, "grad_norm": 0.4271446168422699, "learning_rate": 9.991945328102897e-05, "loss": 0.0909, "step": 4030 }, { "epoch": 1.8080107406578652, "grad_norm": 0.39205896854400635, "learning_rate": 9.991788210509758e-05, "loss": 0.0918, "step": 4040 }, { "epoch": 1.8124860147684045, "grad_norm": 0.40427008271217346, "learning_rate": 9.991629576543163e-05, "loss": 0.0917, "step": 4050 }, { "epoch": 1.816961288878944, "grad_norm": 0.3838525712490082, "learning_rate": 9.9914694262513e-05, "loss": 0.087, "step": 4060 }, { "epoch": 1.821436562989483, "grad_norm": 0.3581480383872986, "learning_rate": 9.991307759682815e-05, "loss": 0.0919, "step": 4070 }, { "epoch": 1.8259118371000223, "grad_norm": 0.3628753125667572, "learning_rate": 9.991144576886823e-05, "loss": 0.0902, "step": 4080 }, { "epoch": 1.8303871112105616, "grad_norm": 0.3332909941673279, "learning_rate": 9.990979877912891e-05, "loss": 0.0866, "step": 4090 }, { "epoch": 1.834862385321101, "grad_norm": 0.34634312987327576, "learning_rate": 9.990813662811051e-05, "loss": 0.0918, "step": 4100 }, { "epoch": 1.8393376594316402, "grad_norm": 0.36149850487709045, "learning_rate": 9.990645931631796e-05, "loss": 0.0834, "step": 4110 }, { "epoch": 1.8438129335421793, "grad_norm": 0.3442123532295227, "learning_rate": 9.990476684426075e-05, "loss": 0.0862, "step": 4120 }, { "epoch": 1.8482882076527187, "grad_norm": 0.43136066198349, "learning_rate": 9.990305921245306e-05, "loss": 0.0907, "step": 4130 }, { "epoch": 1.8527634817632581, "grad_norm": 0.43009284138679504, "learning_rate": 9.990133642141359e-05, "loss": 0.0913, "step": 4140 }, { "epoch": 1.8572387558737973, "grad_norm": 0.44326838850975037, "learning_rate": 9.989959847166567e-05, "loss": 0.0942, "step": 4150 }, { "epoch": 1.8617140299843364, "grad_norm": 0.484286904335022, "learning_rate": 9.989784536373726e-05, "loss": 0.091, "step": 4160 }, { "epoch": 1.8661893040948758, "grad_norm": 0.3957354426383972, "learning_rate": 9.989607709816091e-05, "loss": 0.0911, "step": 4170 }, { "epoch": 1.8706645782054152, "grad_norm": 0.40801769495010376, "learning_rate": 9.989429367547377e-05, "loss": 0.0858, "step": 4180 }, { "epoch": 1.8751398523159544, "grad_norm": 0.391603022813797, "learning_rate": 9.989249509621759e-05, "loss": 0.0838, "step": 4190 }, { "epoch": 1.8796151264264935, "grad_norm": 0.33408427238464355, "learning_rate": 9.989068136093873e-05, "loss": 0.0861, "step": 4200 }, { "epoch": 1.884090400537033, "grad_norm": 0.3747757375240326, "learning_rate": 9.988885247018817e-05, "loss": 0.0918, "step": 4210 }, { "epoch": 1.8885656746475723, "grad_norm": 0.3759569525718689, "learning_rate": 9.988700842452146e-05, "loss": 0.0874, "step": 4220 }, { "epoch": 1.8930409487581115, "grad_norm": 0.36101609468460083, "learning_rate": 9.988514922449879e-05, "loss": 0.0888, "step": 4230 }, { "epoch": 1.8975162228686506, "grad_norm": 0.37712615728378296, "learning_rate": 9.988327487068492e-05, "loss": 0.0866, "step": 4240 }, { "epoch": 1.90199149697919, "grad_norm": 0.32239624857902527, "learning_rate": 9.988138536364922e-05, "loss": 0.0909, "step": 4250 }, { "epoch": 1.9064667710897294, "grad_norm": 0.3223976194858551, "learning_rate": 9.987948070396571e-05, "loss": 0.0877, "step": 4260 }, { "epoch": 1.9109420452002686, "grad_norm": 0.35482683777809143, "learning_rate": 9.987756089221296e-05, "loss": 0.0917, "step": 4270 }, { "epoch": 1.9154173193108077, "grad_norm": 0.37392860651016235, "learning_rate": 9.987562592897413e-05, "loss": 0.0886, "step": 4280 }, { "epoch": 1.919892593421347, "grad_norm": 0.32759276032447815, "learning_rate": 9.987367581483705e-05, "loss": 0.0933, "step": 4290 }, { "epoch": 1.9243678675318865, "grad_norm": 0.32494863867759705, "learning_rate": 9.987171055039408e-05, "loss": 0.0857, "step": 4300 }, { "epoch": 1.9288431416424257, "grad_norm": 0.36674055457115173, "learning_rate": 9.986973013624224e-05, "loss": 0.0874, "step": 4310 }, { "epoch": 1.9333184157529648, "grad_norm": 0.34136077761650085, "learning_rate": 9.986773457298311e-05, "loss": 0.0898, "step": 4320 }, { "epoch": 1.9377936898635042, "grad_norm": 0.3403914272785187, "learning_rate": 9.986572386122291e-05, "loss": 0.0888, "step": 4330 }, { "epoch": 1.9422689639740434, "grad_norm": 0.2663991153240204, "learning_rate": 9.986369800157242e-05, "loss": 0.0942, "step": 4340 }, { "epoch": 1.9467442380845825, "grad_norm": 0.36054807901382446, "learning_rate": 9.986165699464705e-05, "loss": 0.0793, "step": 4350 }, { "epoch": 1.951219512195122, "grad_norm": 0.3225817084312439, "learning_rate": 9.985960084106682e-05, "loss": 0.0882, "step": 4360 }, { "epoch": 1.9556947863056613, "grad_norm": 0.3571658730506897, "learning_rate": 9.98575295414563e-05, "loss": 0.0844, "step": 4370 }, { "epoch": 1.9601700604162005, "grad_norm": 0.4586593806743622, "learning_rate": 9.985544309644475e-05, "loss": 0.0832, "step": 4380 }, { "epoch": 1.9646453345267396, "grad_norm": 0.3480185866355896, "learning_rate": 9.985334150666592e-05, "loss": 0.0916, "step": 4390 }, { "epoch": 1.969120608637279, "grad_norm": 0.3831617534160614, "learning_rate": 9.985122477275824e-05, "loss": 0.0836, "step": 4400 }, { "epoch": 1.9735958827478184, "grad_norm": 0.31325650215148926, "learning_rate": 9.984909289536473e-05, "loss": 0.085, "step": 4410 }, { "epoch": 1.9780711568583575, "grad_norm": 0.3383249342441559, "learning_rate": 9.984694587513298e-05, "loss": 0.0888, "step": 4420 }, { "epoch": 1.9825464309688967, "grad_norm": 0.360294908285141, "learning_rate": 9.984478371271521e-05, "loss": 0.0903, "step": 4430 }, { "epoch": 1.987021705079436, "grad_norm": 0.35939499735832214, "learning_rate": 9.984260640876821e-05, "loss": 0.0876, "step": 4440 }, { "epoch": 1.9914969791899755, "grad_norm": 0.3330692648887634, "learning_rate": 9.984041396395343e-05, "loss": 0.0865, "step": 4450 }, { "epoch": 1.9959722533005146, "grad_norm": 0.42192819714546204, "learning_rate": 9.983820637893681e-05, "loss": 0.0918, "step": 4460 }, { "epoch": 2.000447527411054, "grad_norm": 0.4481101930141449, "learning_rate": 9.983598365438902e-05, "loss": 0.0926, "step": 4470 }, { "epoch": 2.004922801521593, "grad_norm": 0.3795298933982849, "learning_rate": 9.983374579098523e-05, "loss": 0.0892, "step": 4480 }, { "epoch": 2.0093980756321326, "grad_norm": 0.4415881335735321, "learning_rate": 9.983149278940526e-05, "loss": 0.0907, "step": 4490 }, { "epoch": 2.013873349742672, "grad_norm": 0.3356173038482666, "learning_rate": 9.98292246503335e-05, "loss": 0.0901, "step": 4500 }, { "epoch": 2.018348623853211, "grad_norm": 0.3482261300086975, "learning_rate": 9.982694137445896e-05, "loss": 0.0906, "step": 4510 }, { "epoch": 2.0228238979637503, "grad_norm": 0.33225560188293457, "learning_rate": 9.982464296247522e-05, "loss": 0.0888, "step": 4520 }, { "epoch": 2.0272991720742897, "grad_norm": 0.338069885969162, "learning_rate": 9.98223294150805e-05, "loss": 0.0864, "step": 4530 }, { "epoch": 2.0317744461848286, "grad_norm": 0.35138967633247375, "learning_rate": 9.982000073297759e-05, "loss": 0.0842, "step": 4540 }, { "epoch": 2.036249720295368, "grad_norm": 0.3127942979335785, "learning_rate": 9.981765691687388e-05, "loss": 0.0836, "step": 4550 }, { "epoch": 2.0407249944059074, "grad_norm": 0.3468284010887146, "learning_rate": 9.981529796748134e-05, "loss": 0.0841, "step": 4560 }, { "epoch": 2.0452002685164468, "grad_norm": 0.4147009551525116, "learning_rate": 9.981292388551659e-05, "loss": 0.0809, "step": 4570 }, { "epoch": 2.0496755426269857, "grad_norm": 0.3485106825828552, "learning_rate": 9.98105346717008e-05, "loss": 0.0858, "step": 4580 }, { "epoch": 2.054150816737525, "grad_norm": 0.3977799415588379, "learning_rate": 9.980813032675974e-05, "loss": 0.0825, "step": 4590 }, { "epoch": 2.0586260908480645, "grad_norm": 0.376919686794281, "learning_rate": 9.980571085142381e-05, "loss": 0.0893, "step": 4600 }, { "epoch": 2.063101364958604, "grad_norm": 0.33488771319389343, "learning_rate": 9.980327624642795e-05, "loss": 0.0849, "step": 4610 }, { "epoch": 2.067576639069143, "grad_norm": 0.32006049156188965, "learning_rate": 9.980082651251175e-05, "loss": 0.0826, "step": 4620 }, { "epoch": 2.072051913179682, "grad_norm": 0.3895520269870758, "learning_rate": 9.979836165041936e-05, "loss": 0.0835, "step": 4630 }, { "epoch": 2.0765271872902216, "grad_norm": 0.35103559494018555, "learning_rate": 9.979588166089958e-05, "loss": 0.0891, "step": 4640 }, { "epoch": 2.081002461400761, "grad_norm": 0.3499997854232788, "learning_rate": 9.979338654470569e-05, "loss": 0.0866, "step": 4650 }, { "epoch": 2.0854777355113, "grad_norm": 0.32935091853141785, "learning_rate": 9.979087630259572e-05, "loss": 0.0803, "step": 4660 }, { "epoch": 2.0899530096218393, "grad_norm": 0.2995895445346832, "learning_rate": 9.978835093533216e-05, "loss": 0.0878, "step": 4670 }, { "epoch": 2.0944282837323787, "grad_norm": 0.3577474355697632, "learning_rate": 9.97858104436822e-05, "loss": 0.0871, "step": 4680 }, { "epoch": 2.098903557842918, "grad_norm": 0.3258088529109955, "learning_rate": 9.978325482841753e-05, "loss": 0.0797, "step": 4690 }, { "epoch": 2.103378831953457, "grad_norm": 0.33249810338020325, "learning_rate": 9.978068409031449e-05, "loss": 0.0866, "step": 4700 }, { "epoch": 2.1078541060639964, "grad_norm": 0.3204028904438019, "learning_rate": 9.977809823015401e-05, "loss": 0.0825, "step": 4710 }, { "epoch": 2.1123293801745358, "grad_norm": 0.343642920255661, "learning_rate": 9.97754972487216e-05, "loss": 0.0882, "step": 4720 }, { "epoch": 2.116804654285075, "grad_norm": 0.36746057868003845, "learning_rate": 9.977288114680737e-05, "loss": 0.0836, "step": 4730 }, { "epoch": 2.121279928395614, "grad_norm": 0.35592013597488403, "learning_rate": 9.977024992520602e-05, "loss": 0.0871, "step": 4740 }, { "epoch": 2.1257552025061535, "grad_norm": 0.35528361797332764, "learning_rate": 9.976760358471686e-05, "loss": 0.0782, "step": 4750 }, { "epoch": 2.130230476616693, "grad_norm": 0.33605802059173584, "learning_rate": 9.976494212614377e-05, "loss": 0.0862, "step": 4760 }, { "epoch": 2.1347057507272322, "grad_norm": 0.3933661878108978, "learning_rate": 9.976226555029522e-05, "loss": 0.0791, "step": 4770 }, { "epoch": 2.139181024837771, "grad_norm": 0.40230149030685425, "learning_rate": 9.97595738579843e-05, "loss": 0.082, "step": 4780 }, { "epoch": 2.1436562989483106, "grad_norm": 0.31968796253204346, "learning_rate": 9.975686705002867e-05, "loss": 0.0873, "step": 4790 }, { "epoch": 2.14813157305885, "grad_norm": 0.3732394874095917, "learning_rate": 9.975414512725057e-05, "loss": 0.0851, "step": 4800 }, { "epoch": 2.1526068471693893, "grad_norm": 0.3246452212333679, "learning_rate": 9.975140809047687e-05, "loss": 0.0843, "step": 4810 }, { "epoch": 2.1570821212799283, "grad_norm": 0.3508913815021515, "learning_rate": 9.974865594053902e-05, "loss": 0.0856, "step": 4820 }, { "epoch": 2.1615573953904677, "grad_norm": 0.38308584690093994, "learning_rate": 9.974588867827301e-05, "loss": 0.0841, "step": 4830 }, { "epoch": 2.166032669501007, "grad_norm": 0.36763691902160645, "learning_rate": 9.974310630451948e-05, "loss": 0.0807, "step": 4840 }, { "epoch": 2.170507943611546, "grad_norm": 0.3575384020805359, "learning_rate": 9.974030882012367e-05, "loss": 0.0853, "step": 4850 }, { "epoch": 2.1749832177220854, "grad_norm": 0.3295077383518219, "learning_rate": 9.973749622593534e-05, "loss": 0.084, "step": 4860 }, { "epoch": 2.1794584918326247, "grad_norm": 0.36555299162864685, "learning_rate": 9.973466852280889e-05, "loss": 0.0846, "step": 4870 }, { "epoch": 2.183933765943164, "grad_norm": 0.3933255970478058, "learning_rate": 9.973182571160332e-05, "loss": 0.0807, "step": 4880 }, { "epoch": 2.1884090400537035, "grad_norm": 0.33461180329322815, "learning_rate": 9.972896779318219e-05, "loss": 0.0827, "step": 4890 }, { "epoch": 2.1928843141642425, "grad_norm": 0.2955152988433838, "learning_rate": 9.972609476841367e-05, "loss": 0.0817, "step": 4900 }, { "epoch": 2.197359588274782, "grad_norm": 0.4030620753765106, "learning_rate": 9.97232066381705e-05, "loss": 0.081, "step": 4910 }, { "epoch": 2.2018348623853212, "grad_norm": 0.3135741055011749, "learning_rate": 9.972030340333001e-05, "loss": 0.0849, "step": 4920 }, { "epoch": 2.20631013649586, "grad_norm": 0.32419225573539734, "learning_rate": 9.971738506477414e-05, "loss": 0.084, "step": 4930 }, { "epoch": 2.2107854106063995, "grad_norm": 0.3587246537208557, "learning_rate": 9.971445162338939e-05, "loss": 0.0837, "step": 4940 }, { "epoch": 2.215260684716939, "grad_norm": 0.29348820447921753, "learning_rate": 9.97115030800669e-05, "loss": 0.0833, "step": 4950 }, { "epoch": 2.2197359588274783, "grad_norm": 0.32515788078308105, "learning_rate": 9.97085394357023e-05, "loss": 0.0867, "step": 4960 }, { "epoch": 2.2242112329380173, "grad_norm": 0.3232537806034088, "learning_rate": 9.97055606911959e-05, "loss": 0.0874, "step": 4970 }, { "epoch": 2.2286865070485566, "grad_norm": 0.34485089778900146, "learning_rate": 9.970256684745258e-05, "loss": 0.0807, "step": 4980 }, { "epoch": 2.233161781159096, "grad_norm": 0.3318161368370056, "learning_rate": 9.969955790538175e-05, "loss": 0.0794, "step": 4990 }, { "epoch": 2.2376370552696354, "grad_norm": 0.3311028480529785, "learning_rate": 9.969653386589748e-05, "loss": 0.0811, "step": 5000 }, { "epoch": 2.2421123293801744, "grad_norm": 0.38457465171813965, "learning_rate": 9.969349472991838e-05, "loss": 0.0783, "step": 5010 }, { "epoch": 2.2465876034907137, "grad_norm": 0.2888469398021698, "learning_rate": 9.969044049836767e-05, "loss": 0.0826, "step": 5020 }, { "epoch": 2.251062877601253, "grad_norm": 0.32756006717681885, "learning_rate": 9.968737117217313e-05, "loss": 0.0787, "step": 5030 }, { "epoch": 2.2555381517117925, "grad_norm": 0.3572210967540741, "learning_rate": 9.968428675226714e-05, "loss": 0.0846, "step": 5040 }, { "epoch": 2.2600134258223314, "grad_norm": 0.2912848889827728, "learning_rate": 9.968118723958668e-05, "loss": 0.0833, "step": 5050 }, { "epoch": 2.264488699932871, "grad_norm": 0.34924203157424927, "learning_rate": 9.967807263507329e-05, "loss": 0.0839, "step": 5060 }, { "epoch": 2.26896397404341, "grad_norm": 0.3884543180465698, "learning_rate": 9.967494293967312e-05, "loss": 0.0786, "step": 5070 }, { "epoch": 2.2734392481539496, "grad_norm": 0.31203341484069824, "learning_rate": 9.967179815433685e-05, "loss": 0.0763, "step": 5080 }, { "epoch": 2.2779145222644885, "grad_norm": 0.3572942316532135, "learning_rate": 9.966863828001982e-05, "loss": 0.0846, "step": 5090 }, { "epoch": 2.282389796375028, "grad_norm": 0.33447879552841187, "learning_rate": 9.966546331768191e-05, "loss": 0.0766, "step": 5100 }, { "epoch": 2.2868650704855673, "grad_norm": 0.3297809064388275, "learning_rate": 9.966227326828759e-05, "loss": 0.0821, "step": 5110 }, { "epoch": 2.2913403445961067, "grad_norm": 0.31819531321525574, "learning_rate": 9.96590681328059e-05, "loss": 0.0764, "step": 5120 }, { "epoch": 2.2958156187066456, "grad_norm": 0.28865888714790344, "learning_rate": 9.965584791221048e-05, "loss": 0.0756, "step": 5130 }, { "epoch": 2.300290892817185, "grad_norm": 0.3087629973888397, "learning_rate": 9.965261260747956e-05, "loss": 0.0828, "step": 5140 }, { "epoch": 2.3047661669277244, "grad_norm": 0.36358577013015747, "learning_rate": 9.96493622195959e-05, "loss": 0.0815, "step": 5150 }, { "epoch": 2.309241441038264, "grad_norm": 0.28668737411499023, "learning_rate": 9.964609674954696e-05, "loss": 0.074, "step": 5160 }, { "epoch": 2.3137167151488027, "grad_norm": 0.32336127758026123, "learning_rate": 9.964281619832463e-05, "loss": 0.0785, "step": 5170 }, { "epoch": 2.318191989259342, "grad_norm": 0.35738757252693176, "learning_rate": 9.963952056692549e-05, "loss": 0.0782, "step": 5180 }, { "epoch": 2.3226672633698815, "grad_norm": 0.31598925590515137, "learning_rate": 9.963620985635065e-05, "loss": 0.077, "step": 5190 }, { "epoch": 2.3271425374804204, "grad_norm": 0.32844123244285583, "learning_rate": 9.963288406760582e-05, "loss": 0.0814, "step": 5200 }, { "epoch": 2.33161781159096, "grad_norm": 0.3015550971031189, "learning_rate": 9.96295432017013e-05, "loss": 0.0762, "step": 5210 }, { "epoch": 2.336093085701499, "grad_norm": 0.3028024137020111, "learning_rate": 9.962618725965196e-05, "loss": 0.0738, "step": 5220 }, { "epoch": 2.3405683598120386, "grad_norm": 0.30463382601737976, "learning_rate": 9.962281624247722e-05, "loss": 0.0779, "step": 5230 }, { "epoch": 2.345043633922578, "grad_norm": 0.30750560760498047, "learning_rate": 9.961943015120113e-05, "loss": 0.0768, "step": 5240 }, { "epoch": 2.349518908033117, "grad_norm": 0.3371618390083313, "learning_rate": 9.961602898685226e-05, "loss": 0.0835, "step": 5250 }, { "epoch": 2.3539941821436563, "grad_norm": 0.3067393898963928, "learning_rate": 9.961261275046383e-05, "loss": 0.0705, "step": 5260 }, { "epoch": 2.3584694562541957, "grad_norm": 0.28947630524635315, "learning_rate": 9.96091814430736e-05, "loss": 0.0798, "step": 5270 }, { "epoch": 2.3629447303647346, "grad_norm": 0.30749639868736267, "learning_rate": 9.96057350657239e-05, "loss": 0.0806, "step": 5280 }, { "epoch": 2.367420004475274, "grad_norm": 0.36239323019981384, "learning_rate": 9.960227361946164e-05, "loss": 0.0797, "step": 5290 }, { "epoch": 2.3718952785858134, "grad_norm": 0.3314504027366638, "learning_rate": 9.959879710533835e-05, "loss": 0.0804, "step": 5300 }, { "epoch": 2.376370552696353, "grad_norm": 0.2810899615287781, "learning_rate": 9.959530552441005e-05, "loss": 0.0797, "step": 5310 }, { "epoch": 2.380845826806892, "grad_norm": 0.28159859776496887, "learning_rate": 9.959179887773744e-05, "loss": 0.0845, "step": 5320 }, { "epoch": 2.385321100917431, "grad_norm": 0.27524006366729736, "learning_rate": 9.958827716638572e-05, "loss": 0.0773, "step": 5330 }, { "epoch": 2.3897963750279705, "grad_norm": 0.373171865940094, "learning_rate": 9.95847403914247e-05, "loss": 0.0811, "step": 5340 }, { "epoch": 2.39427164913851, "grad_norm": 0.3397611379623413, "learning_rate": 9.958118855392876e-05, "loss": 0.0809, "step": 5350 }, { "epoch": 2.398746923249049, "grad_norm": 0.27546945214271545, "learning_rate": 9.957762165497686e-05, "loss": 0.0824, "step": 5360 }, { "epoch": 2.403222197359588, "grad_norm": 0.3060997724533081, "learning_rate": 9.95740396956525e-05, "loss": 0.0802, "step": 5370 }, { "epoch": 2.4076974714701276, "grad_norm": 0.32319891452789307, "learning_rate": 9.957044267704384e-05, "loss": 0.0788, "step": 5380 }, { "epoch": 2.412172745580667, "grad_norm": 0.3265884220600128, "learning_rate": 9.95668306002435e-05, "loss": 0.0732, "step": 5390 }, { "epoch": 2.416648019691206, "grad_norm": 0.3401384949684143, "learning_rate": 9.956320346634876e-05, "loss": 0.0789, "step": 5400 }, { "epoch": 2.4211232938017453, "grad_norm": 0.3208257853984833, "learning_rate": 9.955956127646147e-05, "loss": 0.0789, "step": 5410 }, { "epoch": 2.4255985679122847, "grad_norm": 0.3704995810985565, "learning_rate": 9.955590403168799e-05, "loss": 0.079, "step": 5420 }, { "epoch": 2.430073842022824, "grad_norm": 0.31078988313674927, "learning_rate": 9.955223173313931e-05, "loss": 0.0839, "step": 5430 }, { "epoch": 2.434549116133363, "grad_norm": 0.31910568475723267, "learning_rate": 9.9548544381931e-05, "loss": 0.0757, "step": 5440 }, { "epoch": 2.4390243902439024, "grad_norm": 0.30220308899879456, "learning_rate": 9.954484197918315e-05, "loss": 0.079, "step": 5450 }, { "epoch": 2.4434996643544418, "grad_norm": 0.271886944770813, "learning_rate": 9.954112452602045e-05, "loss": 0.0772, "step": 5460 }, { "epoch": 2.447974938464981, "grad_norm": 0.32382968068122864, "learning_rate": 9.953739202357218e-05, "loss": 0.0818, "step": 5470 }, { "epoch": 2.45245021257552, "grad_norm": 0.3263242244720459, "learning_rate": 9.953364447297219e-05, "loss": 0.0774, "step": 5480 }, { "epoch": 2.4569254866860595, "grad_norm": 0.3204214572906494, "learning_rate": 9.952988187535886e-05, "loss": 0.0813, "step": 5490 }, { "epoch": 2.461400760796599, "grad_norm": 0.31618815660476685, "learning_rate": 9.952610423187516e-05, "loss": 0.0804, "step": 5500 }, { "epoch": 2.4658760349071382, "grad_norm": 0.29035091400146484, "learning_rate": 9.952231154366868e-05, "loss": 0.0793, "step": 5510 }, { "epoch": 2.470351309017677, "grad_norm": 0.31878265738487244, "learning_rate": 9.95185038118915e-05, "loss": 0.079, "step": 5520 }, { "epoch": 2.4748265831282166, "grad_norm": 0.3045737147331238, "learning_rate": 9.951468103770032e-05, "loss": 0.0767, "step": 5530 }, { "epoch": 2.479301857238756, "grad_norm": 0.289536714553833, "learning_rate": 9.951084322225641e-05, "loss": 0.0782, "step": 5540 }, { "epoch": 2.483777131349295, "grad_norm": 0.28606390953063965, "learning_rate": 9.950699036672559e-05, "loss": 0.0768, "step": 5550 }, { "epoch": 2.4882524054598343, "grad_norm": 0.31454768776893616, "learning_rate": 9.950312247227825e-05, "loss": 0.0794, "step": 5560 }, { "epoch": 2.4927276795703737, "grad_norm": 0.294856458902359, "learning_rate": 9.949923954008935e-05, "loss": 0.0799, "step": 5570 }, { "epoch": 2.497202953680913, "grad_norm": 0.3237718343734741, "learning_rate": 9.949534157133844e-05, "loss": 0.0781, "step": 5580 }, { "epoch": 2.5016782277914524, "grad_norm": 0.30115634202957153, "learning_rate": 9.949142856720961e-05, "loss": 0.0774, "step": 5590 }, { "epoch": 2.5061535019019914, "grad_norm": 0.3007535934448242, "learning_rate": 9.94875005288915e-05, "loss": 0.08, "step": 5600 }, { "epoch": 2.5106287760125308, "grad_norm": 0.2885611355304718, "learning_rate": 9.948355745757741e-05, "loss": 0.0781, "step": 5610 }, { "epoch": 2.51510405012307, "grad_norm": 0.28336694836616516, "learning_rate": 9.947959935446507e-05, "loss": 0.0811, "step": 5620 }, { "epoch": 2.519579324233609, "grad_norm": 0.3414939343929291, "learning_rate": 9.94756262207569e-05, "loss": 0.0758, "step": 5630 }, { "epoch": 2.5240545983441485, "grad_norm": 0.2914351224899292, "learning_rate": 9.94716380576598e-05, "loss": 0.072, "step": 5640 }, { "epoch": 2.528529872454688, "grad_norm": 0.331012099981308, "learning_rate": 9.946763486638528e-05, "loss": 0.074, "step": 5650 }, { "epoch": 2.5330051465652272, "grad_norm": 0.3256583511829376, "learning_rate": 9.946361664814943e-05, "loss": 0.0731, "step": 5660 }, { "epoch": 2.5374804206757666, "grad_norm": 0.3443710505962372, "learning_rate": 9.945958340417283e-05, "loss": 0.0757, "step": 5670 }, { "epoch": 2.5419556947863056, "grad_norm": 0.3120281398296356, "learning_rate": 9.945553513568068e-05, "loss": 0.0813, "step": 5680 }, { "epoch": 2.546430968896845, "grad_norm": 0.2677285373210907, "learning_rate": 9.945147184390278e-05, "loss": 0.0806, "step": 5690 }, { "epoch": 2.5509062430073843, "grad_norm": 0.27898502349853516, "learning_rate": 9.944739353007344e-05, "loss": 0.0774, "step": 5700 }, { "epoch": 2.5553815171179233, "grad_norm": 0.27394160628318787, "learning_rate": 9.944330019543149e-05, "loss": 0.0771, "step": 5710 }, { "epoch": 2.5598567912284627, "grad_norm": 0.26690733432769775, "learning_rate": 9.943919184122043e-05, "loss": 0.075, "step": 5720 }, { "epoch": 2.564332065339002, "grad_norm": 0.30302849411964417, "learning_rate": 9.943506846868826e-05, "loss": 0.0767, "step": 5730 }, { "epoch": 2.5688073394495414, "grad_norm": 0.29628923535346985, "learning_rate": 9.943093007908755e-05, "loss": 0.0771, "step": 5740 }, { "epoch": 2.573282613560081, "grad_norm": 0.27419641613960266, "learning_rate": 9.942677667367541e-05, "loss": 0.0718, "step": 5750 }, { "epoch": 2.5777578876706198, "grad_norm": 0.32251209020614624, "learning_rate": 9.942260825371358e-05, "loss": 0.0813, "step": 5760 }, { "epoch": 2.582233161781159, "grad_norm": 0.2857881486415863, "learning_rate": 9.941842482046828e-05, "loss": 0.081, "step": 5770 }, { "epoch": 2.5867084358916985, "grad_norm": 0.30672022700309753, "learning_rate": 9.941422637521035e-05, "loss": 0.0816, "step": 5780 }, { "epoch": 2.5911837100022375, "grad_norm": 0.31661978363990784, "learning_rate": 9.941001291921512e-05, "loss": 0.0801, "step": 5790 }, { "epoch": 2.595658984112777, "grad_norm": 0.33465299010276794, "learning_rate": 9.940578445376258e-05, "loss": 0.0755, "step": 5800 }, { "epoch": 2.6001342582233162, "grad_norm": 0.3488026261329651, "learning_rate": 9.940154098013723e-05, "loss": 0.0803, "step": 5810 }, { "epoch": 2.6046095323338556, "grad_norm": 0.31015029549598694, "learning_rate": 9.939728249962807e-05, "loss": 0.0846, "step": 5820 }, { "epoch": 2.6090848064443946, "grad_norm": 0.3456787168979645, "learning_rate": 9.939300901352876e-05, "loss": 0.0771, "step": 5830 }, { "epoch": 2.613560080554934, "grad_norm": 0.31527090072631836, "learning_rate": 9.938872052313746e-05, "loss": 0.0783, "step": 5840 }, { "epoch": 2.6180353546654733, "grad_norm": 0.30701133608818054, "learning_rate": 9.938441702975689e-05, "loss": 0.0789, "step": 5850 }, { "epoch": 2.6225106287760127, "grad_norm": 0.30617719888687134, "learning_rate": 9.938009853469436e-05, "loss": 0.0759, "step": 5860 }, { "epoch": 2.6269859028865516, "grad_norm": 0.34478726983070374, "learning_rate": 9.93757650392617e-05, "loss": 0.0772, "step": 5870 }, { "epoch": 2.631461176997091, "grad_norm": 0.3009147644042969, "learning_rate": 9.937141654477528e-05, "loss": 0.0796, "step": 5880 }, { "epoch": 2.6359364511076304, "grad_norm": 0.3495965003967285, "learning_rate": 9.936705305255612e-05, "loss": 0.0718, "step": 5890 }, { "epoch": 2.6404117252181694, "grad_norm": 0.32810577750205994, "learning_rate": 9.936267456392971e-05, "loss": 0.0804, "step": 5900 }, { "epoch": 2.6448869993287087, "grad_norm": 0.32465365529060364, "learning_rate": 9.93582810802261e-05, "loss": 0.0814, "step": 5910 }, { "epoch": 2.649362273439248, "grad_norm": 0.35824212431907654, "learning_rate": 9.935387260277993e-05, "loss": 0.0806, "step": 5920 }, { "epoch": 2.6538375475497875, "grad_norm": 0.32496771216392517, "learning_rate": 9.934944913293038e-05, "loss": 0.0772, "step": 5930 }, { "epoch": 2.658312821660327, "grad_norm": 0.3154754042625427, "learning_rate": 9.934501067202117e-05, "loss": 0.0791, "step": 5940 }, { "epoch": 2.662788095770866, "grad_norm": 0.29464471340179443, "learning_rate": 9.934055722140061e-05, "loss": 0.0752, "step": 5950 }, { "epoch": 2.667263369881405, "grad_norm": 0.29302698373794556, "learning_rate": 9.933608878242153e-05, "loss": 0.076, "step": 5960 }, { "epoch": 2.6717386439919446, "grad_norm": 0.3047963082790375, "learning_rate": 9.93316053564413e-05, "loss": 0.0866, "step": 5970 }, { "epoch": 2.6762139181024835, "grad_norm": 0.3148934543132782, "learning_rate": 9.932710694482191e-05, "loss": 0.0774, "step": 5980 }, { "epoch": 2.680689192213023, "grad_norm": 0.27235594391822815, "learning_rate": 9.932259354892984e-05, "loss": 0.0756, "step": 5990 }, { "epoch": 2.6851644663235623, "grad_norm": 0.2746899127960205, "learning_rate": 9.931806517013612e-05, "loss": 0.0754, "step": 6000 }, { "epoch": 2.6896397404341017, "grad_norm": 0.27094316482543945, "learning_rate": 9.931352180981637e-05, "loss": 0.077, "step": 6010 }, { "epoch": 2.694115014544641, "grad_norm": 0.3208261728286743, "learning_rate": 9.930896346935077e-05, "loss": 0.077, "step": 6020 }, { "epoch": 2.69859028865518, "grad_norm": 0.3757467567920685, "learning_rate": 9.930439015012396e-05, "loss": 0.0793, "step": 6030 }, { "epoch": 2.7030655627657194, "grad_norm": 0.2833695709705353, "learning_rate": 9.929980185352526e-05, "loss": 0.0733, "step": 6040 }, { "epoch": 2.707540836876259, "grad_norm": 0.3514404892921448, "learning_rate": 9.929519858094843e-05, "loss": 0.0767, "step": 6050 }, { "epoch": 2.7120161109867977, "grad_norm": 0.2909387946128845, "learning_rate": 9.929058033379181e-05, "loss": 0.079, "step": 6060 }, { "epoch": 2.716491385097337, "grad_norm": 0.3214477002620697, "learning_rate": 9.928594711345836e-05, "loss": 0.0754, "step": 6070 }, { "epoch": 2.7209666592078765, "grad_norm": 0.2969396412372589, "learning_rate": 9.92812989213555e-05, "loss": 0.0752, "step": 6080 }, { "epoch": 2.725441933318416, "grad_norm": 0.35903823375701904, "learning_rate": 9.927663575889521e-05, "loss": 0.0798, "step": 6090 }, { "epoch": 2.7299172074289553, "grad_norm": 0.3545173406600952, "learning_rate": 9.927195762749405e-05, "loss": 0.0774, "step": 6100 }, { "epoch": 2.734392481539494, "grad_norm": 0.3112255334854126, "learning_rate": 9.926726452857312e-05, "loss": 0.0776, "step": 6110 }, { "epoch": 2.7388677556500336, "grad_norm": 0.29919204115867615, "learning_rate": 9.926255646355804e-05, "loss": 0.0804, "step": 6120 }, { "epoch": 2.743343029760573, "grad_norm": 0.3097529709339142, "learning_rate": 9.925783343387903e-05, "loss": 0.0728, "step": 6130 }, { "epoch": 2.747818303871112, "grad_norm": 0.28891441226005554, "learning_rate": 9.925309544097078e-05, "loss": 0.0755, "step": 6140 }, { "epoch": 2.7522935779816513, "grad_norm": 0.29816174507141113, "learning_rate": 9.92483424862726e-05, "loss": 0.0752, "step": 6150 }, { "epoch": 2.7567688520921907, "grad_norm": 0.2541561722755432, "learning_rate": 9.924357457122828e-05, "loss": 0.0729, "step": 6160 }, { "epoch": 2.76124412620273, "grad_norm": 0.2715972065925598, "learning_rate": 9.923879169728622e-05, "loss": 0.0737, "step": 6170 }, { "epoch": 2.7657194003132695, "grad_norm": 0.26995357871055603, "learning_rate": 9.923399386589933e-05, "loss": 0.0779, "step": 6180 }, { "epoch": 2.7701946744238084, "grad_norm": 0.3160754442214966, "learning_rate": 9.922918107852504e-05, "loss": 0.0696, "step": 6190 }, { "epoch": 2.774669948534348, "grad_norm": 0.31955283880233765, "learning_rate": 9.922435333662536e-05, "loss": 0.0731, "step": 6200 }, { "epoch": 2.779145222644887, "grad_norm": 0.2881999611854553, "learning_rate": 9.921951064166684e-05, "loss": 0.0721, "step": 6210 }, { "epoch": 2.783620496755426, "grad_norm": 0.3545488119125366, "learning_rate": 9.921465299512054e-05, "loss": 0.0752, "step": 6220 }, { "epoch": 2.7880957708659655, "grad_norm": 0.3176058530807495, "learning_rate": 9.92097803984621e-05, "loss": 0.0729, "step": 6230 }, { "epoch": 2.792571044976505, "grad_norm": 0.2988502085208893, "learning_rate": 9.92048928531717e-05, "loss": 0.0772, "step": 6240 }, { "epoch": 2.797046319087044, "grad_norm": 0.2555900514125824, "learning_rate": 9.9199990360734e-05, "loss": 0.0752, "step": 6250 }, { "epoch": 2.801521593197583, "grad_norm": 0.26692187786102295, "learning_rate": 9.91950729226383e-05, "loss": 0.0749, "step": 6260 }, { "epoch": 2.8059968673081226, "grad_norm": 0.3149475157260895, "learning_rate": 9.919014054037836e-05, "loss": 0.0691, "step": 6270 }, { "epoch": 2.810472141418662, "grad_norm": 0.324413925409317, "learning_rate": 9.918519321545251e-05, "loss": 0.0766, "step": 6280 }, { "epoch": 2.8149474155292014, "grad_norm": 0.3231767416000366, "learning_rate": 9.918023094936363e-05, "loss": 0.0735, "step": 6290 }, { "epoch": 2.8194226896397403, "grad_norm": 0.2373916208744049, "learning_rate": 9.917525374361912e-05, "loss": 0.07, "step": 6300 }, { "epoch": 2.8238979637502797, "grad_norm": 0.3145451843738556, "learning_rate": 9.917026159973092e-05, "loss": 0.0694, "step": 6310 }, { "epoch": 2.828373237860819, "grad_norm": 0.2793519198894501, "learning_rate": 9.91652545192155e-05, "loss": 0.0758, "step": 6320 }, { "epoch": 2.832848511971358, "grad_norm": 0.2636408805847168, "learning_rate": 9.91602325035939e-05, "loss": 0.0741, "step": 6330 }, { "epoch": 2.8373237860818974, "grad_norm": 0.28599560260772705, "learning_rate": 9.915519555439166e-05, "loss": 0.0726, "step": 6340 }, { "epoch": 2.8417990601924368, "grad_norm": 0.26618364453315735, "learning_rate": 9.915014367313888e-05, "loss": 0.0739, "step": 6350 }, { "epoch": 2.846274334302976, "grad_norm": 0.34492430090904236, "learning_rate": 9.914507686137019e-05, "loss": 0.0785, "step": 6360 }, { "epoch": 2.8507496084135155, "grad_norm": 0.27743589878082275, "learning_rate": 9.913999512062475e-05, "loss": 0.077, "step": 6370 }, { "epoch": 2.8552248825240545, "grad_norm": 0.2968171536922455, "learning_rate": 9.913489845244626e-05, "loss": 0.0748, "step": 6380 }, { "epoch": 2.859700156634594, "grad_norm": 0.2790394127368927, "learning_rate": 9.912978685838294e-05, "loss": 0.0767, "step": 6390 }, { "epoch": 2.8641754307451333, "grad_norm": 0.2897612750530243, "learning_rate": 9.912466033998757e-05, "loss": 0.0771, "step": 6400 }, { "epoch": 2.868650704855672, "grad_norm": 0.29056957364082336, "learning_rate": 9.911951889881747e-05, "loss": 0.0722, "step": 6410 }, { "epoch": 2.8731259789662116, "grad_norm": 0.268126517534256, "learning_rate": 9.911436253643445e-05, "loss": 0.0666, "step": 6420 }, { "epoch": 2.877601253076751, "grad_norm": 0.2950541079044342, "learning_rate": 9.910919125440485e-05, "loss": 0.0782, "step": 6430 }, { "epoch": 2.8820765271872903, "grad_norm": 0.23625068366527557, "learning_rate": 9.910400505429965e-05, "loss": 0.0738, "step": 6440 }, { "epoch": 2.8865518012978297, "grad_norm": 0.32226458191871643, "learning_rate": 9.90988039376942e-05, "loss": 0.0756, "step": 6450 }, { "epoch": 2.8910270754083687, "grad_norm": 0.28867530822753906, "learning_rate": 9.909358790616849e-05, "loss": 0.0687, "step": 6460 }, { "epoch": 2.895502349518908, "grad_norm": 0.3015729486942291, "learning_rate": 9.908835696130701e-05, "loss": 0.0763, "step": 6470 }, { "epoch": 2.8999776236294474, "grad_norm": 0.2226191759109497, "learning_rate": 9.90831111046988e-05, "loss": 0.0718, "step": 6480 }, { "epoch": 2.9044528977399864, "grad_norm": 0.27334854006767273, "learning_rate": 9.90778503379374e-05, "loss": 0.0705, "step": 6490 }, { "epoch": 2.9089281718505258, "grad_norm": 0.2853389084339142, "learning_rate": 9.90725746626209e-05, "loss": 0.0732, "step": 6500 }, { "epoch": 2.913403445961065, "grad_norm": 0.22213901579380035, "learning_rate": 9.90672840803519e-05, "loss": 0.0731, "step": 6510 }, { "epoch": 2.9178787200716045, "grad_norm": 0.29410696029663086, "learning_rate": 9.906197859273753e-05, "loss": 0.0686, "step": 6520 }, { "epoch": 2.922353994182144, "grad_norm": 0.2521035373210907, "learning_rate": 9.905665820138949e-05, "loss": 0.0751, "step": 6530 }, { "epoch": 2.926829268292683, "grad_norm": 0.3413904905319214, "learning_rate": 9.905132290792394e-05, "loss": 0.0713, "step": 6540 }, { "epoch": 2.9313045424032222, "grad_norm": 0.2752014398574829, "learning_rate": 9.904597271396162e-05, "loss": 0.0709, "step": 6550 }, { "epoch": 2.9357798165137616, "grad_norm": 0.3035432994365692, "learning_rate": 9.904060762112777e-05, "loss": 0.0731, "step": 6560 }, { "epoch": 2.9402550906243006, "grad_norm": 0.259872704744339, "learning_rate": 9.903522763105218e-05, "loss": 0.0722, "step": 6570 }, { "epoch": 2.94473036473484, "grad_norm": 0.259944349527359, "learning_rate": 9.902983274536912e-05, "loss": 0.0744, "step": 6580 }, { "epoch": 2.9492056388453793, "grad_norm": 0.33066585659980774, "learning_rate": 9.902442296571743e-05, "loss": 0.0785, "step": 6590 }, { "epoch": 2.9536809129559183, "grad_norm": 0.26501792669296265, "learning_rate": 9.901899829374047e-05, "loss": 0.0723, "step": 6600 }, { "epoch": 2.9581561870664577, "grad_norm": 0.33029329776763916, "learning_rate": 9.901355873108609e-05, "loss": 0.0729, "step": 6610 }, { "epoch": 2.962631461176997, "grad_norm": 0.29893314838409424, "learning_rate": 9.90081042794067e-05, "loss": 0.0696, "step": 6620 }, { "epoch": 2.9671067352875364, "grad_norm": 0.30573374032974243, "learning_rate": 9.900263494035921e-05, "loss": 0.0742, "step": 6630 }, { "epoch": 2.971582009398076, "grad_norm": 0.28400084376335144, "learning_rate": 9.899715071560508e-05, "loss": 0.0721, "step": 6640 }, { "epoch": 2.9760572835086148, "grad_norm": 0.30813509225845337, "learning_rate": 9.899165160681025e-05, "loss": 0.0748, "step": 6650 }, { "epoch": 2.980532557619154, "grad_norm": 0.30129873752593994, "learning_rate": 9.89861376156452e-05, "loss": 0.075, "step": 6660 }, { "epoch": 2.9850078317296935, "grad_norm": 0.2526872754096985, "learning_rate": 9.898060874378496e-05, "loss": 0.0719, "step": 6670 }, { "epoch": 2.9894831058402325, "grad_norm": 0.2820201516151428, "learning_rate": 9.897506499290902e-05, "loss": 0.0772, "step": 6680 }, { "epoch": 2.993958379950772, "grad_norm": 0.3512851297855377, "learning_rate": 9.896950636470147e-05, "loss": 0.0671, "step": 6690 }, { "epoch": 2.9984336540613112, "grad_norm": 0.2897195816040039, "learning_rate": 9.896393286085084e-05, "loss": 0.0724, "step": 6700 }, { "epoch": 3.0029089281718506, "grad_norm": 0.3385540544986725, "learning_rate": 9.895834448305024e-05, "loss": 0.0778, "step": 6710 }, { "epoch": 3.00738420228239, "grad_norm": 0.2660006284713745, "learning_rate": 9.895274123299723e-05, "loss": 0.0758, "step": 6720 }, { "epoch": 3.011859476392929, "grad_norm": 0.3346986770629883, "learning_rate": 9.894712311239398e-05, "loss": 0.0695, "step": 6730 }, { "epoch": 3.0163347505034683, "grad_norm": 0.3059203624725342, "learning_rate": 9.894149012294708e-05, "loss": 0.0748, "step": 6740 }, { "epoch": 3.0208100246140077, "grad_norm": 0.3684989809989929, "learning_rate": 9.893584226636772e-05, "loss": 0.0738, "step": 6750 }, { "epoch": 3.0252852987245467, "grad_norm": 0.28487175703048706, "learning_rate": 9.893017954437156e-05, "loss": 0.0716, "step": 6760 }, { "epoch": 3.029760572835086, "grad_norm": 0.3045671880245209, "learning_rate": 9.892450195867877e-05, "loss": 0.0701, "step": 6770 }, { "epoch": 3.0342358469456254, "grad_norm": 0.34374454617500305, "learning_rate": 9.891880951101407e-05, "loss": 0.0754, "step": 6780 }, { "epoch": 3.038711121056165, "grad_norm": 0.27149227261543274, "learning_rate": 9.891310220310666e-05, "loss": 0.0753, "step": 6790 }, { "epoch": 3.0431863951667038, "grad_norm": 0.3376479744911194, "learning_rate": 9.890738003669029e-05, "loss": 0.0759, "step": 6800 }, { "epoch": 3.047661669277243, "grad_norm": 0.3163354694843292, "learning_rate": 9.890164301350318e-05, "loss": 0.0698, "step": 6810 }, { "epoch": 3.0521369433877825, "grad_norm": 0.29092973470687866, "learning_rate": 9.889589113528809e-05, "loss": 0.071, "step": 6820 }, { "epoch": 3.056612217498322, "grad_norm": 0.26039573550224304, "learning_rate": 9.88901244037923e-05, "loss": 0.0708, "step": 6830 }, { "epoch": 3.061087491608861, "grad_norm": 0.2804190516471863, "learning_rate": 9.888434282076758e-05, "loss": 0.0763, "step": 6840 }, { "epoch": 3.0655627657194002, "grad_norm": 0.3095707297325134, "learning_rate": 9.887854638797023e-05, "loss": 0.0743, "step": 6850 }, { "epoch": 3.0700380398299396, "grad_norm": 0.28687432408332825, "learning_rate": 9.887273510716107e-05, "loss": 0.0692, "step": 6860 }, { "epoch": 3.074513313940479, "grad_norm": 0.2823196053504944, "learning_rate": 9.886690898010535e-05, "loss": 0.0687, "step": 6870 }, { "epoch": 3.078988588051018, "grad_norm": 0.24166768789291382, "learning_rate": 9.886106800857298e-05, "loss": 0.0783, "step": 6880 }, { "epoch": 3.0834638621615573, "grad_norm": 0.3099563717842102, "learning_rate": 9.885521219433823e-05, "loss": 0.0713, "step": 6890 }, { "epoch": 3.0879391362720967, "grad_norm": 0.2829977869987488, "learning_rate": 9.884934153917997e-05, "loss": 0.0717, "step": 6900 }, { "epoch": 3.092414410382636, "grad_norm": 0.24152520298957825, "learning_rate": 9.884345604488155e-05, "loss": 0.0726, "step": 6910 }, { "epoch": 3.096889684493175, "grad_norm": 0.31917035579681396, "learning_rate": 9.88375557132308e-05, "loss": 0.0726, "step": 6920 }, { "epoch": 3.1013649586037144, "grad_norm": 0.31663912534713745, "learning_rate": 9.883164054602012e-05, "loss": 0.0663, "step": 6930 }, { "epoch": 3.105840232714254, "grad_norm": 0.26323023438453674, "learning_rate": 9.882571054504636e-05, "loss": 0.072, "step": 6940 }, { "epoch": 3.110315506824793, "grad_norm": 0.26969102025032043, "learning_rate": 9.88197657121109e-05, "loss": 0.0705, "step": 6950 }, { "epoch": 3.114790780935332, "grad_norm": 0.31261560320854187, "learning_rate": 9.881380604901964e-05, "loss": 0.0734, "step": 6960 }, { "epoch": 3.1192660550458715, "grad_norm": 0.2545627951622009, "learning_rate": 9.880783155758296e-05, "loss": 0.0636, "step": 6970 }, { "epoch": 3.123741329156411, "grad_norm": 0.2754337191581726, "learning_rate": 9.880184223961573e-05, "loss": 0.0727, "step": 6980 }, { "epoch": 3.1282166032669503, "grad_norm": 0.30316564440727234, "learning_rate": 9.879583809693738e-05, "loss": 0.0709, "step": 6990 }, { "epoch": 3.132691877377489, "grad_norm": 0.27912238240242004, "learning_rate": 9.878981913137179e-05, "loss": 0.0679, "step": 7000 }, { "epoch": 3.1371671514880286, "grad_norm": 0.3141554594039917, "learning_rate": 9.878378534474737e-05, "loss": 0.0723, "step": 7010 }, { "epoch": 3.141642425598568, "grad_norm": 0.26099032163619995, "learning_rate": 9.877773673889701e-05, "loss": 0.065, "step": 7020 }, { "epoch": 3.1461176997091074, "grad_norm": 0.29787710309028625, "learning_rate": 9.877167331565816e-05, "loss": 0.0733, "step": 7030 }, { "epoch": 3.1505929738196463, "grad_norm": 0.2506696879863739, "learning_rate": 9.876559507687267e-05, "loss": 0.064, "step": 7040 }, { "epoch": 3.1550682479301857, "grad_norm": 0.267813116312027, "learning_rate": 9.8759502024387e-05, "loss": 0.0677, "step": 7050 }, { "epoch": 3.159543522040725, "grad_norm": 0.24661880731582642, "learning_rate": 9.875339416005202e-05, "loss": 0.0717, "step": 7060 }, { "epoch": 3.1640187961512645, "grad_norm": 0.3037547469139099, "learning_rate": 9.874727148572315e-05, "loss": 0.0685, "step": 7070 }, { "epoch": 3.1684940702618034, "grad_norm": 0.31977611780166626, "learning_rate": 9.87411340032603e-05, "loss": 0.0665, "step": 7080 }, { "epoch": 3.172969344372343, "grad_norm": 0.26763203740119934, "learning_rate": 9.873498171452789e-05, "loss": 0.0705, "step": 7090 }, { "epoch": 3.177444618482882, "grad_norm": 0.29169347882270813, "learning_rate": 9.872881462139479e-05, "loss": 0.0678, "step": 7100 }, { "epoch": 3.181919892593421, "grad_norm": 0.27255311608314514, "learning_rate": 9.872263272573443e-05, "loss": 0.0731, "step": 7110 }, { "epoch": 3.1863951667039605, "grad_norm": 0.3009626567363739, "learning_rate": 9.871643602942469e-05, "loss": 0.0703, "step": 7120 }, { "epoch": 3.1908704408145, "grad_norm": 0.28894951939582825, "learning_rate": 9.871022453434798e-05, "loss": 0.0724, "step": 7130 }, { "epoch": 3.1953457149250393, "grad_norm": 0.31481418013572693, "learning_rate": 9.870399824239117e-05, "loss": 0.0677, "step": 7140 }, { "epoch": 3.1998209890355787, "grad_norm": 0.28347089886665344, "learning_rate": 9.869775715544562e-05, "loss": 0.0678, "step": 7150 }, { "epoch": 3.2042962631461176, "grad_norm": 0.3254918158054352, "learning_rate": 9.869150127540727e-05, "loss": 0.0719, "step": 7160 }, { "epoch": 3.208771537256657, "grad_norm": 0.26531657576560974, "learning_rate": 9.868523060417646e-05, "loss": 0.07, "step": 7170 }, { "epoch": 3.2132468113671964, "grad_norm": 0.3653613328933716, "learning_rate": 9.867894514365802e-05, "loss": 0.0671, "step": 7180 }, { "epoch": 3.2177220854777353, "grad_norm": 0.2814737558364868, "learning_rate": 9.867264489576135e-05, "loss": 0.0686, "step": 7190 }, { "epoch": 3.2221973595882747, "grad_norm": 0.2912384271621704, "learning_rate": 9.86663298624003e-05, "loss": 0.0734, "step": 7200 }, { "epoch": 3.226672633698814, "grad_norm": 0.28067702054977417, "learning_rate": 9.866000004549318e-05, "loss": 0.0627, "step": 7210 }, { "epoch": 3.2311479078093535, "grad_norm": 0.32230547070503235, "learning_rate": 9.865365544696285e-05, "loss": 0.0705, "step": 7220 }, { "epoch": 3.2356231819198924, "grad_norm": 0.2854677140712738, "learning_rate": 9.864729606873663e-05, "loss": 0.0686, "step": 7230 }, { "epoch": 3.240098456030432, "grad_norm": 0.31066375970840454, "learning_rate": 9.864092191274632e-05, "loss": 0.0645, "step": 7240 }, { "epoch": 3.244573730140971, "grad_norm": 0.236763134598732, "learning_rate": 9.86345329809282e-05, "loss": 0.0679, "step": 7250 }, { "epoch": 3.2490490042515106, "grad_norm": 0.2650910019874573, "learning_rate": 9.862812927522309e-05, "loss": 0.0694, "step": 7260 }, { "epoch": 3.2535242783620495, "grad_norm": 0.28757843375205994, "learning_rate": 9.862171079757628e-05, "loss": 0.0676, "step": 7270 }, { "epoch": 3.257999552472589, "grad_norm": 0.3071646988391876, "learning_rate": 9.861527754993749e-05, "loss": 0.0719, "step": 7280 }, { "epoch": 3.2624748265831283, "grad_norm": 0.289948046207428, "learning_rate": 9.860882953426099e-05, "loss": 0.0632, "step": 7290 }, { "epoch": 3.2669501006936676, "grad_norm": 0.2739786207675934, "learning_rate": 9.860236675250552e-05, "loss": 0.0712, "step": 7300 }, { "epoch": 3.2714253748042066, "grad_norm": 0.3159315288066864, "learning_rate": 9.859588920663432e-05, "loss": 0.0695, "step": 7310 }, { "epoch": 3.275900648914746, "grad_norm": 0.2642971873283386, "learning_rate": 9.858939689861506e-05, "loss": 0.0653, "step": 7320 }, { "epoch": 3.2803759230252854, "grad_norm": 0.2521320581436157, "learning_rate": 9.858288983041996e-05, "loss": 0.0637, "step": 7330 }, { "epoch": 3.2848511971358247, "grad_norm": 0.2843557894229889, "learning_rate": 9.857636800402568e-05, "loss": 0.0689, "step": 7340 }, { "epoch": 3.2893264712463637, "grad_norm": 0.303556889295578, "learning_rate": 9.856983142141339e-05, "loss": 0.0696, "step": 7350 }, { "epoch": 3.293801745356903, "grad_norm": 0.3403227627277374, "learning_rate": 9.856328008456872e-05, "loss": 0.0678, "step": 7360 }, { "epoch": 3.2982770194674425, "grad_norm": 0.23043465614318848, "learning_rate": 9.855671399548181e-05, "loss": 0.0679, "step": 7370 }, { "epoch": 3.302752293577982, "grad_norm": 0.2969086170196533, "learning_rate": 9.855013315614725e-05, "loss": 0.0667, "step": 7380 }, { "epoch": 3.3072275676885208, "grad_norm": 0.3012472689151764, "learning_rate": 9.854353756856412e-05, "loss": 0.0673, "step": 7390 }, { "epoch": 3.31170284179906, "grad_norm": 0.3042696714401245, "learning_rate": 9.8536927234736e-05, "loss": 0.0665, "step": 7400 }, { "epoch": 3.3161781159095995, "grad_norm": 0.31525081396102905, "learning_rate": 9.853030215667093e-05, "loss": 0.0633, "step": 7410 }, { "epoch": 3.320653390020139, "grad_norm": 0.3002654016017914, "learning_rate": 9.852366233638144e-05, "loss": 0.0699, "step": 7420 }, { "epoch": 3.325128664130678, "grad_norm": 0.2816012501716614, "learning_rate": 9.851700777588453e-05, "loss": 0.066, "step": 7430 }, { "epoch": 3.3296039382412173, "grad_norm": 0.21993428468704224, "learning_rate": 9.851033847720166e-05, "loss": 0.07, "step": 7440 }, { "epoch": 3.3340792123517566, "grad_norm": 0.28951191902160645, "learning_rate": 9.85036544423588e-05, "loss": 0.0707, "step": 7450 }, { "epoch": 3.3385544864622956, "grad_norm": 0.2690865695476532, "learning_rate": 9.849695567338639e-05, "loss": 0.069, "step": 7460 }, { "epoch": 3.343029760572835, "grad_norm": 0.2619618773460388, "learning_rate": 9.849024217231935e-05, "loss": 0.0734, "step": 7470 }, { "epoch": 3.3475050346833743, "grad_norm": 0.32857707142829895, "learning_rate": 9.848351394119704e-05, "loss": 0.0639, "step": 7480 }, { "epoch": 3.3519803087939137, "grad_norm": 0.27684956789016724, "learning_rate": 9.847677098206332e-05, "loss": 0.0714, "step": 7490 }, { "epoch": 3.356455582904453, "grad_norm": 0.2774549424648285, "learning_rate": 9.847001329696653e-05, "loss": 0.0662, "step": 7500 }, { "epoch": 3.360930857014992, "grad_norm": 0.2916870713233948, "learning_rate": 9.846324088795948e-05, "loss": 0.0711, "step": 7510 }, { "epoch": 3.3654061311255314, "grad_norm": 0.23969431221485138, "learning_rate": 9.845645375709945e-05, "loss": 0.0654, "step": 7520 }, { "epoch": 3.369881405236071, "grad_norm": 0.2814083993434906, "learning_rate": 9.844965190644817e-05, "loss": 0.0667, "step": 7530 }, { "epoch": 3.3743566793466098, "grad_norm": 0.29183074831962585, "learning_rate": 9.84428353380719e-05, "loss": 0.0768, "step": 7540 }, { "epoch": 3.378831953457149, "grad_norm": 0.24261219799518585, "learning_rate": 9.843600405404131e-05, "loss": 0.0711, "step": 7550 }, { "epoch": 3.3833072275676885, "grad_norm": 0.30989038944244385, "learning_rate": 9.842915805643155e-05, "loss": 0.0681, "step": 7560 }, { "epoch": 3.387782501678228, "grad_norm": 0.3178860545158386, "learning_rate": 9.842229734732229e-05, "loss": 0.0675, "step": 7570 }, { "epoch": 3.3922577757887673, "grad_norm": 0.24820458889007568, "learning_rate": 9.841542192879762e-05, "loss": 0.0642, "step": 7580 }, { "epoch": 3.3967330498993062, "grad_norm": 0.2619808614253998, "learning_rate": 9.840853180294608e-05, "loss": 0.0672, "step": 7590 }, { "epoch": 3.4012083240098456, "grad_norm": 0.2985442280769348, "learning_rate": 9.840162697186075e-05, "loss": 0.0694, "step": 7600 }, { "epoch": 3.405683598120385, "grad_norm": 0.2810680568218231, "learning_rate": 9.83947074376391e-05, "loss": 0.0705, "step": 7610 }, { "epoch": 3.410158872230924, "grad_norm": 0.25649964809417725, "learning_rate": 9.838777320238312e-05, "loss": 0.0666, "step": 7620 }, { "epoch": 3.4146341463414633, "grad_norm": 0.24997064471244812, "learning_rate": 9.838082426819926e-05, "loss": 0.0673, "step": 7630 }, { "epoch": 3.4191094204520027, "grad_norm": 0.29423895478248596, "learning_rate": 9.83738606371984e-05, "loss": 0.071, "step": 7640 }, { "epoch": 3.423584694562542, "grad_norm": 0.3046830892562866, "learning_rate": 9.836688231149592e-05, "loss": 0.0688, "step": 7650 }, { "epoch": 3.428059968673081, "grad_norm": 0.28289347887039185, "learning_rate": 9.835988929321165e-05, "loss": 0.061, "step": 7660 }, { "epoch": 3.4325352427836204, "grad_norm": 0.2968411445617676, "learning_rate": 9.835288158446986e-05, "loss": 0.0658, "step": 7670 }, { "epoch": 3.43701051689416, "grad_norm": 0.2681797444820404, "learning_rate": 9.834585918739936e-05, "loss": 0.067, "step": 7680 }, { "epoch": 3.441485791004699, "grad_norm": 0.2733508050441742, "learning_rate": 9.833882210413332e-05, "loss": 0.0665, "step": 7690 }, { "epoch": 3.445961065115238, "grad_norm": 0.24902908504009247, "learning_rate": 9.833177033680944e-05, "loss": 0.0673, "step": 7700 }, { "epoch": 3.4504363392257775, "grad_norm": 0.28776735067367554, "learning_rate": 9.832470388756987e-05, "loss": 0.0648, "step": 7710 }, { "epoch": 3.454911613336317, "grad_norm": 0.27565211057662964, "learning_rate": 9.831762275856118e-05, "loss": 0.0694, "step": 7720 }, { "epoch": 3.4593868874468563, "grad_norm": 0.2657743990421295, "learning_rate": 9.831052695193445e-05, "loss": 0.0627, "step": 7730 }, { "epoch": 3.4638621615573952, "grad_norm": 0.24752621352672577, "learning_rate": 9.830341646984521e-05, "loss": 0.0631, "step": 7740 }, { "epoch": 3.4683374356679346, "grad_norm": 0.26454678177833557, "learning_rate": 9.829629131445342e-05, "loss": 0.071, "step": 7750 }, { "epoch": 3.472812709778474, "grad_norm": 0.28523892164230347, "learning_rate": 9.828915148792352e-05, "loss": 0.0685, "step": 7760 }, { "epoch": 3.4772879838890134, "grad_norm": 0.3040837049484253, "learning_rate": 9.82819969924244e-05, "loss": 0.064, "step": 7770 }, { "epoch": 3.4817632579995523, "grad_norm": 0.24729987978935242, "learning_rate": 9.82748278301294e-05, "loss": 0.0738, "step": 7780 }, { "epoch": 3.4862385321100917, "grad_norm": 0.32873162627220154, "learning_rate": 9.826764400321633e-05, "loss": 0.0701, "step": 7790 }, { "epoch": 3.490713806220631, "grad_norm": 0.2868390381336212, "learning_rate": 9.826044551386744e-05, "loss": 0.0638, "step": 7800 }, { "epoch": 3.49518908033117, "grad_norm": 0.32356342673301697, "learning_rate": 9.825323236426944e-05, "loss": 0.0651, "step": 7810 }, { "epoch": 3.4996643544417094, "grad_norm": 0.2583141326904297, "learning_rate": 9.824600455661353e-05, "loss": 0.0657, "step": 7820 }, { "epoch": 3.504139628552249, "grad_norm": 0.3120787739753723, "learning_rate": 9.823876209309527e-05, "loss": 0.0654, "step": 7830 }, { "epoch": 3.508614902662788, "grad_norm": 0.2732532322406769, "learning_rate": 9.823150497591476e-05, "loss": 0.0642, "step": 7840 }, { "epoch": 3.5130901767733276, "grad_norm": 0.20645299553871155, "learning_rate": 9.822423320727654e-05, "loss": 0.0618, "step": 7850 }, { "epoch": 3.5175654508838665, "grad_norm": 0.3084970712661743, "learning_rate": 9.821694678938953e-05, "loss": 0.0637, "step": 7860 }, { "epoch": 3.522040724994406, "grad_norm": 0.2425844520330429, "learning_rate": 9.82096457244672e-05, "loss": 0.0624, "step": 7870 }, { "epoch": 3.5265159991049453, "grad_norm": 0.2865842878818512, "learning_rate": 9.820233001472738e-05, "loss": 0.0671, "step": 7880 }, { "epoch": 3.5309912732154842, "grad_norm": 0.2720085680484772, "learning_rate": 9.819499966239243e-05, "loss": 0.068, "step": 7890 }, { "epoch": 3.5354665473260236, "grad_norm": 0.22250548005104065, "learning_rate": 9.818765466968909e-05, "loss": 0.0718, "step": 7900 }, { "epoch": 3.539941821436563, "grad_norm": 0.3251212537288666, "learning_rate": 9.818029503884859e-05, "loss": 0.0647, "step": 7910 }, { "epoch": 3.5444170955471024, "grad_norm": 0.24566210806369781, "learning_rate": 9.817292077210659e-05, "loss": 0.0631, "step": 7920 }, { "epoch": 3.5488923696576418, "grad_norm": 0.2554638087749481, "learning_rate": 9.816553187170317e-05, "loss": 0.0701, "step": 7930 }, { "epoch": 3.5533676437681807, "grad_norm": 0.3093981444835663, "learning_rate": 9.815812833988291e-05, "loss": 0.0659, "step": 7940 }, { "epoch": 3.55784291787872, "grad_norm": 0.25043150782585144, "learning_rate": 9.815071017889482e-05, "loss": 0.0681, "step": 7950 }, { "epoch": 3.5623181919892595, "grad_norm": 0.29701370000839233, "learning_rate": 9.81432773909923e-05, "loss": 0.0669, "step": 7960 }, { "epoch": 3.5667934660997984, "grad_norm": 0.2573421001434326, "learning_rate": 9.813582997843327e-05, "loss": 0.0642, "step": 7970 }, { "epoch": 3.571268740210338, "grad_norm": 0.30171576142311096, "learning_rate": 9.812836794348004e-05, "loss": 0.0655, "step": 7980 }, { "epoch": 3.575744014320877, "grad_norm": 0.3029201924800873, "learning_rate": 9.812089128839938e-05, "loss": 0.0689, "step": 7990 }, { "epoch": 3.5802192884314166, "grad_norm": 0.2622665464878082, "learning_rate": 9.811340001546251e-05, "loss": 0.0635, "step": 8000 }, { "epoch": 3.584694562541956, "grad_norm": 0.30391743779182434, "learning_rate": 9.81058941269451e-05, "loss": 0.065, "step": 8010 }, { "epoch": 3.589169836652495, "grad_norm": 0.251321017742157, "learning_rate": 9.80983736251272e-05, "loss": 0.063, "step": 8020 }, { "epoch": 3.5936451107630343, "grad_norm": 0.28762638568878174, "learning_rate": 9.809083851229335e-05, "loss": 0.0674, "step": 8030 }, { "epoch": 3.5981203848735737, "grad_norm": 0.266948938369751, "learning_rate": 9.808328879073251e-05, "loss": 0.0651, "step": 8040 }, { "epoch": 3.6025956589841126, "grad_norm": 0.20133982598781586, "learning_rate": 9.807572446273814e-05, "loss": 0.0628, "step": 8050 }, { "epoch": 3.607070933094652, "grad_norm": 0.3533587157726288, "learning_rate": 9.806814553060801e-05, "loss": 0.0664, "step": 8060 }, { "epoch": 3.6115462072051914, "grad_norm": 0.3735819160938263, "learning_rate": 9.806055199664446e-05, "loss": 0.061, "step": 8070 }, { "epoch": 3.6160214813157303, "grad_norm": 0.26862242817878723, "learning_rate": 9.805294386315415e-05, "loss": 0.0627, "step": 8080 }, { "epoch": 3.6204967554262697, "grad_norm": 0.3111754357814789, "learning_rate": 9.804532113244828e-05, "loss": 0.068, "step": 8090 }, { "epoch": 3.624972029536809, "grad_norm": 0.2801678776741028, "learning_rate": 9.803768380684242e-05, "loss": 0.0697, "step": 8100 }, { "epoch": 3.6294473036473485, "grad_norm": 0.2752600312232971, "learning_rate": 9.803003188865656e-05, "loss": 0.0694, "step": 8110 }, { "epoch": 3.633922577757888, "grad_norm": 0.2729775905609131, "learning_rate": 9.802236538021518e-05, "loss": 0.06, "step": 8120 }, { "epoch": 3.638397851868427, "grad_norm": 0.24939078092575073, "learning_rate": 9.801468428384716e-05, "loss": 0.0668, "step": 8130 }, { "epoch": 3.642873125978966, "grad_norm": 0.23295491933822632, "learning_rate": 9.80069886018858e-05, "loss": 0.0674, "step": 8140 }, { "epoch": 3.6473484000895056, "grad_norm": 0.2683183550834656, "learning_rate": 9.799927833666887e-05, "loss": 0.0696, "step": 8150 }, { "epoch": 3.6518236742000445, "grad_norm": 0.2508954703807831, "learning_rate": 9.799155349053851e-05, "loss": 0.0628, "step": 8160 }, { "epoch": 3.656298948310584, "grad_norm": 0.3360709249973297, "learning_rate": 9.798381406584135e-05, "loss": 0.0683, "step": 8170 }, { "epoch": 3.6607742224211233, "grad_norm": 0.28711459040641785, "learning_rate": 9.797606006492841e-05, "loss": 0.0655, "step": 8180 }, { "epoch": 3.6652494965316627, "grad_norm": 0.30931442975997925, "learning_rate": 9.796829149015517e-05, "loss": 0.0633, "step": 8190 }, { "epoch": 3.669724770642202, "grad_norm": 0.238079234957695, "learning_rate": 9.796050834388149e-05, "loss": 0.0663, "step": 8200 }, { "epoch": 3.674200044752741, "grad_norm": 0.30720195174217224, "learning_rate": 9.795271062847172e-05, "loss": 0.0648, "step": 8210 }, { "epoch": 3.6786753188632804, "grad_norm": 0.19506627321243286, "learning_rate": 9.794489834629455e-05, "loss": 0.0639, "step": 8220 }, { "epoch": 3.6831505929738197, "grad_norm": 0.27752581238746643, "learning_rate": 9.79370714997232e-05, "loss": 0.0626, "step": 8230 }, { "epoch": 3.6876258670843587, "grad_norm": 0.24365244805812836, "learning_rate": 9.792923009113522e-05, "loss": 0.0631, "step": 8240 }, { "epoch": 3.692101141194898, "grad_norm": 0.29671546816825867, "learning_rate": 9.792137412291265e-05, "loss": 0.065, "step": 8250 }, { "epoch": 3.6965764153054375, "grad_norm": 0.24501805007457733, "learning_rate": 9.791350359744189e-05, "loss": 0.0622, "step": 8260 }, { "epoch": 3.701051689415977, "grad_norm": 0.31776800751686096, "learning_rate": 9.790561851711383e-05, "loss": 0.0621, "step": 8270 }, { "epoch": 3.7055269635265162, "grad_norm": 0.26878321170806885, "learning_rate": 9.789771888432375e-05, "loss": 0.0616, "step": 8280 }, { "epoch": 3.710002237637055, "grad_norm": 0.26216307282447815, "learning_rate": 9.788980470147132e-05, "loss": 0.0632, "step": 8290 }, { "epoch": 3.7144775117475946, "grad_norm": 0.3125775456428528, "learning_rate": 9.788187597096069e-05, "loss": 0.067, "step": 8300 }, { "epoch": 3.718952785858134, "grad_norm": 0.32283759117126465, "learning_rate": 9.787393269520039e-05, "loss": 0.0633, "step": 8310 }, { "epoch": 3.723428059968673, "grad_norm": 0.26198849081993103, "learning_rate": 9.786597487660337e-05, "loss": 0.0633, "step": 8320 }, { "epoch": 3.7279033340792123, "grad_norm": 0.2797652781009674, "learning_rate": 9.785800251758701e-05, "loss": 0.0614, "step": 8330 }, { "epoch": 3.7323786081897516, "grad_norm": 0.2806423604488373, "learning_rate": 9.785001562057309e-05, "loss": 0.0648, "step": 8340 }, { "epoch": 3.736853882300291, "grad_norm": 0.2818640172481537, "learning_rate": 9.784201418798786e-05, "loss": 0.0643, "step": 8350 }, { "epoch": 3.7413291564108304, "grad_norm": 0.26862162351608276, "learning_rate": 9.783399822226189e-05, "loss": 0.0606, "step": 8360 }, { "epoch": 3.7458044305213694, "grad_norm": 0.28002697229385376, "learning_rate": 9.782596772583026e-05, "loss": 0.0609, "step": 8370 }, { "epoch": 3.7502797046319087, "grad_norm": 0.2619195282459259, "learning_rate": 9.781792270113241e-05, "loss": 0.0646, "step": 8380 }, { "epoch": 3.754754978742448, "grad_norm": 0.26814213395118713, "learning_rate": 9.780986315061218e-05, "loss": 0.0651, "step": 8390 }, { "epoch": 3.759230252852987, "grad_norm": 0.26579034328460693, "learning_rate": 9.780178907671789e-05, "loss": 0.0666, "step": 8400 }, { "epoch": 3.7637055269635264, "grad_norm": 0.24903492629528046, "learning_rate": 9.77937004819022e-05, "loss": 0.0634, "step": 8410 }, { "epoch": 3.768180801074066, "grad_norm": 0.2347663938999176, "learning_rate": 9.778559736862223e-05, "loss": 0.0638, "step": 8420 }, { "epoch": 3.772656075184605, "grad_norm": 0.27628958225250244, "learning_rate": 9.777747973933948e-05, "loss": 0.0611, "step": 8430 }, { "epoch": 3.7771313492951446, "grad_norm": 0.2684421241283417, "learning_rate": 9.776934759651988e-05, "loss": 0.0603, "step": 8440 }, { "epoch": 3.7816066234056835, "grad_norm": 0.23000222444534302, "learning_rate": 9.776120094263376e-05, "loss": 0.062, "step": 8450 }, { "epoch": 3.786081897516223, "grad_norm": 0.25908249616622925, "learning_rate": 9.775303978015585e-05, "loss": 0.0616, "step": 8460 }, { "epoch": 3.7905571716267623, "grad_norm": 0.29622700810432434, "learning_rate": 9.77448641115653e-05, "loss": 0.0573, "step": 8470 }, { "epoch": 3.7950324457373013, "grad_norm": 0.2343520224094391, "learning_rate": 9.773667393934567e-05, "loss": 0.0592, "step": 8480 }, { "epoch": 3.7995077198478406, "grad_norm": 0.25802695751190186, "learning_rate": 9.772846926598491e-05, "loss": 0.0594, "step": 8490 }, { "epoch": 3.80398299395838, "grad_norm": 0.35119253396987915, "learning_rate": 9.772025009397537e-05, "loss": 0.0655, "step": 8500 }, { "epoch": 3.808458268068919, "grad_norm": 0.33468008041381836, "learning_rate": 9.771201642581385e-05, "loss": 0.0633, "step": 8510 }, { "epoch": 3.8129335421794583, "grad_norm": 0.2975739538669586, "learning_rate": 9.77037682640015e-05, "loss": 0.0615, "step": 8520 }, { "epoch": 3.8174088162899977, "grad_norm": 0.2293076068162918, "learning_rate": 9.769550561104388e-05, "loss": 0.062, "step": 8530 }, { "epoch": 3.821884090400537, "grad_norm": 0.2834729850292206, "learning_rate": 9.7687228469451e-05, "loss": 0.0605, "step": 8540 }, { "epoch": 3.8263593645110765, "grad_norm": 0.2692764699459076, "learning_rate": 9.767893684173721e-05, "loss": 0.0631, "step": 8550 }, { "epoch": 3.8308346386216154, "grad_norm": 0.2700265049934387, "learning_rate": 9.76706307304213e-05, "loss": 0.0635, "step": 8560 }, { "epoch": 3.835309912732155, "grad_norm": 0.32148829102516174, "learning_rate": 9.766231013802645e-05, "loss": 0.0648, "step": 8570 }, { "epoch": 3.839785186842694, "grad_norm": 0.3093864917755127, "learning_rate": 9.765397506708023e-05, "loss": 0.0603, "step": 8580 }, { "epoch": 3.844260460953233, "grad_norm": 0.2933785319328308, "learning_rate": 9.76456255201146e-05, "loss": 0.064, "step": 8590 }, { "epoch": 3.8487357350637725, "grad_norm": 0.3018898367881775, "learning_rate": 9.763726149966596e-05, "loss": 0.0616, "step": 8600 }, { "epoch": 3.853211009174312, "grad_norm": 0.2560597062110901, "learning_rate": 9.762888300827507e-05, "loss": 0.0646, "step": 8610 }, { "epoch": 3.8576862832848513, "grad_norm": 0.22179432213306427, "learning_rate": 9.762049004848706e-05, "loss": 0.0623, "step": 8620 }, { "epoch": 3.8621615573953907, "grad_norm": 0.2642495632171631, "learning_rate": 9.761208262285155e-05, "loss": 0.064, "step": 8630 }, { "epoch": 3.8666368315059296, "grad_norm": 0.22237837314605713, "learning_rate": 9.760366073392246e-05, "loss": 0.0606, "step": 8640 }, { "epoch": 3.871112105616469, "grad_norm": 0.2695241868495941, "learning_rate": 9.759522438425813e-05, "loss": 0.0616, "step": 8650 }, { "epoch": 3.8755873797270084, "grad_norm": 0.29388490319252014, "learning_rate": 9.758677357642131e-05, "loss": 0.0655, "step": 8660 }, { "epoch": 3.8800626538375473, "grad_norm": 0.27349936962127686, "learning_rate": 9.757830831297914e-05, "loss": 0.0594, "step": 8670 }, { "epoch": 3.8845379279480867, "grad_norm": 0.2507762908935547, "learning_rate": 9.756982859650314e-05, "loss": 0.0601, "step": 8680 }, { "epoch": 3.889013202058626, "grad_norm": 0.34833410382270813, "learning_rate": 9.756133442956923e-05, "loss": 0.0596, "step": 8690 }, { "epoch": 3.8934884761691655, "grad_norm": 0.22958588600158691, "learning_rate": 9.755282581475769e-05, "loss": 0.0642, "step": 8700 }, { "epoch": 3.897963750279705, "grad_norm": 0.25646132230758667, "learning_rate": 9.754430275465323e-05, "loss": 0.0579, "step": 8710 }, { "epoch": 3.902439024390244, "grad_norm": 0.2559359669685364, "learning_rate": 9.753576525184492e-05, "loss": 0.0561, "step": 8720 }, { "epoch": 3.906914298500783, "grad_norm": 0.30206695199012756, "learning_rate": 9.752721330892624e-05, "loss": 0.0627, "step": 8730 }, { "epoch": 3.9113895726113226, "grad_norm": 0.27838486433029175, "learning_rate": 9.751864692849504e-05, "loss": 0.06, "step": 8740 }, { "epoch": 3.9158648467218615, "grad_norm": 0.30146801471710205, "learning_rate": 9.751006611315356e-05, "loss": 0.0595, "step": 8750 }, { "epoch": 3.920340120832401, "grad_norm": 0.2654535174369812, "learning_rate": 9.750147086550844e-05, "loss": 0.0636, "step": 8760 }, { "epoch": 3.9248153949429403, "grad_norm": 0.2640177011489868, "learning_rate": 9.749286118817067e-05, "loss": 0.0643, "step": 8770 }, { "epoch": 3.9292906690534797, "grad_norm": 0.27625924348831177, "learning_rate": 9.748423708375563e-05, "loss": 0.0621, "step": 8780 }, { "epoch": 3.933765943164019, "grad_norm": 0.27478793263435364, "learning_rate": 9.747559855488313e-05, "loss": 0.0648, "step": 8790 }, { "epoch": 3.938241217274558, "grad_norm": 0.2877417802810669, "learning_rate": 9.746694560417731e-05, "loss": 0.0628, "step": 8800 }, { "epoch": 3.9427164913850974, "grad_norm": 0.30414798855781555, "learning_rate": 9.74582782342667e-05, "loss": 0.0581, "step": 8810 }, { "epoch": 3.9471917654956368, "grad_norm": 0.2788531184196472, "learning_rate": 9.744959644778422e-05, "loss": 0.0661, "step": 8820 }, { "epoch": 3.9516670396061757, "grad_norm": 0.28016605973243713, "learning_rate": 9.744090024736719e-05, "loss": 0.0658, "step": 8830 }, { "epoch": 3.956142313716715, "grad_norm": 0.2957974076271057, "learning_rate": 9.743218963565725e-05, "loss": 0.0591, "step": 8840 }, { "epoch": 3.9606175878272545, "grad_norm": 0.24873599410057068, "learning_rate": 9.742346461530048e-05, "loss": 0.0607, "step": 8850 }, { "epoch": 3.9650928619377934, "grad_norm": 0.2468767911195755, "learning_rate": 9.74147251889473e-05, "loss": 0.0582, "step": 8860 }, { "epoch": 3.969568136048333, "grad_norm": 0.24718919396400452, "learning_rate": 9.740597135925253e-05, "loss": 0.0581, "step": 8870 }, { "epoch": 3.974043410158872, "grad_norm": 0.24922460317611694, "learning_rate": 9.739720312887535e-05, "loss": 0.0558, "step": 8880 }, { "epoch": 3.9785186842694116, "grad_norm": 0.2873542010784149, "learning_rate": 9.73884205004793e-05, "loss": 0.0596, "step": 8890 }, { "epoch": 3.982993958379951, "grad_norm": 0.2852094769477844, "learning_rate": 9.737962347673231e-05, "loss": 0.0608, "step": 8900 }, { "epoch": 3.98746923249049, "grad_norm": 0.2771070599555969, "learning_rate": 9.73708120603067e-05, "loss": 0.0622, "step": 8910 }, { "epoch": 3.9919445066010293, "grad_norm": 0.28079748153686523, "learning_rate": 9.736198625387916e-05, "loss": 0.0617, "step": 8920 }, { "epoch": 3.9964197807115687, "grad_norm": 0.26392948627471924, "learning_rate": 9.735314606013068e-05, "loss": 0.0611, "step": 8930 }, { "epoch": 4.000895054822108, "grad_norm": 0.2087579071521759, "learning_rate": 9.734429148174675e-05, "loss": 0.0571, "step": 8940 }, { "epoch": 4.005370328932647, "grad_norm": 0.26374635100364685, "learning_rate": 9.733542252141711e-05, "loss": 0.0567, "step": 8950 }, { "epoch": 4.009845603043186, "grad_norm": 0.2572246193885803, "learning_rate": 9.732653918183592e-05, "loss": 0.0618, "step": 8960 }, { "epoch": 4.014320877153725, "grad_norm": 0.2643309235572815, "learning_rate": 9.731764146570173e-05, "loss": 0.0573, "step": 8970 }, { "epoch": 4.018796151264265, "grad_norm": 0.24151448905467987, "learning_rate": 9.730872937571739e-05, "loss": 0.0571, "step": 8980 }, { "epoch": 4.023271425374804, "grad_norm": 0.2648807764053345, "learning_rate": 9.729980291459019e-05, "loss": 0.0561, "step": 8990 }, { "epoch": 4.027746699485344, "grad_norm": 0.2660192847251892, "learning_rate": 9.729086208503174e-05, "loss": 0.0583, "step": 9000 }, { "epoch": 4.032221973595883, "grad_norm": 0.20925603806972504, "learning_rate": 9.728190688975804e-05, "loss": 0.0578, "step": 9010 }, { "epoch": 4.036697247706422, "grad_norm": 0.2601310610771179, "learning_rate": 9.727293733148942e-05, "loss": 0.0604, "step": 9020 }, { "epoch": 4.041172521816962, "grad_norm": 0.28208667039871216, "learning_rate": 9.726395341295062e-05, "loss": 0.0563, "step": 9030 }, { "epoch": 4.045647795927501, "grad_norm": 0.28783804178237915, "learning_rate": 9.72549551368707e-05, "loss": 0.0607, "step": 9040 }, { "epoch": 4.0501230700380395, "grad_norm": 0.2802649438381195, "learning_rate": 9.724594250598311e-05, "loss": 0.0564, "step": 9050 }, { "epoch": 4.054598344148579, "grad_norm": 0.2522059679031372, "learning_rate": 9.723691552302562e-05, "loss": 0.0559, "step": 9060 }, { "epoch": 4.059073618259118, "grad_norm": 0.2843928635120392, "learning_rate": 9.722787419074044e-05, "loss": 0.0624, "step": 9070 }, { "epoch": 4.063548892369657, "grad_norm": 0.2546488642692566, "learning_rate": 9.721881851187406e-05, "loss": 0.0557, "step": 9080 }, { "epoch": 4.068024166480197, "grad_norm": 0.2836596369743347, "learning_rate": 9.720974848917735e-05, "loss": 0.0616, "step": 9090 }, { "epoch": 4.072499440590736, "grad_norm": 0.2614564597606659, "learning_rate": 9.720066412540554e-05, "loss": 0.0594, "step": 9100 }, { "epoch": 4.076974714701276, "grad_norm": 0.23317354917526245, "learning_rate": 9.719156542331824e-05, "loss": 0.063, "step": 9110 }, { "epoch": 4.081449988811815, "grad_norm": 0.24743646383285522, "learning_rate": 9.718245238567939e-05, "loss": 0.056, "step": 9120 }, { "epoch": 4.085925262922354, "grad_norm": 0.2501859664916992, "learning_rate": 9.717332501525729e-05, "loss": 0.0599, "step": 9130 }, { "epoch": 4.0904005370328935, "grad_norm": 0.23015186190605164, "learning_rate": 9.716418331482458e-05, "loss": 0.0595, "step": 9140 }, { "epoch": 4.0948758111434325, "grad_norm": 0.29163306951522827, "learning_rate": 9.715502728715826e-05, "loss": 0.0615, "step": 9150 }, { "epoch": 4.099351085253971, "grad_norm": 0.3633784353733063, "learning_rate": 9.714585693503974e-05, "loss": 0.0588, "step": 9160 }, { "epoch": 4.103826359364511, "grad_norm": 0.2541772127151489, "learning_rate": 9.713667226125467e-05, "loss": 0.0646, "step": 9170 }, { "epoch": 4.10830163347505, "grad_norm": 0.3086049258708954, "learning_rate": 9.712747326859315e-05, "loss": 0.0563, "step": 9180 }, { "epoch": 4.11277690758559, "grad_norm": 0.31434279680252075, "learning_rate": 9.711825995984957e-05, "loss": 0.0644, "step": 9190 }, { "epoch": 4.117252181696129, "grad_norm": 0.23047895729541779, "learning_rate": 9.710903233782272e-05, "loss": 0.0572, "step": 9200 }, { "epoch": 4.121727455806668, "grad_norm": 0.23937386274337769, "learning_rate": 9.709979040531569e-05, "loss": 0.0607, "step": 9210 }, { "epoch": 4.126202729917208, "grad_norm": 0.21885406970977783, "learning_rate": 9.709053416513592e-05, "loss": 0.0569, "step": 9220 }, { "epoch": 4.130678004027747, "grad_norm": 0.2703443169593811, "learning_rate": 9.708126362009522e-05, "loss": 0.0581, "step": 9230 }, { "epoch": 4.135153278138286, "grad_norm": 0.2598564624786377, "learning_rate": 9.707197877300974e-05, "loss": 0.0548, "step": 9240 }, { "epoch": 4.139628552248825, "grad_norm": 0.29065975546836853, "learning_rate": 9.706267962669998e-05, "loss": 0.0573, "step": 9250 }, { "epoch": 4.144103826359364, "grad_norm": 0.23526322841644287, "learning_rate": 9.705336618399077e-05, "loss": 0.062, "step": 9260 }, { "epoch": 4.148579100469904, "grad_norm": 0.24105599522590637, "learning_rate": 9.704403844771128e-05, "loss": 0.0594, "step": 9270 }, { "epoch": 4.153054374580443, "grad_norm": 0.27282264828681946, "learning_rate": 9.703469642069503e-05, "loss": 0.0573, "step": 9280 }, { "epoch": 4.157529648690982, "grad_norm": 0.2771025598049164, "learning_rate": 9.702534010577991e-05, "loss": 0.0577, "step": 9290 }, { "epoch": 4.162004922801522, "grad_norm": 0.29029327630996704, "learning_rate": 9.701596950580806e-05, "loss": 0.061, "step": 9300 }, { "epoch": 4.166480196912061, "grad_norm": 0.2776515483856201, "learning_rate": 9.700658462362608e-05, "loss": 0.0565, "step": 9310 }, { "epoch": 4.1709554710226, "grad_norm": 0.27885767817497253, "learning_rate": 9.699718546208484e-05, "loss": 0.0563, "step": 9320 }, { "epoch": 4.17543074513314, "grad_norm": 0.23760324716567993, "learning_rate": 9.698777202403953e-05, "loss": 0.0533, "step": 9330 }, { "epoch": 4.1799060192436785, "grad_norm": 0.22923870384693146, "learning_rate": 9.697834431234973e-05, "loss": 0.0555, "step": 9340 }, { "epoch": 4.1843812933542175, "grad_norm": 0.2303307205438614, "learning_rate": 9.696890232987931e-05, "loss": 0.0606, "step": 9350 }, { "epoch": 4.188856567464757, "grad_norm": 0.3091556429862976, "learning_rate": 9.695944607949649e-05, "loss": 0.0579, "step": 9360 }, { "epoch": 4.193331841575296, "grad_norm": 0.30779290199279785, "learning_rate": 9.694997556407387e-05, "loss": 0.0573, "step": 9370 }, { "epoch": 4.197807115685836, "grad_norm": 0.24137601256370544, "learning_rate": 9.69404907864883e-05, "loss": 0.0586, "step": 9380 }, { "epoch": 4.202282389796375, "grad_norm": 0.255460262298584, "learning_rate": 9.693099174962103e-05, "loss": 0.0582, "step": 9390 }, { "epoch": 4.206757663906914, "grad_norm": 0.2953519821166992, "learning_rate": 9.692147845635761e-05, "loss": 0.0562, "step": 9400 }, { "epoch": 4.211232938017454, "grad_norm": 0.2844688296318054, "learning_rate": 9.691195090958791e-05, "loss": 0.0566, "step": 9410 }, { "epoch": 4.215708212127993, "grad_norm": 0.21874798834323883, "learning_rate": 9.690240911220618e-05, "loss": 0.0576, "step": 9420 }, { "epoch": 4.220183486238533, "grad_norm": 0.2694944739341736, "learning_rate": 9.689285306711094e-05, "loss": 0.0581, "step": 9430 }, { "epoch": 4.2246587603490715, "grad_norm": 0.26777052879333496, "learning_rate": 9.688328277720507e-05, "loss": 0.0534, "step": 9440 }, { "epoch": 4.2291340344596104, "grad_norm": 0.3157203495502472, "learning_rate": 9.687369824539577e-05, "loss": 0.0537, "step": 9450 }, { "epoch": 4.23360930857015, "grad_norm": 0.25405165553092957, "learning_rate": 9.686409947459458e-05, "loss": 0.0591, "step": 9460 }, { "epoch": 4.238084582680689, "grad_norm": 0.23942634463310242, "learning_rate": 9.685448646771734e-05, "loss": 0.0532, "step": 9470 }, { "epoch": 4.242559856791228, "grad_norm": 0.2629034221172333, "learning_rate": 9.684485922768422e-05, "loss": 0.0556, "step": 9480 }, { "epoch": 4.247035130901768, "grad_norm": 0.2712942957878113, "learning_rate": 9.683521775741977e-05, "loss": 0.0608, "step": 9490 }, { "epoch": 4.251510405012307, "grad_norm": 0.23765400052070618, "learning_rate": 9.682556205985274e-05, "loss": 0.0579, "step": 9500 }, { "epoch": 4.255985679122846, "grad_norm": 0.2850038409233093, "learning_rate": 9.681589213791633e-05, "loss": 0.0548, "step": 9510 }, { "epoch": 4.260460953233386, "grad_norm": 0.2680826187133789, "learning_rate": 9.6806207994548e-05, "loss": 0.0646, "step": 9520 }, { "epoch": 4.264936227343925, "grad_norm": 0.2378014624118805, "learning_rate": 9.679650963268951e-05, "loss": 0.0609, "step": 9530 }, { "epoch": 4.2694115014544645, "grad_norm": 0.24389344453811646, "learning_rate": 9.6786797055287e-05, "loss": 0.0585, "step": 9540 }, { "epoch": 4.273886775565003, "grad_norm": 0.2472943365573883, "learning_rate": 9.677707026529086e-05, "loss": 0.0604, "step": 9550 }, { "epoch": 4.278362049675542, "grad_norm": 0.2839109003543854, "learning_rate": 9.676732926565585e-05, "loss": 0.06, "step": 9560 }, { "epoch": 4.282837323786082, "grad_norm": 0.28025633096694946, "learning_rate": 9.675757405934103e-05, "loss": 0.0595, "step": 9570 }, { "epoch": 4.287312597896621, "grad_norm": 0.2736935019493103, "learning_rate": 9.674780464930979e-05, "loss": 0.0576, "step": 9580 }, { "epoch": 4.29178787200716, "grad_norm": 0.24747468531131744, "learning_rate": 9.673802103852979e-05, "loss": 0.055, "step": 9590 }, { "epoch": 4.2962631461177, "grad_norm": 0.2718241214752197, "learning_rate": 9.672822322997305e-05, "loss": 0.06, "step": 9600 }, { "epoch": 4.300738420228239, "grad_norm": 0.2597699761390686, "learning_rate": 9.671841122661587e-05, "loss": 0.0577, "step": 9610 }, { "epoch": 4.305213694338779, "grad_norm": 0.3072591722011566, "learning_rate": 9.67085850314389e-05, "loss": 0.0598, "step": 9620 }, { "epoch": 4.309688968449318, "grad_norm": 0.235366553068161, "learning_rate": 9.669874464742705e-05, "loss": 0.0633, "step": 9630 }, { "epoch": 4.3141642425598565, "grad_norm": 0.2954546809196472, "learning_rate": 9.668889007756961e-05, "loss": 0.0571, "step": 9640 }, { "epoch": 4.318639516670396, "grad_norm": 0.3087657690048218, "learning_rate": 9.667902132486009e-05, "loss": 0.0597, "step": 9650 }, { "epoch": 4.323114790780935, "grad_norm": 0.26112520694732666, "learning_rate": 9.66691383922964e-05, "loss": 0.0594, "step": 9660 }, { "epoch": 4.327590064891474, "grad_norm": 0.2933204472064972, "learning_rate": 9.665924128288068e-05, "loss": 0.0576, "step": 9670 }, { "epoch": 4.332065339002014, "grad_norm": 0.23702220618724823, "learning_rate": 9.664932999961942e-05, "loss": 0.055, "step": 9680 }, { "epoch": 4.336540613112553, "grad_norm": 0.21880537271499634, "learning_rate": 9.663940454552342e-05, "loss": 0.0569, "step": 9690 }, { "epoch": 4.341015887223092, "grad_norm": 0.2630373239517212, "learning_rate": 9.662946492360776e-05, "loss": 0.0573, "step": 9700 }, { "epoch": 4.345491161333632, "grad_norm": 0.3107565641403198, "learning_rate": 9.661951113689182e-05, "loss": 0.0581, "step": 9710 }, { "epoch": 4.349966435444171, "grad_norm": 0.22515152394771576, "learning_rate": 9.660954318839933e-05, "loss": 0.0593, "step": 9720 }, { "epoch": 4.3544417095547105, "grad_norm": 0.2754456400871277, "learning_rate": 9.659956108115827e-05, "loss": 0.0517, "step": 9730 }, { "epoch": 4.3589169836652495, "grad_norm": 0.24447691440582275, "learning_rate": 9.658956481820094e-05, "loss": 0.0546, "step": 9740 }, { "epoch": 4.363392257775788, "grad_norm": 0.23009465634822845, "learning_rate": 9.657955440256395e-05, "loss": 0.052, "step": 9750 }, { "epoch": 4.367867531886328, "grad_norm": 0.2584233283996582, "learning_rate": 9.65695298372882e-05, "loss": 0.06, "step": 9760 }, { "epoch": 4.372342805996867, "grad_norm": 0.25300052762031555, "learning_rate": 9.655949112541887e-05, "loss": 0.0555, "step": 9770 }, { "epoch": 4.376818080107407, "grad_norm": 0.23312827944755554, "learning_rate": 9.654943827000548e-05, "loss": 0.0558, "step": 9780 }, { "epoch": 4.381293354217946, "grad_norm": 0.23699155449867249, "learning_rate": 9.65393712741018e-05, "loss": 0.0562, "step": 9790 }, { "epoch": 4.385768628328485, "grad_norm": 0.22687511146068573, "learning_rate": 9.652929014076593e-05, "loss": 0.0542, "step": 9800 }, { "epoch": 4.390243902439025, "grad_norm": 0.256837397813797, "learning_rate": 9.651919487306025e-05, "loss": 0.0554, "step": 9810 }, { "epoch": 4.394719176549564, "grad_norm": 0.24327430129051208, "learning_rate": 9.650908547405144e-05, "loss": 0.0567, "step": 9820 }, { "epoch": 4.399194450660103, "grad_norm": 0.2213292121887207, "learning_rate": 9.649896194681045e-05, "loss": 0.0602, "step": 9830 }, { "epoch": 4.4036697247706424, "grad_norm": 0.26796481013298035, "learning_rate": 9.648882429441257e-05, "loss": 0.0582, "step": 9840 }, { "epoch": 4.408144998881181, "grad_norm": 0.2649553418159485, "learning_rate": 9.647867251993734e-05, "loss": 0.05, "step": 9850 }, { "epoch": 4.41262027299172, "grad_norm": 0.27581554651260376, "learning_rate": 9.646850662646859e-05, "loss": 0.0504, "step": 9860 }, { "epoch": 4.41709554710226, "grad_norm": 0.3080006539821625, "learning_rate": 9.645832661709444e-05, "loss": 0.0557, "step": 9870 }, { "epoch": 4.421570821212799, "grad_norm": 0.2673134505748749, "learning_rate": 9.644813249490735e-05, "loss": 0.0576, "step": 9880 }, { "epoch": 4.426046095323339, "grad_norm": 0.2442914843559265, "learning_rate": 9.6437924263004e-05, "loss": 0.0567, "step": 9890 }, { "epoch": 4.430521369433878, "grad_norm": 0.2818883955478668, "learning_rate": 9.642770192448536e-05, "loss": 0.0601, "step": 9900 }, { "epoch": 4.434996643544417, "grad_norm": 0.29218706488609314, "learning_rate": 9.641746548245673e-05, "loss": 0.0595, "step": 9910 }, { "epoch": 4.439471917654957, "grad_norm": 0.24421058595180511, "learning_rate": 9.640721494002769e-05, "loss": 0.0566, "step": 9920 }, { "epoch": 4.443947191765496, "grad_norm": 0.25770872831344604, "learning_rate": 9.639695030031204e-05, "loss": 0.0572, "step": 9930 }, { "epoch": 4.4484224658760345, "grad_norm": 0.2313278466463089, "learning_rate": 9.638667156642794e-05, "loss": 0.0579, "step": 9940 }, { "epoch": 4.452897739986574, "grad_norm": 0.24046361446380615, "learning_rate": 9.637637874149779e-05, "loss": 0.0543, "step": 9950 }, { "epoch": 4.457373014097113, "grad_norm": 0.20515382289886475, "learning_rate": 9.636607182864827e-05, "loss": 0.0579, "step": 9960 }, { "epoch": 4.461848288207653, "grad_norm": 0.2420370727777481, "learning_rate": 9.635575083101036e-05, "loss": 0.0597, "step": 9970 }, { "epoch": 4.466323562318192, "grad_norm": 0.27035099267959595, "learning_rate": 9.634541575171929e-05, "loss": 0.0578, "step": 9980 }, { "epoch": 4.470798836428731, "grad_norm": 0.21404042840003967, "learning_rate": 9.63350665939146e-05, "loss": 0.0547, "step": 9990 }, { "epoch": 4.475274110539271, "grad_norm": 0.22910374402999878, "learning_rate": 9.632470336074009e-05, "loss": 0.0525, "step": 10000 }, { "epoch": 4.47974938464981, "grad_norm": 0.21561434864997864, "learning_rate": 9.631432605534383e-05, "loss": 0.0543, "step": 10010 }, { "epoch": 4.484224658760349, "grad_norm": 0.25823140144348145, "learning_rate": 9.630393468087818e-05, "loss": 0.0549, "step": 10020 }, { "epoch": 4.4886999328708885, "grad_norm": 0.1991884708404541, "learning_rate": 9.629352924049975e-05, "loss": 0.0499, "step": 10030 }, { "epoch": 4.4931752069814275, "grad_norm": 0.24088208377361298, "learning_rate": 9.628310973736943e-05, "loss": 0.0552, "step": 10040 }, { "epoch": 4.497650481091966, "grad_norm": 0.24108943343162537, "learning_rate": 9.627267617465243e-05, "loss": 0.0515, "step": 10050 }, { "epoch": 4.502125755202506, "grad_norm": 0.20222212374210358, "learning_rate": 9.626222855551816e-05, "loss": 0.0543, "step": 10060 }, { "epoch": 4.506601029313045, "grad_norm": 0.263415664434433, "learning_rate": 9.625176688314035e-05, "loss": 0.0555, "step": 10070 }, { "epoch": 4.511076303423585, "grad_norm": 0.20600029826164246, "learning_rate": 9.624129116069694e-05, "loss": 0.0582, "step": 10080 }, { "epoch": 4.515551577534124, "grad_norm": 0.23254333436489105, "learning_rate": 9.623080139137023e-05, "loss": 0.0552, "step": 10090 }, { "epoch": 4.520026851644663, "grad_norm": 0.23826387524604797, "learning_rate": 9.62202975783467e-05, "loss": 0.054, "step": 10100 }, { "epoch": 4.524502125755203, "grad_norm": 0.2677384316921234, "learning_rate": 9.620977972481716e-05, "loss": 0.0618, "step": 10110 }, { "epoch": 4.528977399865742, "grad_norm": 0.20585915446281433, "learning_rate": 9.619924783397661e-05, "loss": 0.0563, "step": 10120 }, { "epoch": 4.5334526739762815, "grad_norm": 0.24688221514225006, "learning_rate": 9.61887019090244e-05, "loss": 0.0536, "step": 10130 }, { "epoch": 4.53792794808682, "grad_norm": 0.20961406826972961, "learning_rate": 9.617814195316411e-05, "loss": 0.0546, "step": 10140 }, { "epoch": 4.542403222197359, "grad_norm": 0.2331228256225586, "learning_rate": 9.616756796960353e-05, "loss": 0.0566, "step": 10150 }, { "epoch": 4.546878496307899, "grad_norm": 0.29044824838638306, "learning_rate": 9.61569799615548e-05, "loss": 0.0534, "step": 10160 }, { "epoch": 4.551353770418438, "grad_norm": 0.2603375017642975, "learning_rate": 9.614637793223425e-05, "loss": 0.0534, "step": 10170 }, { "epoch": 4.555829044528977, "grad_norm": 0.28418660163879395, "learning_rate": 9.613576188486253e-05, "loss": 0.0542, "step": 10180 }, { "epoch": 4.560304318639517, "grad_norm": 0.21613077819347382, "learning_rate": 9.612513182266447e-05, "loss": 0.0604, "step": 10190 }, { "epoch": 4.564779592750056, "grad_norm": 0.22265012562274933, "learning_rate": 9.611448774886924e-05, "loss": 0.0554, "step": 10200 }, { "epoch": 4.569254866860595, "grad_norm": 0.22337763011455536, "learning_rate": 9.610382966671021e-05, "loss": 0.0496, "step": 10210 }, { "epoch": 4.573730140971135, "grad_norm": 0.2599366009235382, "learning_rate": 9.609315757942503e-05, "loss": 0.0516, "step": 10220 }, { "epoch": 4.578205415081674, "grad_norm": 0.22128795087337494, "learning_rate": 9.60824714902556e-05, "loss": 0.0531, "step": 10230 }, { "epoch": 4.582680689192213, "grad_norm": 0.24537494778633118, "learning_rate": 9.607177140244806e-05, "loss": 0.0557, "step": 10240 }, { "epoch": 4.587155963302752, "grad_norm": 0.2764444947242737, "learning_rate": 9.606105731925283e-05, "loss": 0.053, "step": 10250 }, { "epoch": 4.591631237413291, "grad_norm": 0.26711276173591614, "learning_rate": 9.605032924392457e-05, "loss": 0.0544, "step": 10260 }, { "epoch": 4.596106511523831, "grad_norm": 0.2569831311702728, "learning_rate": 9.603958717972214e-05, "loss": 0.0553, "step": 10270 }, { "epoch": 4.60058178563437, "grad_norm": 0.22038322687149048, "learning_rate": 9.602883112990875e-05, "loss": 0.0525, "step": 10280 }, { "epoch": 4.60505705974491, "grad_norm": 0.17705503106117249, "learning_rate": 9.601806109775179e-05, "loss": 0.0572, "step": 10290 }, { "epoch": 4.609532333855449, "grad_norm": 0.2496725469827652, "learning_rate": 9.600727708652289e-05, "loss": 0.0553, "step": 10300 }, { "epoch": 4.614007607965988, "grad_norm": 0.24831072986125946, "learning_rate": 9.599647909949798e-05, "loss": 0.0507, "step": 10310 }, { "epoch": 4.618482882076528, "grad_norm": 0.22404256463050842, "learning_rate": 9.598566713995718e-05, "loss": 0.0543, "step": 10320 }, { "epoch": 4.6229581561870665, "grad_norm": 0.22090521454811096, "learning_rate": 9.597484121118487e-05, "loss": 0.053, "step": 10330 }, { "epoch": 4.6274334302976055, "grad_norm": 0.2713897228240967, "learning_rate": 9.596400131646972e-05, "loss": 0.0523, "step": 10340 }, { "epoch": 4.631908704408145, "grad_norm": 0.22385922074317932, "learning_rate": 9.595314745910456e-05, "loss": 0.0565, "step": 10350 }, { "epoch": 4.636383978518684, "grad_norm": 0.1929776519536972, "learning_rate": 9.594227964238653e-05, "loss": 0.0569, "step": 10360 }, { "epoch": 4.640859252629223, "grad_norm": 0.24381612241268158, "learning_rate": 9.593139786961697e-05, "loss": 0.0525, "step": 10370 }, { "epoch": 4.645334526739763, "grad_norm": 0.2332475334405899, "learning_rate": 9.59205021441015e-05, "loss": 0.0515, "step": 10380 }, { "epoch": 4.649809800850302, "grad_norm": 0.2474587857723236, "learning_rate": 9.590959246914995e-05, "loss": 0.0529, "step": 10390 }, { "epoch": 4.654285074960841, "grad_norm": 0.2686334252357483, "learning_rate": 9.589866884807635e-05, "loss": 0.0492, "step": 10400 }, { "epoch": 4.658760349071381, "grad_norm": 0.2519616484642029, "learning_rate": 9.588773128419906e-05, "loss": 0.0519, "step": 10410 }, { "epoch": 4.66323562318192, "grad_norm": 0.1766405701637268, "learning_rate": 9.58767797808406e-05, "loss": 0.0516, "step": 10420 }, { "epoch": 4.6677108972924595, "grad_norm": 0.20664215087890625, "learning_rate": 9.586581434132775e-05, "loss": 0.053, "step": 10430 }, { "epoch": 4.672186171402998, "grad_norm": 0.2059725522994995, "learning_rate": 9.58548349689915e-05, "loss": 0.0537, "step": 10440 }, { "epoch": 4.676661445513537, "grad_norm": 0.22378763556480408, "learning_rate": 9.584384166716714e-05, "loss": 0.0562, "step": 10450 }, { "epoch": 4.681136719624077, "grad_norm": 0.26983365416526794, "learning_rate": 9.583283443919409e-05, "loss": 0.0566, "step": 10460 }, { "epoch": 4.685611993734616, "grad_norm": 0.19978022575378418, "learning_rate": 9.582181328841611e-05, "loss": 0.0541, "step": 10470 }, { "epoch": 4.690087267845156, "grad_norm": 0.22685207426548004, "learning_rate": 9.581077821818109e-05, "loss": 0.0565, "step": 10480 }, { "epoch": 4.694562541955695, "grad_norm": 0.25033092498779297, "learning_rate": 9.579972923184122e-05, "loss": 0.0545, "step": 10490 }, { "epoch": 4.699037816066234, "grad_norm": 0.26767346262931824, "learning_rate": 9.578866633275288e-05, "loss": 0.0511, "step": 10500 }, { "epoch": 4.703513090176774, "grad_norm": 0.22659511864185333, "learning_rate": 9.577758952427669e-05, "loss": 0.0589, "step": 10510 }, { "epoch": 4.707988364287313, "grad_norm": 0.2503576874732971, "learning_rate": 9.576649880977748e-05, "loss": 0.0556, "step": 10520 }, { "epoch": 4.7124636383978515, "grad_norm": 0.19737689197063446, "learning_rate": 9.575539419262434e-05, "loss": 0.052, "step": 10530 }, { "epoch": 4.716938912508391, "grad_norm": 0.21432526409626007, "learning_rate": 9.574427567619053e-05, "loss": 0.0515, "step": 10540 }, { "epoch": 4.72141418661893, "grad_norm": 0.2154179960489273, "learning_rate": 9.573314326385359e-05, "loss": 0.0537, "step": 10550 }, { "epoch": 4.725889460729469, "grad_norm": 0.2674520015716553, "learning_rate": 9.572199695899522e-05, "loss": 0.0497, "step": 10560 }, { "epoch": 4.730364734840009, "grad_norm": 0.20934461057186127, "learning_rate": 9.571083676500141e-05, "loss": 0.0553, "step": 10570 }, { "epoch": 4.734840008950548, "grad_norm": 0.24892187118530273, "learning_rate": 9.569966268526232e-05, "loss": 0.0516, "step": 10580 }, { "epoch": 4.739315283061088, "grad_norm": 0.22739887237548828, "learning_rate": 9.568847472317232e-05, "loss": 0.0555, "step": 10590 }, { "epoch": 4.743790557171627, "grad_norm": 0.2236451655626297, "learning_rate": 9.567727288213005e-05, "loss": 0.0529, "step": 10600 }, { "epoch": 4.748265831282166, "grad_norm": 0.20754890143871307, "learning_rate": 9.566605716553831e-05, "loss": 0.0503, "step": 10610 }, { "epoch": 4.752741105392706, "grad_norm": 0.2598242461681366, "learning_rate": 9.565482757680415e-05, "loss": 0.0541, "step": 10620 }, { "epoch": 4.7572163795032445, "grad_norm": 0.24581129848957062, "learning_rate": 9.56435841193388e-05, "loss": 0.0462, "step": 10630 }, { "epoch": 4.761691653613784, "grad_norm": 0.24896709620952606, "learning_rate": 9.563232679655776e-05, "loss": 0.051, "step": 10640 }, { "epoch": 4.766166927724323, "grad_norm": 0.24935537576675415, "learning_rate": 9.562105561188069e-05, "loss": 0.0517, "step": 10650 }, { "epoch": 4.770642201834862, "grad_norm": 0.21170563995838165, "learning_rate": 9.560977056873149e-05, "loss": 0.0546, "step": 10660 }, { "epoch": 4.775117475945402, "grad_norm": 0.22386832535266876, "learning_rate": 9.559847167053823e-05, "loss": 0.0542, "step": 10670 }, { "epoch": 4.779592750055941, "grad_norm": 0.27254152297973633, "learning_rate": 9.558715892073323e-05, "loss": 0.0574, "step": 10680 }, { "epoch": 4.78406802416648, "grad_norm": 0.25849759578704834, "learning_rate": 9.557583232275303e-05, "loss": 0.0525, "step": 10690 }, { "epoch": 4.78854329827702, "grad_norm": 0.21827207505702972, "learning_rate": 9.556449188003831e-05, "loss": 0.0515, "step": 10700 }, { "epoch": 4.793018572387559, "grad_norm": 0.274384081363678, "learning_rate": 9.555313759603402e-05, "loss": 0.0535, "step": 10710 }, { "epoch": 4.797493846498098, "grad_norm": 0.2635228931903839, "learning_rate": 9.554176947418931e-05, "loss": 0.0522, "step": 10720 }, { "epoch": 4.8019691206086375, "grad_norm": 0.2222270369529724, "learning_rate": 9.553038751795746e-05, "loss": 0.0532, "step": 10730 }, { "epoch": 4.806444394719176, "grad_norm": 0.2773084342479706, "learning_rate": 9.551899173079607e-05, "loss": 0.0566, "step": 10740 }, { "epoch": 4.810919668829715, "grad_norm": 0.19339749217033386, "learning_rate": 9.550758211616684e-05, "loss": 0.0506, "step": 10750 }, { "epoch": 4.815394942940255, "grad_norm": 0.2794245183467865, "learning_rate": 9.549615867753573e-05, "loss": 0.0608, "step": 10760 }, { "epoch": 4.819870217050794, "grad_norm": 0.20750465989112854, "learning_rate": 9.548472141837286e-05, "loss": 0.0506, "step": 10770 }, { "epoch": 4.824345491161334, "grad_norm": 0.2664441764354706, "learning_rate": 9.54732703421526e-05, "loss": 0.0535, "step": 10780 }, { "epoch": 4.828820765271873, "grad_norm": 0.20051917433738708, "learning_rate": 9.546180545235344e-05, "loss": 0.0543, "step": 10790 }, { "epoch": 4.833296039382412, "grad_norm": 0.2242336869239807, "learning_rate": 9.545032675245813e-05, "loss": 0.0531, "step": 10800 }, { "epoch": 4.837771313492952, "grad_norm": 0.23257504403591156, "learning_rate": 9.543883424595361e-05, "loss": 0.0538, "step": 10810 }, { "epoch": 4.842246587603491, "grad_norm": 0.23518428206443787, "learning_rate": 9.542732793633098e-05, "loss": 0.0512, "step": 10820 }, { "epoch": 4.84672186171403, "grad_norm": 0.2135966420173645, "learning_rate": 9.541580782708557e-05, "loss": 0.0505, "step": 10830 }, { "epoch": 4.851197135824569, "grad_norm": 0.24368475377559662, "learning_rate": 9.540427392171688e-05, "loss": 0.0526, "step": 10840 }, { "epoch": 4.855672409935108, "grad_norm": 0.26603952050209045, "learning_rate": 9.539272622372858e-05, "loss": 0.051, "step": 10850 }, { "epoch": 4.860147684045648, "grad_norm": 0.25276967883110046, "learning_rate": 9.538116473662861e-05, "loss": 0.0531, "step": 10860 }, { "epoch": 4.864622958156187, "grad_norm": 0.2795259356498718, "learning_rate": 9.5369589463929e-05, "loss": 0.0535, "step": 10870 }, { "epoch": 4.869098232266726, "grad_norm": 0.30648741126060486, "learning_rate": 9.535800040914601e-05, "loss": 0.0568, "step": 10880 }, { "epoch": 4.873573506377266, "grad_norm": 0.21677008271217346, "learning_rate": 9.534639757580013e-05, "loss": 0.0531, "step": 10890 }, { "epoch": 4.878048780487805, "grad_norm": 0.21497292816638947, "learning_rate": 9.533478096741597e-05, "loss": 0.0501, "step": 10900 }, { "epoch": 4.882524054598344, "grad_norm": 0.25450655817985535, "learning_rate": 9.532315058752232e-05, "loss": 0.0552, "step": 10910 }, { "epoch": 4.8869993287088835, "grad_norm": 0.29273080825805664, "learning_rate": 9.531150643965223e-05, "loss": 0.0541, "step": 10920 }, { "epoch": 4.8914746028194225, "grad_norm": 0.2293233871459961, "learning_rate": 9.529984852734285e-05, "loss": 0.0522, "step": 10930 }, { "epoch": 4.895949876929962, "grad_norm": 0.21604092419147491, "learning_rate": 9.528817685413558e-05, "loss": 0.0553, "step": 10940 }, { "epoch": 4.900425151040501, "grad_norm": 0.18777231872081757, "learning_rate": 9.527649142357596e-05, "loss": 0.0513, "step": 10950 }, { "epoch": 4.90490042515104, "grad_norm": 0.2783224582672119, "learning_rate": 9.526479223921366e-05, "loss": 0.0526, "step": 10960 }, { "epoch": 4.90937569926158, "grad_norm": 0.25982755422592163, "learning_rate": 9.525307930460266e-05, "loss": 0.0584, "step": 10970 }, { "epoch": 4.913850973372119, "grad_norm": 0.2574670612812042, "learning_rate": 9.524135262330098e-05, "loss": 0.0496, "step": 10980 }, { "epoch": 4.918326247482659, "grad_norm": 0.19419275224208832, "learning_rate": 9.522961219887092e-05, "loss": 0.0558, "step": 10990 }, { "epoch": 4.922801521593198, "grad_norm": 0.2619624435901642, "learning_rate": 9.521785803487889e-05, "loss": 0.0528, "step": 11000 }, { "epoch": 4.927276795703737, "grad_norm": 0.2254522293806076, "learning_rate": 9.520609013489547e-05, "loss": 0.053, "step": 11010 }, { "epoch": 4.9317520698142765, "grad_norm": 0.19613121449947357, "learning_rate": 9.51943085024955e-05, "loss": 0.0534, "step": 11020 }, { "epoch": 4.936227343924815, "grad_norm": 0.22455435991287231, "learning_rate": 9.518251314125788e-05, "loss": 0.0496, "step": 11030 }, { "epoch": 4.940702618035354, "grad_norm": 0.21523891389369965, "learning_rate": 9.517070405476575e-05, "loss": 0.0511, "step": 11040 }, { "epoch": 4.945177892145894, "grad_norm": 0.20727460086345673, "learning_rate": 9.515888124660638e-05, "loss": 0.0516, "step": 11050 }, { "epoch": 4.949653166256433, "grad_norm": 0.1863197684288025, "learning_rate": 9.514704472037123e-05, "loss": 0.0514, "step": 11060 }, { "epoch": 4.954128440366972, "grad_norm": 0.20250675082206726, "learning_rate": 9.513519447965595e-05, "loss": 0.054, "step": 11070 }, { "epoch": 4.958603714477512, "grad_norm": 0.22080934047698975, "learning_rate": 9.512333052806033e-05, "loss": 0.0561, "step": 11080 }, { "epoch": 4.963078988588051, "grad_norm": 0.2101924568414688, "learning_rate": 9.511145286918828e-05, "loss": 0.0546, "step": 11090 }, { "epoch": 4.96755426269859, "grad_norm": 0.20323804020881653, "learning_rate": 9.509956150664796e-05, "loss": 0.0494, "step": 11100 }, { "epoch": 4.97202953680913, "grad_norm": 0.2571505308151245, "learning_rate": 9.508765644405162e-05, "loss": 0.0529, "step": 11110 }, { "epoch": 4.976504810919669, "grad_norm": 0.25417378544807434, "learning_rate": 9.507573768501574e-05, "loss": 0.0505, "step": 11120 }, { "epoch": 4.980980085030208, "grad_norm": 0.24169062077999115, "learning_rate": 9.50638052331609e-05, "loss": 0.0501, "step": 11130 }, { "epoch": 4.985455359140747, "grad_norm": 0.23826964199543, "learning_rate": 9.505185909211188e-05, "loss": 0.0556, "step": 11140 }, { "epoch": 4.989930633251286, "grad_norm": 0.17638371884822845, "learning_rate": 9.50398992654976e-05, "loss": 0.0519, "step": 11150 }, { "epoch": 4.994405907361826, "grad_norm": 0.20734313130378723, "learning_rate": 9.502792575695112e-05, "loss": 0.0499, "step": 11160 }, { "epoch": 4.998881181472365, "grad_norm": 0.22528144717216492, "learning_rate": 9.501593857010969e-05, "loss": 0.0555, "step": 11170 }, { "epoch": 5.003356455582905, "grad_norm": 0.21009255945682526, "learning_rate": 9.50039377086147e-05, "loss": 0.0523, "step": 11180 }, { "epoch": 5.007831729693444, "grad_norm": 0.20401059091091156, "learning_rate": 9.499192317611167e-05, "loss": 0.0519, "step": 11190 }, { "epoch": 5.012307003803983, "grad_norm": 0.21557004749774933, "learning_rate": 9.497989497625035e-05, "loss": 0.0545, "step": 11200 }, { "epoch": 5.016782277914523, "grad_norm": 0.23922067880630493, "learning_rate": 9.496785311268457e-05, "loss": 0.0563, "step": 11210 }, { "epoch": 5.0212575520250615, "grad_norm": 0.24187473952770233, "learning_rate": 9.49557975890723e-05, "loss": 0.0524, "step": 11220 }, { "epoch": 5.0257328261356005, "grad_norm": 0.199630007147789, "learning_rate": 9.494372840907572e-05, "loss": 0.0537, "step": 11230 }, { "epoch": 5.03020810024614, "grad_norm": 0.19466882944107056, "learning_rate": 9.493164557636112e-05, "loss": 0.0486, "step": 11240 }, { "epoch": 5.034683374356679, "grad_norm": 0.1810058355331421, "learning_rate": 9.491954909459895e-05, "loss": 0.0489, "step": 11250 }, { "epoch": 5.039158648467218, "grad_norm": 0.2027171403169632, "learning_rate": 9.49074389674638e-05, "loss": 0.054, "step": 11260 }, { "epoch": 5.043633922577758, "grad_norm": 0.21586036682128906, "learning_rate": 9.48953151986344e-05, "loss": 0.0483, "step": 11270 }, { "epoch": 5.048109196688297, "grad_norm": 0.2147798389196396, "learning_rate": 9.488317779179361e-05, "loss": 0.0517, "step": 11280 }, { "epoch": 5.052584470798837, "grad_norm": 0.21428924798965454, "learning_rate": 9.487102675062851e-05, "loss": 0.0508, "step": 11290 }, { "epoch": 5.057059744909376, "grad_norm": 0.17959052324295044, "learning_rate": 9.485886207883022e-05, "loss": 0.0481, "step": 11300 }, { "epoch": 5.061535019019915, "grad_norm": 0.2434990406036377, "learning_rate": 9.484668378009408e-05, "loss": 0.054, "step": 11310 }, { "epoch": 5.0660102931304545, "grad_norm": 0.22875070571899414, "learning_rate": 9.483449185811948e-05, "loss": 0.0496, "step": 11320 }, { "epoch": 5.070485567240993, "grad_norm": 0.2941875755786896, "learning_rate": 9.482228631661005e-05, "loss": 0.052, "step": 11330 }, { "epoch": 5.074960841351532, "grad_norm": 0.21440055966377258, "learning_rate": 9.481006715927351e-05, "loss": 0.051, "step": 11340 }, { "epoch": 5.079436115462072, "grad_norm": 0.22733141481876373, "learning_rate": 9.479783438982172e-05, "loss": 0.0512, "step": 11350 }, { "epoch": 5.083911389572611, "grad_norm": 0.2254066914319992, "learning_rate": 9.478558801197065e-05, "loss": 0.0545, "step": 11360 }, { "epoch": 5.088386663683151, "grad_norm": 0.2168971747159958, "learning_rate": 9.477332802944044e-05, "loss": 0.0502, "step": 11370 }, { "epoch": 5.09286193779369, "grad_norm": 0.22835278511047363, "learning_rate": 9.476105444595534e-05, "loss": 0.0523, "step": 11380 }, { "epoch": 5.097337211904229, "grad_norm": 0.2240561991930008, "learning_rate": 9.474876726524374e-05, "loss": 0.0502, "step": 11390 }, { "epoch": 5.101812486014769, "grad_norm": 0.22353364527225494, "learning_rate": 9.473646649103818e-05, "loss": 0.0484, "step": 11400 }, { "epoch": 5.106287760125308, "grad_norm": 0.2488842010498047, "learning_rate": 9.47241521270753e-05, "loss": 0.05, "step": 11410 }, { "epoch": 5.1107630342358465, "grad_norm": 0.21628059446811676, "learning_rate": 9.471182417709587e-05, "loss": 0.044, "step": 11420 }, { "epoch": 5.115238308346386, "grad_norm": 0.20322981476783752, "learning_rate": 9.46994826448448e-05, "loss": 0.0509, "step": 11430 }, { "epoch": 5.119713582456925, "grad_norm": 0.24497012794017792, "learning_rate": 9.468712753407112e-05, "loss": 0.0526, "step": 11440 }, { "epoch": 5.124188856567465, "grad_norm": 0.18894140422344208, "learning_rate": 9.4674758848528e-05, "loss": 0.0478, "step": 11450 }, { "epoch": 5.128664130678004, "grad_norm": 0.15635353326797485, "learning_rate": 9.46623765919727e-05, "loss": 0.0532, "step": 11460 }, { "epoch": 5.133139404788543, "grad_norm": 0.187982439994812, "learning_rate": 9.464998076816664e-05, "loss": 0.0535, "step": 11470 }, { "epoch": 5.137614678899083, "grad_norm": 0.20060020685195923, "learning_rate": 9.463757138087535e-05, "loss": 0.0474, "step": 11480 }, { "epoch": 5.142089953009622, "grad_norm": 0.23887500166893005, "learning_rate": 9.462514843386845e-05, "loss": 0.0469, "step": 11490 }, { "epoch": 5.146565227120161, "grad_norm": 0.22446706891059875, "learning_rate": 9.46127119309197e-05, "loss": 0.0496, "step": 11500 }, { "epoch": 5.151040501230701, "grad_norm": 0.23625801503658295, "learning_rate": 9.460026187580702e-05, "loss": 0.0493, "step": 11510 }, { "epoch": 5.1555157753412395, "grad_norm": 0.2047010064125061, "learning_rate": 9.458779827231237e-05, "loss": 0.0524, "step": 11520 }, { "epoch": 5.159991049451779, "grad_norm": 0.19676990807056427, "learning_rate": 9.457532112422187e-05, "loss": 0.0525, "step": 11530 }, { "epoch": 5.164466323562318, "grad_norm": 0.22218270599842072, "learning_rate": 9.456283043532576e-05, "loss": 0.0463, "step": 11540 }, { "epoch": 5.168941597672857, "grad_norm": 0.19876323640346527, "learning_rate": 9.45503262094184e-05, "loss": 0.052, "step": 11550 }, { "epoch": 5.173416871783397, "grad_norm": 0.23600813746452332, "learning_rate": 9.453780845029821e-05, "loss": 0.0535, "step": 11560 }, { "epoch": 5.177892145893936, "grad_norm": 0.20529134571552277, "learning_rate": 9.452527716176776e-05, "loss": 0.0474, "step": 11570 }, { "epoch": 5.182367420004475, "grad_norm": 0.1934622824192047, "learning_rate": 9.451273234763371e-05, "loss": 0.0555, "step": 11580 }, { "epoch": 5.186842694115015, "grad_norm": 0.23026640713214874, "learning_rate": 9.450017401170689e-05, "loss": 0.05, "step": 11590 }, { "epoch": 5.191317968225554, "grad_norm": 0.22378131747245789, "learning_rate": 9.448760215780217e-05, "loss": 0.0494, "step": 11600 }, { "epoch": 5.195793242336093, "grad_norm": 0.23083056509494781, "learning_rate": 9.447501678973852e-05, "loss": 0.0481, "step": 11610 }, { "epoch": 5.2002685164466325, "grad_norm": 0.24726121127605438, "learning_rate": 9.446241791133907e-05, "loss": 0.0496, "step": 11620 }, { "epoch": 5.204743790557171, "grad_norm": 0.20926973223686218, "learning_rate": 9.444980552643103e-05, "loss": 0.0506, "step": 11630 }, { "epoch": 5.209219064667711, "grad_norm": 0.18582668900489807, "learning_rate": 9.443717963884569e-05, "loss": 0.0504, "step": 11640 }, { "epoch": 5.21369433877825, "grad_norm": 0.2814835011959076, "learning_rate": 9.442454025241847e-05, "loss": 0.051, "step": 11650 }, { "epoch": 5.218169612888789, "grad_norm": 0.22467274963855743, "learning_rate": 9.441188737098889e-05, "loss": 0.051, "step": 11660 }, { "epoch": 5.222644886999329, "grad_norm": 0.23449361324310303, "learning_rate": 9.439922099840054e-05, "loss": 0.053, "step": 11670 }, { "epoch": 5.227120161109868, "grad_norm": 0.21743959188461304, "learning_rate": 9.438654113850118e-05, "loss": 0.0499, "step": 11680 }, { "epoch": 5.231595435220407, "grad_norm": 0.24982912838459015, "learning_rate": 9.437384779514256e-05, "loss": 0.0517, "step": 11690 }, { "epoch": 5.236070709330947, "grad_norm": 0.2680239975452423, "learning_rate": 9.43611409721806e-05, "loss": 0.0511, "step": 11700 }, { "epoch": 5.240545983441486, "grad_norm": 0.21764490008354187, "learning_rate": 9.43484206734753e-05, "loss": 0.053, "step": 11710 }, { "epoch": 5.245021257552025, "grad_norm": 0.2644926607608795, "learning_rate": 9.433568690289075e-05, "loss": 0.052, "step": 11720 }, { "epoch": 5.249496531662564, "grad_norm": 0.16799037158489227, "learning_rate": 9.432293966429514e-05, "loss": 0.0483, "step": 11730 }, { "epoch": 5.253971805773103, "grad_norm": 0.2747966945171356, "learning_rate": 9.431017896156074e-05, "loss": 0.0546, "step": 11740 }, { "epoch": 5.258447079883643, "grad_norm": 0.21256914734840393, "learning_rate": 9.42974047985639e-05, "loss": 0.0478, "step": 11750 }, { "epoch": 5.262922353994182, "grad_norm": 0.22289220988750458, "learning_rate": 9.428461717918511e-05, "loss": 0.0531, "step": 11760 }, { "epoch": 5.267397628104721, "grad_norm": 0.21855075657367706, "learning_rate": 9.427181610730888e-05, "loss": 0.0544, "step": 11770 }, { "epoch": 5.271872902215261, "grad_norm": 0.22455374896526337, "learning_rate": 9.425900158682385e-05, "loss": 0.0556, "step": 11780 }, { "epoch": 5.2763481763258, "grad_norm": 0.22301901876926422, "learning_rate": 9.424617362162271e-05, "loss": 0.0537, "step": 11790 }, { "epoch": 5.28082345043634, "grad_norm": 0.24258540570735931, "learning_rate": 9.42333322156023e-05, "loss": 0.0508, "step": 11800 }, { "epoch": 5.2852987245468785, "grad_norm": 0.22380582988262177, "learning_rate": 9.422047737266347e-05, "loss": 0.0532, "step": 11810 }, { "epoch": 5.2897739986574175, "grad_norm": 0.20364278554916382, "learning_rate": 9.420760909671118e-05, "loss": 0.0489, "step": 11820 }, { "epoch": 5.294249272767957, "grad_norm": 0.23438726365566254, "learning_rate": 9.419472739165449e-05, "loss": 0.0564, "step": 11830 }, { "epoch": 5.298724546878496, "grad_norm": 0.1873813271522522, "learning_rate": 9.41818322614065e-05, "loss": 0.0465, "step": 11840 }, { "epoch": 5.303199820989035, "grad_norm": 0.20513059198856354, "learning_rate": 9.416892370988444e-05, "loss": 0.0488, "step": 11850 }, { "epoch": 5.307675095099575, "grad_norm": 0.22137634456157684, "learning_rate": 9.415600174100956e-05, "loss": 0.0525, "step": 11860 }, { "epoch": 5.312150369210114, "grad_norm": 0.24335788190364838, "learning_rate": 9.414306635870722e-05, "loss": 0.0495, "step": 11870 }, { "epoch": 5.316625643320654, "grad_norm": 0.2328237146139145, "learning_rate": 9.413011756690685e-05, "loss": 0.0516, "step": 11880 }, { "epoch": 5.321100917431193, "grad_norm": 0.22849954664707184, "learning_rate": 9.411715536954196e-05, "loss": 0.0496, "step": 11890 }, { "epoch": 5.325576191541732, "grad_norm": 0.2344713807106018, "learning_rate": 9.410417977055011e-05, "loss": 0.0507, "step": 11900 }, { "epoch": 5.3300514656522715, "grad_norm": 0.21410098671913147, "learning_rate": 9.409119077387294e-05, "loss": 0.0544, "step": 11910 }, { "epoch": 5.33452673976281, "grad_norm": 0.20980581641197205, "learning_rate": 9.407818838345619e-05, "loss": 0.0515, "step": 11920 }, { "epoch": 5.339002013873349, "grad_norm": 0.2154228240251541, "learning_rate": 9.40651726032496e-05, "loss": 0.0529, "step": 11930 }, { "epoch": 5.343477287983889, "grad_norm": 0.20793640613555908, "learning_rate": 9.405214343720707e-05, "loss": 0.0482, "step": 11940 }, { "epoch": 5.347952562094428, "grad_norm": 0.18256552517414093, "learning_rate": 9.403910088928651e-05, "loss": 0.0544, "step": 11950 }, { "epoch": 5.352427836204967, "grad_norm": 0.21489699184894562, "learning_rate": 9.402604496344984e-05, "loss": 0.0464, "step": 11960 }, { "epoch": 5.356903110315507, "grad_norm": 0.20587298274040222, "learning_rate": 9.401297566366318e-05, "loss": 0.0495, "step": 11970 }, { "epoch": 5.361378384426046, "grad_norm": 0.25509634613990784, "learning_rate": 9.399989299389661e-05, "loss": 0.0565, "step": 11980 }, { "epoch": 5.365853658536586, "grad_norm": 0.24807193875312805, "learning_rate": 9.39867969581243e-05, "loss": 0.055, "step": 11990 }, { "epoch": 5.370328932647125, "grad_norm": 0.19540193676948547, "learning_rate": 9.397368756032445e-05, "loss": 0.046, "step": 12000 }, { "epoch": 5.374804206757664, "grad_norm": 0.1912635713815689, "learning_rate": 9.39605648044794e-05, "loss": 0.0527, "step": 12010 }, { "epoch": 5.379279480868203, "grad_norm": 0.20776204764842987, "learning_rate": 9.394742869457547e-05, "loss": 0.0469, "step": 12020 }, { "epoch": 5.383754754978742, "grad_norm": 0.20063914358615875, "learning_rate": 9.393427923460308e-05, "loss": 0.0498, "step": 12030 }, { "epoch": 5.388230029089282, "grad_norm": 0.2091071903705597, "learning_rate": 9.392111642855665e-05, "loss": 0.0492, "step": 12040 }, { "epoch": 5.392705303199821, "grad_norm": 0.20866911113262177, "learning_rate": 9.390794028043474e-05, "loss": 0.0504, "step": 12050 }, { "epoch": 5.39718057731036, "grad_norm": 0.19992321729660034, "learning_rate": 9.389475079423988e-05, "loss": 0.0479, "step": 12060 }, { "epoch": 5.4016558514209, "grad_norm": 0.1831560581922531, "learning_rate": 9.388154797397871e-05, "loss": 0.0508, "step": 12070 }, { "epoch": 5.406131125531439, "grad_norm": 0.17890678346157074, "learning_rate": 9.38683318236619e-05, "loss": 0.0479, "step": 12080 }, { "epoch": 5.410606399641978, "grad_norm": 0.2477792501449585, "learning_rate": 9.385510234730415e-05, "loss": 0.0522, "step": 12090 }, { "epoch": 5.415081673752518, "grad_norm": 0.22895941138267517, "learning_rate": 9.384185954892422e-05, "loss": 0.0453, "step": 12100 }, { "epoch": 5.4195569478630565, "grad_norm": 0.19731943309307098, "learning_rate": 9.382860343254496e-05, "loss": 0.0499, "step": 12110 }, { "epoch": 5.4240322219735955, "grad_norm": 0.19517767429351807, "learning_rate": 9.381533400219318e-05, "loss": 0.0463, "step": 12120 }, { "epoch": 5.428507496084135, "grad_norm": 0.2206287980079651, "learning_rate": 9.380205126189983e-05, "loss": 0.0477, "step": 12130 }, { "epoch": 5.432982770194674, "grad_norm": 0.2075619250535965, "learning_rate": 9.378875521569981e-05, "loss": 0.0509, "step": 12140 }, { "epoch": 5.437458044305214, "grad_norm": 0.22856418788433075, "learning_rate": 9.377544586763215e-05, "loss": 0.0543, "step": 12150 }, { "epoch": 5.441933318415753, "grad_norm": 0.21619604527950287, "learning_rate": 9.376212322173985e-05, "loss": 0.053, "step": 12160 }, { "epoch": 5.446408592526292, "grad_norm": 0.21785415709018707, "learning_rate": 9.374878728206999e-05, "loss": 0.0508, "step": 12170 }, { "epoch": 5.450883866636832, "grad_norm": 0.20401157438755035, "learning_rate": 9.373543805267368e-05, "loss": 0.0512, "step": 12180 }, { "epoch": 5.455359140747371, "grad_norm": 0.21712632477283478, "learning_rate": 9.372207553760603e-05, "loss": 0.0491, "step": 12190 }, { "epoch": 5.45983441485791, "grad_norm": 0.21762795746326447, "learning_rate": 9.370869974092629e-05, "loss": 0.0526, "step": 12200 }, { "epoch": 5.4643096889684495, "grad_norm": 0.23712295293807983, "learning_rate": 9.369531066669758e-05, "loss": 0.0514, "step": 12210 }, { "epoch": 5.468784963078988, "grad_norm": 0.19422072172164917, "learning_rate": 9.368190831898724e-05, "loss": 0.0516, "step": 12220 }, { "epoch": 5.473260237189528, "grad_norm": 0.255489319562912, "learning_rate": 9.366849270186649e-05, "loss": 0.0514, "step": 12230 }, { "epoch": 5.477735511300067, "grad_norm": 0.21552325785160065, "learning_rate": 9.365506381941066e-05, "loss": 0.0494, "step": 12240 }, { "epoch": 5.482210785410606, "grad_norm": 0.2300003170967102, "learning_rate": 9.364162167569907e-05, "loss": 0.0474, "step": 12250 }, { "epoch": 5.486686059521146, "grad_norm": 0.24377156794071198, "learning_rate": 9.362816627481512e-05, "loss": 0.0469, "step": 12260 }, { "epoch": 5.491161333631685, "grad_norm": 0.2148071527481079, "learning_rate": 9.36146976208462e-05, "loss": 0.0484, "step": 12270 }, { "epoch": 5.495636607742224, "grad_norm": 0.20738567411899567, "learning_rate": 9.360121571788371e-05, "loss": 0.0484, "step": 12280 }, { "epoch": 5.500111881852764, "grad_norm": 0.2366109937429428, "learning_rate": 9.358772057002312e-05, "loss": 0.0472, "step": 12290 }, { "epoch": 5.504587155963303, "grad_norm": 0.1862984597682953, "learning_rate": 9.357421218136386e-05, "loss": 0.0488, "step": 12300 }, { "epoch": 5.5090624300738416, "grad_norm": 0.2156638354063034, "learning_rate": 9.356069055600948e-05, "loss": 0.0484, "step": 12310 }, { "epoch": 5.513537704184381, "grad_norm": 0.22329257428646088, "learning_rate": 9.354715569806744e-05, "loss": 0.0479, "step": 12320 }, { "epoch": 5.51801297829492, "grad_norm": 0.2129819095134735, "learning_rate": 9.353360761164931e-05, "loss": 0.0475, "step": 12330 }, { "epoch": 5.52248825240546, "grad_norm": 0.23148798942565918, "learning_rate": 9.352004630087062e-05, "loss": 0.042, "step": 12340 }, { "epoch": 5.526963526515999, "grad_norm": 0.3115878105163574, "learning_rate": 9.350647176985095e-05, "loss": 0.0495, "step": 12350 }, { "epoch": 5.531438800626538, "grad_norm": 0.22075380384922028, "learning_rate": 9.349288402271388e-05, "loss": 0.046, "step": 12360 }, { "epoch": 5.535914074737078, "grad_norm": 0.24007263779640198, "learning_rate": 9.347928306358699e-05, "loss": 0.0476, "step": 12370 }, { "epoch": 5.540389348847617, "grad_norm": 0.21069885790348053, "learning_rate": 9.346566889660193e-05, "loss": 0.0469, "step": 12380 }, { "epoch": 5.544864622958157, "grad_norm": 0.19413995742797852, "learning_rate": 9.345204152589428e-05, "loss": 0.0515, "step": 12390 }, { "epoch": 5.549339897068696, "grad_norm": 0.1936689019203186, "learning_rate": 9.343840095560372e-05, "loss": 0.0523, "step": 12400 }, { "epoch": 5.5538151711792345, "grad_norm": 0.20570313930511475, "learning_rate": 9.342474718987386e-05, "loss": 0.0501, "step": 12410 }, { "epoch": 5.558290445289774, "grad_norm": 0.2430773377418518, "learning_rate": 9.341108023285238e-05, "loss": 0.0497, "step": 12420 }, { "epoch": 5.562765719400313, "grad_norm": 0.23150259256362915, "learning_rate": 9.339740008869092e-05, "loss": 0.0546, "step": 12430 }, { "epoch": 5.567240993510852, "grad_norm": 0.22070106863975525, "learning_rate": 9.338370676154516e-05, "loss": 0.0471, "step": 12440 }, { "epoch": 5.571716267621392, "grad_norm": 0.2445363700389862, "learning_rate": 9.337000025557476e-05, "loss": 0.0496, "step": 12450 }, { "epoch": 5.576191541731931, "grad_norm": 0.2273276448249817, "learning_rate": 9.335628057494341e-05, "loss": 0.055, "step": 12460 }, { "epoch": 5.58066681584247, "grad_norm": 0.22906428575515747, "learning_rate": 9.334254772381876e-05, "loss": 0.0498, "step": 12470 }, { "epoch": 5.58514208995301, "grad_norm": 0.21529696881771088, "learning_rate": 9.332880170637252e-05, "loss": 0.0501, "step": 12480 }, { "epoch": 5.589617364063549, "grad_norm": 0.2760084271430969, "learning_rate": 9.331504252678037e-05, "loss": 0.0492, "step": 12490 }, { "epoch": 5.5940926381740885, "grad_norm": 0.26806485652923584, "learning_rate": 9.330127018922194e-05, "loss": 0.0491, "step": 12500 }, { "epoch": 5.5985679122846275, "grad_norm": 0.20341867208480835, "learning_rate": 9.328748469788093e-05, "loss": 0.0491, "step": 12510 }, { "epoch": 5.603043186395166, "grad_norm": 0.22836028039455414, "learning_rate": 9.327368605694502e-05, "loss": 0.0526, "step": 12520 }, { "epoch": 5.607518460505706, "grad_norm": 0.2312297374010086, "learning_rate": 9.325987427060586e-05, "loss": 0.0501, "step": 12530 }, { "epoch": 5.611993734616245, "grad_norm": 0.20873434841632843, "learning_rate": 9.32460493430591e-05, "loss": 0.0486, "step": 12540 }, { "epoch": 5.616469008726785, "grad_norm": 0.16824068129062653, "learning_rate": 9.323221127850441e-05, "loss": 0.0485, "step": 12550 }, { "epoch": 5.620944282837324, "grad_norm": 0.22477909922599792, "learning_rate": 9.321836008114539e-05, "loss": 0.0484, "step": 12560 }, { "epoch": 5.625419556947863, "grad_norm": 0.23175379633903503, "learning_rate": 9.320449575518972e-05, "loss": 0.0495, "step": 12570 }, { "epoch": 5.629894831058403, "grad_norm": 0.23248590528964996, "learning_rate": 9.319061830484898e-05, "loss": 0.0488, "step": 12580 }, { "epoch": 5.634370105168942, "grad_norm": 0.19837740063667297, "learning_rate": 9.317672773433876e-05, "loss": 0.0509, "step": 12590 }, { "epoch": 5.638845379279481, "grad_norm": 0.21171174943447113, "learning_rate": 9.316282404787871e-05, "loss": 0.0463, "step": 12600 }, { "epoch": 5.64332065339002, "grad_norm": 0.22161062061786652, "learning_rate": 9.314890724969232e-05, "loss": 0.0495, "step": 12610 }, { "epoch": 5.647795927500559, "grad_norm": 0.20274871587753296, "learning_rate": 9.313497734400722e-05, "loss": 0.0467, "step": 12620 }, { "epoch": 5.652271201611098, "grad_norm": 0.18073520064353943, "learning_rate": 9.31210343350549e-05, "loss": 0.0509, "step": 12630 }, { "epoch": 5.656746475721638, "grad_norm": 0.1822454035282135, "learning_rate": 9.31070782270709e-05, "loss": 0.0483, "step": 12640 }, { "epoch": 5.661221749832177, "grad_norm": 0.2143086940050125, "learning_rate": 9.309310902429472e-05, "loss": 0.0456, "step": 12650 }, { "epoch": 5.665697023942716, "grad_norm": 0.21196234226226807, "learning_rate": 9.30791267309698e-05, "loss": 0.0483, "step": 12660 }, { "epoch": 5.670172298053256, "grad_norm": 0.22483086585998535, "learning_rate": 9.306513135134362e-05, "loss": 0.051, "step": 12670 }, { "epoch": 5.674647572163795, "grad_norm": 0.23647232353687286, "learning_rate": 9.305112288966761e-05, "loss": 0.0472, "step": 12680 }, { "epoch": 5.679122846274335, "grad_norm": 0.19354937970638275, "learning_rate": 9.30371013501972e-05, "loss": 0.0497, "step": 12690 }, { "epoch": 5.6835981203848736, "grad_norm": 0.1974339336156845, "learning_rate": 9.30230667371917e-05, "loss": 0.0456, "step": 12700 }, { "epoch": 5.6880733944954125, "grad_norm": 0.22961778938770294, "learning_rate": 9.300901905491447e-05, "loss": 0.0543, "step": 12710 }, { "epoch": 5.692548668605952, "grad_norm": 0.24553552269935608, "learning_rate": 9.299495830763286e-05, "loss": 0.0471, "step": 12720 }, { "epoch": 5.697023942716491, "grad_norm": 0.19014185667037964, "learning_rate": 9.298088449961813e-05, "loss": 0.0484, "step": 12730 }, { "epoch": 5.701499216827031, "grad_norm": 0.21002914011478424, "learning_rate": 9.296679763514552e-05, "loss": 0.0526, "step": 12740 }, { "epoch": 5.70597449093757, "grad_norm": 0.2061125636100769, "learning_rate": 9.295269771849427e-05, "loss": 0.0517, "step": 12750 }, { "epoch": 5.710449765048109, "grad_norm": 0.19812734425067902, "learning_rate": 9.293858475394754e-05, "loss": 0.0481, "step": 12760 }, { "epoch": 5.714925039158649, "grad_norm": 0.1664070188999176, "learning_rate": 9.292445874579251e-05, "loss": 0.049, "step": 12770 }, { "epoch": 5.719400313269188, "grad_norm": 0.2155786156654358, "learning_rate": 9.291031969832026e-05, "loss": 0.0487, "step": 12780 }, { "epoch": 5.723875587379727, "grad_norm": 0.2102098912000656, "learning_rate": 9.289616761582587e-05, "loss": 0.0505, "step": 12790 }, { "epoch": 5.7283508614902665, "grad_norm": 0.1981678307056427, "learning_rate": 9.288200250260836e-05, "loss": 0.0514, "step": 12800 }, { "epoch": 5.7328261356008055, "grad_norm": 0.1859380155801773, "learning_rate": 9.286782436297073e-05, "loss": 0.0456, "step": 12810 }, { "epoch": 5.737301409711344, "grad_norm": 0.1982182413339615, "learning_rate": 9.285363320121992e-05, "loss": 0.0465, "step": 12820 }, { "epoch": 5.741776683821884, "grad_norm": 0.21469788253307343, "learning_rate": 9.283942902166681e-05, "loss": 0.0466, "step": 12830 }, { "epoch": 5.746251957932423, "grad_norm": 0.18780435621738434, "learning_rate": 9.282521182862629e-05, "loss": 0.0504, "step": 12840 }, { "epoch": 5.750727232042963, "grad_norm": 0.22261862456798553, "learning_rate": 9.281098162641714e-05, "loss": 0.0484, "step": 12850 }, { "epoch": 5.755202506153502, "grad_norm": 0.23778130114078522, "learning_rate": 9.279673841936214e-05, "loss": 0.0489, "step": 12860 }, { "epoch": 5.759677780264041, "grad_norm": 0.209798663854599, "learning_rate": 9.278248221178798e-05, "loss": 0.0484, "step": 12870 }, { "epoch": 5.764153054374581, "grad_norm": 0.174835667014122, "learning_rate": 9.276821300802534e-05, "loss": 0.0458, "step": 12880 }, { "epoch": 5.76862832848512, "grad_norm": 0.19941803812980652, "learning_rate": 9.275393081240882e-05, "loss": 0.0471, "step": 12890 }, { "epoch": 5.7731036025956595, "grad_norm": 0.2158508151769638, "learning_rate": 9.273963562927695e-05, "loss": 0.053, "step": 12900 }, { "epoch": 5.777578876706198, "grad_norm": 0.22232221066951752, "learning_rate": 9.272532746297227e-05, "loss": 0.047, "step": 12910 }, { "epoch": 5.782054150816737, "grad_norm": 0.1855051964521408, "learning_rate": 9.27110063178412e-05, "loss": 0.0492, "step": 12920 }, { "epoch": 5.786529424927277, "grad_norm": 0.21868827939033508, "learning_rate": 9.269667219823412e-05, "loss": 0.0483, "step": 12930 }, { "epoch": 5.791004699037816, "grad_norm": 0.17684322595596313, "learning_rate": 9.268232510850539e-05, "loss": 0.0482, "step": 12940 }, { "epoch": 5.795479973148355, "grad_norm": 0.22365310788154602, "learning_rate": 9.266796505301322e-05, "loss": 0.0486, "step": 12950 }, { "epoch": 5.799955247258895, "grad_norm": 0.18967099487781525, "learning_rate": 9.265359203611987e-05, "loss": 0.0456, "step": 12960 }, { "epoch": 5.804430521369434, "grad_norm": 0.25677669048309326, "learning_rate": 9.263920606219147e-05, "loss": 0.0487, "step": 12970 }, { "epoch": 5.808905795479973, "grad_norm": 0.19891799986362457, "learning_rate": 9.262480713559808e-05, "loss": 0.0455, "step": 12980 }, { "epoch": 5.813381069590513, "grad_norm": 0.23020821809768677, "learning_rate": 9.261039526071374e-05, "loss": 0.0495, "step": 12990 }, { "epoch": 5.8178563437010515, "grad_norm": 0.2282811552286148, "learning_rate": 9.259597044191636e-05, "loss": 0.0496, "step": 13000 }, { "epoch": 5.8223316178115905, "grad_norm": 0.19122476875782013, "learning_rate": 9.258153268358785e-05, "loss": 0.0495, "step": 13010 }, { "epoch": 5.82680689192213, "grad_norm": 0.19568875432014465, "learning_rate": 9.256708199011401e-05, "loss": 0.0485, "step": 13020 }, { "epoch": 5.831282166032669, "grad_norm": 0.2183283418416977, "learning_rate": 9.255261836588458e-05, "loss": 0.0502, "step": 13030 }, { "epoch": 5.835757440143209, "grad_norm": 0.2324369102716446, "learning_rate": 9.253814181529323e-05, "loss": 0.0473, "step": 13040 }, { "epoch": 5.840232714253748, "grad_norm": 0.21526357531547546, "learning_rate": 9.252365234273755e-05, "loss": 0.0463, "step": 13050 }, { "epoch": 5.844707988364287, "grad_norm": 0.23216550052165985, "learning_rate": 9.250914995261905e-05, "loss": 0.0545, "step": 13060 }, { "epoch": 5.849183262474827, "grad_norm": 0.19618897140026093, "learning_rate": 9.249463464934321e-05, "loss": 0.044, "step": 13070 }, { "epoch": 5.853658536585366, "grad_norm": 0.19227701425552368, "learning_rate": 9.248010643731935e-05, "loss": 0.0471, "step": 13080 }, { "epoch": 5.8581338106959056, "grad_norm": 0.22573326528072357, "learning_rate": 9.246556532096078e-05, "loss": 0.047, "step": 13090 }, { "epoch": 5.8626090848064445, "grad_norm": 0.19718167185783386, "learning_rate": 9.24510113046847e-05, "loss": 0.0486, "step": 13100 }, { "epoch": 5.867084358916983, "grad_norm": 0.23797045648097992, "learning_rate": 9.243644439291223e-05, "loss": 0.0494, "step": 13110 }, { "epoch": 5.871559633027523, "grad_norm": 0.22680717706680298, "learning_rate": 9.242186459006845e-05, "loss": 0.0477, "step": 13120 }, { "epoch": 5.876034907138062, "grad_norm": 0.1806063950061798, "learning_rate": 9.240727190058227e-05, "loss": 0.0518, "step": 13130 }, { "epoch": 5.880510181248601, "grad_norm": 0.20738841593265533, "learning_rate": 9.239266632888659e-05, "loss": 0.0491, "step": 13140 }, { "epoch": 5.884985455359141, "grad_norm": 0.2181163728237152, "learning_rate": 9.237804787941819e-05, "loss": 0.0453, "step": 13150 }, { "epoch": 5.88946072946968, "grad_norm": 0.2038734257221222, "learning_rate": 9.236341655661778e-05, "loss": 0.0523, "step": 13160 }, { "epoch": 5.893936003580219, "grad_norm": 0.24431608617305756, "learning_rate": 9.234877236492997e-05, "loss": 0.0535, "step": 13170 }, { "epoch": 5.898411277690759, "grad_norm": 0.24287256598472595, "learning_rate": 9.233411530880326e-05, "loss": 0.0468, "step": 13180 }, { "epoch": 5.902886551801298, "grad_norm": 0.18742318451404572, "learning_rate": 9.231944539269009e-05, "loss": 0.0503, "step": 13190 }, { "epoch": 5.9073618259118374, "grad_norm": 0.2022157609462738, "learning_rate": 9.230476262104677e-05, "loss": 0.0526, "step": 13200 }, { "epoch": 5.911837100022376, "grad_norm": 0.2031514048576355, "learning_rate": 9.229006699833358e-05, "loss": 0.0442, "step": 13210 }, { "epoch": 5.916312374132915, "grad_norm": 0.2233230471611023, "learning_rate": 9.227535852901463e-05, "loss": 0.0488, "step": 13220 }, { "epoch": 5.920787648243455, "grad_norm": 0.20201826095581055, "learning_rate": 9.226063721755799e-05, "loss": 0.0452, "step": 13230 }, { "epoch": 5.925262922353994, "grad_norm": 0.21156004071235657, "learning_rate": 9.224590306843558e-05, "loss": 0.0529, "step": 13240 }, { "epoch": 5.929738196464534, "grad_norm": 0.24461151659488678, "learning_rate": 9.223115608612325e-05, "loss": 0.0493, "step": 13250 }, { "epoch": 5.934213470575073, "grad_norm": 0.2412540167570114, "learning_rate": 9.221639627510076e-05, "loss": 0.0485, "step": 13260 }, { "epoch": 5.938688744685612, "grad_norm": 0.1721092313528061, "learning_rate": 9.220162363985174e-05, "loss": 0.0469, "step": 13270 }, { "epoch": 5.943164018796152, "grad_norm": 0.2022211253643036, "learning_rate": 9.218683818486372e-05, "loss": 0.0473, "step": 13280 }, { "epoch": 5.947639292906691, "grad_norm": 0.20367282629013062, "learning_rate": 9.217203991462815e-05, "loss": 0.0548, "step": 13290 }, { "epoch": 5.9521145670172295, "grad_norm": 0.21488544344902039, "learning_rate": 9.215722883364033e-05, "loss": 0.0501, "step": 13300 }, { "epoch": 5.956589841127769, "grad_norm": 0.1887718290090561, "learning_rate": 9.214240494639949e-05, "loss": 0.0444, "step": 13310 }, { "epoch": 5.961065115238308, "grad_norm": 0.18350791931152344, "learning_rate": 9.212756825740873e-05, "loss": 0.0503, "step": 13320 }, { "epoch": 5.965540389348847, "grad_norm": 0.17065678536891937, "learning_rate": 9.211271877117507e-05, "loss": 0.0496, "step": 13330 }, { "epoch": 5.970015663459387, "grad_norm": 0.19211028516292572, "learning_rate": 9.209785649220935e-05, "loss": 0.0474, "step": 13340 }, { "epoch": 5.974490937569926, "grad_norm": 0.18783444166183472, "learning_rate": 9.208298142502636e-05, "loss": 0.048, "step": 13350 }, { "epoch": 5.978966211680465, "grad_norm": 0.22281689941883087, "learning_rate": 9.206809357414474e-05, "loss": 0.0467, "step": 13360 }, { "epoch": 5.983441485791005, "grad_norm": 0.2635248303413391, "learning_rate": 9.205319294408705e-05, "loss": 0.0501, "step": 13370 }, { "epoch": 5.987916759901544, "grad_norm": 0.18985670804977417, "learning_rate": 9.20382795393797e-05, "loss": 0.0491, "step": 13380 }, { "epoch": 5.9923920340120835, "grad_norm": 0.1898002177476883, "learning_rate": 9.202335336455296e-05, "loss": 0.0484, "step": 13390 }, { "epoch": 5.9968673081226225, "grad_norm": 0.227137491106987, "learning_rate": 9.200841442414106e-05, "loss": 0.0544, "step": 13400 }, { "epoch": 6.001342582233161, "grad_norm": 0.21438324451446533, "learning_rate": 9.199346272268199e-05, "loss": 0.0442, "step": 13410 }, { "epoch": 6.005817856343701, "grad_norm": 0.20868484675884247, "learning_rate": 9.197849826471774e-05, "loss": 0.0469, "step": 13420 }, { "epoch": 6.01029313045424, "grad_norm": 0.17521192133426666, "learning_rate": 9.196352105479409e-05, "loss": 0.0473, "step": 13430 }, { "epoch": 6.01476840456478, "grad_norm": 0.17984789609909058, "learning_rate": 9.194853109746074e-05, "loss": 0.0479, "step": 13440 }, { "epoch": 6.019243678675319, "grad_norm": 0.23771309852600098, "learning_rate": 9.193352839727121e-05, "loss": 0.0477, "step": 13450 }, { "epoch": 6.023718952785858, "grad_norm": 0.22797030210494995, "learning_rate": 9.191851295878295e-05, "loss": 0.0448, "step": 13460 }, { "epoch": 6.028194226896398, "grad_norm": 0.2085992991924286, "learning_rate": 9.190348478655724e-05, "loss": 0.044, "step": 13470 }, { "epoch": 6.032669501006937, "grad_norm": 0.19927796721458435, "learning_rate": 9.188844388515926e-05, "loss": 0.0466, "step": 13480 }, { "epoch": 6.037144775117476, "grad_norm": 0.20748595893383026, "learning_rate": 9.187339025915802e-05, "loss": 0.0476, "step": 13490 }, { "epoch": 6.041620049228015, "grad_norm": 0.25835174322128296, "learning_rate": 9.185832391312644e-05, "loss": 0.0487, "step": 13500 }, { "epoch": 6.046095323338554, "grad_norm": 0.20990467071533203, "learning_rate": 9.184324485164124e-05, "loss": 0.0443, "step": 13510 }, { "epoch": 6.050570597449093, "grad_norm": 0.18215397000312805, "learning_rate": 9.182815307928307e-05, "loss": 0.0454, "step": 13520 }, { "epoch": 6.055045871559633, "grad_norm": 0.19444115459918976, "learning_rate": 9.18130486006364e-05, "loss": 0.0474, "step": 13530 }, { "epoch": 6.059521145670172, "grad_norm": 0.2503259479999542, "learning_rate": 9.179793142028959e-05, "loss": 0.0532, "step": 13540 }, { "epoch": 6.063996419780712, "grad_norm": 0.20335064828395844, "learning_rate": 9.17828015428348e-05, "loss": 0.0436, "step": 13550 }, { "epoch": 6.068471693891251, "grad_norm": 0.16907264292240143, "learning_rate": 9.176765897286813e-05, "loss": 0.0469, "step": 13560 }, { "epoch": 6.07294696800179, "grad_norm": 0.19459868967533112, "learning_rate": 9.175250371498946e-05, "loss": 0.0464, "step": 13570 }, { "epoch": 6.07742224211233, "grad_norm": 0.20377546548843384, "learning_rate": 9.173733577380258e-05, "loss": 0.042, "step": 13580 }, { "epoch": 6.081897516222869, "grad_norm": 0.25805944204330444, "learning_rate": 9.17221551539151e-05, "loss": 0.0496, "step": 13590 }, { "epoch": 6.0863727903334075, "grad_norm": 0.21011966466903687, "learning_rate": 9.17069618599385e-05, "loss": 0.0483, "step": 13600 }, { "epoch": 6.090848064443947, "grad_norm": 0.22706210613250732, "learning_rate": 9.169175589648809e-05, "loss": 0.0466, "step": 13610 }, { "epoch": 6.095323338554486, "grad_norm": 0.21207641065120697, "learning_rate": 9.167653726818305e-05, "loss": 0.0454, "step": 13620 }, { "epoch": 6.099798612665026, "grad_norm": 0.16991089284420013, "learning_rate": 9.16613059796464e-05, "loss": 0.0491, "step": 13630 }, { "epoch": 6.104273886775565, "grad_norm": 0.1876649558544159, "learning_rate": 9.164606203550497e-05, "loss": 0.0462, "step": 13640 }, { "epoch": 6.108749160886104, "grad_norm": 0.21167807281017303, "learning_rate": 9.163080544038952e-05, "loss": 0.0493, "step": 13650 }, { "epoch": 6.113224434996644, "grad_norm": 0.2512698471546173, "learning_rate": 9.161553619893457e-05, "loss": 0.0486, "step": 13660 }, { "epoch": 6.117699709107183, "grad_norm": 0.23190931975841522, "learning_rate": 9.160025431577851e-05, "loss": 0.0488, "step": 13670 }, { "epoch": 6.122174983217722, "grad_norm": 0.1732003390789032, "learning_rate": 9.158495979556358e-05, "loss": 0.0444, "step": 13680 }, { "epoch": 6.1266502573282615, "grad_norm": 0.19667194783687592, "learning_rate": 9.156965264293586e-05, "loss": 0.0423, "step": 13690 }, { "epoch": 6.1311255314388005, "grad_norm": 0.2102915495634079, "learning_rate": 9.155433286254525e-05, "loss": 0.0435, "step": 13700 }, { "epoch": 6.13560080554934, "grad_norm": 0.203181192278862, "learning_rate": 9.153900045904549e-05, "loss": 0.0433, "step": 13710 }, { "epoch": 6.140076079659879, "grad_norm": 0.20502960681915283, "learning_rate": 9.152365543709416e-05, "loss": 0.0473, "step": 13720 }, { "epoch": 6.144551353770418, "grad_norm": 0.1968383640050888, "learning_rate": 9.150829780135269e-05, "loss": 0.0517, "step": 13730 }, { "epoch": 6.149026627880958, "grad_norm": 0.22402982413768768, "learning_rate": 9.14929275564863e-05, "loss": 0.0423, "step": 13740 }, { "epoch": 6.153501901991497, "grad_norm": 0.18875516951084137, "learning_rate": 9.147754470716408e-05, "loss": 0.0428, "step": 13750 }, { "epoch": 6.157977176102036, "grad_norm": 0.21956737339496613, "learning_rate": 9.146214925805891e-05, "loss": 0.0505, "step": 13760 }, { "epoch": 6.162452450212576, "grad_norm": 0.18030880391597748, "learning_rate": 9.144674121384757e-05, "loss": 0.0446, "step": 13770 }, { "epoch": 6.166927724323115, "grad_norm": 0.232269287109375, "learning_rate": 9.143132057921058e-05, "loss": 0.0507, "step": 13780 }, { "epoch": 6.1714029984336545, "grad_norm": 0.19155369699001312, "learning_rate": 9.141588735883232e-05, "loss": 0.0455, "step": 13790 }, { "epoch": 6.175878272544193, "grad_norm": 0.1853640079498291, "learning_rate": 9.140044155740101e-05, "loss": 0.0452, "step": 13800 }, { "epoch": 6.180353546654732, "grad_norm": 0.19158171117305756, "learning_rate": 9.138498317960867e-05, "loss": 0.0474, "step": 13810 }, { "epoch": 6.184828820765272, "grad_norm": 0.17654862999916077, "learning_rate": 9.136951223015113e-05, "loss": 0.049, "step": 13820 }, { "epoch": 6.189304094875811, "grad_norm": 0.18133533000946045, "learning_rate": 9.135402871372808e-05, "loss": 0.0477, "step": 13830 }, { "epoch": 6.19377936898635, "grad_norm": 0.2263277769088745, "learning_rate": 9.133853263504302e-05, "loss": 0.047, "step": 13840 }, { "epoch": 6.19825464309689, "grad_norm": 0.1842632293701172, "learning_rate": 9.132302399880321e-05, "loss": 0.047, "step": 13850 }, { "epoch": 6.202729917207429, "grad_norm": 0.2118486911058426, "learning_rate": 9.130750280971978e-05, "loss": 0.0523, "step": 13860 }, { "epoch": 6.207205191317968, "grad_norm": 0.19419611990451813, "learning_rate": 9.129196907250765e-05, "loss": 0.0438, "step": 13870 }, { "epoch": 6.211680465428508, "grad_norm": 0.18961367011070251, "learning_rate": 9.127642279188558e-05, "loss": 0.049, "step": 13880 }, { "epoch": 6.2161557395390465, "grad_norm": 0.19554993510246277, "learning_rate": 9.126086397257612e-05, "loss": 0.045, "step": 13890 }, { "epoch": 6.220631013649586, "grad_norm": 0.19207070767879486, "learning_rate": 9.124529261930559e-05, "loss": 0.0512, "step": 13900 }, { "epoch": 6.225106287760125, "grad_norm": 0.17535705864429474, "learning_rate": 9.122970873680419e-05, "loss": 0.0505, "step": 13910 }, { "epoch": 6.229581561870664, "grad_norm": 0.2042308896780014, "learning_rate": 9.121411232980588e-05, "loss": 0.0447, "step": 13920 }, { "epoch": 6.234056835981204, "grad_norm": 0.1869090348482132, "learning_rate": 9.119850340304843e-05, "loss": 0.0437, "step": 13930 }, { "epoch": 6.238532110091743, "grad_norm": 0.19733303785324097, "learning_rate": 9.118288196127345e-05, "loss": 0.0443, "step": 13940 }, { "epoch": 6.243007384202282, "grad_norm": 0.2057299166917801, "learning_rate": 9.116724800922629e-05, "loss": 0.05, "step": 13950 }, { "epoch": 6.247482658312822, "grad_norm": 0.22464431822299957, "learning_rate": 9.115160155165614e-05, "loss": 0.0512, "step": 13960 }, { "epoch": 6.251957932423361, "grad_norm": 0.2474597841501236, "learning_rate": 9.1135942593316e-05, "loss": 0.0437, "step": 13970 }, { "epoch": 6.256433206533901, "grad_norm": 0.2033524364233017, "learning_rate": 9.112027113896262e-05, "loss": 0.0478, "step": 13980 }, { "epoch": 6.2609084806444395, "grad_norm": 0.14931447803974152, "learning_rate": 9.110458719335659e-05, "loss": 0.044, "step": 13990 }, { "epoch": 6.265383754754978, "grad_norm": 0.16722825169563293, "learning_rate": 9.108889076126226e-05, "loss": 0.0497, "step": 14000 }, { "epoch": 6.269859028865518, "grad_norm": 0.22637279331684113, "learning_rate": 9.107318184744781e-05, "loss": 0.0482, "step": 14010 }, { "epoch": 6.274334302976057, "grad_norm": 0.2296632081270218, "learning_rate": 9.105746045668521e-05, "loss": 0.0474, "step": 14020 }, { "epoch": 6.278809577086596, "grad_norm": 0.1968671679496765, "learning_rate": 9.104172659375017e-05, "loss": 0.0486, "step": 14030 }, { "epoch": 6.283284851197136, "grad_norm": 0.1741342842578888, "learning_rate": 9.102598026342222e-05, "loss": 0.0478, "step": 14040 }, { "epoch": 6.287760125307675, "grad_norm": 0.15113016963005066, "learning_rate": 9.101022147048473e-05, "loss": 0.0494, "step": 14050 }, { "epoch": 6.292235399418215, "grad_norm": 0.17822885513305664, "learning_rate": 9.099445021972473e-05, "loss": 0.0504, "step": 14060 }, { "epoch": 6.296710673528754, "grad_norm": 0.15728485584259033, "learning_rate": 9.097866651593317e-05, "loss": 0.049, "step": 14070 }, { "epoch": 6.301185947639293, "grad_norm": 0.19928182661533356, "learning_rate": 9.09628703639047e-05, "loss": 0.0471, "step": 14080 }, { "epoch": 6.3056612217498325, "grad_norm": 0.20045238733291626, "learning_rate": 9.094706176843777e-05, "loss": 0.0479, "step": 14090 }, { "epoch": 6.310136495860371, "grad_norm": 0.19196805357933044, "learning_rate": 9.093124073433463e-05, "loss": 0.0421, "step": 14100 }, { "epoch": 6.31461176997091, "grad_norm": 0.1525011658668518, "learning_rate": 9.091540726640126e-05, "loss": 0.0446, "step": 14110 }, { "epoch": 6.31908704408145, "grad_norm": 0.21144013106822968, "learning_rate": 9.089956136944751e-05, "loss": 0.0438, "step": 14120 }, { "epoch": 6.323562318191989, "grad_norm": 0.22392235696315765, "learning_rate": 9.088370304828685e-05, "loss": 0.0523, "step": 14130 }, { "epoch": 6.328037592302529, "grad_norm": 0.1992444097995758, "learning_rate": 9.086783230773672e-05, "loss": 0.0457, "step": 14140 }, { "epoch": 6.332512866413068, "grad_norm": 0.13526609539985657, "learning_rate": 9.085194915261818e-05, "loss": 0.0458, "step": 14150 }, { "epoch": 6.336988140523607, "grad_norm": 0.20806245505809784, "learning_rate": 9.083605358775612e-05, "loss": 0.0503, "step": 14160 }, { "epoch": 6.341463414634147, "grad_norm": 0.1874699592590332, "learning_rate": 9.082014561797918e-05, "loss": 0.0472, "step": 14170 }, { "epoch": 6.345938688744686, "grad_norm": 0.2045682817697525, "learning_rate": 9.080422524811982e-05, "loss": 0.0451, "step": 14180 }, { "epoch": 6.3504139628552245, "grad_norm": 0.1658376008272171, "learning_rate": 9.078829248301417e-05, "loss": 0.0436, "step": 14190 }, { "epoch": 6.354889236965764, "grad_norm": 0.17986880242824554, "learning_rate": 9.077234732750224e-05, "loss": 0.0436, "step": 14200 }, { "epoch": 6.359364511076303, "grad_norm": 0.18365758657455444, "learning_rate": 9.075638978642771e-05, "loss": 0.0434, "step": 14210 }, { "epoch": 6.363839785186842, "grad_norm": 0.20398174226284027, "learning_rate": 9.074041986463808e-05, "loss": 0.0449, "step": 14220 }, { "epoch": 6.368315059297382, "grad_norm": 0.1832636594772339, "learning_rate": 9.072443756698459e-05, "loss": 0.0427, "step": 14230 }, { "epoch": 6.372790333407921, "grad_norm": 0.17306652665138245, "learning_rate": 9.070844289832224e-05, "loss": 0.0472, "step": 14240 }, { "epoch": 6.377265607518461, "grad_norm": 0.197926327586174, "learning_rate": 9.069243586350975e-05, "loss": 0.0473, "step": 14250 }, { "epoch": 6.381740881629, "grad_norm": 0.1675863415002823, "learning_rate": 9.067641646740968e-05, "loss": 0.0433, "step": 14260 }, { "epoch": 6.386216155739539, "grad_norm": 0.1979970782995224, "learning_rate": 9.066038471488829e-05, "loss": 0.043, "step": 14270 }, { "epoch": 6.3906914298500785, "grad_norm": 0.1851215362548828, "learning_rate": 9.064434061081562e-05, "loss": 0.0445, "step": 14280 }, { "epoch": 6.3951667039606175, "grad_norm": 0.214791938662529, "learning_rate": 9.062828416006539e-05, "loss": 0.05, "step": 14290 }, { "epoch": 6.399641978071157, "grad_norm": 0.20266325771808624, "learning_rate": 9.061221536751517e-05, "loss": 0.0444, "step": 14300 }, { "epoch": 6.404117252181696, "grad_norm": 0.1952340304851532, "learning_rate": 9.059613423804623e-05, "loss": 0.0443, "step": 14310 }, { "epoch": 6.408592526292235, "grad_norm": 0.22699907422065735, "learning_rate": 9.058004077654359e-05, "loss": 0.0468, "step": 14320 }, { "epoch": 6.413067800402775, "grad_norm": 0.16590279340744019, "learning_rate": 9.056393498789602e-05, "loss": 0.0432, "step": 14330 }, { "epoch": 6.417543074513314, "grad_norm": 0.18743430078029633, "learning_rate": 9.0547816876996e-05, "loss": 0.0464, "step": 14340 }, { "epoch": 6.422018348623853, "grad_norm": 0.22955380380153656, "learning_rate": 9.053168644873984e-05, "loss": 0.0487, "step": 14350 }, { "epoch": 6.426493622734393, "grad_norm": 0.2228045016527176, "learning_rate": 9.05155437080275e-05, "loss": 0.0451, "step": 14360 }, { "epoch": 6.430968896844932, "grad_norm": 0.19705824553966522, "learning_rate": 9.049938865976275e-05, "loss": 0.0422, "step": 14370 }, { "epoch": 6.435444170955471, "grad_norm": 0.19816070795059204, "learning_rate": 9.048322130885305e-05, "loss": 0.0506, "step": 14380 }, { "epoch": 6.43991944506601, "grad_norm": 0.17605958878993988, "learning_rate": 9.046704166020961e-05, "loss": 0.0473, "step": 14390 }, { "epoch": 6.444394719176549, "grad_norm": 0.17804081737995148, "learning_rate": 9.045084971874738e-05, "loss": 0.0443, "step": 14400 }, { "epoch": 6.448869993287089, "grad_norm": 0.17259113490581512, "learning_rate": 9.043464548938506e-05, "loss": 0.0478, "step": 14410 }, { "epoch": 6.453345267397628, "grad_norm": 0.1727050244808197, "learning_rate": 9.041842897704502e-05, "loss": 0.0454, "step": 14420 }, { "epoch": 6.457820541508167, "grad_norm": 0.20974421501159668, "learning_rate": 9.040220018665347e-05, "loss": 0.0483, "step": 14430 }, { "epoch": 6.462295815618707, "grad_norm": 0.17625847458839417, "learning_rate": 9.038595912314027e-05, "loss": 0.0525, "step": 14440 }, { "epoch": 6.466771089729246, "grad_norm": 0.2083590179681778, "learning_rate": 9.0369705791439e-05, "loss": 0.0448, "step": 14450 }, { "epoch": 6.471246363839785, "grad_norm": 0.2261252999305725, "learning_rate": 9.035344019648702e-05, "loss": 0.0486, "step": 14460 }, { "epoch": 6.475721637950325, "grad_norm": 0.21949057281017303, "learning_rate": 9.033716234322538e-05, "loss": 0.0448, "step": 14470 }, { "epoch": 6.480196912060864, "grad_norm": 0.20337744057178497, "learning_rate": 9.032087223659885e-05, "loss": 0.0514, "step": 14480 }, { "epoch": 6.484672186171403, "grad_norm": 0.19663836061954498, "learning_rate": 9.030456988155596e-05, "loss": 0.0472, "step": 14490 }, { "epoch": 6.489147460281942, "grad_norm": 0.21431311964988708, "learning_rate": 9.028825528304892e-05, "loss": 0.0446, "step": 14500 }, { "epoch": 6.493622734392481, "grad_norm": 0.20889507234096527, "learning_rate": 9.027192844603365e-05, "loss": 0.0456, "step": 14510 }, { "epoch": 6.498098008503021, "grad_norm": 0.1914851814508438, "learning_rate": 9.025558937546988e-05, "loss": 0.0465, "step": 14520 }, { "epoch": 6.50257328261356, "grad_norm": 0.17527368664741516, "learning_rate": 9.02392380763209e-05, "loss": 0.0502, "step": 14530 }, { "epoch": 6.507048556724099, "grad_norm": 0.18775124847888947, "learning_rate": 9.022287455355387e-05, "loss": 0.0458, "step": 14540 }, { "epoch": 6.511523830834639, "grad_norm": 0.18063649535179138, "learning_rate": 9.020649881213958e-05, "loss": 0.0445, "step": 14550 }, { "epoch": 6.515999104945178, "grad_norm": 0.2091149389743805, "learning_rate": 9.019011085705253e-05, "loss": 0.0476, "step": 14560 }, { "epoch": 6.520474379055717, "grad_norm": 0.16797935962677002, "learning_rate": 9.017371069327096e-05, "loss": 0.0474, "step": 14570 }, { "epoch": 6.5249496531662565, "grad_norm": 0.1953924596309662, "learning_rate": 9.015729832577681e-05, "loss": 0.048, "step": 14580 }, { "epoch": 6.5294249272767955, "grad_norm": 0.20655331015586853, "learning_rate": 9.014087375955573e-05, "loss": 0.0464, "step": 14590 }, { "epoch": 6.533900201387335, "grad_norm": 0.21441911160945892, "learning_rate": 9.012443699959705e-05, "loss": 0.0409, "step": 14600 }, { "epoch": 6.538375475497874, "grad_norm": 0.1934572458267212, "learning_rate": 9.010798805089384e-05, "loss": 0.0474, "step": 14610 }, { "epoch": 6.542850749608413, "grad_norm": 0.20565283298492432, "learning_rate": 9.009152691844285e-05, "loss": 0.0476, "step": 14620 }, { "epoch": 6.547326023718953, "grad_norm": 0.21143430471420288, "learning_rate": 9.007505360724453e-05, "loss": 0.0452, "step": 14630 }, { "epoch": 6.551801297829492, "grad_norm": 0.19268907606601715, "learning_rate": 9.005856812230304e-05, "loss": 0.0464, "step": 14640 }, { "epoch": 6.556276571940032, "grad_norm": 0.18461844325065613, "learning_rate": 9.004207046862624e-05, "loss": 0.0447, "step": 14650 }, { "epoch": 6.560751846050571, "grad_norm": 0.19239485263824463, "learning_rate": 9.002556065122571e-05, "loss": 0.0458, "step": 14660 }, { "epoch": 6.56522712016111, "grad_norm": 0.21910732984542847, "learning_rate": 9.000903867511666e-05, "loss": 0.0453, "step": 14670 }, { "epoch": 6.5697023942716495, "grad_norm": 0.18605443835258484, "learning_rate": 8.999250454531802e-05, "loss": 0.0469, "step": 14680 }, { "epoch": 6.574177668382188, "grad_norm": 0.18585406243801117, "learning_rate": 8.997595826685243e-05, "loss": 0.0427, "step": 14690 }, { "epoch": 6.578652942492727, "grad_norm": 0.19513756036758423, "learning_rate": 8.995939984474624e-05, "loss": 0.0448, "step": 14700 }, { "epoch": 6.583128216603267, "grad_norm": 0.18467475473880768, "learning_rate": 8.994282928402944e-05, "loss": 0.0461, "step": 14710 }, { "epoch": 6.587603490713806, "grad_norm": 0.1657634675502777, "learning_rate": 8.992624658973574e-05, "loss": 0.0458, "step": 14720 }, { "epoch": 6.592078764824345, "grad_norm": 0.20386487245559692, "learning_rate": 8.990965176690252e-05, "loss": 0.0447, "step": 14730 }, { "epoch": 6.596554038934885, "grad_norm": 0.16122180223464966, "learning_rate": 8.989304482057084e-05, "loss": 0.0432, "step": 14740 }, { "epoch": 6.601029313045424, "grad_norm": 0.17129896581172943, "learning_rate": 8.987642575578545e-05, "loss": 0.044, "step": 14750 }, { "epoch": 6.605504587155964, "grad_norm": 0.15892256796360016, "learning_rate": 8.98597945775948e-05, "loss": 0.047, "step": 14760 }, { "epoch": 6.609979861266503, "grad_norm": 0.17794613540172577, "learning_rate": 8.984315129105099e-05, "loss": 0.0458, "step": 14770 }, { "epoch": 6.6144551353770416, "grad_norm": 0.18322476744651794, "learning_rate": 8.982649590120982e-05, "loss": 0.0419, "step": 14780 }, { "epoch": 6.618930409487581, "grad_norm": 0.1922733634710312, "learning_rate": 8.980982841313074e-05, "loss": 0.0481, "step": 14790 }, { "epoch": 6.62340568359812, "grad_norm": 0.15588252246379852, "learning_rate": 8.979314883187693e-05, "loss": 0.044, "step": 14800 }, { "epoch": 6.62788095770866, "grad_norm": 0.2035282403230667, "learning_rate": 8.977645716251518e-05, "loss": 0.0447, "step": 14810 }, { "epoch": 6.632356231819199, "grad_norm": 0.20976032316684723, "learning_rate": 8.975975341011596e-05, "loss": 0.049, "step": 14820 }, { "epoch": 6.636831505929738, "grad_norm": 0.22799095511436462, "learning_rate": 8.974303757975345e-05, "loss": 0.0445, "step": 14830 }, { "epoch": 6.641306780040278, "grad_norm": 0.16483870148658752, "learning_rate": 8.972630967650548e-05, "loss": 0.0442, "step": 14840 }, { "epoch": 6.645782054150817, "grad_norm": 0.16737355291843414, "learning_rate": 8.970956970545355e-05, "loss": 0.044, "step": 14850 }, { "epoch": 6.650257328261356, "grad_norm": 0.2258603572845459, "learning_rate": 8.969281767168283e-05, "loss": 0.0439, "step": 14860 }, { "epoch": 6.654732602371896, "grad_norm": 0.2098829597234726, "learning_rate": 8.967605358028211e-05, "loss": 0.0432, "step": 14870 }, { "epoch": 6.6592078764824345, "grad_norm": 0.21005117893218994, "learning_rate": 8.965927743634391e-05, "loss": 0.0463, "step": 14880 }, { "epoch": 6.6636831505929734, "grad_norm": 0.165131613612175, "learning_rate": 8.964248924496435e-05, "loss": 0.0478, "step": 14890 }, { "epoch": 6.668158424703513, "grad_norm": 0.2237686961889267, "learning_rate": 8.962568901124327e-05, "loss": 0.0493, "step": 14900 }, { "epoch": 6.672633698814052, "grad_norm": 0.22907885909080505, "learning_rate": 8.96088767402841e-05, "loss": 0.0458, "step": 14910 }, { "epoch": 6.677108972924591, "grad_norm": 0.21021617949008942, "learning_rate": 8.959205243719402e-05, "loss": 0.0455, "step": 14920 }, { "epoch": 6.681584247035131, "grad_norm": 0.1649412214756012, "learning_rate": 8.957521610708375e-05, "loss": 0.049, "step": 14930 }, { "epoch": 6.68605952114567, "grad_norm": 0.20815981924533844, "learning_rate": 8.955836775506776e-05, "loss": 0.0453, "step": 14940 }, { "epoch": 6.69053479525621, "grad_norm": 0.19232314825057983, "learning_rate": 8.954150738626414e-05, "loss": 0.0496, "step": 14950 }, { "epoch": 6.695010069366749, "grad_norm": 0.16739322245121002, "learning_rate": 8.95246350057946e-05, "loss": 0.051, "step": 14960 }, { "epoch": 6.699485343477288, "grad_norm": 0.18557411432266235, "learning_rate": 8.950775061878453e-05, "loss": 0.0477, "step": 14970 }, { "epoch": 6.7039606175878275, "grad_norm": 0.21641458570957184, "learning_rate": 8.949085423036296e-05, "loss": 0.0414, "step": 14980 }, { "epoch": 6.708435891698366, "grad_norm": 0.1578938215970993, "learning_rate": 8.947394584566258e-05, "loss": 0.0456, "step": 14990 }, { "epoch": 6.712911165808906, "grad_norm": 0.21308869123458862, "learning_rate": 8.945702546981969e-05, "loss": 0.0407, "step": 15000 }, { "epoch": 6.717386439919445, "grad_norm": 0.18396435678005219, "learning_rate": 8.944009310797426e-05, "loss": 0.0497, "step": 15010 }, { "epoch": 6.721861714029984, "grad_norm": 0.1782134622335434, "learning_rate": 8.942314876526992e-05, "loss": 0.0457, "step": 15020 }, { "epoch": 6.726336988140524, "grad_norm": 0.21728824079036713, "learning_rate": 8.940619244685388e-05, "loss": 0.0446, "step": 15030 }, { "epoch": 6.730812262251063, "grad_norm": 0.19327379763126373, "learning_rate": 8.938922415787703e-05, "loss": 0.047, "step": 15040 }, { "epoch": 6.735287536361602, "grad_norm": 0.15362538397312164, "learning_rate": 8.93722439034939e-05, "loss": 0.0462, "step": 15050 }, { "epoch": 6.739762810472142, "grad_norm": 0.19821400940418243, "learning_rate": 8.935525168886262e-05, "loss": 0.0446, "step": 15060 }, { "epoch": 6.744238084582681, "grad_norm": 0.20626495778560638, "learning_rate": 8.933824751914502e-05, "loss": 0.0498, "step": 15070 }, { "epoch": 6.7487133586932195, "grad_norm": 0.1577313393354416, "learning_rate": 8.932123139950648e-05, "loss": 0.0445, "step": 15080 }, { "epoch": 6.753188632803759, "grad_norm": 0.192116841673851, "learning_rate": 8.930420333511606e-05, "loss": 0.042, "step": 15090 }, { "epoch": 6.757663906914298, "grad_norm": 0.1676388680934906, "learning_rate": 8.928716333114643e-05, "loss": 0.0411, "step": 15100 }, { "epoch": 6.762139181024838, "grad_norm": 0.16645444929599762, "learning_rate": 8.927011139277389e-05, "loss": 0.0426, "step": 15110 }, { "epoch": 6.766614455135377, "grad_norm": 0.1861289143562317, "learning_rate": 8.92530475251784e-05, "loss": 0.0462, "step": 15120 }, { "epoch": 6.771089729245916, "grad_norm": 0.20708808302879333, "learning_rate": 8.923597173354345e-05, "loss": 0.0449, "step": 15130 }, { "epoch": 6.775565003356456, "grad_norm": 0.21114665269851685, "learning_rate": 8.921888402305628e-05, "loss": 0.0446, "step": 15140 }, { "epoch": 6.780040277466995, "grad_norm": 0.2035851925611496, "learning_rate": 8.920178439890765e-05, "loss": 0.0432, "step": 15150 }, { "epoch": 6.784515551577535, "grad_norm": 0.2445325404405594, "learning_rate": 8.9184672866292e-05, "loss": 0.045, "step": 15160 }, { "epoch": 6.7889908256880735, "grad_norm": 0.2059822827577591, "learning_rate": 8.916754943040732e-05, "loss": 0.0468, "step": 15170 }, { "epoch": 6.7934660997986125, "grad_norm": 0.20079027116298676, "learning_rate": 8.91504140964553e-05, "loss": 0.0447, "step": 15180 }, { "epoch": 6.797941373909152, "grad_norm": 0.23105764389038086, "learning_rate": 8.913326686964117e-05, "loss": 0.042, "step": 15190 }, { "epoch": 6.802416648019691, "grad_norm": 0.22076530754566193, "learning_rate": 8.911610775517382e-05, "loss": 0.0468, "step": 15200 }, { "epoch": 6.80689192213023, "grad_norm": 0.2381335347890854, "learning_rate": 8.909893675826574e-05, "loss": 0.046, "step": 15210 }, { "epoch": 6.81136719624077, "grad_norm": 0.1742953360080719, "learning_rate": 8.908175388413304e-05, "loss": 0.0415, "step": 15220 }, { "epoch": 6.815842470351309, "grad_norm": 0.26625779271125793, "learning_rate": 8.906455913799538e-05, "loss": 0.0451, "step": 15230 }, { "epoch": 6.820317744461848, "grad_norm": 0.18230749666690826, "learning_rate": 8.90473525250761e-05, "loss": 0.0419, "step": 15240 }, { "epoch": 6.824793018572388, "grad_norm": 0.1443922519683838, "learning_rate": 8.903013405060211e-05, "loss": 0.0397, "step": 15250 }, { "epoch": 6.829268292682927, "grad_norm": 0.16845917701721191, "learning_rate": 8.901290371980393e-05, "loss": 0.0452, "step": 15260 }, { "epoch": 6.833743566793466, "grad_norm": 0.1697920560836792, "learning_rate": 8.899566153791566e-05, "loss": 0.0457, "step": 15270 }, { "epoch": 6.8382188409040054, "grad_norm": 0.15973469614982605, "learning_rate": 8.897840751017506e-05, "loss": 0.0419, "step": 15280 }, { "epoch": 6.842694115014544, "grad_norm": 0.20272567868232727, "learning_rate": 8.89611416418234e-05, "loss": 0.0423, "step": 15290 }, { "epoch": 6.847169389125084, "grad_norm": 0.16808508336544037, "learning_rate": 8.894386393810563e-05, "loss": 0.048, "step": 15300 }, { "epoch": 6.851644663235623, "grad_norm": 0.19438420236110687, "learning_rate": 8.892657440427025e-05, "loss": 0.0426, "step": 15310 }, { "epoch": 6.856119937346162, "grad_norm": 0.15160071849822998, "learning_rate": 8.890927304556935e-05, "loss": 0.0437, "step": 15320 }, { "epoch": 6.860595211456702, "grad_norm": 0.19614970684051514, "learning_rate": 8.889195986725865e-05, "loss": 0.043, "step": 15330 }, { "epoch": 6.865070485567241, "grad_norm": 0.2032114565372467, "learning_rate": 8.887463487459742e-05, "loss": 0.0471, "step": 15340 }, { "epoch": 6.869545759677781, "grad_norm": 0.16525055468082428, "learning_rate": 8.885729807284856e-05, "loss": 0.0446, "step": 15350 }, { "epoch": 6.87402103378832, "grad_norm": 0.1547035127878189, "learning_rate": 8.883994946727849e-05, "loss": 0.0473, "step": 15360 }, { "epoch": 6.878496307898859, "grad_norm": 0.1846599280834198, "learning_rate": 8.882258906315729e-05, "loss": 0.0448, "step": 15370 }, { "epoch": 6.882971582009398, "grad_norm": 0.14728431403636932, "learning_rate": 8.880521686575857e-05, "loss": 0.0467, "step": 15380 }, { "epoch": 6.887446856119937, "grad_norm": 0.17067483067512512, "learning_rate": 8.878783288035957e-05, "loss": 0.0421, "step": 15390 }, { "epoch": 6.891922130230476, "grad_norm": 0.17859451472759247, "learning_rate": 8.877043711224108e-05, "loss": 0.045, "step": 15400 }, { "epoch": 6.896397404341016, "grad_norm": 0.17596520483493805, "learning_rate": 8.875302956668747e-05, "loss": 0.043, "step": 15410 }, { "epoch": 6.900872678451555, "grad_norm": 0.18049860000610352, "learning_rate": 8.873561024898668e-05, "loss": 0.0445, "step": 15420 }, { "epoch": 6.905347952562094, "grad_norm": 0.15894076228141785, "learning_rate": 8.871817916443025e-05, "loss": 0.0437, "step": 15430 }, { "epoch": 6.909823226672634, "grad_norm": 0.17266546189785004, "learning_rate": 8.87007363183133e-05, "loss": 0.041, "step": 15440 }, { "epoch": 6.914298500783173, "grad_norm": 0.16335953772068024, "learning_rate": 8.868328171593448e-05, "loss": 0.0501, "step": 15450 }, { "epoch": 6.918773774893713, "grad_norm": 0.1775519698858261, "learning_rate": 8.866581536259605e-05, "loss": 0.038, "step": 15460 }, { "epoch": 6.9232490490042515, "grad_norm": 0.18678520619869232, "learning_rate": 8.864833726360383e-05, "loss": 0.0426, "step": 15470 }, { "epoch": 6.9277243231147905, "grad_norm": 0.18908576667308807, "learning_rate": 8.863084742426719e-05, "loss": 0.048, "step": 15480 }, { "epoch": 6.93219959722533, "grad_norm": 0.17495889961719513, "learning_rate": 8.86133458498991e-05, "loss": 0.0401, "step": 15490 }, { "epoch": 6.936674871335869, "grad_norm": 0.1653512418270111, "learning_rate": 8.859583254581605e-05, "loss": 0.0442, "step": 15500 }, { "epoch": 6.941150145446409, "grad_norm": 0.17557084560394287, "learning_rate": 8.857830751733815e-05, "loss": 0.0466, "step": 15510 }, { "epoch": 6.945625419556948, "grad_norm": 0.19334204494953156, "learning_rate": 8.856077076978902e-05, "loss": 0.0449, "step": 15520 }, { "epoch": 6.950100693667487, "grad_norm": 0.164515420794487, "learning_rate": 8.854322230849588e-05, "loss": 0.0467, "step": 15530 }, { "epoch": 6.954575967778027, "grad_norm": 0.15594199299812317, "learning_rate": 8.852566213878947e-05, "loss": 0.0433, "step": 15540 }, { "epoch": 6.959051241888566, "grad_norm": 0.18270200490951538, "learning_rate": 8.85080902660041e-05, "loss": 0.044, "step": 15550 }, { "epoch": 6.963526515999105, "grad_norm": 0.18020159006118774, "learning_rate": 8.849050669547768e-05, "loss": 0.0482, "step": 15560 }, { "epoch": 6.9680017901096445, "grad_norm": 0.16634701192378998, "learning_rate": 8.84729114325516e-05, "loss": 0.0448, "step": 15570 }, { "epoch": 6.972477064220183, "grad_norm": 0.15182901918888092, "learning_rate": 8.845530448257085e-05, "loss": 0.041, "step": 15580 }, { "epoch": 6.976952338330722, "grad_norm": 0.21236705780029297, "learning_rate": 8.843768585088393e-05, "loss": 0.0455, "step": 15590 }, { "epoch": 6.981427612441262, "grad_norm": 0.1812094748020172, "learning_rate": 8.842005554284296e-05, "loss": 0.0469, "step": 15600 }, { "epoch": 6.985902886551801, "grad_norm": 0.2244059145450592, "learning_rate": 8.840241356380352e-05, "loss": 0.0437, "step": 15610 }, { "epoch": 6.99037816066234, "grad_norm": 0.19640545547008514, "learning_rate": 8.838475991912482e-05, "loss": 0.0423, "step": 15620 }, { "epoch": 6.99485343477288, "grad_norm": 0.18812911212444305, "learning_rate": 8.836709461416952e-05, "loss": 0.0436, "step": 15630 }, { "epoch": 6.999328708883419, "grad_norm": 0.22442927956581116, "learning_rate": 8.834941765430391e-05, "loss": 0.044, "step": 15640 }, { "epoch": 7.003803982993959, "grad_norm": 0.17310523986816406, "learning_rate": 8.83317290448978e-05, "loss": 0.0448, "step": 15650 }, { "epoch": 7.008279257104498, "grad_norm": 0.20389701426029205, "learning_rate": 8.831402879132446e-05, "loss": 0.046, "step": 15660 }, { "epoch": 7.012754531215037, "grad_norm": 0.2242773324251175, "learning_rate": 8.829631689896082e-05, "loss": 0.0472, "step": 15670 }, { "epoch": 7.017229805325576, "grad_norm": 0.2034761756658554, "learning_rate": 8.827859337318725e-05, "loss": 0.0437, "step": 15680 }, { "epoch": 7.021705079436115, "grad_norm": 0.18542225658893585, "learning_rate": 8.82608582193877e-05, "loss": 0.0434, "step": 15690 }, { "epoch": 7.026180353546655, "grad_norm": 0.17154033482074738, "learning_rate": 8.824311144294965e-05, "loss": 0.0474, "step": 15700 }, { "epoch": 7.030655627657194, "grad_norm": 0.1941189020872116, "learning_rate": 8.822535304926409e-05, "loss": 0.0463, "step": 15710 }, { "epoch": 7.035130901767733, "grad_norm": 0.1894596368074417, "learning_rate": 8.820758304372557e-05, "loss": 0.0451, "step": 15720 }, { "epoch": 7.039606175878273, "grad_norm": 0.1743895560503006, "learning_rate": 8.818980143173213e-05, "loss": 0.041, "step": 15730 }, { "epoch": 7.044081449988812, "grad_norm": 0.163924902677536, "learning_rate": 8.817200821868533e-05, "loss": 0.0434, "step": 15740 }, { "epoch": 7.048556724099351, "grad_norm": 0.1780611276626587, "learning_rate": 8.815420340999033e-05, "loss": 0.041, "step": 15750 }, { "epoch": 7.053031998209891, "grad_norm": 0.16410252451896667, "learning_rate": 8.813638701105573e-05, "loss": 0.045, "step": 15760 }, { "epoch": 7.0575072723204295, "grad_norm": 0.15698955953121185, "learning_rate": 8.81185590272937e-05, "loss": 0.0394, "step": 15770 }, { "epoch": 7.0619825464309685, "grad_norm": 0.18656708300113678, "learning_rate": 8.810071946411989e-05, "loss": 0.0449, "step": 15780 }, { "epoch": 7.066457820541508, "grad_norm": 0.20082242786884308, "learning_rate": 8.80828683269535e-05, "loss": 0.0417, "step": 15790 }, { "epoch": 7.070933094652047, "grad_norm": 0.1746295541524887, "learning_rate": 8.806500562121723e-05, "loss": 0.0416, "step": 15800 }, { "epoch": 7.075408368762587, "grad_norm": 0.17419084906578064, "learning_rate": 8.804713135233731e-05, "loss": 0.0462, "step": 15810 }, { "epoch": 7.079883642873126, "grad_norm": 0.20463700592517853, "learning_rate": 8.802924552574345e-05, "loss": 0.0436, "step": 15820 }, { "epoch": 7.084358916983665, "grad_norm": 0.18996745347976685, "learning_rate": 8.801134814686891e-05, "loss": 0.0444, "step": 15830 }, { "epoch": 7.088834191094205, "grad_norm": 0.19516333937644958, "learning_rate": 8.799343922115044e-05, "loss": 0.0424, "step": 15840 }, { "epoch": 7.093309465204744, "grad_norm": 0.1836620569229126, "learning_rate": 8.797551875402827e-05, "loss": 0.0434, "step": 15850 }, { "epoch": 7.097784739315283, "grad_norm": 0.17299355566501617, "learning_rate": 8.795758675094621e-05, "loss": 0.0413, "step": 15860 }, { "epoch": 7.1022600134258225, "grad_norm": 0.17561103403568268, "learning_rate": 8.79396432173515e-05, "loss": 0.0428, "step": 15870 }, { "epoch": 7.106735287536361, "grad_norm": 0.18160013854503632, "learning_rate": 8.792168815869493e-05, "loss": 0.0419, "step": 15880 }, { "epoch": 7.111210561646901, "grad_norm": 0.15974262356758118, "learning_rate": 8.790372158043074e-05, "loss": 0.0442, "step": 15890 }, { "epoch": 7.11568583575744, "grad_norm": 0.1809634417295456, "learning_rate": 8.788574348801675e-05, "loss": 0.0465, "step": 15900 }, { "epoch": 7.120161109867979, "grad_norm": 0.23166052997112274, "learning_rate": 8.786775388691418e-05, "loss": 0.0452, "step": 15910 }, { "epoch": 7.124636383978519, "grad_norm": 0.1842937171459198, "learning_rate": 8.784975278258783e-05, "loss": 0.0443, "step": 15920 }, { "epoch": 7.129111658089058, "grad_norm": 0.18714065849781036, "learning_rate": 8.783174018050594e-05, "loss": 0.0482, "step": 15930 }, { "epoch": 7.133586932199597, "grad_norm": 0.16849878430366516, "learning_rate": 8.781371608614029e-05, "loss": 0.0436, "step": 15940 }, { "epoch": 7.138062206310137, "grad_norm": 0.1988891065120697, "learning_rate": 8.77956805049661e-05, "loss": 0.0454, "step": 15950 }, { "epoch": 7.142537480420676, "grad_norm": 0.17557691037654877, "learning_rate": 8.77776334424621e-05, "loss": 0.0398, "step": 15960 }, { "epoch": 7.147012754531215, "grad_norm": 0.18510566651821136, "learning_rate": 8.775957490411053e-05, "loss": 0.046, "step": 15970 }, { "epoch": 7.151488028641754, "grad_norm": 0.1661657840013504, "learning_rate": 8.774150489539707e-05, "loss": 0.0471, "step": 15980 }, { "epoch": 7.155963302752293, "grad_norm": 0.20188313722610474, "learning_rate": 8.772342342181095e-05, "loss": 0.0475, "step": 15990 }, { "epoch": 7.160438576862833, "grad_norm": 0.16972927749156952, "learning_rate": 8.770533048884482e-05, "loss": 0.0407, "step": 16000 }, { "epoch": 7.164913850973372, "grad_norm": 0.18506084382534027, "learning_rate": 8.768722610199484e-05, "loss": 0.0464, "step": 16010 }, { "epoch": 7.169389125083911, "grad_norm": 0.1858856976032257, "learning_rate": 8.766911026676064e-05, "loss": 0.0468, "step": 16020 }, { "epoch": 7.173864399194451, "grad_norm": 0.1778232902288437, "learning_rate": 8.765098298864533e-05, "loss": 0.0418, "step": 16030 }, { "epoch": 7.17833967330499, "grad_norm": 0.17278295755386353, "learning_rate": 8.763284427315551e-05, "loss": 0.0414, "step": 16040 }, { "epoch": 7.18281494741553, "grad_norm": 0.17676317691802979, "learning_rate": 8.761469412580125e-05, "loss": 0.0462, "step": 16050 }, { "epoch": 7.187290221526069, "grad_norm": 0.20161843299865723, "learning_rate": 8.759653255209606e-05, "loss": 0.0433, "step": 16060 }, { "epoch": 7.1917654956366075, "grad_norm": 0.16336897015571594, "learning_rate": 8.757835955755695e-05, "loss": 0.0446, "step": 16070 }, { "epoch": 7.196240769747147, "grad_norm": 0.1747824102640152, "learning_rate": 8.756017514770443e-05, "loss": 0.0419, "step": 16080 }, { "epoch": 7.200716043857686, "grad_norm": 0.19372302293777466, "learning_rate": 8.75419793280624e-05, "loss": 0.0403, "step": 16090 }, { "epoch": 7.205191317968225, "grad_norm": 0.15870335698127747, "learning_rate": 8.75237721041583e-05, "loss": 0.0427, "step": 16100 }, { "epoch": 7.209666592078765, "grad_norm": 0.19079169631004333, "learning_rate": 8.750555348152298e-05, "loss": 0.0449, "step": 16110 }, { "epoch": 7.214141866189304, "grad_norm": 0.17138484120368958, "learning_rate": 8.74873234656908e-05, "loss": 0.0451, "step": 16120 }, { "epoch": 7.218617140299843, "grad_norm": 0.18301965296268463, "learning_rate": 8.746908206219955e-05, "loss": 0.0429, "step": 16130 }, { "epoch": 7.223092414410383, "grad_norm": 0.22340670228004456, "learning_rate": 8.745082927659047e-05, "loss": 0.0422, "step": 16140 }, { "epoch": 7.227567688520922, "grad_norm": 0.189294695854187, "learning_rate": 8.74325651144083e-05, "loss": 0.0467, "step": 16150 }, { "epoch": 7.2320429626314615, "grad_norm": 0.2037849724292755, "learning_rate": 8.741428958120118e-05, "loss": 0.0453, "step": 16160 }, { "epoch": 7.2365182367420005, "grad_norm": 0.2144768387079239, "learning_rate": 8.739600268252078e-05, "loss": 0.0445, "step": 16170 }, { "epoch": 7.240993510852539, "grad_norm": 0.19419555366039276, "learning_rate": 8.737770442392212e-05, "loss": 0.0438, "step": 16180 }, { "epoch": 7.245468784963079, "grad_norm": 0.22810891270637512, "learning_rate": 8.735939481096378e-05, "loss": 0.0463, "step": 16190 }, { "epoch": 7.249944059073618, "grad_norm": 0.23461391031742096, "learning_rate": 8.73410738492077e-05, "loss": 0.0443, "step": 16200 }, { "epoch": 7.254419333184158, "grad_norm": 0.20922911167144775, "learning_rate": 8.732274154421933e-05, "loss": 0.0466, "step": 16210 }, { "epoch": 7.258894607294697, "grad_norm": 0.1700456142425537, "learning_rate": 8.730439790156752e-05, "loss": 0.0467, "step": 16220 }, { "epoch": 7.263369881405236, "grad_norm": 0.18849657475948334, "learning_rate": 8.728604292682459e-05, "loss": 0.0443, "step": 16230 }, { "epoch": 7.267845155515776, "grad_norm": 0.1934780329465866, "learning_rate": 8.72676766255663e-05, "loss": 0.0449, "step": 16240 }, { "epoch": 7.272320429626315, "grad_norm": 0.23827648162841797, "learning_rate": 8.724929900337186e-05, "loss": 0.0459, "step": 16250 }, { "epoch": 7.276795703736854, "grad_norm": 0.18592847883701324, "learning_rate": 8.723091006582389e-05, "loss": 0.0424, "step": 16260 }, { "epoch": 7.281270977847393, "grad_norm": 0.19018904864788055, "learning_rate": 8.721250981850846e-05, "loss": 0.0442, "step": 16270 }, { "epoch": 7.285746251957932, "grad_norm": 0.18602502346038818, "learning_rate": 8.719409826701508e-05, "loss": 0.0442, "step": 16280 }, { "epoch": 7.290221526068471, "grad_norm": 0.2161768227815628, "learning_rate": 8.717567541693673e-05, "loss": 0.0465, "step": 16290 }, { "epoch": 7.294696800179011, "grad_norm": 0.16373614966869354, "learning_rate": 8.715724127386972e-05, "loss": 0.0402, "step": 16300 }, { "epoch": 7.29917207428955, "grad_norm": 0.1667058914899826, "learning_rate": 8.713879584341391e-05, "loss": 0.0457, "step": 16310 }, { "epoch": 7.30364734840009, "grad_norm": 0.17959056794643402, "learning_rate": 8.71203391311725e-05, "loss": 0.0431, "step": 16320 }, { "epoch": 7.308122622510629, "grad_norm": 0.1842210441827774, "learning_rate": 8.710187114275219e-05, "loss": 0.044, "step": 16330 }, { "epoch": 7.312597896621168, "grad_norm": 0.16418246924877167, "learning_rate": 8.708339188376302e-05, "loss": 0.0422, "step": 16340 }, { "epoch": 7.317073170731708, "grad_norm": 0.15530180931091309, "learning_rate": 8.706490135981855e-05, "loss": 0.0446, "step": 16350 }, { "epoch": 7.3215484448422465, "grad_norm": 0.15536758303642273, "learning_rate": 8.704639957653567e-05, "loss": 0.0437, "step": 16360 }, { "epoch": 7.3260237189527855, "grad_norm": 0.19470180571079254, "learning_rate": 8.702788653953477e-05, "loss": 0.0424, "step": 16370 }, { "epoch": 7.330498993063325, "grad_norm": 0.15725161135196686, "learning_rate": 8.700936225443959e-05, "loss": 0.0455, "step": 16380 }, { "epoch": 7.334974267173864, "grad_norm": 0.19228880107402802, "learning_rate": 8.699082672687734e-05, "loss": 0.0457, "step": 16390 }, { "epoch": 7.339449541284404, "grad_norm": 0.1972973346710205, "learning_rate": 8.697227996247861e-05, "loss": 0.0433, "step": 16400 }, { "epoch": 7.343924815394943, "grad_norm": 0.18495553731918335, "learning_rate": 8.695372196687743e-05, "loss": 0.042, "step": 16410 }, { "epoch": 7.348400089505482, "grad_norm": 0.15847991406917572, "learning_rate": 8.693515274571123e-05, "loss": 0.0408, "step": 16420 }, { "epoch": 7.352875363616022, "grad_norm": 0.1886957585811615, "learning_rate": 8.691657230462083e-05, "loss": 0.0424, "step": 16430 }, { "epoch": 7.357350637726561, "grad_norm": 0.16319288313388824, "learning_rate": 8.689798064925049e-05, "loss": 0.0442, "step": 16440 }, { "epoch": 7.3618259118371, "grad_norm": 0.15909825265407562, "learning_rate": 8.687937778524786e-05, "loss": 0.0435, "step": 16450 }, { "epoch": 7.3663011859476395, "grad_norm": 0.19271248579025269, "learning_rate": 8.686076371826401e-05, "loss": 0.0399, "step": 16460 }, { "epoch": 7.370776460058178, "grad_norm": 0.1631445437669754, "learning_rate": 8.684213845395339e-05, "loss": 0.041, "step": 16470 }, { "epoch": 7.375251734168717, "grad_norm": 0.171257883310318, "learning_rate": 8.682350199797388e-05, "loss": 0.0426, "step": 16480 }, { "epoch": 7.379727008279257, "grad_norm": 0.1587861180305481, "learning_rate": 8.680485435598673e-05, "loss": 0.0408, "step": 16490 }, { "epoch": 7.384202282389796, "grad_norm": 0.17213420569896698, "learning_rate": 8.678619553365659e-05, "loss": 0.0425, "step": 16500 }, { "epoch": 7.388677556500336, "grad_norm": 0.18147645890712738, "learning_rate": 8.676752553665153e-05, "loss": 0.0429, "step": 16510 }, { "epoch": 7.393152830610875, "grad_norm": 0.16909494996070862, "learning_rate": 8.674884437064302e-05, "loss": 0.0412, "step": 16520 }, { "epoch": 7.397628104721414, "grad_norm": 0.15589812397956848, "learning_rate": 8.673015204130586e-05, "loss": 0.0412, "step": 16530 }, { "epoch": 7.402103378831954, "grad_norm": 0.19294369220733643, "learning_rate": 8.671144855431833e-05, "loss": 0.0417, "step": 16540 }, { "epoch": 7.406578652942493, "grad_norm": 0.18416227400302887, "learning_rate": 8.669273391536204e-05, "loss": 0.0423, "step": 16550 }, { "epoch": 7.4110539270530325, "grad_norm": 0.1584472358226776, "learning_rate": 8.6674008130122e-05, "loss": 0.0389, "step": 16560 }, { "epoch": 7.415529201163571, "grad_norm": 0.14288325607776642, "learning_rate": 8.66552712042866e-05, "loss": 0.0403, "step": 16570 }, { "epoch": 7.42000447527411, "grad_norm": 0.19715119898319244, "learning_rate": 8.663652314354765e-05, "loss": 0.042, "step": 16580 }, { "epoch": 7.42447974938465, "grad_norm": 0.1492798626422882, "learning_rate": 8.661776395360029e-05, "loss": 0.0433, "step": 16590 }, { "epoch": 7.428955023495189, "grad_norm": 0.18083225190639496, "learning_rate": 8.659899364014309e-05, "loss": 0.0409, "step": 16600 }, { "epoch": 7.433430297605728, "grad_norm": 0.1597113460302353, "learning_rate": 8.658021220887795e-05, "loss": 0.0455, "step": 16610 }, { "epoch": 7.437905571716268, "grad_norm": 0.20218351483345032, "learning_rate": 8.656141966551019e-05, "loss": 0.0479, "step": 16620 }, { "epoch": 7.442380845826807, "grad_norm": 0.20538906753063202, "learning_rate": 8.654261601574849e-05, "loss": 0.0451, "step": 16630 }, { "epoch": 7.446856119937346, "grad_norm": 0.1455278843641281, "learning_rate": 8.652380126530488e-05, "loss": 0.0385, "step": 16640 }, { "epoch": 7.451331394047886, "grad_norm": 0.19799132645130157, "learning_rate": 8.650497541989482e-05, "loss": 0.041, "step": 16650 }, { "epoch": 7.4558066681584245, "grad_norm": 0.23019756376743317, "learning_rate": 8.648613848523707e-05, "loss": 0.0456, "step": 16660 }, { "epoch": 7.460281942268964, "grad_norm": 0.17233970761299133, "learning_rate": 8.646729046705382e-05, "loss": 0.0427, "step": 16670 }, { "epoch": 7.464757216379503, "grad_norm": 0.15758445858955383, "learning_rate": 8.644843137107059e-05, "loss": 0.0454, "step": 16680 }, { "epoch": 7.469232490490042, "grad_norm": 0.16284237802028656, "learning_rate": 8.642956120301626e-05, "loss": 0.0413, "step": 16690 }, { "epoch": 7.473707764600582, "grad_norm": 0.1793171912431717, "learning_rate": 8.641067996862311e-05, "loss": 0.0394, "step": 16700 }, { "epoch": 7.478183038711121, "grad_norm": 0.1956944465637207, "learning_rate": 8.639178767362676e-05, "loss": 0.0451, "step": 16710 }, { "epoch": 7.48265831282166, "grad_norm": 0.19033560156822205, "learning_rate": 8.637288432376618e-05, "loss": 0.0451, "step": 16720 }, { "epoch": 7.4871335869322, "grad_norm": 0.1762709617614746, "learning_rate": 8.635396992478371e-05, "loss": 0.045, "step": 16730 }, { "epoch": 7.491608861042739, "grad_norm": 0.17516069114208221, "learning_rate": 8.633504448242505e-05, "loss": 0.0454, "step": 16740 }, { "epoch": 7.4960841351532785, "grad_norm": 0.18152958154678345, "learning_rate": 8.631610800243926e-05, "loss": 0.044, "step": 16750 }, { "epoch": 7.5005594092638175, "grad_norm": 0.2020217627286911, "learning_rate": 8.629716049057872e-05, "loss": 0.0444, "step": 16760 }, { "epoch": 7.505034683374356, "grad_norm": 0.18790242075920105, "learning_rate": 8.627820195259918e-05, "loss": 0.0458, "step": 16770 }, { "epoch": 7.509509957484896, "grad_norm": 0.21251825988292694, "learning_rate": 8.625923239425978e-05, "loss": 0.0434, "step": 16780 }, { "epoch": 7.513985231595435, "grad_norm": 0.19011180102825165, "learning_rate": 8.624025182132292e-05, "loss": 0.0409, "step": 16790 }, { "epoch": 7.518460505705974, "grad_norm": 0.16822294890880585, "learning_rate": 8.622126023955446e-05, "loss": 0.0404, "step": 16800 }, { "epoch": 7.522935779816514, "grad_norm": 0.215826615691185, "learning_rate": 8.620225765472348e-05, "loss": 0.0417, "step": 16810 }, { "epoch": 7.527411053927053, "grad_norm": 0.164056196808815, "learning_rate": 8.61832440726025e-05, "loss": 0.0455, "step": 16820 }, { "epoch": 7.531886328037592, "grad_norm": 0.2370225340127945, "learning_rate": 8.616421949896734e-05, "loss": 0.0444, "step": 16830 }, { "epoch": 7.536361602148132, "grad_norm": 0.19042357802391052, "learning_rate": 8.614518393959714e-05, "loss": 0.0412, "step": 16840 }, { "epoch": 7.540836876258671, "grad_norm": 0.1378353089094162, "learning_rate": 8.612613740027443e-05, "loss": 0.0426, "step": 16850 }, { "epoch": 7.54531215036921, "grad_norm": 0.1943700909614563, "learning_rate": 8.610707988678503e-05, "loss": 0.0433, "step": 16860 }, { "epoch": 7.549787424479749, "grad_norm": 0.1669510006904602, "learning_rate": 8.608801140491811e-05, "loss": 0.0394, "step": 16870 }, { "epoch": 7.554262698590288, "grad_norm": 0.1591033786535263, "learning_rate": 8.606893196046619e-05, "loss": 0.0433, "step": 16880 }, { "epoch": 7.558737972700828, "grad_norm": 0.17582523822784424, "learning_rate": 8.604984155922506e-05, "loss": 0.0431, "step": 16890 }, { "epoch": 7.563213246811367, "grad_norm": 0.17239385843276978, "learning_rate": 8.603074020699393e-05, "loss": 0.0416, "step": 16900 }, { "epoch": 7.567688520921907, "grad_norm": 0.17805996537208557, "learning_rate": 8.601162790957525e-05, "loss": 0.0399, "step": 16910 }, { "epoch": 7.572163795032446, "grad_norm": 0.16721537709236145, "learning_rate": 8.599250467277483e-05, "loss": 0.0404, "step": 16920 }, { "epoch": 7.576639069142985, "grad_norm": 0.1505529284477234, "learning_rate": 8.597337050240184e-05, "loss": 0.0402, "step": 16930 }, { "epoch": 7.581114343253525, "grad_norm": 0.16781403124332428, "learning_rate": 8.595422540426869e-05, "loss": 0.0429, "step": 16940 }, { "epoch": 7.585589617364064, "grad_norm": 0.1980757713317871, "learning_rate": 8.59350693841912e-05, "loss": 0.0438, "step": 16950 }, { "epoch": 7.5900648914746025, "grad_norm": 0.19137543439865112, "learning_rate": 8.591590244798844e-05, "loss": 0.0394, "step": 16960 }, { "epoch": 7.594540165585142, "grad_norm": 0.20097194612026215, "learning_rate": 8.589672460148281e-05, "loss": 0.0427, "step": 16970 }, { "epoch": 7.599015439695681, "grad_norm": 0.16263742744922638, "learning_rate": 8.587753585050004e-05, "loss": 0.0433, "step": 16980 }, { "epoch": 7.60349071380622, "grad_norm": 0.17394129931926727, "learning_rate": 8.585833620086918e-05, "loss": 0.0415, "step": 16990 }, { "epoch": 7.60796598791676, "grad_norm": 0.17386116087436676, "learning_rate": 8.583912565842257e-05, "loss": 0.0381, "step": 17000 }, { "epoch": 7.612441262027299, "grad_norm": 0.1500382423400879, "learning_rate": 8.581990422899585e-05, "loss": 0.0412, "step": 17010 }, { "epoch": 7.616916536137839, "grad_norm": 0.20136655867099762, "learning_rate": 8.5800671918428e-05, "loss": 0.0454, "step": 17020 }, { "epoch": 7.621391810248378, "grad_norm": 0.20008623600006104, "learning_rate": 8.578142873256129e-05, "loss": 0.0456, "step": 17030 }, { "epoch": 7.625867084358917, "grad_norm": 0.1783890277147293, "learning_rate": 8.576217467724128e-05, "loss": 0.0403, "step": 17040 }, { "epoch": 7.6303423584694565, "grad_norm": 0.15848557651042938, "learning_rate": 8.574290975831685e-05, "loss": 0.0411, "step": 17050 }, { "epoch": 7.6348176325799955, "grad_norm": 0.16151514649391174, "learning_rate": 8.572363398164017e-05, "loss": 0.0404, "step": 17060 }, { "epoch": 7.639292906690535, "grad_norm": 0.1674869805574417, "learning_rate": 8.570434735306671e-05, "loss": 0.0474, "step": 17070 }, { "epoch": 7.643768180801074, "grad_norm": 0.15377040207386017, "learning_rate": 8.568504987845525e-05, "loss": 0.0427, "step": 17080 }, { "epoch": 7.648243454911613, "grad_norm": 0.19523070752620697, "learning_rate": 8.566574156366784e-05, "loss": 0.0404, "step": 17090 }, { "epoch": 7.652718729022153, "grad_norm": 0.15839941799640656, "learning_rate": 8.564642241456986e-05, "loss": 0.0436, "step": 17100 }, { "epoch": 7.657194003132692, "grad_norm": 0.1871986985206604, "learning_rate": 8.562709243702993e-05, "loss": 0.0423, "step": 17110 }, { "epoch": 7.661669277243231, "grad_norm": 0.1688004583120346, "learning_rate": 8.560775163691999e-05, "loss": 0.0393, "step": 17120 }, { "epoch": 7.666144551353771, "grad_norm": 0.18599019944667816, "learning_rate": 8.558840002011528e-05, "loss": 0.0419, "step": 17130 }, { "epoch": 7.67061982546431, "grad_norm": 0.1732429414987564, "learning_rate": 8.556903759249428e-05, "loss": 0.0434, "step": 17140 }, { "epoch": 7.675095099574849, "grad_norm": 0.15659387409687042, "learning_rate": 8.554966435993882e-05, "loss": 0.0385, "step": 17150 }, { "epoch": 7.679570373685388, "grad_norm": 0.19928762316703796, "learning_rate": 8.553028032833397e-05, "loss": 0.046, "step": 17160 }, { "epoch": 7.684045647795927, "grad_norm": 0.18971875309944153, "learning_rate": 8.551088550356807e-05, "loss": 0.0483, "step": 17170 }, { "epoch": 7.688520921906466, "grad_norm": 0.19395747780799866, "learning_rate": 8.549147989153276e-05, "loss": 0.0465, "step": 17180 }, { "epoch": 7.692996196017006, "grad_norm": 0.16702601313591003, "learning_rate": 8.547206349812298e-05, "loss": 0.0443, "step": 17190 }, { "epoch": 7.697471470127545, "grad_norm": 0.1702394187450409, "learning_rate": 8.545263632923687e-05, "loss": 0.0415, "step": 17200 }, { "epoch": 7.701946744238085, "grad_norm": 0.1592932641506195, "learning_rate": 8.543319839077593e-05, "loss": 0.043, "step": 17210 }, { "epoch": 7.706422018348624, "grad_norm": 0.18852868676185608, "learning_rate": 8.541374968864487e-05, "loss": 0.0408, "step": 17220 }, { "epoch": 7.710897292459163, "grad_norm": 0.1877792775630951, "learning_rate": 8.539429022875169e-05, "loss": 0.0423, "step": 17230 }, { "epoch": 7.715372566569703, "grad_norm": 0.17186564207077026, "learning_rate": 8.537482001700769e-05, "loss": 0.0425, "step": 17240 }, { "epoch": 7.7198478406802415, "grad_norm": 0.14808891713619232, "learning_rate": 8.535533905932738e-05, "loss": 0.0445, "step": 17250 }, { "epoch": 7.724323114790781, "grad_norm": 0.15498071908950806, "learning_rate": 8.533584736162857e-05, "loss": 0.0404, "step": 17260 }, { "epoch": 7.72879838890132, "grad_norm": 0.1496400386095047, "learning_rate": 8.531634492983232e-05, "loss": 0.0493, "step": 17270 }, { "epoch": 7.733273663011859, "grad_norm": 0.18842670321464539, "learning_rate": 8.529683176986295e-05, "loss": 0.0456, "step": 17280 }, { "epoch": 7.737748937122399, "grad_norm": 0.1803567111492157, "learning_rate": 8.527730788764805e-05, "loss": 0.0422, "step": 17290 }, { "epoch": 7.742224211232938, "grad_norm": 0.16920949518680573, "learning_rate": 8.525777328911846e-05, "loss": 0.0485, "step": 17300 }, { "epoch": 7.746699485343477, "grad_norm": 0.1650346964597702, "learning_rate": 8.523822798020827e-05, "loss": 0.0412, "step": 17310 }, { "epoch": 7.751174759454017, "grad_norm": 0.21293410658836365, "learning_rate": 8.521867196685482e-05, "loss": 0.0411, "step": 17320 }, { "epoch": 7.755650033564556, "grad_norm": 0.1987268477678299, "learning_rate": 8.519910525499874e-05, "loss": 0.0435, "step": 17330 }, { "epoch": 7.760125307675095, "grad_norm": 0.16560161113739014, "learning_rate": 8.517952785058385e-05, "loss": 0.0412, "step": 17340 }, { "epoch": 7.7646005817856345, "grad_norm": 0.23699040710926056, "learning_rate": 8.515993975955727e-05, "loss": 0.0417, "step": 17350 }, { "epoch": 7.769075855896173, "grad_norm": 0.20150898396968842, "learning_rate": 8.514034098786933e-05, "loss": 0.0395, "step": 17360 }, { "epoch": 7.773551130006713, "grad_norm": 0.18132302165031433, "learning_rate": 8.512073154147362e-05, "loss": 0.0443, "step": 17370 }, { "epoch": 7.778026404117252, "grad_norm": 0.17051373422145844, "learning_rate": 8.510111142632698e-05, "loss": 0.0475, "step": 17380 }, { "epoch": 7.782501678227791, "grad_norm": 0.21078500151634216, "learning_rate": 8.508148064838948e-05, "loss": 0.0415, "step": 17390 }, { "epoch": 7.786976952338331, "grad_norm": 0.17400966584682465, "learning_rate": 8.506183921362443e-05, "loss": 0.0428, "step": 17400 }, { "epoch": 7.79145222644887, "grad_norm": 0.19757600128650665, "learning_rate": 8.504218712799839e-05, "loss": 0.0438, "step": 17410 }, { "epoch": 7.79592750055941, "grad_norm": 0.16608186066150665, "learning_rate": 8.502252439748113e-05, "loss": 0.0403, "step": 17420 }, { "epoch": 7.800402774669949, "grad_norm": 0.19138208031654358, "learning_rate": 8.500285102804568e-05, "loss": 0.0415, "step": 17430 }, { "epoch": 7.804878048780488, "grad_norm": 0.17776839435100555, "learning_rate": 8.498316702566828e-05, "loss": 0.0407, "step": 17440 }, { "epoch": 7.8093533228910275, "grad_norm": 0.19174635410308838, "learning_rate": 8.49634723963284e-05, "loss": 0.0413, "step": 17450 }, { "epoch": 7.813828597001566, "grad_norm": 0.17764894664287567, "learning_rate": 8.494376714600878e-05, "loss": 0.0434, "step": 17460 }, { "epoch": 7.818303871112105, "grad_norm": 0.19904199242591858, "learning_rate": 8.492405128069534e-05, "loss": 0.0424, "step": 17470 }, { "epoch": 7.822779145222645, "grad_norm": 0.17637892067432404, "learning_rate": 8.490432480637723e-05, "loss": 0.0406, "step": 17480 }, { "epoch": 7.827254419333184, "grad_norm": 0.1908082813024521, "learning_rate": 8.488458772904684e-05, "loss": 0.0404, "step": 17490 }, { "epoch": 7.831729693443723, "grad_norm": 0.19079619646072388, "learning_rate": 8.486484005469977e-05, "loss": 0.044, "step": 17500 }, { "epoch": 7.836204967554263, "grad_norm": 0.1778876930475235, "learning_rate": 8.484508178933486e-05, "loss": 0.0411, "step": 17510 }, { "epoch": 7.840680241664802, "grad_norm": 0.1759481430053711, "learning_rate": 8.482531293895412e-05, "loss": 0.039, "step": 17520 }, { "epoch": 7.845155515775341, "grad_norm": 0.19925157725811005, "learning_rate": 8.480553350956282e-05, "loss": 0.041, "step": 17530 }, { "epoch": 7.849630789885881, "grad_norm": 0.1850070059299469, "learning_rate": 8.478574350716941e-05, "loss": 0.0457, "step": 17540 }, { "epoch": 7.8541060639964195, "grad_norm": 0.15439249575138092, "learning_rate": 8.476594293778561e-05, "loss": 0.0412, "step": 17550 }, { "epoch": 7.858581338106959, "grad_norm": 0.17578186094760895, "learning_rate": 8.474613180742628e-05, "loss": 0.0418, "step": 17560 }, { "epoch": 7.863056612217498, "grad_norm": 0.16213715076446533, "learning_rate": 8.472631012210953e-05, "loss": 0.0377, "step": 17570 }, { "epoch": 7.867531886328037, "grad_norm": 0.13764812052249908, "learning_rate": 8.470647788785665e-05, "loss": 0.0419, "step": 17580 }, { "epoch": 7.872007160438577, "grad_norm": 0.15505650639533997, "learning_rate": 8.468663511069217e-05, "loss": 0.0381, "step": 17590 }, { "epoch": 7.876482434549116, "grad_norm": 0.17459222674369812, "learning_rate": 8.466678179664379e-05, "loss": 0.0377, "step": 17600 }, { "epoch": 7.880957708659656, "grad_norm": 0.16940656304359436, "learning_rate": 8.46469179517424e-05, "loss": 0.0435, "step": 17610 }, { "epoch": 7.885432982770195, "grad_norm": 0.20491908490657806, "learning_rate": 8.462704358202216e-05, "loss": 0.0446, "step": 17620 }, { "epoch": 7.889908256880734, "grad_norm": 0.21064355969429016, "learning_rate": 8.460715869352035e-05, "loss": 0.0375, "step": 17630 }, { "epoch": 7.8943835309912735, "grad_norm": 0.16503319144248962, "learning_rate": 8.458726329227747e-05, "loss": 0.0421, "step": 17640 }, { "epoch": 7.8988588051018125, "grad_norm": 0.17245417833328247, "learning_rate": 8.456735738433723e-05, "loss": 0.0438, "step": 17650 }, { "epoch": 7.903334079212351, "grad_norm": 0.1671958714723587, "learning_rate": 8.454744097574652e-05, "loss": 0.0411, "step": 17660 }, { "epoch": 7.907809353322891, "grad_norm": 0.1745990365743637, "learning_rate": 8.452751407255541e-05, "loss": 0.044, "step": 17670 }, { "epoch": 7.91228462743343, "grad_norm": 0.207766592502594, "learning_rate": 8.450757668081716e-05, "loss": 0.0413, "step": 17680 }, { "epoch": 7.916759901543969, "grad_norm": 0.2363077849149704, "learning_rate": 8.448762880658825e-05, "loss": 0.0445, "step": 17690 }, { "epoch": 7.921235175654509, "grad_norm": 0.21152162551879883, "learning_rate": 8.44676704559283e-05, "loss": 0.0407, "step": 17700 }, { "epoch": 7.925710449765048, "grad_norm": 0.16946792602539062, "learning_rate": 8.444770163490012e-05, "loss": 0.0458, "step": 17710 }, { "epoch": 7.930185723875588, "grad_norm": 0.22300297021865845, "learning_rate": 8.442772234956972e-05, "loss": 0.0439, "step": 17720 }, { "epoch": 7.934660997986127, "grad_norm": 0.15968844294548035, "learning_rate": 8.44077326060063e-05, "loss": 0.0406, "step": 17730 }, { "epoch": 7.939136272096666, "grad_norm": 0.1568620353937149, "learning_rate": 8.438773241028219e-05, "loss": 0.0394, "step": 17740 }, { "epoch": 7.943611546207205, "grad_norm": 0.13806656002998352, "learning_rate": 8.436772176847294e-05, "loss": 0.0406, "step": 17750 }, { "epoch": 7.948086820317744, "grad_norm": 0.20839311182498932, "learning_rate": 8.434770068665723e-05, "loss": 0.0423, "step": 17760 }, { "epoch": 7.952562094428284, "grad_norm": 0.13916338980197906, "learning_rate": 8.432766917091694e-05, "loss": 0.0408, "step": 17770 }, { "epoch": 7.957037368538823, "grad_norm": 0.20485934615135193, "learning_rate": 8.430762722733714e-05, "loss": 0.0453, "step": 17780 }, { "epoch": 7.961512642649362, "grad_norm": 0.1515527218580246, "learning_rate": 8.428757486200603e-05, "loss": 0.0403, "step": 17790 }, { "epoch": 7.965987916759902, "grad_norm": 0.19497904181480408, "learning_rate": 8.4267512081015e-05, "loss": 0.0424, "step": 17800 }, { "epoch": 7.970463190870441, "grad_norm": 0.21014785766601562, "learning_rate": 8.424743889045856e-05, "loss": 0.0441, "step": 17810 }, { "epoch": 7.97493846498098, "grad_norm": 0.19362302124500275, "learning_rate": 8.422735529643444e-05, "loss": 0.0422, "step": 17820 }, { "epoch": 7.97941373909152, "grad_norm": 0.2240544706583023, "learning_rate": 8.420726130504351e-05, "loss": 0.045, "step": 17830 }, { "epoch": 7.983889013202059, "grad_norm": 0.20906029641628265, "learning_rate": 8.418715692238978e-05, "loss": 0.0422, "step": 17840 }, { "epoch": 7.9883642873125975, "grad_norm": 0.15100215375423431, "learning_rate": 8.416704215458043e-05, "loss": 0.0416, "step": 17850 }, { "epoch": 7.992839561423137, "grad_norm": 0.16829240322113037, "learning_rate": 8.41469170077258e-05, "loss": 0.0389, "step": 17860 }, { "epoch": 7.997314835533676, "grad_norm": 0.18242640793323517, "learning_rate": 8.41267814879394e-05, "loss": 0.0415, "step": 17870 }, { "epoch": 8.001790109644215, "grad_norm": 0.16094349324703217, "learning_rate": 8.410663560133784e-05, "loss": 0.0388, "step": 17880 }, { "epoch": 8.006265383754755, "grad_norm": 0.16191032528877258, "learning_rate": 8.40864793540409e-05, "loss": 0.0411, "step": 17890 }, { "epoch": 8.010740657865295, "grad_norm": 0.17440935969352722, "learning_rate": 8.406631275217156e-05, "loss": 0.0382, "step": 17900 }, { "epoch": 8.015215931975833, "grad_norm": 0.16645868122577667, "learning_rate": 8.404613580185585e-05, "loss": 0.0416, "step": 17910 }, { "epoch": 8.019691206086373, "grad_norm": 0.15888004004955292, "learning_rate": 8.402594850922305e-05, "loss": 0.0382, "step": 17920 }, { "epoch": 8.024166480196913, "grad_norm": 0.13496889173984528, "learning_rate": 8.400575088040548e-05, "loss": 0.0375, "step": 17930 }, { "epoch": 8.02864175430745, "grad_norm": 0.16777344048023224, "learning_rate": 8.398554292153866e-05, "loss": 0.0398, "step": 17940 }, { "epoch": 8.03311702841799, "grad_norm": 0.17544814944267273, "learning_rate": 8.396532463876124e-05, "loss": 0.0413, "step": 17950 }, { "epoch": 8.03759230252853, "grad_norm": 0.18175847828388214, "learning_rate": 8.394509603821499e-05, "loss": 0.041, "step": 17960 }, { "epoch": 8.042067576639068, "grad_norm": 0.1981024444103241, "learning_rate": 8.392485712604483e-05, "loss": 0.0389, "step": 17970 }, { "epoch": 8.046542850749608, "grad_norm": 0.18800285458564758, "learning_rate": 8.390460790839882e-05, "loss": 0.0388, "step": 17980 }, { "epoch": 8.051018124860148, "grad_norm": 0.1572868973016739, "learning_rate": 8.388434839142813e-05, "loss": 0.0441, "step": 17990 }, { "epoch": 8.055493398970688, "grad_norm": 0.1601586639881134, "learning_rate": 8.386407858128706e-05, "loss": 0.043, "step": 18000 }, { "epoch": 8.059968673081226, "grad_norm": 0.15234774351119995, "learning_rate": 8.384379848413304e-05, "loss": 0.0401, "step": 18010 }, { "epoch": 8.064443947191766, "grad_norm": 0.1440565586090088, "learning_rate": 8.382350810612663e-05, "loss": 0.0394, "step": 18020 }, { "epoch": 8.068919221302306, "grad_norm": 0.13715986907482147, "learning_rate": 8.380320745343153e-05, "loss": 0.0397, "step": 18030 }, { "epoch": 8.073394495412844, "grad_norm": 0.15956589579582214, "learning_rate": 8.378289653221452e-05, "loss": 0.0401, "step": 18040 }, { "epoch": 8.077869769523383, "grad_norm": 0.1480378359556198, "learning_rate": 8.376257534864553e-05, "loss": 0.0407, "step": 18050 }, { "epoch": 8.082345043633923, "grad_norm": 0.15881630778312683, "learning_rate": 8.37422439088976e-05, "loss": 0.0398, "step": 18060 }, { "epoch": 8.086820317744461, "grad_norm": 0.1258513480424881, "learning_rate": 8.372190221914689e-05, "loss": 0.037, "step": 18070 }, { "epoch": 8.091295591855001, "grad_norm": 0.19019441306591034, "learning_rate": 8.370155028557265e-05, "loss": 0.0393, "step": 18080 }, { "epoch": 8.095770865965541, "grad_norm": 0.1779077798128128, "learning_rate": 8.368118811435726e-05, "loss": 0.0426, "step": 18090 }, { "epoch": 8.100246140076079, "grad_norm": 0.16241447627544403, "learning_rate": 8.366081571168625e-05, "loss": 0.0384, "step": 18100 }, { "epoch": 8.104721414186619, "grad_norm": 0.15235112607479095, "learning_rate": 8.364043308374816e-05, "loss": 0.0438, "step": 18110 }, { "epoch": 8.109196688297159, "grad_norm": 0.15695619583129883, "learning_rate": 8.362004023673474e-05, "loss": 0.0434, "step": 18120 }, { "epoch": 8.113671962407697, "grad_norm": 0.19786696135997772, "learning_rate": 8.359963717684077e-05, "loss": 0.0419, "step": 18130 }, { "epoch": 8.118147236518237, "grad_norm": 0.16591528058052063, "learning_rate": 8.357922391026418e-05, "loss": 0.0428, "step": 18140 }, { "epoch": 8.122622510628776, "grad_norm": 0.21581807732582092, "learning_rate": 8.355880044320598e-05, "loss": 0.0454, "step": 18150 }, { "epoch": 8.127097784739314, "grad_norm": 0.19248178601264954, "learning_rate": 8.353836678187027e-05, "loss": 0.0415, "step": 18160 }, { "epoch": 8.131573058849854, "grad_norm": 0.1579846441745758, "learning_rate": 8.351792293246427e-05, "loss": 0.0389, "step": 18170 }, { "epoch": 8.136048332960394, "grad_norm": 0.17178694903850555, "learning_rate": 8.349746890119826e-05, "loss": 0.0394, "step": 18180 }, { "epoch": 8.140523607070934, "grad_norm": 0.16406051814556122, "learning_rate": 8.347700469428564e-05, "loss": 0.0438, "step": 18190 }, { "epoch": 8.144998881181472, "grad_norm": 0.16487258672714233, "learning_rate": 8.345653031794292e-05, "loss": 0.0477, "step": 18200 }, { "epoch": 8.149474155292012, "grad_norm": 0.17692898213863373, "learning_rate": 8.343604577838964e-05, "loss": 0.0431, "step": 18210 }, { "epoch": 8.153949429402552, "grad_norm": 0.19551418721675873, "learning_rate": 8.34155510818485e-05, "loss": 0.0392, "step": 18220 }, { "epoch": 8.15842470351309, "grad_norm": 0.18588562309741974, "learning_rate": 8.339504623454521e-05, "loss": 0.0433, "step": 18230 }, { "epoch": 8.16289997762363, "grad_norm": 0.16267746686935425, "learning_rate": 8.337453124270863e-05, "loss": 0.0417, "step": 18240 }, { "epoch": 8.16737525173417, "grad_norm": 0.17258945107460022, "learning_rate": 8.335400611257067e-05, "loss": 0.0456, "step": 18250 }, { "epoch": 8.171850525844707, "grad_norm": 0.14395061135292053, "learning_rate": 8.33334708503663e-05, "loss": 0.0393, "step": 18260 }, { "epoch": 8.176325799955247, "grad_norm": 0.177877277135849, "learning_rate": 8.331292546233362e-05, "loss": 0.0421, "step": 18270 }, { "epoch": 8.180801074065787, "grad_norm": 0.15900270640850067, "learning_rate": 8.329236995471373e-05, "loss": 0.0426, "step": 18280 }, { "epoch": 8.185276348176325, "grad_norm": 0.24632515013217926, "learning_rate": 8.327180433375091e-05, "loss": 0.0408, "step": 18290 }, { "epoch": 8.189751622286865, "grad_norm": 0.1613364964723587, "learning_rate": 8.32512286056924e-05, "loss": 0.0399, "step": 18300 }, { "epoch": 8.194226896397405, "grad_norm": 0.14324787259101868, "learning_rate": 8.323064277678862e-05, "loss": 0.0405, "step": 18310 }, { "epoch": 8.198702170507943, "grad_norm": 0.20135872066020966, "learning_rate": 8.321004685329296e-05, "loss": 0.0444, "step": 18320 }, { "epoch": 8.203177444618483, "grad_norm": 0.17169424891471863, "learning_rate": 8.318944084146192e-05, "loss": 0.04, "step": 18330 }, { "epoch": 8.207652718729022, "grad_norm": 0.16640296578407288, "learning_rate": 8.316882474755507e-05, "loss": 0.0398, "step": 18340 }, { "epoch": 8.212127992839562, "grad_norm": 0.16181710362434387, "learning_rate": 8.314819857783503e-05, "loss": 0.0412, "step": 18350 }, { "epoch": 8.2166032669501, "grad_norm": 0.17021405696868896, "learning_rate": 8.31275623385675e-05, "loss": 0.0414, "step": 18360 }, { "epoch": 8.22107854106064, "grad_norm": 0.15465416014194489, "learning_rate": 8.31069160360212e-05, "loss": 0.0428, "step": 18370 }, { "epoch": 8.22555381517118, "grad_norm": 0.17887504398822784, "learning_rate": 8.308625967646795e-05, "loss": 0.0364, "step": 18380 }, { "epoch": 8.230029089281718, "grad_norm": 0.16517432034015656, "learning_rate": 8.306559326618259e-05, "loss": 0.0422, "step": 18390 }, { "epoch": 8.234504363392258, "grad_norm": 0.1745709627866745, "learning_rate": 8.304491681144306e-05, "loss": 0.0376, "step": 18400 }, { "epoch": 8.238979637502798, "grad_norm": 0.19964104890823364, "learning_rate": 8.30242303185303e-05, "loss": 0.0428, "step": 18410 }, { "epoch": 8.243454911613336, "grad_norm": 0.15670078992843628, "learning_rate": 8.300353379372834e-05, "loss": 0.0382, "step": 18420 }, { "epoch": 8.247930185723876, "grad_norm": 0.16640672087669373, "learning_rate": 8.298282724332419e-05, "loss": 0.043, "step": 18430 }, { "epoch": 8.252405459834415, "grad_norm": 0.15126392245292664, "learning_rate": 8.2962110673608e-05, "loss": 0.043, "step": 18440 }, { "epoch": 8.256880733944953, "grad_norm": 0.1445978581905365, "learning_rate": 8.29413840908729e-05, "loss": 0.0401, "step": 18450 }, { "epoch": 8.261356008055493, "grad_norm": 0.18673735857009888, "learning_rate": 8.292064750141509e-05, "loss": 0.0423, "step": 18460 }, { "epoch": 8.265831282166033, "grad_norm": 0.16346865892410278, "learning_rate": 8.289990091153376e-05, "loss": 0.0407, "step": 18470 }, { "epoch": 8.270306556276571, "grad_norm": 0.14446404576301575, "learning_rate": 8.287914432753123e-05, "loss": 0.0413, "step": 18480 }, { "epoch": 8.274781830387111, "grad_norm": 0.20406803488731384, "learning_rate": 8.285837775571276e-05, "loss": 0.0411, "step": 18490 }, { "epoch": 8.27925710449765, "grad_norm": 0.1835588812828064, "learning_rate": 8.283760120238672e-05, "loss": 0.0382, "step": 18500 }, { "epoch": 8.283732378608189, "grad_norm": 0.19173060357570648, "learning_rate": 8.281681467386446e-05, "loss": 0.0399, "step": 18510 }, { "epoch": 8.288207652718729, "grad_norm": 0.19215047359466553, "learning_rate": 8.279601817646036e-05, "loss": 0.0452, "step": 18520 }, { "epoch": 8.292682926829269, "grad_norm": 0.17664578557014465, "learning_rate": 8.277521171649189e-05, "loss": 0.0428, "step": 18530 }, { "epoch": 8.297158200939808, "grad_norm": 0.17697691917419434, "learning_rate": 8.275439530027948e-05, "loss": 0.0418, "step": 18540 }, { "epoch": 8.301633475050346, "grad_norm": 0.18025827407836914, "learning_rate": 8.273356893414659e-05, "loss": 0.0421, "step": 18550 }, { "epoch": 8.306108749160886, "grad_norm": 0.16333407163619995, "learning_rate": 8.271273262441975e-05, "loss": 0.039, "step": 18560 }, { "epoch": 8.310584023271426, "grad_norm": 0.17039284110069275, "learning_rate": 8.269188637742846e-05, "loss": 0.0412, "step": 18570 }, { "epoch": 8.315059297381964, "grad_norm": 0.18630781769752502, "learning_rate": 8.267103019950529e-05, "loss": 0.0401, "step": 18580 }, { "epoch": 8.319534571492504, "grad_norm": 0.24738165736198425, "learning_rate": 8.265016409698573e-05, "loss": 0.0389, "step": 18590 }, { "epoch": 8.324009845603044, "grad_norm": 0.1588653326034546, "learning_rate": 8.262928807620843e-05, "loss": 0.0424, "step": 18600 }, { "epoch": 8.328485119713582, "grad_norm": 0.17721807956695557, "learning_rate": 8.260840214351493e-05, "loss": 0.038, "step": 18610 }, { "epoch": 8.332960393824122, "grad_norm": 0.17049318552017212, "learning_rate": 8.258750630524984e-05, "loss": 0.0431, "step": 18620 }, { "epoch": 8.337435667934662, "grad_norm": 0.15850508213043213, "learning_rate": 8.256660056776076e-05, "loss": 0.042, "step": 18630 }, { "epoch": 8.3419109420452, "grad_norm": 0.13138726353645325, "learning_rate": 8.254568493739828e-05, "loss": 0.0403, "step": 18640 }, { "epoch": 8.34638621615574, "grad_norm": 0.15832962095737457, "learning_rate": 8.252475942051605e-05, "loss": 0.0422, "step": 18650 }, { "epoch": 8.35086149026628, "grad_norm": 0.16532593965530396, "learning_rate": 8.250382402347065e-05, "loss": 0.0396, "step": 18660 }, { "epoch": 8.355336764376817, "grad_norm": 0.15333673357963562, "learning_rate": 8.248287875262176e-05, "loss": 0.0409, "step": 18670 }, { "epoch": 8.359812038487357, "grad_norm": 0.17704284191131592, "learning_rate": 8.246192361433196e-05, "loss": 0.0395, "step": 18680 }, { "epoch": 8.364287312597897, "grad_norm": 0.1785801649093628, "learning_rate": 8.244095861496686e-05, "loss": 0.0402, "step": 18690 }, { "epoch": 8.368762586708435, "grad_norm": 0.1447179913520813, "learning_rate": 8.241998376089508e-05, "loss": 0.0395, "step": 18700 }, { "epoch": 8.373237860818975, "grad_norm": 0.16925294697284698, "learning_rate": 8.239899905848825e-05, "loss": 0.0418, "step": 18710 }, { "epoch": 8.377713134929515, "grad_norm": 0.16454386711120605, "learning_rate": 8.237800451412095e-05, "loss": 0.0381, "step": 18720 }, { "epoch": 8.382188409040054, "grad_norm": 0.1487918496131897, "learning_rate": 8.235700013417076e-05, "loss": 0.0463, "step": 18730 }, { "epoch": 8.386663683150593, "grad_norm": 0.14021310210227966, "learning_rate": 8.233598592501828e-05, "loss": 0.0386, "step": 18740 }, { "epoch": 8.391138957261132, "grad_norm": 0.14230288565158844, "learning_rate": 8.231496189304704e-05, "loss": 0.0418, "step": 18750 }, { "epoch": 8.395614231371672, "grad_norm": 0.16103212535381317, "learning_rate": 8.229392804464362e-05, "loss": 0.0458, "step": 18760 }, { "epoch": 8.40008950548221, "grad_norm": 0.16708262264728546, "learning_rate": 8.227288438619754e-05, "loss": 0.0439, "step": 18770 }, { "epoch": 8.40456477959275, "grad_norm": 0.13513882458209991, "learning_rate": 8.225183092410128e-05, "loss": 0.0413, "step": 18780 }, { "epoch": 8.40904005370329, "grad_norm": 0.20790116488933563, "learning_rate": 8.223076766475035e-05, "loss": 0.0408, "step": 18790 }, { "epoch": 8.413515327813828, "grad_norm": 0.25717219710350037, "learning_rate": 8.220969461454322e-05, "loss": 0.0468, "step": 18800 }, { "epoch": 8.417990601924368, "grad_norm": 0.19646687805652618, "learning_rate": 8.218861177988129e-05, "loss": 0.041, "step": 18810 }, { "epoch": 8.422465876034908, "grad_norm": 0.16942273080348969, "learning_rate": 8.2167519167169e-05, "loss": 0.0407, "step": 18820 }, { "epoch": 8.426941150145446, "grad_norm": 0.19712670147418976, "learning_rate": 8.214641678281374e-05, "loss": 0.0404, "step": 18830 }, { "epoch": 8.431416424255985, "grad_norm": 0.19591599702835083, "learning_rate": 8.212530463322583e-05, "loss": 0.0387, "step": 18840 }, { "epoch": 8.435891698366525, "grad_norm": 0.15266451239585876, "learning_rate": 8.210418272481859e-05, "loss": 0.041, "step": 18850 }, { "epoch": 8.440366972477065, "grad_norm": 0.16012690961360931, "learning_rate": 8.20830510640083e-05, "loss": 0.0417, "step": 18860 }, { "epoch": 8.444842246587603, "grad_norm": 0.19774295389652252, "learning_rate": 8.206190965721419e-05, "loss": 0.0463, "step": 18870 }, { "epoch": 8.449317520698143, "grad_norm": 0.18598641455173492, "learning_rate": 8.204075851085849e-05, "loss": 0.0433, "step": 18880 }, { "epoch": 8.453792794808683, "grad_norm": 0.18102110922336578, "learning_rate": 8.201959763136633e-05, "loss": 0.0414, "step": 18890 }, { "epoch": 8.458268068919221, "grad_norm": 0.18005061149597168, "learning_rate": 8.199842702516583e-05, "loss": 0.0394, "step": 18900 }, { "epoch": 8.46274334302976, "grad_norm": 0.18674880266189575, "learning_rate": 8.197724669868807e-05, "loss": 0.0412, "step": 18910 }, { "epoch": 8.4672186171403, "grad_norm": 0.1428016722202301, "learning_rate": 8.19560566583671e-05, "loss": 0.0418, "step": 18920 }, { "epoch": 8.471693891250839, "grad_norm": 0.1658412218093872, "learning_rate": 8.193485691063985e-05, "loss": 0.0395, "step": 18930 }, { "epoch": 8.476169165361378, "grad_norm": 0.17503370344638824, "learning_rate": 8.191364746194625e-05, "loss": 0.0435, "step": 18940 }, { "epoch": 8.480644439471918, "grad_norm": 0.19122684001922607, "learning_rate": 8.18924283187292e-05, "loss": 0.0384, "step": 18950 }, { "epoch": 8.485119713582456, "grad_norm": 0.16498567163944244, "learning_rate": 8.18711994874345e-05, "loss": 0.0425, "step": 18960 }, { "epoch": 8.489594987692996, "grad_norm": 0.15283435583114624, "learning_rate": 8.184996097451089e-05, "loss": 0.0421, "step": 18970 }, { "epoch": 8.494070261803536, "grad_norm": 0.18122726678848267, "learning_rate": 8.182871278641009e-05, "loss": 0.0401, "step": 18980 }, { "epoch": 8.498545535914074, "grad_norm": 0.17255358397960663, "learning_rate": 8.180745492958674e-05, "loss": 0.0401, "step": 18990 }, { "epoch": 8.503020810024614, "grad_norm": 0.18882501125335693, "learning_rate": 8.178618741049842e-05, "loss": 0.0377, "step": 19000 }, { "epoch": 8.507496084135154, "grad_norm": 0.15743666887283325, "learning_rate": 8.17649102356056e-05, "loss": 0.0389, "step": 19010 }, { "epoch": 8.511971358245692, "grad_norm": 0.1576993614435196, "learning_rate": 8.174362341137177e-05, "loss": 0.0379, "step": 19020 }, { "epoch": 8.516446632356232, "grad_norm": 0.14886151254177094, "learning_rate": 8.172232694426329e-05, "loss": 0.0404, "step": 19030 }, { "epoch": 8.520921906466771, "grad_norm": 0.1576187163591385, "learning_rate": 8.170102084074946e-05, "loss": 0.0474, "step": 19040 }, { "epoch": 8.525397180577311, "grad_norm": 0.1579693704843521, "learning_rate": 8.167970510730253e-05, "loss": 0.038, "step": 19050 }, { "epoch": 8.52987245468785, "grad_norm": 0.1731068193912506, "learning_rate": 8.165837975039763e-05, "loss": 0.0385, "step": 19060 }, { "epoch": 8.534347728798389, "grad_norm": 0.1793048083782196, "learning_rate": 8.163704477651287e-05, "loss": 0.0389, "step": 19070 }, { "epoch": 8.538823002908929, "grad_norm": 0.1719820201396942, "learning_rate": 8.161570019212921e-05, "loss": 0.0442, "step": 19080 }, { "epoch": 8.543298277019467, "grad_norm": 0.1507747769355774, "learning_rate": 8.159434600373061e-05, "loss": 0.0408, "step": 19090 }, { "epoch": 8.547773551130007, "grad_norm": 0.1786649525165558, "learning_rate": 8.157298221780389e-05, "loss": 0.0399, "step": 19100 }, { "epoch": 8.552248825240547, "grad_norm": 0.1830642819404602, "learning_rate": 8.155160884083881e-05, "loss": 0.0425, "step": 19110 }, { "epoch": 8.556724099351085, "grad_norm": 0.17668302357196808, "learning_rate": 8.153022587932803e-05, "loss": 0.037, "step": 19120 }, { "epoch": 8.561199373461625, "grad_norm": 0.17067810893058777, "learning_rate": 8.150883333976713e-05, "loss": 0.0382, "step": 19130 }, { "epoch": 8.565674647572164, "grad_norm": 0.20199036598205566, "learning_rate": 8.148743122865463e-05, "loss": 0.0398, "step": 19140 }, { "epoch": 8.570149921682702, "grad_norm": 0.19285623729228973, "learning_rate": 8.146601955249188e-05, "loss": 0.0415, "step": 19150 }, { "epoch": 8.574625195793242, "grad_norm": 0.15654705464839935, "learning_rate": 8.14445983177832e-05, "loss": 0.0402, "step": 19160 }, { "epoch": 8.579100469903782, "grad_norm": 0.2264958620071411, "learning_rate": 8.14231675310358e-05, "loss": 0.0401, "step": 19170 }, { "epoch": 8.58357574401432, "grad_norm": 0.17234380543231964, "learning_rate": 8.140172719875979e-05, "loss": 0.0386, "step": 19180 }, { "epoch": 8.58805101812486, "grad_norm": 0.16621969640254974, "learning_rate": 8.138027732746818e-05, "loss": 0.0365, "step": 19190 }, { "epoch": 8.5925262922354, "grad_norm": 0.12648606300354004, "learning_rate": 8.135881792367686e-05, "loss": 0.0416, "step": 19200 }, { "epoch": 8.597001566345938, "grad_norm": 0.1707962155342102, "learning_rate": 8.133734899390464e-05, "loss": 0.0386, "step": 19210 }, { "epoch": 8.601476840456478, "grad_norm": 0.15935488045215607, "learning_rate": 8.13158705446732e-05, "loss": 0.0405, "step": 19220 }, { "epoch": 8.605952114567017, "grad_norm": 0.15911057591438293, "learning_rate": 8.129438258250712e-05, "loss": 0.0448, "step": 19230 }, { "epoch": 8.610427388677557, "grad_norm": 0.14056889712810516, "learning_rate": 8.127288511393392e-05, "loss": 0.0384, "step": 19240 }, { "epoch": 8.614902662788095, "grad_norm": 0.19229966402053833, "learning_rate": 8.125137814548393e-05, "loss": 0.0385, "step": 19250 }, { "epoch": 8.619377936898635, "grad_norm": 0.15280242264270782, "learning_rate": 8.12298616836904e-05, "loss": 0.0411, "step": 19260 }, { "epoch": 8.623853211009175, "grad_norm": 0.16818220913410187, "learning_rate": 8.120833573508948e-05, "loss": 0.0398, "step": 19270 }, { "epoch": 8.628328485119713, "grad_norm": 0.18204472959041595, "learning_rate": 8.118680030622014e-05, "loss": 0.0407, "step": 19280 }, { "epoch": 8.632803759230253, "grad_norm": 0.1771439015865326, "learning_rate": 8.116525540362434e-05, "loss": 0.0371, "step": 19290 }, { "epoch": 8.637279033340793, "grad_norm": 0.15769553184509277, "learning_rate": 8.114370103384681e-05, "loss": 0.0436, "step": 19300 }, { "epoch": 8.64175430745133, "grad_norm": 0.1917450875043869, "learning_rate": 8.11221372034352e-05, "loss": 0.0403, "step": 19310 }, { "epoch": 8.64622958156187, "grad_norm": 0.16497264802455902, "learning_rate": 8.110056391894005e-05, "loss": 0.0384, "step": 19320 }, { "epoch": 8.65070485567241, "grad_norm": 0.17460960149765015, "learning_rate": 8.107898118691473e-05, "loss": 0.0411, "step": 19330 }, { "epoch": 8.655180129782948, "grad_norm": 0.15202867984771729, "learning_rate": 8.105738901391552e-05, "loss": 0.0409, "step": 19340 }, { "epoch": 8.659655403893488, "grad_norm": 0.1843234896659851, "learning_rate": 8.103578740650156e-05, "loss": 0.0402, "step": 19350 }, { "epoch": 8.664130678004028, "grad_norm": 0.13112342357635498, "learning_rate": 8.101417637123484e-05, "loss": 0.0382, "step": 19360 }, { "epoch": 8.668605952114566, "grad_norm": 0.17027601599693298, "learning_rate": 8.099255591468022e-05, "loss": 0.0432, "step": 19370 }, { "epoch": 8.673081226225106, "grad_norm": 0.15578390657901764, "learning_rate": 8.097092604340542e-05, "loss": 0.0409, "step": 19380 }, { "epoch": 8.677556500335646, "grad_norm": 0.15828321874141693, "learning_rate": 8.094928676398101e-05, "loss": 0.0364, "step": 19390 }, { "epoch": 8.682031774446184, "grad_norm": 0.1752287745475769, "learning_rate": 8.092763808298048e-05, "loss": 0.0382, "step": 19400 }, { "epoch": 8.686507048556724, "grad_norm": 0.16957317292690277, "learning_rate": 8.090598000698009e-05, "loss": 0.0403, "step": 19410 }, { "epoch": 8.690982322667264, "grad_norm": 0.15065675973892212, "learning_rate": 8.088431254255899e-05, "loss": 0.0418, "step": 19420 }, { "epoch": 8.695457596777803, "grad_norm": 0.17671872675418854, "learning_rate": 8.086263569629919e-05, "loss": 0.0374, "step": 19430 }, { "epoch": 8.699932870888341, "grad_norm": 0.14198876917362213, "learning_rate": 8.084094947478556e-05, "loss": 0.0415, "step": 19440 }, { "epoch": 8.704408144998881, "grad_norm": 0.16078819334506989, "learning_rate": 8.081925388460578e-05, "loss": 0.0391, "step": 19450 }, { "epoch": 8.708883419109421, "grad_norm": 0.16146573424339294, "learning_rate": 8.07975489323504e-05, "loss": 0.0377, "step": 19460 }, { "epoch": 8.71335869321996, "grad_norm": 0.18861082196235657, "learning_rate": 8.077583462461283e-05, "loss": 0.042, "step": 19470 }, { "epoch": 8.717833967330499, "grad_norm": 0.16737928986549377, "learning_rate": 8.075411096798928e-05, "loss": 0.0374, "step": 19480 }, { "epoch": 8.722309241441039, "grad_norm": 0.15085355937480927, "learning_rate": 8.073237796907882e-05, "loss": 0.0355, "step": 19490 }, { "epoch": 8.726784515551577, "grad_norm": 0.18412497639656067, "learning_rate": 8.07106356344834e-05, "loss": 0.0427, "step": 19500 }, { "epoch": 8.731259789662117, "grad_norm": 0.1631443351507187, "learning_rate": 8.068888397080772e-05, "loss": 0.0379, "step": 19510 }, { "epoch": 8.735735063772657, "grad_norm": 0.15567100048065186, "learning_rate": 8.06671229846594e-05, "loss": 0.0413, "step": 19520 }, { "epoch": 8.740210337883195, "grad_norm": 0.16631631553173065, "learning_rate": 8.064535268264883e-05, "loss": 0.0378, "step": 19530 }, { "epoch": 8.744685611993734, "grad_norm": 0.15178993344306946, "learning_rate": 8.062357307138926e-05, "loss": 0.0378, "step": 19540 }, { "epoch": 8.749160886104274, "grad_norm": 0.14977361261844635, "learning_rate": 8.060178415749674e-05, "loss": 0.0376, "step": 19550 }, { "epoch": 8.753636160214814, "grad_norm": 0.15547321736812592, "learning_rate": 8.057998594759022e-05, "loss": 0.0377, "step": 19560 }, { "epoch": 8.758111434325352, "grad_norm": 0.1816289722919464, "learning_rate": 8.055817844829136e-05, "loss": 0.0367, "step": 19570 }, { "epoch": 8.762586708435892, "grad_norm": 0.2102845311164856, "learning_rate": 8.053636166622476e-05, "loss": 0.0397, "step": 19580 }, { "epoch": 8.767061982546432, "grad_norm": 0.1652090698480606, "learning_rate": 8.051453560801772e-05, "loss": 0.0424, "step": 19590 }, { "epoch": 8.77153725665697, "grad_norm": 0.1739332675933838, "learning_rate": 8.049270028030046e-05, "loss": 0.0384, "step": 19600 }, { "epoch": 8.77601253076751, "grad_norm": 0.15926113724708557, "learning_rate": 8.047085568970598e-05, "loss": 0.0394, "step": 19610 }, { "epoch": 8.78048780487805, "grad_norm": 0.1572796106338501, "learning_rate": 8.044900184287007e-05, "loss": 0.0393, "step": 19620 }, { "epoch": 8.784963078988588, "grad_norm": 0.16606032848358154, "learning_rate": 8.042713874643136e-05, "loss": 0.0403, "step": 19630 }, { "epoch": 8.789438353099127, "grad_norm": 0.14822310209274292, "learning_rate": 8.040526640703128e-05, "loss": 0.0402, "step": 19640 }, { "epoch": 8.793913627209667, "grad_norm": 0.15091931819915771, "learning_rate": 8.038338483131407e-05, "loss": 0.0354, "step": 19650 }, { "epoch": 8.798388901320205, "grad_norm": 0.1432708352804184, "learning_rate": 8.036149402592676e-05, "loss": 0.0436, "step": 19660 }, { "epoch": 8.802864175430745, "grad_norm": 0.1818658411502838, "learning_rate": 8.033959399751924e-05, "loss": 0.0417, "step": 19670 }, { "epoch": 8.807339449541285, "grad_norm": 0.16857483983039856, "learning_rate": 8.031768475274413e-05, "loss": 0.0415, "step": 19680 }, { "epoch": 8.811814723651823, "grad_norm": 0.17114350199699402, "learning_rate": 8.029576629825687e-05, "loss": 0.0408, "step": 19690 }, { "epoch": 8.816289997762363, "grad_norm": 0.17276006937026978, "learning_rate": 8.027383864071573e-05, "loss": 0.0426, "step": 19700 }, { "epoch": 8.820765271872903, "grad_norm": 0.15119463205337524, "learning_rate": 8.025190178678175e-05, "loss": 0.0431, "step": 19710 }, { "epoch": 8.82524054598344, "grad_norm": 0.15425275266170502, "learning_rate": 8.022995574311876e-05, "loss": 0.0383, "step": 19720 }, { "epoch": 8.82971582009398, "grad_norm": 0.19350172579288483, "learning_rate": 8.020800051639337e-05, "loss": 0.044, "step": 19730 }, { "epoch": 8.83419109420452, "grad_norm": 0.1555086225271225, "learning_rate": 8.018603611327504e-05, "loss": 0.0384, "step": 19740 }, { "epoch": 8.83866636831506, "grad_norm": 0.1573864072561264, "learning_rate": 8.016406254043595e-05, "loss": 0.0346, "step": 19750 }, { "epoch": 8.843141642425598, "grad_norm": 0.13461627066135406, "learning_rate": 8.01420798045511e-05, "loss": 0.038, "step": 19760 }, { "epoch": 8.847616916536138, "grad_norm": 0.1446070820093155, "learning_rate": 8.012008791229826e-05, "loss": 0.0413, "step": 19770 }, { "epoch": 8.852092190646678, "grad_norm": 0.16040769219398499, "learning_rate": 8.009808687035798e-05, "loss": 0.0377, "step": 19780 }, { "epoch": 8.856567464757216, "grad_norm": 0.17050565779209137, "learning_rate": 8.007607668541362e-05, "loss": 0.0412, "step": 19790 }, { "epoch": 8.861042738867756, "grad_norm": 0.1698427051305771, "learning_rate": 8.005405736415126e-05, "loss": 0.039, "step": 19800 }, { "epoch": 8.865518012978296, "grad_norm": 0.13622227311134338, "learning_rate": 8.00320289132598e-05, "loss": 0.0432, "step": 19810 }, { "epoch": 8.869993287088834, "grad_norm": 0.12565262615680695, "learning_rate": 8.000999133943093e-05, "loss": 0.0391, "step": 19820 }, { "epoch": 8.874468561199373, "grad_norm": 0.1646106243133545, "learning_rate": 7.998794464935904e-05, "loss": 0.0435, "step": 19830 }, { "epoch": 8.878943835309913, "grad_norm": 0.16406939923763275, "learning_rate": 7.996588884974135e-05, "loss": 0.0379, "step": 19840 }, { "epoch": 8.883419109420451, "grad_norm": 0.18040162324905396, "learning_rate": 7.994382394727784e-05, "loss": 0.0442, "step": 19850 }, { "epoch": 8.887894383530991, "grad_norm": 0.18886779248714447, "learning_rate": 7.992174994867123e-05, "loss": 0.0398, "step": 19860 }, { "epoch": 8.892369657641531, "grad_norm": 0.16738328337669373, "learning_rate": 7.989966686062702e-05, "loss": 0.0398, "step": 19870 }, { "epoch": 8.896844931752069, "grad_norm": 0.19806918501853943, "learning_rate": 7.987757468985348e-05, "loss": 0.0387, "step": 19880 }, { "epoch": 8.901320205862609, "grad_norm": 0.15400654077529907, "learning_rate": 7.985547344306161e-05, "loss": 0.0387, "step": 19890 }, { "epoch": 8.905795479973149, "grad_norm": 0.17706851661205292, "learning_rate": 7.983336312696522e-05, "loss": 0.0427, "step": 19900 }, { "epoch": 8.910270754083687, "grad_norm": 0.15494975447654724, "learning_rate": 7.98112437482808e-05, "loss": 0.0415, "step": 19910 }, { "epoch": 8.914746028194227, "grad_norm": 0.1616833359003067, "learning_rate": 7.978911531372765e-05, "loss": 0.0403, "step": 19920 }, { "epoch": 8.919221302304766, "grad_norm": 0.15933284163475037, "learning_rate": 7.97669778300278e-05, "loss": 0.0399, "step": 19930 }, { "epoch": 8.923696576415306, "grad_norm": 0.17767131328582764, "learning_rate": 7.974483130390604e-05, "loss": 0.0418, "step": 19940 }, { "epoch": 8.928171850525844, "grad_norm": 0.17365208268165588, "learning_rate": 7.972267574208991e-05, "loss": 0.0404, "step": 19950 }, { "epoch": 8.932647124636384, "grad_norm": 0.1366056650876999, "learning_rate": 7.970051115130966e-05, "loss": 0.0405, "step": 19960 }, { "epoch": 8.937122398746924, "grad_norm": 0.16909989714622498, "learning_rate": 7.96783375382983e-05, "loss": 0.036, "step": 19970 }, { "epoch": 8.941597672857462, "grad_norm": 0.15397602319717407, "learning_rate": 7.965615490979163e-05, "loss": 0.0388, "step": 19980 }, { "epoch": 8.946072946968002, "grad_norm": 0.167382150888443, "learning_rate": 7.963396327252812e-05, "loss": 0.0415, "step": 19990 }, { "epoch": 8.950548221078542, "grad_norm": 0.16557110846042633, "learning_rate": 7.961176263324901e-05, "loss": 0.0432, "step": 20000 }, { "epoch": 8.95502349518908, "grad_norm": 0.1581326127052307, "learning_rate": 7.958955299869825e-05, "loss": 0.0406, "step": 20010 }, { "epoch": 8.95949876929962, "grad_norm": 0.16891135275363922, "learning_rate": 7.956733437562259e-05, "loss": 0.0407, "step": 20020 }, { "epoch": 8.96397404341016, "grad_norm": 0.17293046414852142, "learning_rate": 7.954510677077138e-05, "loss": 0.0405, "step": 20030 }, { "epoch": 8.968449317520697, "grad_norm": 0.17652945220470428, "learning_rate": 7.952287019089685e-05, "loss": 0.0444, "step": 20040 }, { "epoch": 8.972924591631237, "grad_norm": 0.15324997901916504, "learning_rate": 7.950062464275387e-05, "loss": 0.0415, "step": 20050 }, { "epoch": 8.977399865741777, "grad_norm": 0.16913047432899475, "learning_rate": 7.947837013310005e-05, "loss": 0.0342, "step": 20060 }, { "epoch": 8.981875139852317, "grad_norm": 0.1839042752981186, "learning_rate": 7.945610666869568e-05, "loss": 0.0404, "step": 20070 }, { "epoch": 8.986350413962855, "grad_norm": 0.15709927678108215, "learning_rate": 7.943383425630387e-05, "loss": 0.0413, "step": 20080 }, { "epoch": 8.990825688073395, "grad_norm": 0.15723516047000885, "learning_rate": 7.941155290269038e-05, "loss": 0.04, "step": 20090 }, { "epoch": 8.995300962183933, "grad_norm": 0.1660551130771637, "learning_rate": 7.938926261462366e-05, "loss": 0.0402, "step": 20100 }, { "epoch": 8.999776236294473, "grad_norm": 0.1746511310338974, "learning_rate": 7.936696339887494e-05, "loss": 0.0371, "step": 20110 }, { "epoch": 9.004251510405012, "grad_norm": 0.17823758721351624, "learning_rate": 7.934465526221815e-05, "loss": 0.0399, "step": 20120 }, { "epoch": 9.008726784515552, "grad_norm": 0.17889602482318878, "learning_rate": 7.932233821142987e-05, "loss": 0.0438, "step": 20130 }, { "epoch": 9.01320205862609, "grad_norm": 0.1629742980003357, "learning_rate": 7.930001225328946e-05, "loss": 0.0363, "step": 20140 }, { "epoch": 9.01767733273663, "grad_norm": 0.1573507934808731, "learning_rate": 7.927767739457897e-05, "loss": 0.0424, "step": 20150 }, { "epoch": 9.02215260684717, "grad_norm": 0.18235711753368378, "learning_rate": 7.925533364208309e-05, "loss": 0.0393, "step": 20160 }, { "epoch": 9.026627880957708, "grad_norm": 0.16671296954154968, "learning_rate": 7.923298100258929e-05, "loss": 0.0387, "step": 20170 }, { "epoch": 9.031103155068248, "grad_norm": 0.17357133328914642, "learning_rate": 7.921061948288773e-05, "loss": 0.0433, "step": 20180 }, { "epoch": 9.035578429178788, "grad_norm": 0.1585456281900406, "learning_rate": 7.918824908977123e-05, "loss": 0.0412, "step": 20190 }, { "epoch": 9.040053703289326, "grad_norm": 0.19196617603302002, "learning_rate": 7.916586983003533e-05, "loss": 0.0377, "step": 20200 }, { "epoch": 9.044528977399866, "grad_norm": 0.1557106375694275, "learning_rate": 7.914348171047825e-05, "loss": 0.0388, "step": 20210 }, { "epoch": 9.049004251510405, "grad_norm": 0.1792958378791809, "learning_rate": 7.912108473790092e-05, "loss": 0.0441, "step": 20220 }, { "epoch": 9.053479525620943, "grad_norm": 0.15843865275382996, "learning_rate": 7.909867891910694e-05, "loss": 0.0383, "step": 20230 }, { "epoch": 9.057954799731483, "grad_norm": 0.17598851025104523, "learning_rate": 7.907626426090262e-05, "loss": 0.0412, "step": 20240 }, { "epoch": 9.062430073842023, "grad_norm": 0.18964408338069916, "learning_rate": 7.905384077009693e-05, "loss": 0.039, "step": 20250 }, { "epoch": 9.066905347952563, "grad_norm": 0.1868501454591751, "learning_rate": 7.903140845350153e-05, "loss": 0.0385, "step": 20260 }, { "epoch": 9.071380622063101, "grad_norm": 0.1877356916666031, "learning_rate": 7.900896731793077e-05, "loss": 0.0397, "step": 20270 }, { "epoch": 9.07585589617364, "grad_norm": 0.17525964975357056, "learning_rate": 7.898651737020166e-05, "loss": 0.0384, "step": 20280 }, { "epoch": 9.08033117028418, "grad_norm": 0.18225665390491486, "learning_rate": 7.896405861713394e-05, "loss": 0.0377, "step": 20290 }, { "epoch": 9.084806444394719, "grad_norm": 0.17323312163352966, "learning_rate": 7.894159106554997e-05, "loss": 0.039, "step": 20300 }, { "epoch": 9.089281718505259, "grad_norm": 0.18675650656223297, "learning_rate": 7.891911472227478e-05, "loss": 0.0402, "step": 20310 }, { "epoch": 9.093756992615798, "grad_norm": 0.15357626974582672, "learning_rate": 7.88966295941361e-05, "loss": 0.0388, "step": 20320 }, { "epoch": 9.098232266726336, "grad_norm": 0.14533217251300812, "learning_rate": 7.887413568796433e-05, "loss": 0.0403, "step": 20330 }, { "epoch": 9.102707540836876, "grad_norm": 0.16868790984153748, "learning_rate": 7.88516330105925e-05, "loss": 0.0377, "step": 20340 }, { "epoch": 9.107182814947416, "grad_norm": 0.20182004570960999, "learning_rate": 7.882912156885637e-05, "loss": 0.041, "step": 20350 }, { "epoch": 9.111658089057954, "grad_norm": 0.16571520268917084, "learning_rate": 7.880660136959428e-05, "loss": 0.0405, "step": 20360 }, { "epoch": 9.116133363168494, "grad_norm": 0.15195785462856293, "learning_rate": 7.878407241964729e-05, "loss": 0.0385, "step": 20370 }, { "epoch": 9.120608637279034, "grad_norm": 0.16694459319114685, "learning_rate": 7.87615347258591e-05, "loss": 0.0417, "step": 20380 }, { "epoch": 9.125083911389572, "grad_norm": 0.146088108420372, "learning_rate": 7.873898829507606e-05, "loss": 0.0386, "step": 20390 }, { "epoch": 9.129559185500112, "grad_norm": 0.13032561540603638, "learning_rate": 7.871643313414718e-05, "loss": 0.0361, "step": 20400 }, { "epoch": 9.134034459610652, "grad_norm": 0.1547931730747223, "learning_rate": 7.869386924992414e-05, "loss": 0.0404, "step": 20410 }, { "epoch": 9.13850973372119, "grad_norm": 0.17684489488601685, "learning_rate": 7.867129664926123e-05, "loss": 0.039, "step": 20420 }, { "epoch": 9.14298500783173, "grad_norm": 0.14027513563632965, "learning_rate": 7.864871533901544e-05, "loss": 0.0375, "step": 20430 }, { "epoch": 9.14746028194227, "grad_norm": 0.17037677764892578, "learning_rate": 7.862612532604632e-05, "loss": 0.0403, "step": 20440 }, { "epoch": 9.151935556052809, "grad_norm": 0.14389541745185852, "learning_rate": 7.860352661721619e-05, "loss": 0.0401, "step": 20450 }, { "epoch": 9.156410830163347, "grad_norm": 0.20722858607769012, "learning_rate": 7.858091921938988e-05, "loss": 0.0386, "step": 20460 }, { "epoch": 9.160886104273887, "grad_norm": 0.16800668835639954, "learning_rate": 7.855830313943497e-05, "loss": 0.0397, "step": 20470 }, { "epoch": 9.165361378384427, "grad_norm": 0.17797493934631348, "learning_rate": 7.85356783842216e-05, "loss": 0.0382, "step": 20480 }, { "epoch": 9.169836652494965, "grad_norm": 0.18523593246936798, "learning_rate": 7.851304496062254e-05, "loss": 0.0422, "step": 20490 }, { "epoch": 9.174311926605505, "grad_norm": 0.1324145644903183, "learning_rate": 7.849040287551331e-05, "loss": 0.0384, "step": 20500 }, { "epoch": 9.178787200716044, "grad_norm": 0.173782080411911, "learning_rate": 7.846775213577192e-05, "loss": 0.0373, "step": 20510 }, { "epoch": 9.183262474826583, "grad_norm": 0.16842181980609894, "learning_rate": 7.844509274827907e-05, "loss": 0.0374, "step": 20520 }, { "epoch": 9.187737748937122, "grad_norm": 0.16673138737678528, "learning_rate": 7.842242471991809e-05, "loss": 0.036, "step": 20530 }, { "epoch": 9.192213023047662, "grad_norm": 0.16775918006896973, "learning_rate": 7.839974805757496e-05, "loss": 0.0379, "step": 20540 }, { "epoch": 9.1966882971582, "grad_norm": 0.1628389209508896, "learning_rate": 7.837706276813819e-05, "loss": 0.0376, "step": 20550 }, { "epoch": 9.20116357126874, "grad_norm": 0.1520315557718277, "learning_rate": 7.835436885849902e-05, "loss": 0.0407, "step": 20560 }, { "epoch": 9.20563884537928, "grad_norm": 0.19041572511196136, "learning_rate": 7.833166633555124e-05, "loss": 0.0388, "step": 20570 }, { "epoch": 9.210114119489818, "grad_norm": 0.15536759793758392, "learning_rate": 7.830895520619128e-05, "loss": 0.0367, "step": 20580 }, { "epoch": 9.214589393600358, "grad_norm": 0.16506603360176086, "learning_rate": 7.828623547731818e-05, "loss": 0.04, "step": 20590 }, { "epoch": 9.219064667710898, "grad_norm": 0.13482585549354553, "learning_rate": 7.82635071558336e-05, "loss": 0.038, "step": 20600 }, { "epoch": 9.223539941821436, "grad_norm": 0.16073353588581085, "learning_rate": 7.824077024864179e-05, "loss": 0.0372, "step": 20610 }, { "epoch": 9.228015215931975, "grad_norm": 0.16771650314331055, "learning_rate": 7.821802476264966e-05, "loss": 0.0376, "step": 20620 }, { "epoch": 9.232490490042515, "grad_norm": 0.19442246854305267, "learning_rate": 7.819527070476665e-05, "loss": 0.0378, "step": 20630 }, { "epoch": 9.236965764153055, "grad_norm": 0.16604088246822357, "learning_rate": 7.817250808190483e-05, "loss": 0.04, "step": 20640 }, { "epoch": 9.241441038263593, "grad_norm": 0.151402547955513, "learning_rate": 7.814973690097893e-05, "loss": 0.0416, "step": 20650 }, { "epoch": 9.245916312374133, "grad_norm": 0.16359202563762665, "learning_rate": 7.81269571689062e-05, "loss": 0.0368, "step": 20660 }, { "epoch": 9.250391586484673, "grad_norm": 0.16010741889476776, "learning_rate": 7.810416889260653e-05, "loss": 0.0401, "step": 20670 }, { "epoch": 9.254866860595211, "grad_norm": 0.1360120326280594, "learning_rate": 7.808137207900241e-05, "loss": 0.0386, "step": 20680 }, { "epoch": 9.25934213470575, "grad_norm": 0.1442594677209854, "learning_rate": 7.80585667350189e-05, "loss": 0.0403, "step": 20690 }, { "epoch": 9.26381740881629, "grad_norm": 0.18411573767662048, "learning_rate": 7.803575286758364e-05, "loss": 0.0432, "step": 20700 }, { "epoch": 9.268292682926829, "grad_norm": 0.19794781506061554, "learning_rate": 7.801293048362691e-05, "loss": 0.0347, "step": 20710 }, { "epoch": 9.272767957037368, "grad_norm": 0.13729265332221985, "learning_rate": 7.799009959008155e-05, "loss": 0.0425, "step": 20720 }, { "epoch": 9.277243231147908, "grad_norm": 0.14450401067733765, "learning_rate": 7.796726019388295e-05, "loss": 0.0372, "step": 20730 }, { "epoch": 9.281718505258446, "grad_norm": 0.1591976433992386, "learning_rate": 7.794441230196913e-05, "loss": 0.0412, "step": 20740 }, { "epoch": 9.286193779368986, "grad_norm": 0.1561484932899475, "learning_rate": 7.79215559212807e-05, "loss": 0.0383, "step": 20750 }, { "epoch": 9.290669053479526, "grad_norm": 0.17065466940402985, "learning_rate": 7.789869105876083e-05, "loss": 0.0377, "step": 20760 }, { "epoch": 9.295144327590064, "grad_norm": 0.19332930445671082, "learning_rate": 7.78758177213552e-05, "loss": 0.0361, "step": 20770 }, { "epoch": 9.299619601700604, "grad_norm": 0.18767450749874115, "learning_rate": 7.785293591601217e-05, "loss": 0.0383, "step": 20780 }, { "epoch": 9.304094875811144, "grad_norm": 0.17139902710914612, "learning_rate": 7.783004564968263e-05, "loss": 0.0403, "step": 20790 }, { "epoch": 9.308570149921684, "grad_norm": 0.16798484325408936, "learning_rate": 7.780714692932002e-05, "loss": 0.0374, "step": 20800 }, { "epoch": 9.313045424032222, "grad_norm": 0.15478159487247467, "learning_rate": 7.77842397618804e-05, "loss": 0.0386, "step": 20810 }, { "epoch": 9.317520698142761, "grad_norm": 0.16696469485759735, "learning_rate": 7.776132415432234e-05, "loss": 0.0394, "step": 20820 }, { "epoch": 9.321995972253301, "grad_norm": 0.15748733282089233, "learning_rate": 7.773840011360698e-05, "loss": 0.037, "step": 20830 }, { "epoch": 9.32647124636384, "grad_norm": 0.18129828572273254, "learning_rate": 7.771546764669807e-05, "loss": 0.0374, "step": 20840 }, { "epoch": 9.33094652047438, "grad_norm": 0.1664133071899414, "learning_rate": 7.769252676056187e-05, "loss": 0.0396, "step": 20850 }, { "epoch": 9.335421794584919, "grad_norm": 0.12690375745296478, "learning_rate": 7.766957746216721e-05, "loss": 0.0357, "step": 20860 }, { "epoch": 9.339897068695457, "grad_norm": 0.15506668388843536, "learning_rate": 7.76466197584855e-05, "loss": 0.0392, "step": 20870 }, { "epoch": 9.344372342805997, "grad_norm": 0.1451631784439087, "learning_rate": 7.762365365649067e-05, "loss": 0.0365, "step": 20880 }, { "epoch": 9.348847616916537, "grad_norm": 0.1312265694141388, "learning_rate": 7.760067916315921e-05, "loss": 0.0408, "step": 20890 }, { "epoch": 9.353322891027075, "grad_norm": 0.1583760678768158, "learning_rate": 7.757769628547018e-05, "loss": 0.0407, "step": 20900 }, { "epoch": 9.357798165137615, "grad_norm": 0.1829310655593872, "learning_rate": 7.755470503040516e-05, "loss": 0.0354, "step": 20910 }, { "epoch": 9.362273439248154, "grad_norm": 0.16932767629623413, "learning_rate": 7.753170540494832e-05, "loss": 0.0401, "step": 20920 }, { "epoch": 9.366748713358692, "grad_norm": 0.1413695067167282, "learning_rate": 7.750869741608628e-05, "loss": 0.0357, "step": 20930 }, { "epoch": 9.371223987469232, "grad_norm": 0.1565260887145996, "learning_rate": 7.748568107080832e-05, "loss": 0.0393, "step": 20940 }, { "epoch": 9.375699261579772, "grad_norm": 0.17270433902740479, "learning_rate": 7.746265637610613e-05, "loss": 0.0418, "step": 20950 }, { "epoch": 9.380174535690312, "grad_norm": 0.1534745693206787, "learning_rate": 7.743962333897405e-05, "loss": 0.0348, "step": 20960 }, { "epoch": 9.38464980980085, "grad_norm": 0.14774189889431, "learning_rate": 7.741658196640892e-05, "loss": 0.0406, "step": 20970 }, { "epoch": 9.38912508391139, "grad_norm": 0.1991555392742157, "learning_rate": 7.739353226541009e-05, "loss": 0.0406, "step": 20980 }, { "epoch": 9.39360035802193, "grad_norm": 0.17126138508319855, "learning_rate": 7.737047424297941e-05, "loss": 0.041, "step": 20990 }, { "epoch": 9.398075632132468, "grad_norm": 0.1619710624217987, "learning_rate": 7.734740790612136e-05, "loss": 0.0333, "step": 21000 }, { "epoch": 9.402550906243007, "grad_norm": 0.16020983457565308, "learning_rate": 7.732433326184283e-05, "loss": 0.0375, "step": 21010 }, { "epoch": 9.407026180353547, "grad_norm": 0.19736595451831818, "learning_rate": 7.730125031715331e-05, "loss": 0.037, "step": 21020 }, { "epoch": 9.411501454464085, "grad_norm": 0.16664858162403107, "learning_rate": 7.727815907906481e-05, "loss": 0.0383, "step": 21030 }, { "epoch": 9.415976728574625, "grad_norm": 0.14720173180103302, "learning_rate": 7.725505955459183e-05, "loss": 0.0438, "step": 21040 }, { "epoch": 9.420452002685165, "grad_norm": 0.1406787782907486, "learning_rate": 7.723195175075136e-05, "loss": 0.0326, "step": 21050 }, { "epoch": 9.424927276795703, "grad_norm": 0.13875225186347961, "learning_rate": 7.720883567456298e-05, "loss": 0.0406, "step": 21060 }, { "epoch": 9.429402550906243, "grad_norm": 0.1695553958415985, "learning_rate": 7.718571133304871e-05, "loss": 0.0362, "step": 21070 }, { "epoch": 9.433877825016783, "grad_norm": 0.15436473488807678, "learning_rate": 7.716257873323316e-05, "loss": 0.0405, "step": 21080 }, { "epoch": 9.43835309912732, "grad_norm": 0.1372680366039276, "learning_rate": 7.713943788214337e-05, "loss": 0.0382, "step": 21090 }, { "epoch": 9.44282837323786, "grad_norm": 0.15681971609592438, "learning_rate": 7.711628878680892e-05, "loss": 0.0384, "step": 21100 }, { "epoch": 9.4473036473484, "grad_norm": 0.14430555701255798, "learning_rate": 7.70931314542619e-05, "loss": 0.037, "step": 21110 }, { "epoch": 9.451778921458938, "grad_norm": 0.14131329953670502, "learning_rate": 7.70699658915369e-05, "loss": 0.0345, "step": 21120 }, { "epoch": 9.456254195569478, "grad_norm": 0.1547747701406479, "learning_rate": 7.7046792105671e-05, "loss": 0.0418, "step": 21130 }, { "epoch": 9.460729469680018, "grad_norm": 0.18311651051044464, "learning_rate": 7.70236101037038e-05, "loss": 0.0403, "step": 21140 }, { "epoch": 9.465204743790558, "grad_norm": 0.17919497191905975, "learning_rate": 7.700041989267736e-05, "loss": 0.0379, "step": 21150 }, { "epoch": 9.469680017901096, "grad_norm": 0.15896692872047424, "learning_rate": 7.697722147963626e-05, "loss": 0.0401, "step": 21160 }, { "epoch": 9.474155292011636, "grad_norm": 0.1546044647693634, "learning_rate": 7.695401487162757e-05, "loss": 0.0394, "step": 21170 }, { "epoch": 9.478630566122176, "grad_norm": 0.12906228005886078, "learning_rate": 7.693080007570084e-05, "loss": 0.0351, "step": 21180 }, { "epoch": 9.483105840232714, "grad_norm": 0.15603135526180267, "learning_rate": 7.690757709890812e-05, "loss": 0.0435, "step": 21190 }, { "epoch": 9.487581114343254, "grad_norm": 0.15124070644378662, "learning_rate": 7.688434594830392e-05, "loss": 0.0403, "step": 21200 }, { "epoch": 9.492056388453793, "grad_norm": 0.17159485816955566, "learning_rate": 7.686110663094525e-05, "loss": 0.0404, "step": 21210 }, { "epoch": 9.496531662564331, "grad_norm": 0.1827392578125, "learning_rate": 7.683785915389164e-05, "loss": 0.0403, "step": 21220 }, { "epoch": 9.501006936674871, "grad_norm": 0.186395525932312, "learning_rate": 7.6814603524205e-05, "loss": 0.0361, "step": 21230 }, { "epoch": 9.505482210785411, "grad_norm": 0.1465705782175064, "learning_rate": 7.679133974894983e-05, "loss": 0.0407, "step": 21240 }, { "epoch": 9.50995748489595, "grad_norm": 0.1551513969898224, "learning_rate": 7.676806783519304e-05, "loss": 0.0402, "step": 21250 }, { "epoch": 9.514432759006489, "grad_norm": 0.18531957268714905, "learning_rate": 7.674478779000398e-05, "loss": 0.0367, "step": 21260 }, { "epoch": 9.518908033117029, "grad_norm": 0.16401049494743347, "learning_rate": 7.672149962045457e-05, "loss": 0.0407, "step": 21270 }, { "epoch": 9.523383307227567, "grad_norm": 0.13175827264785767, "learning_rate": 7.66982033336191e-05, "loss": 0.0396, "step": 21280 }, { "epoch": 9.527858581338107, "grad_norm": 0.1766674816608429, "learning_rate": 7.66748989365744e-05, "loss": 0.0397, "step": 21290 }, { "epoch": 9.532333855448647, "grad_norm": 0.16142278909683228, "learning_rate": 7.66515864363997e-05, "loss": 0.0401, "step": 21300 }, { "epoch": 9.536809129559185, "grad_norm": 0.17765918374061584, "learning_rate": 7.662826584017672e-05, "loss": 0.0377, "step": 21310 }, { "epoch": 9.541284403669724, "grad_norm": 0.16016785800457, "learning_rate": 7.660493715498969e-05, "loss": 0.0416, "step": 21320 }, { "epoch": 9.545759677780264, "grad_norm": 0.17325769364833832, "learning_rate": 7.658160038792518e-05, "loss": 0.0372, "step": 21330 }, { "epoch": 9.550234951890804, "grad_norm": 0.1535911113023758, "learning_rate": 7.655825554607235e-05, "loss": 0.0379, "step": 21340 }, { "epoch": 9.554710226001342, "grad_norm": 0.16835443675518036, "learning_rate": 7.653490263652269e-05, "loss": 0.0397, "step": 21350 }, { "epoch": 9.559185500111882, "grad_norm": 0.11736491322517395, "learning_rate": 7.651154166637025e-05, "loss": 0.0361, "step": 21360 }, { "epoch": 9.563660774222422, "grad_norm": 0.17434458434581757, "learning_rate": 7.648817264271142e-05, "loss": 0.0389, "step": 21370 }, { "epoch": 9.56813604833296, "grad_norm": 0.14016564190387726, "learning_rate": 7.646479557264513e-05, "loss": 0.0322, "step": 21380 }, { "epoch": 9.5726113224435, "grad_norm": 0.14811381697654724, "learning_rate": 7.644141046327271e-05, "loss": 0.0351, "step": 21390 }, { "epoch": 9.57708659655404, "grad_norm": 0.15418559312820435, "learning_rate": 7.641801732169795e-05, "loss": 0.0367, "step": 21400 }, { "epoch": 9.581561870664578, "grad_norm": 0.21156063675880432, "learning_rate": 7.639461615502704e-05, "loss": 0.0366, "step": 21410 }, { "epoch": 9.586037144775117, "grad_norm": 0.1374858170747757, "learning_rate": 7.637120697036866e-05, "loss": 0.0385, "step": 21420 }, { "epoch": 9.590512418885657, "grad_norm": 0.20564617216587067, "learning_rate": 7.634778977483389e-05, "loss": 0.0366, "step": 21430 }, { "epoch": 9.594987692996195, "grad_norm": 0.21555624902248383, "learning_rate": 7.632436457553625e-05, "loss": 0.0377, "step": 21440 }, { "epoch": 9.599462967106735, "grad_norm": 0.19610151648521423, "learning_rate": 7.630093137959171e-05, "loss": 0.038, "step": 21450 }, { "epoch": 9.603938241217275, "grad_norm": 0.15092062950134277, "learning_rate": 7.627749019411866e-05, "loss": 0.0366, "step": 21460 }, { "epoch": 9.608413515327815, "grad_norm": 0.1240912526845932, "learning_rate": 7.625404102623791e-05, "loss": 0.0359, "step": 21470 }, { "epoch": 9.612888789438353, "grad_norm": 0.1254594624042511, "learning_rate": 7.623058388307269e-05, "loss": 0.0353, "step": 21480 }, { "epoch": 9.617364063548893, "grad_norm": 0.15071776509284973, "learning_rate": 7.620711877174866e-05, "loss": 0.039, "step": 21490 }, { "epoch": 9.621839337659432, "grad_norm": 0.14124837517738342, "learning_rate": 7.618364569939391e-05, "loss": 0.0386, "step": 21500 }, { "epoch": 9.62631461176997, "grad_norm": 0.1747540384531021, "learning_rate": 7.616016467313891e-05, "loss": 0.0401, "step": 21510 }, { "epoch": 9.63078988588051, "grad_norm": 0.22139635682106018, "learning_rate": 7.613667570011663e-05, "loss": 0.0409, "step": 21520 }, { "epoch": 9.63526515999105, "grad_norm": 0.18567602336406708, "learning_rate": 7.611317878746238e-05, "loss": 0.0406, "step": 21530 }, { "epoch": 9.639740434101588, "grad_norm": 0.17677251994609833, "learning_rate": 7.608967394231387e-05, "loss": 0.0394, "step": 21540 }, { "epoch": 9.644215708212128, "grad_norm": 0.15533405542373657, "learning_rate": 7.606616117181128e-05, "loss": 0.041, "step": 21550 }, { "epoch": 9.648690982322668, "grad_norm": 0.14677265286445618, "learning_rate": 7.604264048309717e-05, "loss": 0.0373, "step": 21560 }, { "epoch": 9.653166256433206, "grad_norm": 0.139454185962677, "learning_rate": 7.60191118833165e-05, "loss": 0.0392, "step": 21570 }, { "epoch": 9.657641530543746, "grad_norm": 0.13310496509075165, "learning_rate": 7.599557537961663e-05, "loss": 0.0384, "step": 21580 }, { "epoch": 9.662116804654286, "grad_norm": 0.17500916123390198, "learning_rate": 7.597203097914732e-05, "loss": 0.0405, "step": 21590 }, { "epoch": 9.666592078764824, "grad_norm": 0.16521379351615906, "learning_rate": 7.594847868906076e-05, "loss": 0.0413, "step": 21600 }, { "epoch": 9.671067352875363, "grad_norm": 0.15406593680381775, "learning_rate": 7.592491851651151e-05, "loss": 0.0385, "step": 21610 }, { "epoch": 9.675542626985903, "grad_norm": 0.16478314995765686, "learning_rate": 7.590135046865651e-05, "loss": 0.0385, "step": 21620 }, { "epoch": 9.680017901096441, "grad_norm": 0.16185298562049866, "learning_rate": 7.587777455265515e-05, "loss": 0.0391, "step": 21630 }, { "epoch": 9.684493175206981, "grad_norm": 0.15695735812187195, "learning_rate": 7.585419077566912e-05, "loss": 0.0425, "step": 21640 }, { "epoch": 9.688968449317521, "grad_norm": 0.1511046439409256, "learning_rate": 7.583059914486257e-05, "loss": 0.0345, "step": 21650 }, { "epoch": 9.69344372342806, "grad_norm": 0.15333828330039978, "learning_rate": 7.580699966740201e-05, "loss": 0.0395, "step": 21660 }, { "epoch": 9.697918997538599, "grad_norm": 0.1374976933002472, "learning_rate": 7.578339235045637e-05, "loss": 0.0416, "step": 21670 }, { "epoch": 9.702394271649139, "grad_norm": 0.14441601932048798, "learning_rate": 7.57597772011969e-05, "loss": 0.0375, "step": 21680 }, { "epoch": 9.706869545759679, "grad_norm": 0.13103361427783966, "learning_rate": 7.573615422679726e-05, "loss": 0.0422, "step": 21690 }, { "epoch": 9.711344819870217, "grad_norm": 0.17513377964496613, "learning_rate": 7.571252343443349e-05, "loss": 0.0341, "step": 21700 }, { "epoch": 9.715820093980756, "grad_norm": 0.16571323573589325, "learning_rate": 7.5688884831284e-05, "loss": 0.0367, "step": 21710 }, { "epoch": 9.720295368091296, "grad_norm": 0.14549016952514648, "learning_rate": 7.566523842452958e-05, "loss": 0.0366, "step": 21720 }, { "epoch": 9.724770642201834, "grad_norm": 0.1574287861585617, "learning_rate": 7.564158422135337e-05, "loss": 0.0347, "step": 21730 }, { "epoch": 9.729245916312374, "grad_norm": 0.15751226246356964, "learning_rate": 7.561792222894091e-05, "loss": 0.036, "step": 21740 }, { "epoch": 9.733721190422914, "grad_norm": 0.15144960582256317, "learning_rate": 7.559425245448006e-05, "loss": 0.0384, "step": 21750 }, { "epoch": 9.738196464533452, "grad_norm": 0.1471056342124939, "learning_rate": 7.557057490516111e-05, "loss": 0.0393, "step": 21760 }, { "epoch": 9.742671738643992, "grad_norm": 0.13424427807331085, "learning_rate": 7.554688958817664e-05, "loss": 0.0378, "step": 21770 }, { "epoch": 9.747147012754532, "grad_norm": 0.15449731051921844, "learning_rate": 7.552319651072164e-05, "loss": 0.0413, "step": 21780 }, { "epoch": 9.75162228686507, "grad_norm": 0.14389561116695404, "learning_rate": 7.549949567999345e-05, "loss": 0.0376, "step": 21790 }, { "epoch": 9.75609756097561, "grad_norm": 0.13843059539794922, "learning_rate": 7.547578710319174e-05, "loss": 0.043, "step": 21800 }, { "epoch": 9.76057283508615, "grad_norm": 0.154527947306633, "learning_rate": 7.545207078751857e-05, "loss": 0.0401, "step": 21810 }, { "epoch": 9.765048109196687, "grad_norm": 0.1374160200357437, "learning_rate": 7.542834674017831e-05, "loss": 0.04, "step": 21820 }, { "epoch": 9.769523383307227, "grad_norm": 0.15917228162288666, "learning_rate": 7.54046149683777e-05, "loss": 0.0383, "step": 21830 }, { "epoch": 9.773998657417767, "grad_norm": 0.1463654488325119, "learning_rate": 7.538087547932585e-05, "loss": 0.0345, "step": 21840 }, { "epoch": 9.778473931528307, "grad_norm": 0.15602008998394012, "learning_rate": 7.535712828023416e-05, "loss": 0.0401, "step": 21850 }, { "epoch": 9.782949205638845, "grad_norm": 0.1692560464143753, "learning_rate": 7.533337337831642e-05, "loss": 0.0393, "step": 21860 }, { "epoch": 9.787424479749385, "grad_norm": 0.1838880032300949, "learning_rate": 7.530961078078873e-05, "loss": 0.0368, "step": 21870 }, { "epoch": 9.791899753859925, "grad_norm": 0.19387272000312805, "learning_rate": 7.528584049486955e-05, "loss": 0.0425, "step": 21880 }, { "epoch": 9.796375027970463, "grad_norm": 0.18441805243492126, "learning_rate": 7.526206252777968e-05, "loss": 0.0411, "step": 21890 }, { "epoch": 9.800850302081002, "grad_norm": 0.18701903522014618, "learning_rate": 7.52382768867422e-05, "loss": 0.0429, "step": 21900 }, { "epoch": 9.805325576191542, "grad_norm": 0.1817634552717209, "learning_rate": 7.521448357898258e-05, "loss": 0.0376, "step": 21910 }, { "epoch": 9.80980085030208, "grad_norm": 0.14671610295772552, "learning_rate": 7.519068261172859e-05, "loss": 0.038, "step": 21920 }, { "epoch": 9.81427612441262, "grad_norm": 0.13771513104438782, "learning_rate": 7.516687399221037e-05, "loss": 0.0407, "step": 21930 }, { "epoch": 9.81875139852316, "grad_norm": 0.13528786599636078, "learning_rate": 7.514305772766031e-05, "loss": 0.039, "step": 21940 }, { "epoch": 9.823226672633698, "grad_norm": 0.16547441482543945, "learning_rate": 7.511923382531317e-05, "loss": 0.0399, "step": 21950 }, { "epoch": 9.827701946744238, "grad_norm": 0.14710547029972076, "learning_rate": 7.509540229240601e-05, "loss": 0.0412, "step": 21960 }, { "epoch": 9.832177220854778, "grad_norm": 0.14063295722007751, "learning_rate": 7.507156313617827e-05, "loss": 0.0393, "step": 21970 }, { "epoch": 9.836652494965316, "grad_norm": 0.13667500019073486, "learning_rate": 7.504771636387163e-05, "loss": 0.0392, "step": 21980 }, { "epoch": 9.841127769075856, "grad_norm": 0.1521393060684204, "learning_rate": 7.50238619827301e-05, "loss": 0.0406, "step": 21990 }, { "epoch": 9.845603043186395, "grad_norm": 0.11935257911682129, "learning_rate": 7.500000000000001e-05, "loss": 0.0366, "step": 22000 }, { "epoch": 9.850078317296934, "grad_norm": 0.1476140022277832, "learning_rate": 7.497613042293e-05, "loss": 0.0338, "step": 22010 }, { "epoch": 9.854553591407473, "grad_norm": 0.14754192531108856, "learning_rate": 7.495225325877103e-05, "loss": 0.0371, "step": 22020 }, { "epoch": 9.859028865518013, "grad_norm": 0.17684660851955414, "learning_rate": 7.492836851477636e-05, "loss": 0.0387, "step": 22030 }, { "epoch": 9.863504139628553, "grad_norm": 0.14908131957054138, "learning_rate": 7.490447619820152e-05, "loss": 0.0374, "step": 22040 }, { "epoch": 9.867979413739091, "grad_norm": 0.14429309964179993, "learning_rate": 7.488057631630437e-05, "loss": 0.0393, "step": 22050 }, { "epoch": 9.87245468784963, "grad_norm": 0.16863420605659485, "learning_rate": 7.485666887634506e-05, "loss": 0.0404, "step": 22060 }, { "epoch": 9.87692996196017, "grad_norm": 0.14505663514137268, "learning_rate": 7.483275388558607e-05, "loss": 0.0384, "step": 22070 }, { "epoch": 9.881405236070709, "grad_norm": 0.14433899521827698, "learning_rate": 7.480883135129211e-05, "loss": 0.0438, "step": 22080 }, { "epoch": 9.885880510181249, "grad_norm": 0.16300135850906372, "learning_rate": 7.478490128073022e-05, "loss": 0.0387, "step": 22090 }, { "epoch": 9.890355784291788, "grad_norm": 0.15881229937076569, "learning_rate": 7.476096368116974e-05, "loss": 0.0335, "step": 22100 }, { "epoch": 9.894831058402326, "grad_norm": 0.16073143482208252, "learning_rate": 7.473701855988227e-05, "loss": 0.0364, "step": 22110 }, { "epoch": 9.899306332512866, "grad_norm": 0.17678551375865936, "learning_rate": 7.471306592414168e-05, "loss": 0.0429, "step": 22120 }, { "epoch": 9.903781606623406, "grad_norm": 0.15897324681282043, "learning_rate": 7.468910578122418e-05, "loss": 0.0437, "step": 22130 }, { "epoch": 9.908256880733944, "grad_norm": 0.15725526213645935, "learning_rate": 7.466513813840825e-05, "loss": 0.0397, "step": 22140 }, { "epoch": 9.912732154844484, "grad_norm": 0.14936096966266632, "learning_rate": 7.464116300297458e-05, "loss": 0.0365, "step": 22150 }, { "epoch": 9.917207428955024, "grad_norm": 0.16619043052196503, "learning_rate": 7.461718038220621e-05, "loss": 0.0385, "step": 22160 }, { "epoch": 9.921682703065564, "grad_norm": 0.128375843167305, "learning_rate": 7.45931902833884e-05, "loss": 0.0371, "step": 22170 }, { "epoch": 9.926157977176102, "grad_norm": 0.13015858829021454, "learning_rate": 7.456919271380875e-05, "loss": 0.0395, "step": 22180 }, { "epoch": 9.930633251286642, "grad_norm": 0.14782880246639252, "learning_rate": 7.454518768075704e-05, "loss": 0.0377, "step": 22190 }, { "epoch": 9.935108525397181, "grad_norm": 0.12587811052799225, "learning_rate": 7.452117519152542e-05, "loss": 0.0359, "step": 22200 }, { "epoch": 9.93958379950772, "grad_norm": 0.11385670304298401, "learning_rate": 7.44971552534082e-05, "loss": 0.0388, "step": 22210 }, { "epoch": 9.94405907361826, "grad_norm": 0.16549770534038544, "learning_rate": 7.447312787370203e-05, "loss": 0.0355, "step": 22220 }, { "epoch": 9.948534347728799, "grad_norm": 0.14511387050151825, "learning_rate": 7.444909305970578e-05, "loss": 0.0378, "step": 22230 }, { "epoch": 9.953009621839337, "grad_norm": 0.14894072711467743, "learning_rate": 7.44250508187206e-05, "loss": 0.0382, "step": 22240 }, { "epoch": 9.957484895949877, "grad_norm": 0.1347157210111618, "learning_rate": 7.440100115804991e-05, "loss": 0.038, "step": 22250 }, { "epoch": 9.961960170060417, "grad_norm": 0.12958674132823944, "learning_rate": 7.437694408499933e-05, "loss": 0.0368, "step": 22260 }, { "epoch": 9.966435444170955, "grad_norm": 0.1545981466770172, "learning_rate": 7.435287960687675e-05, "loss": 0.0394, "step": 22270 }, { "epoch": 9.970910718281495, "grad_norm": 0.1381291002035141, "learning_rate": 7.432880773099237e-05, "loss": 0.0393, "step": 22280 }, { "epoch": 9.975385992392034, "grad_norm": 0.1523403376340866, "learning_rate": 7.430472846465856e-05, "loss": 0.0405, "step": 22290 }, { "epoch": 9.979861266502573, "grad_norm": 0.1665997952222824, "learning_rate": 7.428064181518997e-05, "loss": 0.0348, "step": 22300 }, { "epoch": 9.984336540613112, "grad_norm": 0.1445692777633667, "learning_rate": 7.42565477899035e-05, "loss": 0.0347, "step": 22310 }, { "epoch": 9.988811814723652, "grad_norm": 0.13672445714473724, "learning_rate": 7.423244639611826e-05, "loss": 0.0351, "step": 22320 }, { "epoch": 9.99328708883419, "grad_norm": 0.14109107851982117, "learning_rate": 7.420833764115561e-05, "loss": 0.0369, "step": 22330 }, { "epoch": 9.99776236294473, "grad_norm": 0.17309033870697021, "learning_rate": 7.418422153233919e-05, "loss": 0.0421, "step": 22340 }, { "epoch": 10.00223763705527, "grad_norm": 0.15318217873573303, "learning_rate": 7.416009807699482e-05, "loss": 0.0421, "step": 22350 }, { "epoch": 10.00671291116581, "grad_norm": 0.15481875836849213, "learning_rate": 7.413596728245054e-05, "loss": 0.0379, "step": 22360 }, { "epoch": 10.011188185276348, "grad_norm": 0.13001978397369385, "learning_rate": 7.411182915603669e-05, "loss": 0.0377, "step": 22370 }, { "epoch": 10.015663459386888, "grad_norm": 0.15249958634376526, "learning_rate": 7.408768370508576e-05, "loss": 0.0339, "step": 22380 }, { "epoch": 10.020138733497427, "grad_norm": 0.16591475903987885, "learning_rate": 7.406353093693253e-05, "loss": 0.0357, "step": 22390 }, { "epoch": 10.024614007607966, "grad_norm": 0.14804746210575104, "learning_rate": 7.403937085891397e-05, "loss": 0.034, "step": 22400 }, { "epoch": 10.029089281718505, "grad_norm": 0.1445431262254715, "learning_rate": 7.401520347836926e-05, "loss": 0.0392, "step": 22410 }, { "epoch": 10.033564555829045, "grad_norm": 0.1532849669456482, "learning_rate": 7.399102880263983e-05, "loss": 0.0371, "step": 22420 }, { "epoch": 10.038039829939583, "grad_norm": 0.1517900675535202, "learning_rate": 7.396684683906928e-05, "loss": 0.0357, "step": 22430 }, { "epoch": 10.042515104050123, "grad_norm": 0.14749202132225037, "learning_rate": 7.394265759500348e-05, "loss": 0.0345, "step": 22440 }, { "epoch": 10.046990378160663, "grad_norm": 0.13713976740837097, "learning_rate": 7.391846107779047e-05, "loss": 0.0384, "step": 22450 }, { "epoch": 10.051465652271201, "grad_norm": 0.1655397266149521, "learning_rate": 7.389425729478051e-05, "loss": 0.0376, "step": 22460 }, { "epoch": 10.05594092638174, "grad_norm": 0.18326354026794434, "learning_rate": 7.387004625332608e-05, "loss": 0.0401, "step": 22470 }, { "epoch": 10.06041620049228, "grad_norm": 0.14487828314304352, "learning_rate": 7.384582796078184e-05, "loss": 0.0341, "step": 22480 }, { "epoch": 10.064891474602819, "grad_norm": 0.14553236961364746, "learning_rate": 7.382160242450469e-05, "loss": 0.0319, "step": 22490 }, { "epoch": 10.069366748713358, "grad_norm": 0.14493724703788757, "learning_rate": 7.379736965185368e-05, "loss": 0.0365, "step": 22500 }, { "epoch": 10.073842022823898, "grad_norm": 0.13138338923454285, "learning_rate": 7.37731296501901e-05, "loss": 0.0412, "step": 22510 }, { "epoch": 10.078317296934436, "grad_norm": 0.14799192547798157, "learning_rate": 7.374888242687746e-05, "loss": 0.0386, "step": 22520 }, { "epoch": 10.082792571044976, "grad_norm": 0.1362508237361908, "learning_rate": 7.372462798928137e-05, "loss": 0.0374, "step": 22530 }, { "epoch": 10.087267845155516, "grad_norm": 0.20392869412899017, "learning_rate": 7.37003663447697e-05, "loss": 0.0364, "step": 22540 }, { "epoch": 10.091743119266056, "grad_norm": 0.1255084127187729, "learning_rate": 7.367609750071252e-05, "loss": 0.0388, "step": 22550 }, { "epoch": 10.096218393376594, "grad_norm": 0.14950473606586456, "learning_rate": 7.365182146448205e-05, "loss": 0.0398, "step": 22560 }, { "epoch": 10.100693667487134, "grad_norm": 0.1698312908411026, "learning_rate": 7.362753824345272e-05, "loss": 0.0385, "step": 22570 }, { "epoch": 10.105168941597674, "grad_norm": 0.13864362239837646, "learning_rate": 7.36032478450011e-05, "loss": 0.036, "step": 22580 }, { "epoch": 10.109644215708212, "grad_norm": 0.1394815295934677, "learning_rate": 7.357895027650598e-05, "loss": 0.0345, "step": 22590 }, { "epoch": 10.114119489818751, "grad_norm": 0.15162207186222076, "learning_rate": 7.355464554534837e-05, "loss": 0.0376, "step": 22600 }, { "epoch": 10.118594763929291, "grad_norm": 0.12756145000457764, "learning_rate": 7.353033365891134e-05, "loss": 0.0364, "step": 22610 }, { "epoch": 10.12307003803983, "grad_norm": 0.12858723104000092, "learning_rate": 7.350601462458024e-05, "loss": 0.0347, "step": 22620 }, { "epoch": 10.12754531215037, "grad_norm": 0.15422047674655914, "learning_rate": 7.348168844974254e-05, "loss": 0.0412, "step": 22630 }, { "epoch": 10.132020586260909, "grad_norm": 0.17546625435352325, "learning_rate": 7.345735514178787e-05, "loss": 0.0378, "step": 22640 }, { "epoch": 10.136495860371447, "grad_norm": 0.15315794944763184, "learning_rate": 7.343301470810808e-05, "loss": 0.0384, "step": 22650 }, { "epoch": 10.140971134481987, "grad_norm": 0.15758016705513, "learning_rate": 7.340866715609712e-05, "loss": 0.038, "step": 22660 }, { "epoch": 10.145446408592527, "grad_norm": 0.1547306329011917, "learning_rate": 7.338431249315115e-05, "loss": 0.036, "step": 22670 }, { "epoch": 10.149921682703065, "grad_norm": 0.13332052528858185, "learning_rate": 7.335995072666848e-05, "loss": 0.0337, "step": 22680 }, { "epoch": 10.154396956813605, "grad_norm": 0.16224223375320435, "learning_rate": 7.333558186404958e-05, "loss": 0.0367, "step": 22690 }, { "epoch": 10.158872230924144, "grad_norm": 0.14634360373020172, "learning_rate": 7.331120591269701e-05, "loss": 0.0348, "step": 22700 }, { "epoch": 10.163347505034684, "grad_norm": 0.1312991976737976, "learning_rate": 7.328682288001561e-05, "loss": 0.0376, "step": 22710 }, { "epoch": 10.167822779145222, "grad_norm": 0.13967706263065338, "learning_rate": 7.326243277341227e-05, "loss": 0.0356, "step": 22720 }, { "epoch": 10.172298053255762, "grad_norm": 0.17657743394374847, "learning_rate": 7.323803560029605e-05, "loss": 0.0403, "step": 22730 }, { "epoch": 10.176773327366302, "grad_norm": 0.1655014604330063, "learning_rate": 7.32136313680782e-05, "loss": 0.0385, "step": 22740 }, { "epoch": 10.18124860147684, "grad_norm": 0.15912781655788422, "learning_rate": 7.318922008417203e-05, "loss": 0.0334, "step": 22750 }, { "epoch": 10.18572387558738, "grad_norm": 0.1636781096458435, "learning_rate": 7.316480175599309e-05, "loss": 0.0411, "step": 22760 }, { "epoch": 10.19019914969792, "grad_norm": 0.13915829360485077, "learning_rate": 7.3140376390959e-05, "loss": 0.0345, "step": 22770 }, { "epoch": 10.194674423808458, "grad_norm": 0.13331948220729828, "learning_rate": 7.311594399648957e-05, "loss": 0.0363, "step": 22780 }, { "epoch": 10.199149697918998, "grad_norm": 0.17854897677898407, "learning_rate": 7.309150458000668e-05, "loss": 0.0366, "step": 22790 }, { "epoch": 10.203624972029537, "grad_norm": 0.16485098004341125, "learning_rate": 7.30670581489344e-05, "loss": 0.0361, "step": 22800 }, { "epoch": 10.208100246140075, "grad_norm": 0.1749841272830963, "learning_rate": 7.30426047106989e-05, "loss": 0.0344, "step": 22810 }, { "epoch": 10.212575520250615, "grad_norm": 0.1681033968925476, "learning_rate": 7.301814427272849e-05, "loss": 0.0392, "step": 22820 }, { "epoch": 10.217050794361155, "grad_norm": 0.13495495915412903, "learning_rate": 7.299367684245362e-05, "loss": 0.0399, "step": 22830 }, { "epoch": 10.221526068471693, "grad_norm": 0.1519542932510376, "learning_rate": 7.296920242730682e-05, "loss": 0.0347, "step": 22840 }, { "epoch": 10.226001342582233, "grad_norm": 0.13909493386745453, "learning_rate": 7.294472103472281e-05, "loss": 0.036, "step": 22850 }, { "epoch": 10.230476616692773, "grad_norm": 0.18337878584861755, "learning_rate": 7.292023267213835e-05, "loss": 0.0381, "step": 22860 }, { "epoch": 10.234951890803313, "grad_norm": 0.135471373796463, "learning_rate": 7.28957373469924e-05, "loss": 0.0326, "step": 22870 }, { "epoch": 10.23942716491385, "grad_norm": 0.14038419723510742, "learning_rate": 7.287123506672595e-05, "loss": 0.0373, "step": 22880 }, { "epoch": 10.24390243902439, "grad_norm": 0.17442883551120758, "learning_rate": 7.284672583878219e-05, "loss": 0.0354, "step": 22890 }, { "epoch": 10.24837771313493, "grad_norm": 0.11974915117025375, "learning_rate": 7.282220967060633e-05, "loss": 0.0359, "step": 22900 }, { "epoch": 10.252852987245468, "grad_norm": 0.14041084051132202, "learning_rate": 7.279768656964577e-05, "loss": 0.0379, "step": 22910 }, { "epoch": 10.257328261356008, "grad_norm": 0.17182037234306335, "learning_rate": 7.277315654334997e-05, "loss": 0.0351, "step": 22920 }, { "epoch": 10.261803535466548, "grad_norm": 0.16106751561164856, "learning_rate": 7.27486195991705e-05, "loss": 0.0346, "step": 22930 }, { "epoch": 10.266278809577086, "grad_norm": 0.15687420964241028, "learning_rate": 7.272407574456103e-05, "loss": 0.0377, "step": 22940 }, { "epoch": 10.270754083687626, "grad_norm": 0.1739259958267212, "learning_rate": 7.269952498697734e-05, "loss": 0.037, "step": 22950 }, { "epoch": 10.275229357798166, "grad_norm": 0.16383785009384155, "learning_rate": 7.267496733387731e-05, "loss": 0.04, "step": 22960 }, { "epoch": 10.279704631908704, "grad_norm": 0.2038682997226715, "learning_rate": 7.265040279272091e-05, "loss": 0.0385, "step": 22970 }, { "epoch": 10.284179906019244, "grad_norm": 0.20603996515274048, "learning_rate": 7.262583137097018e-05, "loss": 0.0334, "step": 22980 }, { "epoch": 10.288655180129783, "grad_norm": 0.16846473515033722, "learning_rate": 7.260125307608929e-05, "loss": 0.0399, "step": 22990 }, { "epoch": 10.293130454240321, "grad_norm": 0.1479729413986206, "learning_rate": 7.257666791554448e-05, "loss": 0.0359, "step": 23000 }, { "epoch": 10.297605728350861, "grad_norm": 0.12614469230175018, "learning_rate": 7.255207589680402e-05, "loss": 0.0373, "step": 23010 }, { "epoch": 10.302081002461401, "grad_norm": 0.1315535455942154, "learning_rate": 7.25274770273384e-05, "loss": 0.0382, "step": 23020 }, { "epoch": 10.30655627657194, "grad_norm": 0.16026756167411804, "learning_rate": 7.250287131462004e-05, "loss": 0.0355, "step": 23030 }, { "epoch": 10.311031550682479, "grad_norm": 0.1394568383693695, "learning_rate": 7.247825876612353e-05, "loss": 0.035, "step": 23040 }, { "epoch": 10.315506824793019, "grad_norm": 0.1681465208530426, "learning_rate": 7.245363938932551e-05, "loss": 0.0318, "step": 23050 }, { "epoch": 10.319982098903559, "grad_norm": 0.15684477984905243, "learning_rate": 7.242901319170471e-05, "loss": 0.0391, "step": 23060 }, { "epoch": 10.324457373014097, "grad_norm": 0.16896012425422668, "learning_rate": 7.240438018074189e-05, "loss": 0.0354, "step": 23070 }, { "epoch": 10.328932647124637, "grad_norm": 0.1636996567249298, "learning_rate": 7.237974036391992e-05, "loss": 0.0376, "step": 23080 }, { "epoch": 10.333407921235176, "grad_norm": 0.17409949004650116, "learning_rate": 7.235509374872373e-05, "loss": 0.0409, "step": 23090 }, { "epoch": 10.337883195345714, "grad_norm": 0.16833746433258057, "learning_rate": 7.233044034264034e-05, "loss": 0.0343, "step": 23100 }, { "epoch": 10.342358469456254, "grad_norm": 0.15382257103919983, "learning_rate": 7.230578015315876e-05, "loss": 0.0366, "step": 23110 }, { "epoch": 10.346833743566794, "grad_norm": 0.14982013404369354, "learning_rate": 7.22811131877701e-05, "loss": 0.0361, "step": 23120 }, { "epoch": 10.351309017677332, "grad_norm": 0.1514611393213272, "learning_rate": 7.225643945396757e-05, "loss": 0.0343, "step": 23130 }, { "epoch": 10.355784291787872, "grad_norm": 0.14807474613189697, "learning_rate": 7.223175895924638e-05, "loss": 0.0346, "step": 23140 }, { "epoch": 10.360259565898412, "grad_norm": 0.14899790287017822, "learning_rate": 7.220707171110382e-05, "loss": 0.0378, "step": 23150 }, { "epoch": 10.36473484000895, "grad_norm": 0.1798473298549652, "learning_rate": 7.218237771703921e-05, "loss": 0.0366, "step": 23160 }, { "epoch": 10.36921011411949, "grad_norm": 0.15826250612735748, "learning_rate": 7.215767698455394e-05, "loss": 0.0386, "step": 23170 }, { "epoch": 10.37368538823003, "grad_norm": 0.1595507711172104, "learning_rate": 7.213296952115144e-05, "loss": 0.0341, "step": 23180 }, { "epoch": 10.378160662340568, "grad_norm": 0.1645360141992569, "learning_rate": 7.210825533433719e-05, "loss": 0.0366, "step": 23190 }, { "epoch": 10.382635936451107, "grad_norm": 0.1677495539188385, "learning_rate": 7.20835344316187e-05, "loss": 0.0374, "step": 23200 }, { "epoch": 10.387111210561647, "grad_norm": 0.1413038820028305, "learning_rate": 7.205880682050554e-05, "loss": 0.0379, "step": 23210 }, { "epoch": 10.391586484672185, "grad_norm": 0.1571611762046814, "learning_rate": 7.203407250850928e-05, "loss": 0.0386, "step": 23220 }, { "epoch": 10.396061758782725, "grad_norm": 0.14342911541461945, "learning_rate": 7.20093315031436e-05, "loss": 0.0354, "step": 23230 }, { "epoch": 10.400537032893265, "grad_norm": 0.16962678730487823, "learning_rate": 7.19845838119241e-05, "loss": 0.0435, "step": 23240 }, { "epoch": 10.405012307003805, "grad_norm": 0.1394611895084381, "learning_rate": 7.195982944236851e-05, "loss": 0.0332, "step": 23250 }, { "epoch": 10.409487581114343, "grad_norm": 0.13873785734176636, "learning_rate": 7.193506840199657e-05, "loss": 0.0366, "step": 23260 }, { "epoch": 10.413962855224883, "grad_norm": 0.13598944246768951, "learning_rate": 7.191030069833001e-05, "loss": 0.0385, "step": 23270 }, { "epoch": 10.418438129335422, "grad_norm": 0.14356189966201782, "learning_rate": 7.188552633889259e-05, "loss": 0.0371, "step": 23280 }, { "epoch": 10.42291340344596, "grad_norm": 0.1611056625843048, "learning_rate": 7.186074533121013e-05, "loss": 0.0338, "step": 23290 }, { "epoch": 10.4273886775565, "grad_norm": 0.14459562301635742, "learning_rate": 7.183595768281043e-05, "loss": 0.0373, "step": 23300 }, { "epoch": 10.43186395166704, "grad_norm": 0.13975372910499573, "learning_rate": 7.181116340122336e-05, "loss": 0.0358, "step": 23310 }, { "epoch": 10.436339225777578, "grad_norm": 0.13238441944122314, "learning_rate": 7.178636249398072e-05, "loss": 0.0346, "step": 23320 }, { "epoch": 10.440814499888118, "grad_norm": 0.1325264871120453, "learning_rate": 7.176155496861638e-05, "loss": 0.0392, "step": 23330 }, { "epoch": 10.445289773998658, "grad_norm": 0.1417611837387085, "learning_rate": 7.173674083266624e-05, "loss": 0.0394, "step": 23340 }, { "epoch": 10.449765048109196, "grad_norm": 0.14096271991729736, "learning_rate": 7.171192009366814e-05, "loss": 0.033, "step": 23350 }, { "epoch": 10.454240322219736, "grad_norm": 0.13820765912532806, "learning_rate": 7.1687092759162e-05, "loss": 0.0338, "step": 23360 }, { "epoch": 10.458715596330276, "grad_norm": 0.147271528840065, "learning_rate": 7.166225883668969e-05, "loss": 0.0367, "step": 23370 }, { "epoch": 10.463190870440814, "grad_norm": 0.15466421842575073, "learning_rate": 7.16374183337951e-05, "loss": 0.0387, "step": 23380 }, { "epoch": 10.467666144551353, "grad_norm": 0.14036767184734344, "learning_rate": 7.161257125802413e-05, "loss": 0.0364, "step": 23390 }, { "epoch": 10.472141418661893, "grad_norm": 0.1449228674173355, "learning_rate": 7.158771761692464e-05, "loss": 0.0356, "step": 23400 }, { "epoch": 10.476616692772433, "grad_norm": 0.13498139381408691, "learning_rate": 7.156285741804656e-05, "loss": 0.0383, "step": 23410 }, { "epoch": 10.481091966882971, "grad_norm": 0.15846197307109833, "learning_rate": 7.153799066894171e-05, "loss": 0.0342, "step": 23420 }, { "epoch": 10.485567240993511, "grad_norm": 0.1754455715417862, "learning_rate": 7.151311737716397e-05, "loss": 0.036, "step": 23430 }, { "epoch": 10.49004251510405, "grad_norm": 0.14509154856204987, "learning_rate": 7.148823755026921e-05, "loss": 0.0374, "step": 23440 }, { "epoch": 10.494517789214589, "grad_norm": 0.17905564606189728, "learning_rate": 7.146335119581523e-05, "loss": 0.0367, "step": 23450 }, { "epoch": 10.498993063325129, "grad_norm": 0.17477257549762726, "learning_rate": 7.143845832136188e-05, "loss": 0.0414, "step": 23460 }, { "epoch": 10.503468337435669, "grad_norm": 0.1334957480430603, "learning_rate": 7.141355893447092e-05, "loss": 0.0361, "step": 23470 }, { "epoch": 10.507943611546207, "grad_norm": 0.12744182348251343, "learning_rate": 7.138865304270616e-05, "loss": 0.0364, "step": 23480 }, { "epoch": 10.512418885656746, "grad_norm": 0.12503176927566528, "learning_rate": 7.136374065363334e-05, "loss": 0.0347, "step": 23490 }, { "epoch": 10.516894159767286, "grad_norm": 0.1616145223379135, "learning_rate": 7.133882177482019e-05, "loss": 0.0415, "step": 23500 }, { "epoch": 10.521369433877824, "grad_norm": 0.19223350286483765, "learning_rate": 7.131389641383639e-05, "loss": 0.0375, "step": 23510 }, { "epoch": 10.525844707988364, "grad_norm": 0.16974583268165588, "learning_rate": 7.128896457825364e-05, "loss": 0.0336, "step": 23520 }, { "epoch": 10.530319982098904, "grad_norm": 0.16545982658863068, "learning_rate": 7.126402627564555e-05, "loss": 0.0344, "step": 23530 }, { "epoch": 10.534795256209442, "grad_norm": 0.14815978705883026, "learning_rate": 7.12390815135877e-05, "loss": 0.0381, "step": 23540 }, { "epoch": 10.539270530319982, "grad_norm": 0.178350031375885, "learning_rate": 7.121413029965769e-05, "loss": 0.0347, "step": 23550 }, { "epoch": 10.543745804430522, "grad_norm": 0.17972926795482635, "learning_rate": 7.118917264143501e-05, "loss": 0.0383, "step": 23560 }, { "epoch": 10.548221078541062, "grad_norm": 0.1654777228832245, "learning_rate": 7.116420854650117e-05, "loss": 0.0371, "step": 23570 }, { "epoch": 10.5526963526516, "grad_norm": 0.13701820373535156, "learning_rate": 7.113923802243957e-05, "loss": 0.0342, "step": 23580 }, { "epoch": 10.55717162676214, "grad_norm": 0.13100340962409973, "learning_rate": 7.11142610768356e-05, "loss": 0.0369, "step": 23590 }, { "epoch": 10.56164690087268, "grad_norm": 0.11870749294757843, "learning_rate": 7.108927771727661e-05, "loss": 0.0341, "step": 23600 }, { "epoch": 10.566122174983217, "grad_norm": 0.188645139336586, "learning_rate": 7.10642879513519e-05, "loss": 0.0381, "step": 23610 }, { "epoch": 10.570597449093757, "grad_norm": 0.1670929342508316, "learning_rate": 7.103929178665266e-05, "loss": 0.0355, "step": 23620 }, { "epoch": 10.575072723204297, "grad_norm": 0.13852931559085846, "learning_rate": 7.101428923077209e-05, "loss": 0.0346, "step": 23630 }, { "epoch": 10.579547997314835, "grad_norm": 0.13423925638198853, "learning_rate": 7.09892802913053e-05, "loss": 0.0382, "step": 23640 }, { "epoch": 10.584023271425375, "grad_norm": 0.1762104481458664, "learning_rate": 7.096426497584933e-05, "loss": 0.0365, "step": 23650 }, { "epoch": 10.588498545535915, "grad_norm": 0.16491267085075378, "learning_rate": 7.093924329200321e-05, "loss": 0.0353, "step": 23660 }, { "epoch": 10.592973819646453, "grad_norm": 0.16007190942764282, "learning_rate": 7.091421524736784e-05, "loss": 0.0363, "step": 23670 }, { "epoch": 10.597449093756993, "grad_norm": 0.14487087726593018, "learning_rate": 7.08891808495461e-05, "loss": 0.0371, "step": 23680 }, { "epoch": 10.601924367867532, "grad_norm": 0.14572344720363617, "learning_rate": 7.086414010614276e-05, "loss": 0.0418, "step": 23690 }, { "epoch": 10.60639964197807, "grad_norm": 0.1585957109928131, "learning_rate": 7.083909302476453e-05, "loss": 0.0377, "step": 23700 }, { "epoch": 10.61087491608861, "grad_norm": 0.15388545393943787, "learning_rate": 7.081403961302006e-05, "loss": 0.0349, "step": 23710 }, { "epoch": 10.61535019019915, "grad_norm": 0.17180073261260986, "learning_rate": 7.078897987851993e-05, "loss": 0.037, "step": 23720 }, { "epoch": 10.619825464309688, "grad_norm": 0.18603608012199402, "learning_rate": 7.076391382887661e-05, "loss": 0.035, "step": 23730 }, { "epoch": 10.624300738420228, "grad_norm": 0.1337353140115738, "learning_rate": 7.073884147170452e-05, "loss": 0.0366, "step": 23740 }, { "epoch": 10.628776012530768, "grad_norm": 0.14762090146541595, "learning_rate": 7.071376281461994e-05, "loss": 0.0398, "step": 23750 }, { "epoch": 10.633251286641308, "grad_norm": 0.11415469646453857, "learning_rate": 7.068867786524116e-05, "loss": 0.038, "step": 23760 }, { "epoch": 10.637726560751846, "grad_norm": 0.12300815433263779, "learning_rate": 7.06635866311883e-05, "loss": 0.033, "step": 23770 }, { "epoch": 10.642201834862385, "grad_norm": 0.135547935962677, "learning_rate": 7.06384891200834e-05, "loss": 0.0398, "step": 23780 }, { "epoch": 10.646677108972925, "grad_norm": 0.14602088928222656, "learning_rate": 7.061338533955043e-05, "loss": 0.0399, "step": 23790 }, { "epoch": 10.651152383083463, "grad_norm": 0.16212597489356995, "learning_rate": 7.058827529721525e-05, "loss": 0.0372, "step": 23800 }, { "epoch": 10.655627657194003, "grad_norm": 0.13481052219867706, "learning_rate": 7.056315900070565e-05, "loss": 0.0372, "step": 23810 }, { "epoch": 10.660102931304543, "grad_norm": 0.14161118865013123, "learning_rate": 7.053803645765128e-05, "loss": 0.0395, "step": 23820 }, { "epoch": 10.664578205415081, "grad_norm": 0.16192996501922607, "learning_rate": 7.051290767568371e-05, "loss": 0.0368, "step": 23830 }, { "epoch": 10.66905347952562, "grad_norm": 0.1522493064403534, "learning_rate": 7.04877726624364e-05, "loss": 0.0356, "step": 23840 }, { "epoch": 10.67352875363616, "grad_norm": 0.138436421751976, "learning_rate": 7.04626314255447e-05, "loss": 0.0364, "step": 23850 }, { "epoch": 10.678004027746699, "grad_norm": 0.1495436280965805, "learning_rate": 7.043748397264587e-05, "loss": 0.0345, "step": 23860 }, { "epoch": 10.682479301857239, "grad_norm": 0.13283726572990417, "learning_rate": 7.041233031137903e-05, "loss": 0.0382, "step": 23870 }, { "epoch": 10.686954575967778, "grad_norm": 0.1404486447572708, "learning_rate": 7.038717044938519e-05, "loss": 0.0354, "step": 23880 }, { "epoch": 10.691429850078316, "grad_norm": 0.18866287171840668, "learning_rate": 7.036200439430725e-05, "loss": 0.0378, "step": 23890 }, { "epoch": 10.695905124188856, "grad_norm": 0.19171808660030365, "learning_rate": 7.033683215379002e-05, "loss": 0.0365, "step": 23900 }, { "epoch": 10.700380398299396, "grad_norm": 0.1599602848291397, "learning_rate": 7.031165373548014e-05, "loss": 0.0385, "step": 23910 }, { "epoch": 10.704855672409934, "grad_norm": 0.16346175968647003, "learning_rate": 7.028646914702614e-05, "loss": 0.0337, "step": 23920 }, { "epoch": 10.709330946520474, "grad_norm": 0.17410805821418762, "learning_rate": 7.026127839607847e-05, "loss": 0.0364, "step": 23930 }, { "epoch": 10.713806220631014, "grad_norm": 0.16704289615154266, "learning_rate": 7.023608149028937e-05, "loss": 0.0367, "step": 23940 }, { "epoch": 10.718281494741554, "grad_norm": 0.16258345544338226, "learning_rate": 7.021087843731302e-05, "loss": 0.0394, "step": 23950 }, { "epoch": 10.722756768852092, "grad_norm": 0.1737159639596939, "learning_rate": 7.018566924480543e-05, "loss": 0.0375, "step": 23960 }, { "epoch": 10.727232042962632, "grad_norm": 0.1474934220314026, "learning_rate": 7.016045392042452e-05, "loss": 0.0374, "step": 23970 }, { "epoch": 10.731707317073171, "grad_norm": 0.15795886516571045, "learning_rate": 7.013523247183e-05, "loss": 0.0383, "step": 23980 }, { "epoch": 10.73618259118371, "grad_norm": 0.17446117103099823, "learning_rate": 7.01100049066835e-05, "loss": 0.0354, "step": 23990 }, { "epoch": 10.74065786529425, "grad_norm": 0.13873018324375153, "learning_rate": 7.008477123264848e-05, "loss": 0.0344, "step": 24000 }, { "epoch": 10.745133139404789, "grad_norm": 0.11835665255784988, "learning_rate": 7.005953145739026e-05, "loss": 0.0345, "step": 24010 }, { "epoch": 10.749608413515327, "grad_norm": 0.13063253462314606, "learning_rate": 7.003428558857604e-05, "loss": 0.0351, "step": 24020 }, { "epoch": 10.754083687625867, "grad_norm": 0.1431019902229309, "learning_rate": 7.000903363387482e-05, "loss": 0.0367, "step": 24030 }, { "epoch": 10.758558961736407, "grad_norm": 0.1573610007762909, "learning_rate": 6.99837756009575e-05, "loss": 0.0344, "step": 24040 }, { "epoch": 10.763034235846945, "grad_norm": 0.14714735746383667, "learning_rate": 6.99585114974968e-05, "loss": 0.0329, "step": 24050 }, { "epoch": 10.767509509957485, "grad_norm": 0.15392041206359863, "learning_rate": 6.993324133116726e-05, "loss": 0.0381, "step": 24060 }, { "epoch": 10.771984784068025, "grad_norm": 0.14518649876117706, "learning_rate": 6.990796510964531e-05, "loss": 0.033, "step": 24070 }, { "epoch": 10.776460058178564, "grad_norm": 0.15604032576084137, "learning_rate": 6.988268284060922e-05, "loss": 0.0363, "step": 24080 }, { "epoch": 10.780935332289102, "grad_norm": 0.1441263109445572, "learning_rate": 6.985739453173903e-05, "loss": 0.0363, "step": 24090 }, { "epoch": 10.785410606399642, "grad_norm": 0.1631411463022232, "learning_rate": 6.98321001907167e-05, "loss": 0.0322, "step": 24100 }, { "epoch": 10.789885880510182, "grad_norm": 0.1514555960893631, "learning_rate": 6.980679982522598e-05, "loss": 0.0348, "step": 24110 }, { "epoch": 10.79436115462072, "grad_norm": 0.17018622159957886, "learning_rate": 6.978149344295242e-05, "loss": 0.0374, "step": 24120 }, { "epoch": 10.79883642873126, "grad_norm": 0.1476484090089798, "learning_rate": 6.975618105158346e-05, "loss": 0.0391, "step": 24130 }, { "epoch": 10.8033117028418, "grad_norm": 0.1746589094400406, "learning_rate": 6.973086265880833e-05, "loss": 0.0374, "step": 24140 }, { "epoch": 10.807786976952338, "grad_norm": 0.14576368033885956, "learning_rate": 6.97055382723181e-05, "loss": 0.0368, "step": 24150 }, { "epoch": 10.812262251062878, "grad_norm": 0.15174373984336853, "learning_rate": 6.968020789980562e-05, "loss": 0.0338, "step": 24160 }, { "epoch": 10.816737525173417, "grad_norm": 0.16628214716911316, "learning_rate": 6.96548715489656e-05, "loss": 0.0323, "step": 24170 }, { "epoch": 10.821212799283956, "grad_norm": 0.12737445533275604, "learning_rate": 6.962952922749457e-05, "loss": 0.0371, "step": 24180 }, { "epoch": 10.825688073394495, "grad_norm": 0.14565549790859222, "learning_rate": 6.960418094309085e-05, "loss": 0.0364, "step": 24190 }, { "epoch": 10.830163347505035, "grad_norm": 0.1639280617237091, "learning_rate": 6.957882670345458e-05, "loss": 0.0334, "step": 24200 }, { "epoch": 10.834638621615573, "grad_norm": 0.1426738053560257, "learning_rate": 6.955346651628771e-05, "loss": 0.0349, "step": 24210 }, { "epoch": 10.839113895726113, "grad_norm": 0.1516444981098175, "learning_rate": 6.952810038929397e-05, "loss": 0.0354, "step": 24220 }, { "epoch": 10.843589169836653, "grad_norm": 0.14474575221538544, "learning_rate": 6.950272833017896e-05, "loss": 0.0364, "step": 24230 }, { "epoch": 10.848064443947191, "grad_norm": 0.1607467532157898, "learning_rate": 6.947735034665002e-05, "loss": 0.0352, "step": 24240 }, { "epoch": 10.85253971805773, "grad_norm": 0.16260457038879395, "learning_rate": 6.94519664464163e-05, "loss": 0.0366, "step": 24250 }, { "epoch": 10.85701499216827, "grad_norm": 0.1692279726266861, "learning_rate": 6.942657663718879e-05, "loss": 0.0366, "step": 24260 }, { "epoch": 10.86149026627881, "grad_norm": 0.15837030112743378, "learning_rate": 6.940118092668022e-05, "loss": 0.0336, "step": 24270 }, { "epoch": 10.865965540389348, "grad_norm": 0.12267782539129257, "learning_rate": 6.937577932260515e-05, "loss": 0.0423, "step": 24280 }, { "epoch": 10.870440814499888, "grad_norm": 0.15483348071575165, "learning_rate": 6.93503718326799e-05, "loss": 0.0366, "step": 24290 }, { "epoch": 10.874916088610428, "grad_norm": 0.15232565999031067, "learning_rate": 6.932495846462261e-05, "loss": 0.0374, "step": 24300 }, { "epoch": 10.879391362720966, "grad_norm": 0.16565893590450287, "learning_rate": 6.929953922615319e-05, "loss": 0.0383, "step": 24310 }, { "epoch": 10.883866636831506, "grad_norm": 0.15683706104755402, "learning_rate": 6.927411412499332e-05, "loss": 0.0368, "step": 24320 }, { "epoch": 10.888341910942046, "grad_norm": 0.1430221050977707, "learning_rate": 6.924868316886649e-05, "loss": 0.0363, "step": 24330 }, { "epoch": 10.892817185052584, "grad_norm": 0.13832293450832367, "learning_rate": 6.922324636549795e-05, "loss": 0.032, "step": 24340 }, { "epoch": 10.897292459163124, "grad_norm": 0.12040139734745026, "learning_rate": 6.91978037226147e-05, "loss": 0.0356, "step": 24350 }, { "epoch": 10.901767733273664, "grad_norm": 0.1569630354642868, "learning_rate": 6.917235524794558e-05, "loss": 0.0349, "step": 24360 }, { "epoch": 10.906243007384202, "grad_norm": 0.15173567831516266, "learning_rate": 6.914690094922116e-05, "loss": 0.0359, "step": 24370 }, { "epoch": 10.910718281494741, "grad_norm": 0.1629522740840912, "learning_rate": 6.912144083417376e-05, "loss": 0.0397, "step": 24380 }, { "epoch": 10.915193555605281, "grad_norm": 0.14132043719291687, "learning_rate": 6.909597491053751e-05, "loss": 0.0386, "step": 24390 }, { "epoch": 10.91966882971582, "grad_norm": 0.13911184668540955, "learning_rate": 6.90705031860483e-05, "loss": 0.0424, "step": 24400 }, { "epoch": 10.92414410382636, "grad_norm": 0.1419563889503479, "learning_rate": 6.904502566844374e-05, "loss": 0.0353, "step": 24410 }, { "epoch": 10.928619377936899, "grad_norm": 0.12884841859340668, "learning_rate": 6.901954236546323e-05, "loss": 0.0376, "step": 24420 }, { "epoch": 10.933094652047437, "grad_norm": 0.14610034227371216, "learning_rate": 6.899405328484794e-05, "loss": 0.0348, "step": 24430 }, { "epoch": 10.937569926157977, "grad_norm": 0.15178443491458893, "learning_rate": 6.896855843434078e-05, "loss": 0.0327, "step": 24440 }, { "epoch": 10.942045200268517, "grad_norm": 0.1703963279724121, "learning_rate": 6.894305782168638e-05, "loss": 0.0341, "step": 24450 }, { "epoch": 10.946520474379057, "grad_norm": 0.1670524775981903, "learning_rate": 6.89175514546312e-05, "loss": 0.0317, "step": 24460 }, { "epoch": 10.950995748489595, "grad_norm": 0.13233128190040588, "learning_rate": 6.889203934092336e-05, "loss": 0.0372, "step": 24470 }, { "epoch": 10.955471022600134, "grad_norm": 0.11338729411363602, "learning_rate": 6.886652148831279e-05, "loss": 0.0358, "step": 24480 }, { "epoch": 10.959946296710674, "grad_norm": 0.16557937860488892, "learning_rate": 6.884099790455113e-05, "loss": 0.0363, "step": 24490 }, { "epoch": 10.964421570821212, "grad_norm": 0.1426076889038086, "learning_rate": 6.881546859739179e-05, "loss": 0.0364, "step": 24500 }, { "epoch": 10.968896844931752, "grad_norm": 0.155308797955513, "learning_rate": 6.878993357458986e-05, "loss": 0.0416, "step": 24510 }, { "epoch": 10.973372119042292, "grad_norm": 0.14865683019161224, "learning_rate": 6.876439284390223e-05, "loss": 0.035, "step": 24520 }, { "epoch": 10.97784739315283, "grad_norm": 0.13962982594966888, "learning_rate": 6.873884641308752e-05, "loss": 0.0351, "step": 24530 }, { "epoch": 10.98232266726337, "grad_norm": 0.1520625203847885, "learning_rate": 6.871329428990602e-05, "loss": 0.0374, "step": 24540 }, { "epoch": 10.98679794137391, "grad_norm": 0.1675470471382141, "learning_rate": 6.868773648211983e-05, "loss": 0.0377, "step": 24550 }, { "epoch": 10.991273215484448, "grad_norm": 0.13617609441280365, "learning_rate": 6.86621729974927e-05, "loss": 0.0366, "step": 24560 }, { "epoch": 10.995748489594988, "grad_norm": 0.14217302203178406, "learning_rate": 6.863660384379017e-05, "loss": 0.0337, "step": 24570 }, { "epoch": 11.000223763705527, "grad_norm": 0.1417115479707718, "learning_rate": 6.861102902877946e-05, "loss": 0.0374, "step": 24580 }, { "epoch": 11.004699037816065, "grad_norm": 0.15593646466732025, "learning_rate": 6.858544856022952e-05, "loss": 0.0347, "step": 24590 }, { "epoch": 11.009174311926605, "grad_norm": 0.16510380804538727, "learning_rate": 6.855986244591104e-05, "loss": 0.0377, "step": 24600 }, { "epoch": 11.013649586037145, "grad_norm": 0.1596868932247162, "learning_rate": 6.85342706935964e-05, "loss": 0.0338, "step": 24610 }, { "epoch": 11.018124860147685, "grad_norm": 0.15237309038639069, "learning_rate": 6.850867331105967e-05, "loss": 0.0395, "step": 24620 }, { "epoch": 11.022600134258223, "grad_norm": 0.13013198971748352, "learning_rate": 6.84830703060767e-05, "loss": 0.0327, "step": 24630 }, { "epoch": 11.027075408368763, "grad_norm": 0.14850053191184998, "learning_rate": 6.845746168642497e-05, "loss": 0.0376, "step": 24640 }, { "epoch": 11.031550682479303, "grad_norm": 0.14042389392852783, "learning_rate": 6.843184745988373e-05, "loss": 0.0354, "step": 24650 }, { "epoch": 11.03602595658984, "grad_norm": 0.17390435934066772, "learning_rate": 6.840622763423391e-05, "loss": 0.031, "step": 24660 }, { "epoch": 11.04050123070038, "grad_norm": 0.15588568150997162, "learning_rate": 6.838060221725811e-05, "loss": 0.0376, "step": 24670 }, { "epoch": 11.04497650481092, "grad_norm": 0.15336549282073975, "learning_rate": 6.835497121674066e-05, "loss": 0.0349, "step": 24680 }, { "epoch": 11.049451778921458, "grad_norm": 0.13964100182056427, "learning_rate": 6.83293346404676e-05, "loss": 0.036, "step": 24690 }, { "epoch": 11.053927053031998, "grad_norm": 0.15514272451400757, "learning_rate": 6.830369249622662e-05, "loss": 0.0332, "step": 24700 }, { "epoch": 11.058402327142538, "grad_norm": 0.1379006803035736, "learning_rate": 6.827804479180716e-05, "loss": 0.0373, "step": 24710 }, { "epoch": 11.062877601253076, "grad_norm": 0.190521240234375, "learning_rate": 6.825239153500029e-05, "loss": 0.0383, "step": 24720 }, { "epoch": 11.067352875363616, "grad_norm": 0.19099292159080505, "learning_rate": 6.82267327335988e-05, "loss": 0.0328, "step": 24730 }, { "epoch": 11.071828149474156, "grad_norm": 0.1479155570268631, "learning_rate": 6.820106839539715e-05, "loss": 0.0347, "step": 24740 }, { "epoch": 11.076303423584694, "grad_norm": 0.19025343656539917, "learning_rate": 6.817539852819149e-05, "loss": 0.0367, "step": 24750 }, { "epoch": 11.080778697695234, "grad_norm": 0.11930341273546219, "learning_rate": 6.814972313977967e-05, "loss": 0.0334, "step": 24760 }, { "epoch": 11.085253971805773, "grad_norm": 0.16407176852226257, "learning_rate": 6.812404223796116e-05, "loss": 0.0355, "step": 24770 }, { "epoch": 11.089729245916313, "grad_norm": 0.16243630647659302, "learning_rate": 6.809835583053715e-05, "loss": 0.0371, "step": 24780 }, { "epoch": 11.094204520026851, "grad_norm": 0.1551225781440735, "learning_rate": 6.80726639253105e-05, "loss": 0.0366, "step": 24790 }, { "epoch": 11.098679794137391, "grad_norm": 0.16974247992038727, "learning_rate": 6.804696653008575e-05, "loss": 0.0369, "step": 24800 }, { "epoch": 11.103155068247931, "grad_norm": 0.15480515360832214, "learning_rate": 6.802126365266905e-05, "loss": 0.035, "step": 24810 }, { "epoch": 11.107630342358469, "grad_norm": 0.15345728397369385, "learning_rate": 6.799555530086828e-05, "loss": 0.0386, "step": 24820 }, { "epoch": 11.112105616469009, "grad_norm": 0.1686081886291504, "learning_rate": 6.796984148249295e-05, "loss": 0.0353, "step": 24830 }, { "epoch": 11.116580890579549, "grad_norm": 0.13989968597888947, "learning_rate": 6.794412220535426e-05, "loss": 0.0343, "step": 24840 }, { "epoch": 11.121056164690087, "grad_norm": 0.1385013610124588, "learning_rate": 6.7918397477265e-05, "loss": 0.0362, "step": 24850 }, { "epoch": 11.125531438800627, "grad_norm": 0.15916647017002106, "learning_rate": 6.789266730603974e-05, "loss": 0.0351, "step": 24860 }, { "epoch": 11.130006712911166, "grad_norm": 0.15058006346225739, "learning_rate": 6.786693169949455e-05, "loss": 0.0342, "step": 24870 }, { "epoch": 11.134481987021704, "grad_norm": 0.13071171939373016, "learning_rate": 6.784119066544727e-05, "loss": 0.0368, "step": 24880 }, { "epoch": 11.138957261132244, "grad_norm": 0.12320343405008316, "learning_rate": 6.781544421171732e-05, "loss": 0.0325, "step": 24890 }, { "epoch": 11.143432535242784, "grad_norm": 0.15384362637996674, "learning_rate": 6.778969234612584e-05, "loss": 0.035, "step": 24900 }, { "epoch": 11.147907809353322, "grad_norm": 0.14803767204284668, "learning_rate": 6.77639350764955e-05, "loss": 0.0359, "step": 24910 }, { "epoch": 11.152383083463862, "grad_norm": 0.15433679521083832, "learning_rate": 6.773817241065072e-05, "loss": 0.0358, "step": 24920 }, { "epoch": 11.156858357574402, "grad_norm": 0.1351025402545929, "learning_rate": 6.771240435641754e-05, "loss": 0.0371, "step": 24930 }, { "epoch": 11.16133363168494, "grad_norm": 0.12596464157104492, "learning_rate": 6.768663092162356e-05, "loss": 0.037, "step": 24940 }, { "epoch": 11.16580890579548, "grad_norm": 0.13331089913845062, "learning_rate": 6.76608521140981e-05, "loss": 0.0357, "step": 24950 }, { "epoch": 11.17028417990602, "grad_norm": 0.19440270960330963, "learning_rate": 6.763506794167208e-05, "loss": 0.0397, "step": 24960 }, { "epoch": 11.17475945401656, "grad_norm": 0.15161000192165375, "learning_rate": 6.7609278412178e-05, "loss": 0.0374, "step": 24970 }, { "epoch": 11.179234728127097, "grad_norm": 0.12446533143520355, "learning_rate": 6.758348353345014e-05, "loss": 0.0335, "step": 24980 }, { "epoch": 11.183710002237637, "grad_norm": 0.13529548048973083, "learning_rate": 6.755768331332424e-05, "loss": 0.0327, "step": 24990 }, { "epoch": 11.188185276348177, "grad_norm": 0.14076381921768188, "learning_rate": 6.753187775963773e-05, "loss": 0.033, "step": 25000 }, { "epoch": 11.192660550458715, "grad_norm": 0.15221212804317474, "learning_rate": 6.750606688022964e-05, "loss": 0.0375, "step": 25010 }, { "epoch": 11.197135824569255, "grad_norm": 0.15706543624401093, "learning_rate": 6.748025068294067e-05, "loss": 0.0331, "step": 25020 }, { "epoch": 11.201611098679795, "grad_norm": 0.16454216837882996, "learning_rate": 6.745442917561309e-05, "loss": 0.0342, "step": 25030 }, { "epoch": 11.206086372790333, "grad_norm": 0.1427244246006012, "learning_rate": 6.742860236609077e-05, "loss": 0.0382, "step": 25040 }, { "epoch": 11.210561646900873, "grad_norm": 0.12087883055210114, "learning_rate": 6.740277026221923e-05, "loss": 0.0356, "step": 25050 }, { "epoch": 11.215036921011412, "grad_norm": 0.15474195778369904, "learning_rate": 6.737693287184557e-05, "loss": 0.0355, "step": 25060 }, { "epoch": 11.21951219512195, "grad_norm": 0.14546196162700653, "learning_rate": 6.735109020281852e-05, "loss": 0.0366, "step": 25070 }, { "epoch": 11.22398746923249, "grad_norm": 0.13363859057426453, "learning_rate": 6.732524226298841e-05, "loss": 0.0363, "step": 25080 }, { "epoch": 11.22846274334303, "grad_norm": 0.12771984934806824, "learning_rate": 6.729938906020713e-05, "loss": 0.0364, "step": 25090 }, { "epoch": 11.232938017453568, "grad_norm": 0.13087189197540283, "learning_rate": 6.727353060232822e-05, "loss": 0.0374, "step": 25100 }, { "epoch": 11.237413291564108, "grad_norm": 0.15207317471504211, "learning_rate": 6.72476668972068e-05, "loss": 0.0334, "step": 25110 }, { "epoch": 11.241888565674648, "grad_norm": 0.14553241431713104, "learning_rate": 6.722179795269956e-05, "loss": 0.0346, "step": 25120 }, { "epoch": 11.246363839785186, "grad_norm": 0.13815806806087494, "learning_rate": 6.719592377666483e-05, "loss": 0.0309, "step": 25130 }, { "epoch": 11.250839113895726, "grad_norm": 0.14831915497779846, "learning_rate": 6.71700443769625e-05, "loss": 0.0364, "step": 25140 }, { "epoch": 11.255314388006266, "grad_norm": 0.11526942998170853, "learning_rate": 6.714415976145402e-05, "loss": 0.0351, "step": 25150 }, { "epoch": 11.259789662116805, "grad_norm": 0.12048986554145813, "learning_rate": 6.711826993800248e-05, "loss": 0.0316, "step": 25160 }, { "epoch": 11.264264936227343, "grad_norm": 0.13066059350967407, "learning_rate": 6.709237491447249e-05, "loss": 0.0335, "step": 25170 }, { "epoch": 11.268740210337883, "grad_norm": 0.12661565840244293, "learning_rate": 6.706647469873031e-05, "loss": 0.0301, "step": 25180 }, { "epoch": 11.273215484448423, "grad_norm": 0.13146917521953583, "learning_rate": 6.704056929864376e-05, "loss": 0.0334, "step": 25190 }, { "epoch": 11.277690758558961, "grad_norm": 0.12546038627624512, "learning_rate": 6.701465872208216e-05, "loss": 0.0353, "step": 25200 }, { "epoch": 11.282166032669501, "grad_norm": 0.13693642616271973, "learning_rate": 6.69887429769165e-05, "loss": 0.0385, "step": 25210 }, { "epoch": 11.28664130678004, "grad_norm": 0.14199501276016235, "learning_rate": 6.696282207101928e-05, "loss": 0.0344, "step": 25220 }, { "epoch": 11.291116580890579, "grad_norm": 0.1368851512670517, "learning_rate": 6.693689601226458e-05, "loss": 0.0361, "step": 25230 }, { "epoch": 11.295591855001119, "grad_norm": 0.12014574557542801, "learning_rate": 6.691096480852808e-05, "loss": 0.0386, "step": 25240 }, { "epoch": 11.300067129111659, "grad_norm": 0.14346694946289062, "learning_rate": 6.688502846768696e-05, "loss": 0.0332, "step": 25250 }, { "epoch": 11.304542403222197, "grad_norm": 0.14308284223079681, "learning_rate": 6.685908699762002e-05, "loss": 0.0353, "step": 25260 }, { "epoch": 11.309017677332736, "grad_norm": 0.14636018872261047, "learning_rate": 6.683314040620759e-05, "loss": 0.0341, "step": 25270 }, { "epoch": 11.313492951443276, "grad_norm": 0.14695607125759125, "learning_rate": 6.680718870133156e-05, "loss": 0.0357, "step": 25280 }, { "epoch": 11.317968225553814, "grad_norm": 0.14309720695018768, "learning_rate": 6.67812318908754e-05, "loss": 0.0347, "step": 25290 }, { "epoch": 11.322443499664354, "grad_norm": 0.14528623223304749, "learning_rate": 6.675526998272405e-05, "loss": 0.0366, "step": 25300 }, { "epoch": 11.326918773774894, "grad_norm": 0.1426745057106018, "learning_rate": 6.672930298476407e-05, "loss": 0.0358, "step": 25310 }, { "epoch": 11.331394047885434, "grad_norm": 0.17992253601551056, "learning_rate": 6.670333090488356e-05, "loss": 0.0367, "step": 25320 }, { "epoch": 11.335869321995972, "grad_norm": 0.13073551654815674, "learning_rate": 6.667735375097214e-05, "loss": 0.0339, "step": 25330 }, { "epoch": 11.340344596106512, "grad_norm": 0.15208585560321808, "learning_rate": 6.6651371530921e-05, "loss": 0.0348, "step": 25340 }, { "epoch": 11.344819870217052, "grad_norm": 0.15515361726284027, "learning_rate": 6.662538425262285e-05, "loss": 0.0342, "step": 25350 }, { "epoch": 11.34929514432759, "grad_norm": 0.1427694857120514, "learning_rate": 6.659939192397192e-05, "loss": 0.0345, "step": 25360 }, { "epoch": 11.35377041843813, "grad_norm": 0.13659171760082245, "learning_rate": 6.6573394552864e-05, "loss": 0.0319, "step": 25370 }, { "epoch": 11.35824569254867, "grad_norm": 0.14450985193252563, "learning_rate": 6.654739214719641e-05, "loss": 0.0321, "step": 25380 }, { "epoch": 11.362720966659207, "grad_norm": 0.15391287207603455, "learning_rate": 6.6521384714868e-05, "loss": 0.0389, "step": 25390 }, { "epoch": 11.367196240769747, "grad_norm": 0.17903493344783783, "learning_rate": 6.649537226377915e-05, "loss": 0.0406, "step": 25400 }, { "epoch": 11.371671514880287, "grad_norm": 0.14185427129268646, "learning_rate": 6.646935480183173e-05, "loss": 0.0333, "step": 25410 }, { "epoch": 11.376146788990825, "grad_norm": 0.16424846649169922, "learning_rate": 6.644333233692916e-05, "loss": 0.0348, "step": 25420 }, { "epoch": 11.380622063101365, "grad_norm": 0.14506468176841736, "learning_rate": 6.641730487697639e-05, "loss": 0.0352, "step": 25430 }, { "epoch": 11.385097337211905, "grad_norm": 0.14604538679122925, "learning_rate": 6.639127242987988e-05, "loss": 0.0337, "step": 25440 }, { "epoch": 11.389572611322443, "grad_norm": 0.15398776531219482, "learning_rate": 6.636523500354759e-05, "loss": 0.0348, "step": 25450 }, { "epoch": 11.394047885432983, "grad_norm": 0.147462397813797, "learning_rate": 6.6339192605889e-05, "loss": 0.035, "step": 25460 }, { "epoch": 11.398523159543522, "grad_norm": 0.1592215895652771, "learning_rate": 6.631314524481513e-05, "loss": 0.0354, "step": 25470 }, { "epoch": 11.402998433654062, "grad_norm": 0.1476496011018753, "learning_rate": 6.628709292823844e-05, "loss": 0.0344, "step": 25480 }, { "epoch": 11.4074737077646, "grad_norm": 0.19405612349510193, "learning_rate": 6.626103566407295e-05, "loss": 0.0347, "step": 25490 }, { "epoch": 11.41194898187514, "grad_norm": 0.15657484531402588, "learning_rate": 6.623497346023418e-05, "loss": 0.0355, "step": 25500 }, { "epoch": 11.41642425598568, "grad_norm": 0.15208199620246887, "learning_rate": 6.620890632463915e-05, "loss": 0.0308, "step": 25510 }, { "epoch": 11.420899530096218, "grad_norm": 0.12684684991836548, "learning_rate": 6.61828342652063e-05, "loss": 0.0373, "step": 25520 }, { "epoch": 11.425374804206758, "grad_norm": 0.11900705844163895, "learning_rate": 6.615675728985572e-05, "loss": 0.0349, "step": 25530 }, { "epoch": 11.429850078317298, "grad_norm": 0.13497784733772278, "learning_rate": 6.613067540650886e-05, "loss": 0.0352, "step": 25540 }, { "epoch": 11.434325352427836, "grad_norm": 0.15486428141593933, "learning_rate": 6.610458862308872e-05, "loss": 0.0341, "step": 25550 }, { "epoch": 11.438800626538375, "grad_norm": 0.13346923887729645, "learning_rate": 6.607849694751977e-05, "loss": 0.0333, "step": 25560 }, { "epoch": 11.443275900648915, "grad_norm": 0.14234763383865356, "learning_rate": 6.6052400387728e-05, "loss": 0.0355, "step": 25570 }, { "epoch": 11.447751174759453, "grad_norm": 0.13669519126415253, "learning_rate": 6.602629895164081e-05, "loss": 0.0372, "step": 25580 }, { "epoch": 11.452226448869993, "grad_norm": 0.11051184684038162, "learning_rate": 6.600019264718713e-05, "loss": 0.0382, "step": 25590 }, { "epoch": 11.456701722980533, "grad_norm": 0.12693026661872864, "learning_rate": 6.59740814822974e-05, "loss": 0.0332, "step": 25600 }, { "epoch": 11.461176997091071, "grad_norm": 0.12451226264238358, "learning_rate": 6.594796546490351e-05, "loss": 0.0358, "step": 25610 }, { "epoch": 11.465652271201611, "grad_norm": 0.170668825507164, "learning_rate": 6.592184460293877e-05, "loss": 0.0338, "step": 25620 }, { "epoch": 11.47012754531215, "grad_norm": 0.15209834277629852, "learning_rate": 6.589571890433803e-05, "loss": 0.0374, "step": 25630 }, { "epoch": 11.474602819422689, "grad_norm": 0.1463908553123474, "learning_rate": 6.586958837703759e-05, "loss": 0.0335, "step": 25640 }, { "epoch": 11.479078093533229, "grad_norm": 0.16246654093265533, "learning_rate": 6.584345302897523e-05, "loss": 0.0341, "step": 25650 }, { "epoch": 11.483553367643768, "grad_norm": 0.14600341022014618, "learning_rate": 6.581731286809014e-05, "loss": 0.0342, "step": 25660 }, { "epoch": 11.488028641754308, "grad_norm": 0.1586671620607376, "learning_rate": 6.579116790232305e-05, "loss": 0.0352, "step": 25670 }, { "epoch": 11.492503915864846, "grad_norm": 0.13028264045715332, "learning_rate": 6.576501813961609e-05, "loss": 0.0309, "step": 25680 }, { "epoch": 11.496979189975386, "grad_norm": 0.15048986673355103, "learning_rate": 6.573886358791285e-05, "loss": 0.0385, "step": 25690 }, { "epoch": 11.501454464085926, "grad_norm": 0.16591264307498932, "learning_rate": 6.571270425515843e-05, "loss": 0.0325, "step": 25700 }, { "epoch": 11.505929738196464, "grad_norm": 0.12659336626529694, "learning_rate": 6.568654014929932e-05, "loss": 0.0342, "step": 25710 }, { "epoch": 11.510405012307004, "grad_norm": 0.1586328148841858, "learning_rate": 6.56603712782835e-05, "loss": 0.0341, "step": 25720 }, { "epoch": 11.514880286417544, "grad_norm": 0.18400245904922485, "learning_rate": 6.563419765006038e-05, "loss": 0.0341, "step": 25730 }, { "epoch": 11.519355560528082, "grad_norm": 0.11995118856430054, "learning_rate": 6.56080192725808e-05, "loss": 0.0341, "step": 25740 }, { "epoch": 11.523830834638622, "grad_norm": 0.14122061431407928, "learning_rate": 6.558183615379707e-05, "loss": 0.0358, "step": 25750 }, { "epoch": 11.528306108749161, "grad_norm": 0.11355946213006973, "learning_rate": 6.555564830166293e-05, "loss": 0.0334, "step": 25760 }, { "epoch": 11.5327813828597, "grad_norm": 0.15025115013122559, "learning_rate": 6.552945572413358e-05, "loss": 0.036, "step": 25770 }, { "epoch": 11.53725665697024, "grad_norm": 0.1511271893978119, "learning_rate": 6.550325842916559e-05, "loss": 0.0344, "step": 25780 }, { "epoch": 11.541731931080779, "grad_norm": 0.14123882353305817, "learning_rate": 6.547705642471703e-05, "loss": 0.0339, "step": 25790 }, { "epoch": 11.546207205191317, "grad_norm": 0.13584092259407043, "learning_rate": 6.545084971874738e-05, "loss": 0.0371, "step": 25800 }, { "epoch": 11.550682479301857, "grad_norm": 0.14756102859973907, "learning_rate": 6.542463831921754e-05, "loss": 0.0358, "step": 25810 }, { "epoch": 11.555157753412397, "grad_norm": 0.1637854427099228, "learning_rate": 6.539842223408984e-05, "loss": 0.0388, "step": 25820 }, { "epoch": 11.559633027522935, "grad_norm": 0.16233329474925995, "learning_rate": 6.537220147132805e-05, "loss": 0.0317, "step": 25830 }, { "epoch": 11.564108301633475, "grad_norm": 0.18384113907814026, "learning_rate": 6.534597603889732e-05, "loss": 0.0362, "step": 25840 }, { "epoch": 11.568583575744015, "grad_norm": 0.14943650364875793, "learning_rate": 6.531974594476425e-05, "loss": 0.0342, "step": 25850 }, { "epoch": 11.573058849854554, "grad_norm": 0.15245263278484344, "learning_rate": 6.529351119689688e-05, "loss": 0.036, "step": 25860 }, { "epoch": 11.577534123965092, "grad_norm": 0.1370159387588501, "learning_rate": 6.52672718032646e-05, "loss": 0.0356, "step": 25870 }, { "epoch": 11.582009398075632, "grad_norm": 0.13449902832508087, "learning_rate": 6.524102777183825e-05, "loss": 0.0342, "step": 25880 }, { "epoch": 11.586484672186172, "grad_norm": 0.13498549163341522, "learning_rate": 6.521477911059008e-05, "loss": 0.0314, "step": 25890 }, { "epoch": 11.59095994629671, "grad_norm": 0.1342642605304718, "learning_rate": 6.518852582749373e-05, "loss": 0.0321, "step": 25900 }, { "epoch": 11.59543522040725, "grad_norm": 0.11038573086261749, "learning_rate": 6.516226793052428e-05, "loss": 0.0328, "step": 25910 }, { "epoch": 11.59991049451779, "grad_norm": 0.13829825818538666, "learning_rate": 6.513600542765817e-05, "loss": 0.037, "step": 25920 }, { "epoch": 11.604385768628328, "grad_norm": 0.13756701350212097, "learning_rate": 6.510973832687323e-05, "loss": 0.0348, "step": 25930 }, { "epoch": 11.608861042738868, "grad_norm": 0.13891129195690155, "learning_rate": 6.508346663614878e-05, "loss": 0.0361, "step": 25940 }, { "epoch": 11.613336316849407, "grad_norm": 0.1486867368221283, "learning_rate": 6.505719036346539e-05, "loss": 0.0345, "step": 25950 }, { "epoch": 11.617811590959946, "grad_norm": 0.14223915338516235, "learning_rate": 6.503090951680512e-05, "loss": 0.0333, "step": 25960 }, { "epoch": 11.622286865070485, "grad_norm": 0.135994091629982, "learning_rate": 6.500462410415142e-05, "loss": 0.0368, "step": 25970 }, { "epoch": 11.626762139181025, "grad_norm": 0.12289035320281982, "learning_rate": 6.497833413348909e-05, "loss": 0.0359, "step": 25980 }, { "epoch": 11.631237413291565, "grad_norm": 0.12501156330108643, "learning_rate": 6.495203961280434e-05, "loss": 0.0343, "step": 25990 }, { "epoch": 11.635712687402103, "grad_norm": 0.13643740117549896, "learning_rate": 6.492574055008473e-05, "loss": 0.035, "step": 26000 }, { "epoch": 11.640187961512643, "grad_norm": 0.1505189687013626, "learning_rate": 6.489943695331923e-05, "loss": 0.0305, "step": 26010 }, { "epoch": 11.644663235623183, "grad_norm": 0.16179721057415009, "learning_rate": 6.487312883049819e-05, "loss": 0.0348, "step": 26020 }, { "epoch": 11.64913850973372, "grad_norm": 0.12674610316753387, "learning_rate": 6.484681618961331e-05, "loss": 0.0333, "step": 26030 }, { "epoch": 11.65361378384426, "grad_norm": 0.11363918334245682, "learning_rate": 6.48204990386577e-05, "loss": 0.0305, "step": 26040 }, { "epoch": 11.6580890579548, "grad_norm": 0.16479937732219696, "learning_rate": 6.479417738562576e-05, "loss": 0.0338, "step": 26050 }, { "epoch": 11.662564332065338, "grad_norm": 0.13171368837356567, "learning_rate": 6.476785123851336e-05, "loss": 0.0356, "step": 26060 }, { "epoch": 11.667039606175878, "grad_norm": 0.12685547769069672, "learning_rate": 6.474152060531768e-05, "loss": 0.0337, "step": 26070 }, { "epoch": 11.671514880286418, "grad_norm": 0.13773025572299957, "learning_rate": 6.471518549403726e-05, "loss": 0.0317, "step": 26080 }, { "epoch": 11.675990154396956, "grad_norm": 0.1395733654499054, "learning_rate": 6.468884591267204e-05, "loss": 0.0324, "step": 26090 }, { "epoch": 11.680465428507496, "grad_norm": 0.15377593040466309, "learning_rate": 6.466250186922325e-05, "loss": 0.0338, "step": 26100 }, { "epoch": 11.684940702618036, "grad_norm": 0.12084140628576279, "learning_rate": 6.463615337169353e-05, "loss": 0.0358, "step": 26110 }, { "epoch": 11.689415976728574, "grad_norm": 0.14681826531887054, "learning_rate": 6.460980042808687e-05, "loss": 0.0347, "step": 26120 }, { "epoch": 11.693891250839114, "grad_norm": 0.16079089045524597, "learning_rate": 6.458344304640858e-05, "loss": 0.0339, "step": 26130 }, { "epoch": 11.698366524949654, "grad_norm": 0.1489315778017044, "learning_rate": 6.455708123466536e-05, "loss": 0.0325, "step": 26140 }, { "epoch": 11.702841799060192, "grad_norm": 0.15368975698947906, "learning_rate": 6.45307150008652e-05, "loss": 0.0352, "step": 26150 }, { "epoch": 11.707317073170731, "grad_norm": 0.15037581324577332, "learning_rate": 6.450434435301751e-05, "loss": 0.0329, "step": 26160 }, { "epoch": 11.711792347281271, "grad_norm": 0.1399565190076828, "learning_rate": 6.447796929913294e-05, "loss": 0.0354, "step": 26170 }, { "epoch": 11.716267621391811, "grad_norm": 0.13828669488430023, "learning_rate": 6.445158984722358e-05, "loss": 0.0343, "step": 26180 }, { "epoch": 11.72074289550235, "grad_norm": 0.14452090859413147, "learning_rate": 6.44252060053028e-05, "loss": 0.0367, "step": 26190 }, { "epoch": 11.725218169612889, "grad_norm": 0.15425193309783936, "learning_rate": 6.439881778138531e-05, "loss": 0.0367, "step": 26200 }, { "epoch": 11.729693443723429, "grad_norm": 0.16490031778812408, "learning_rate": 6.437242518348716e-05, "loss": 0.0361, "step": 26210 }, { "epoch": 11.734168717833967, "grad_norm": 0.15057140588760376, "learning_rate": 6.43460282196257e-05, "loss": 0.0342, "step": 26220 }, { "epoch": 11.738643991944507, "grad_norm": 0.12398771196603775, "learning_rate": 6.431962689781969e-05, "loss": 0.0349, "step": 26230 }, { "epoch": 11.743119266055047, "grad_norm": 0.14128939807415009, "learning_rate": 6.42932212260891e-05, "loss": 0.0335, "step": 26240 }, { "epoch": 11.747594540165585, "grad_norm": 0.15261244773864746, "learning_rate": 6.426681121245527e-05, "loss": 0.0376, "step": 26250 }, { "epoch": 11.752069814276124, "grad_norm": 0.14929087460041046, "learning_rate": 6.42403968649409e-05, "loss": 0.034, "step": 26260 }, { "epoch": 11.756545088386664, "grad_norm": 0.15189720690250397, "learning_rate": 6.421397819156996e-05, "loss": 0.0321, "step": 26270 }, { "epoch": 11.761020362497202, "grad_norm": 0.1276979148387909, "learning_rate": 6.418755520036775e-05, "loss": 0.0308, "step": 26280 }, { "epoch": 11.765495636607742, "grad_norm": 0.1543702334165573, "learning_rate": 6.416112789936086e-05, "loss": 0.0352, "step": 26290 }, { "epoch": 11.769970910718282, "grad_norm": 0.16182424128055573, "learning_rate": 6.413469629657723e-05, "loss": 0.0362, "step": 26300 }, { "epoch": 11.77444618482882, "grad_norm": 0.14141897857189178, "learning_rate": 6.410826040004607e-05, "loss": 0.035, "step": 26310 }, { "epoch": 11.77892145893936, "grad_norm": 0.14014311134815216, "learning_rate": 6.408182021779791e-05, "loss": 0.0369, "step": 26320 }, { "epoch": 11.7833967330499, "grad_norm": 0.13123509287834167, "learning_rate": 6.405537575786456e-05, "loss": 0.0361, "step": 26330 }, { "epoch": 11.787872007160438, "grad_norm": 0.15077434480190277, "learning_rate": 6.402892702827916e-05, "loss": 0.0384, "step": 26340 }, { "epoch": 11.792347281270978, "grad_norm": 0.14237043261528015, "learning_rate": 6.400247403707617e-05, "loss": 0.0351, "step": 26350 }, { "epoch": 11.796822555381517, "grad_norm": 0.15475639700889587, "learning_rate": 6.397601679229126e-05, "loss": 0.0357, "step": 26360 }, { "epoch": 11.801297829492057, "grad_norm": 0.1516905128955841, "learning_rate": 6.394955530196147e-05, "loss": 0.0343, "step": 26370 }, { "epoch": 11.805773103602595, "grad_norm": 0.16059468686580658, "learning_rate": 6.39230895741251e-05, "loss": 0.0354, "step": 26380 }, { "epoch": 11.810248377713135, "grad_norm": 0.1319226175546646, "learning_rate": 6.389661961682173e-05, "loss": 0.0329, "step": 26390 }, { "epoch": 11.814723651823675, "grad_norm": 0.1663103699684143, "learning_rate": 6.387014543809223e-05, "loss": 0.0329, "step": 26400 }, { "epoch": 11.819198925934213, "grad_norm": 0.17339015007019043, "learning_rate": 6.384366704597879e-05, "loss": 0.0349, "step": 26410 }, { "epoch": 11.823674200044753, "grad_norm": 0.13199003040790558, "learning_rate": 6.38171844485248e-05, "loss": 0.0342, "step": 26420 }, { "epoch": 11.828149474155293, "grad_norm": 0.13272427022457123, "learning_rate": 6.3790697653775e-05, "loss": 0.0355, "step": 26430 }, { "epoch": 11.83262474826583, "grad_norm": 0.15193693339824677, "learning_rate": 6.376420666977538e-05, "loss": 0.0334, "step": 26440 }, { "epoch": 11.83710002237637, "grad_norm": 0.14204517006874084, "learning_rate": 6.37377115045732e-05, "loss": 0.036, "step": 26450 }, { "epoch": 11.84157529648691, "grad_norm": 0.1520778387784958, "learning_rate": 6.371121216621698e-05, "loss": 0.0312, "step": 26460 }, { "epoch": 11.846050570597448, "grad_norm": 0.16536349058151245, "learning_rate": 6.368470866275654e-05, "loss": 0.0337, "step": 26470 }, { "epoch": 11.850525844707988, "grad_norm": 0.14914217591285706, "learning_rate": 6.365820100224292e-05, "loss": 0.0347, "step": 26480 }, { "epoch": 11.855001118818528, "grad_norm": 0.15048010647296906, "learning_rate": 6.363168919272846e-05, "loss": 0.0341, "step": 26490 }, { "epoch": 11.859476392929066, "grad_norm": 0.14016474783420563, "learning_rate": 6.360517324226676e-05, "loss": 0.0333, "step": 26500 }, { "epoch": 11.863951667039606, "grad_norm": 0.1349562108516693, "learning_rate": 6.357865315891265e-05, "loss": 0.0334, "step": 26510 }, { "epoch": 11.868426941150146, "grad_norm": 0.1480182409286499, "learning_rate": 6.355212895072223e-05, "loss": 0.0393, "step": 26520 }, { "epoch": 11.872902215260684, "grad_norm": 0.1497170776128769, "learning_rate": 6.352560062575284e-05, "loss": 0.0361, "step": 26530 }, { "epoch": 11.877377489371224, "grad_norm": 0.13099320232868195, "learning_rate": 6.349906819206313e-05, "loss": 0.0344, "step": 26540 }, { "epoch": 11.881852763481763, "grad_norm": 0.12749093770980835, "learning_rate": 6.34725316577129e-05, "loss": 0.0317, "step": 26550 }, { "epoch": 11.886328037592303, "grad_norm": 0.12297697365283966, "learning_rate": 6.344599103076329e-05, "loss": 0.0362, "step": 26560 }, { "epoch": 11.890803311702841, "grad_norm": 0.13208705186843872, "learning_rate": 6.341944631927662e-05, "loss": 0.0354, "step": 26570 }, { "epoch": 11.895278585813381, "grad_norm": 0.1312793791294098, "learning_rate": 6.339289753131649e-05, "loss": 0.0293, "step": 26580 }, { "epoch": 11.899753859923921, "grad_norm": 0.14549461007118225, "learning_rate": 6.336634467494768e-05, "loss": 0.0349, "step": 26590 }, { "epoch": 11.904229134034459, "grad_norm": 0.12442035973072052, "learning_rate": 6.333978775823631e-05, "loss": 0.0335, "step": 26600 }, { "epoch": 11.908704408144999, "grad_norm": 0.1409069448709488, "learning_rate": 6.331322678924962e-05, "loss": 0.0307, "step": 26610 }, { "epoch": 11.913179682255539, "grad_norm": 0.13896483182907104, "learning_rate": 6.328666177605616e-05, "loss": 0.0361, "step": 26620 }, { "epoch": 11.917654956366077, "grad_norm": 0.18492722511291504, "learning_rate": 6.326009272672564e-05, "loss": 0.0346, "step": 26630 }, { "epoch": 11.922130230476617, "grad_norm": 0.14462216198444366, "learning_rate": 6.323351964932908e-05, "loss": 0.0341, "step": 26640 }, { "epoch": 11.926605504587156, "grad_norm": 0.16631685197353363, "learning_rate": 6.320694255193867e-05, "loss": 0.0327, "step": 26650 }, { "epoch": 11.931080778697694, "grad_norm": 0.13757991790771484, "learning_rate": 6.31803614426278e-05, "loss": 0.0398, "step": 26660 }, { "epoch": 11.935556052808234, "grad_norm": 0.16235221922397614, "learning_rate": 6.315377632947115e-05, "loss": 0.0337, "step": 26670 }, { "epoch": 11.940031326918774, "grad_norm": 0.1802998185157776, "learning_rate": 6.312718722054454e-05, "loss": 0.0346, "step": 26680 }, { "epoch": 11.944506601029314, "grad_norm": 0.16240736842155457, "learning_rate": 6.310059412392505e-05, "loss": 0.0338, "step": 26690 }, { "epoch": 11.948981875139852, "grad_norm": 0.15333721041679382, "learning_rate": 6.307399704769099e-05, "loss": 0.035, "step": 26700 }, { "epoch": 11.953457149250392, "grad_norm": 0.14632906019687653, "learning_rate": 6.304739599992183e-05, "loss": 0.0334, "step": 26710 }, { "epoch": 11.957932423360932, "grad_norm": 0.15769152343273163, "learning_rate": 6.302079098869824e-05, "loss": 0.0319, "step": 26720 }, { "epoch": 11.96240769747147, "grad_norm": 0.16147537529468536, "learning_rate": 6.299418202210214e-05, "loss": 0.0329, "step": 26730 }, { "epoch": 11.96688297158201, "grad_norm": 0.14361435174942017, "learning_rate": 6.296756910821666e-05, "loss": 0.0334, "step": 26740 }, { "epoch": 11.97135824569255, "grad_norm": 0.13140283524990082, "learning_rate": 6.294095225512603e-05, "loss": 0.0321, "step": 26750 }, { "epoch": 11.975833519803087, "grad_norm": 0.125149667263031, "learning_rate": 6.291433147091583e-05, "loss": 0.0369, "step": 26760 }, { "epoch": 11.980308793913627, "grad_norm": 0.10604999214410782, "learning_rate": 6.28877067636727e-05, "loss": 0.0319, "step": 26770 }, { "epoch": 11.984784068024167, "grad_norm": 0.14866483211517334, "learning_rate": 6.286107814148454e-05, "loss": 0.0348, "step": 26780 }, { "epoch": 11.989259342134705, "grad_norm": 0.169707790017128, "learning_rate": 6.283444561244042e-05, "loss": 0.0356, "step": 26790 }, { "epoch": 11.993734616245245, "grad_norm": 0.1437348872423172, "learning_rate": 6.280780918463057e-05, "loss": 0.0334, "step": 26800 }, { "epoch": 11.998209890355785, "grad_norm": 0.14065244793891907, "learning_rate": 6.278116886614648e-05, "loss": 0.0352, "step": 26810 }, { "epoch": 12.002685164466323, "grad_norm": 0.13404229283332825, "learning_rate": 6.275452466508077e-05, "loss": 0.034, "step": 26820 }, { "epoch": 12.007160438576863, "grad_norm": 0.1054052785038948, "learning_rate": 6.27278765895272e-05, "loss": 0.0328, "step": 26830 }, { "epoch": 12.011635712687402, "grad_norm": 0.13936401903629303, "learning_rate": 6.27012246475808e-05, "loss": 0.0369, "step": 26840 }, { "epoch": 12.01611098679794, "grad_norm": 0.16576418280601501, "learning_rate": 6.26745688473377e-05, "loss": 0.0346, "step": 26850 }, { "epoch": 12.02058626090848, "grad_norm": 0.1246430054306984, "learning_rate": 6.264790919689525e-05, "loss": 0.0363, "step": 26860 }, { "epoch": 12.02506153501902, "grad_norm": 0.13933642208576202, "learning_rate": 6.262124570435193e-05, "loss": 0.0337, "step": 26870 }, { "epoch": 12.02953680912956, "grad_norm": 0.12121830135583878, "learning_rate": 6.259457837780742e-05, "loss": 0.0335, "step": 26880 }, { "epoch": 12.034012083240098, "grad_norm": 0.1638423651456833, "learning_rate": 6.256790722536251e-05, "loss": 0.0335, "step": 26890 }, { "epoch": 12.038487357350638, "grad_norm": 0.12263534218072891, "learning_rate": 6.254123225511923e-05, "loss": 0.0341, "step": 26900 }, { "epoch": 12.042962631461178, "grad_norm": 0.14917494356632233, "learning_rate": 6.251455347518073e-05, "loss": 0.0339, "step": 26910 }, { "epoch": 12.047437905571716, "grad_norm": 0.127987802028656, "learning_rate": 6.248787089365133e-05, "loss": 0.0312, "step": 26920 }, { "epoch": 12.051913179682256, "grad_norm": 0.16247743368148804, "learning_rate": 6.246118451863646e-05, "loss": 0.0342, "step": 26930 }, { "epoch": 12.056388453792795, "grad_norm": 0.1315723955631256, "learning_rate": 6.243449435824276e-05, "loss": 0.0363, "step": 26940 }, { "epoch": 12.060863727903333, "grad_norm": 0.12546125054359436, "learning_rate": 6.240780042057796e-05, "loss": 0.0342, "step": 26950 }, { "epoch": 12.065339002013873, "grad_norm": 0.15190638601779938, "learning_rate": 6.238110271375102e-05, "loss": 0.0364, "step": 26960 }, { "epoch": 12.069814276124413, "grad_norm": 0.15725989639759064, "learning_rate": 6.235440124587198e-05, "loss": 0.0343, "step": 26970 }, { "epoch": 12.074289550234951, "grad_norm": 0.12996239960193634, "learning_rate": 6.232769602505203e-05, "loss": 0.0321, "step": 26980 }, { "epoch": 12.078764824345491, "grad_norm": 0.1236506849527359, "learning_rate": 6.230098705940354e-05, "loss": 0.0401, "step": 26990 }, { "epoch": 12.08324009845603, "grad_norm": 0.13141527771949768, "learning_rate": 6.227427435703997e-05, "loss": 0.0338, "step": 27000 }, { "epoch": 12.087715372566569, "grad_norm": 0.12013858556747437, "learning_rate": 6.22475579260759e-05, "loss": 0.0334, "step": 27010 }, { "epoch": 12.092190646677109, "grad_norm": 0.1286034733057022, "learning_rate": 6.222083777462715e-05, "loss": 0.0337, "step": 27020 }, { "epoch": 12.096665920787649, "grad_norm": 0.1590978354215622, "learning_rate": 6.219411391081055e-05, "loss": 0.0345, "step": 27030 }, { "epoch": 12.101141194898187, "grad_norm": 0.14670534431934357, "learning_rate": 6.216738634274411e-05, "loss": 0.032, "step": 27040 }, { "epoch": 12.105616469008726, "grad_norm": 0.1502721756696701, "learning_rate": 6.214065507854693e-05, "loss": 0.0345, "step": 27050 }, { "epoch": 12.110091743119266, "grad_norm": 0.12306280434131622, "learning_rate": 6.211392012633932e-05, "loss": 0.0363, "step": 27060 }, { "epoch": 12.114567017229806, "grad_norm": 0.13092218339443207, "learning_rate": 6.208718149424263e-05, "loss": 0.0343, "step": 27070 }, { "epoch": 12.119042291340344, "grad_norm": 0.14480498433113098, "learning_rate": 6.206043919037933e-05, "loss": 0.0332, "step": 27080 }, { "epoch": 12.123517565450884, "grad_norm": 0.13440020382404327, "learning_rate": 6.203369322287306e-05, "loss": 0.0323, "step": 27090 }, { "epoch": 12.127992839561424, "grad_norm": 0.12711089849472046, "learning_rate": 6.200694359984849e-05, "loss": 0.0319, "step": 27100 }, { "epoch": 12.132468113671962, "grad_norm": 0.11393361538648605, "learning_rate": 6.19801903294315e-05, "loss": 0.0337, "step": 27110 }, { "epoch": 12.136943387782502, "grad_norm": 0.1633966714143753, "learning_rate": 6.195343341974899e-05, "loss": 0.0338, "step": 27120 }, { "epoch": 12.141418661893042, "grad_norm": 0.12967947125434875, "learning_rate": 6.192667287892905e-05, "loss": 0.0333, "step": 27130 }, { "epoch": 12.14589393600358, "grad_norm": 0.15102066099643707, "learning_rate": 6.189990871510078e-05, "loss": 0.0315, "step": 27140 }, { "epoch": 12.15036921011412, "grad_norm": 0.1345064640045166, "learning_rate": 6.187314093639444e-05, "loss": 0.0327, "step": 27150 }, { "epoch": 12.15484448422466, "grad_norm": 0.1622081696987152, "learning_rate": 6.184636955094138e-05, "loss": 0.0337, "step": 27160 }, { "epoch": 12.159319758335197, "grad_norm": 0.15409547090530396, "learning_rate": 6.181959456687402e-05, "loss": 0.0343, "step": 27170 }, { "epoch": 12.163795032445737, "grad_norm": 0.14494600892066956, "learning_rate": 6.179281599232591e-05, "loss": 0.0342, "step": 27180 }, { "epoch": 12.168270306556277, "grad_norm": 0.15778981149196625, "learning_rate": 6.17660338354317e-05, "loss": 0.0311, "step": 27190 }, { "epoch": 12.172745580666815, "grad_norm": 0.13488571345806122, "learning_rate": 6.173924810432705e-05, "loss": 0.0332, "step": 27200 }, { "epoch": 12.177220854777355, "grad_norm": 0.13885949552059174, "learning_rate": 6.17124588071488e-05, "loss": 0.0344, "step": 27210 }, { "epoch": 12.181696128887895, "grad_norm": 0.12352212518453598, "learning_rate": 6.168566595203479e-05, "loss": 0.0331, "step": 27220 }, { "epoch": 12.186171402998434, "grad_norm": 0.1179497167468071, "learning_rate": 6.165886954712401e-05, "loss": 0.0337, "step": 27230 }, { "epoch": 12.190646677108973, "grad_norm": 0.13728632032871246, "learning_rate": 6.163206960055651e-05, "loss": 0.037, "step": 27240 }, { "epoch": 12.195121951219512, "grad_norm": 0.12236477434635162, "learning_rate": 6.16052661204734e-05, "loss": 0.0337, "step": 27250 }, { "epoch": 12.199597225330052, "grad_norm": 0.1175297200679779, "learning_rate": 6.157845911501684e-05, "loss": 0.0331, "step": 27260 }, { "epoch": 12.20407249944059, "grad_norm": 0.14147742092609406, "learning_rate": 6.155164859233012e-05, "loss": 0.0304, "step": 27270 }, { "epoch": 12.20854777355113, "grad_norm": 0.11988963186740875, "learning_rate": 6.152483456055756e-05, "loss": 0.0338, "step": 27280 }, { "epoch": 12.21302304766167, "grad_norm": 0.14300580322742462, "learning_rate": 6.149801702784456e-05, "loss": 0.0322, "step": 27290 }, { "epoch": 12.217498321772208, "grad_norm": 0.1385045200586319, "learning_rate": 6.147119600233758e-05, "loss": 0.0351, "step": 27300 }, { "epoch": 12.221973595882748, "grad_norm": 0.1739538013935089, "learning_rate": 6.144437149218415e-05, "loss": 0.0332, "step": 27310 }, { "epoch": 12.226448869993288, "grad_norm": 0.1381087452173233, "learning_rate": 6.141754350553279e-05, "loss": 0.0329, "step": 27320 }, { "epoch": 12.230924144103826, "grad_norm": 0.12457727640867233, "learning_rate": 6.13907120505332e-05, "loss": 0.0359, "step": 27330 }, { "epoch": 12.235399418214365, "grad_norm": 0.12884686887264252, "learning_rate": 6.136387713533603e-05, "loss": 0.0331, "step": 27340 }, { "epoch": 12.239874692324905, "grad_norm": 0.11705426126718521, "learning_rate": 6.133703876809305e-05, "loss": 0.0313, "step": 27350 }, { "epoch": 12.244349966435443, "grad_norm": 0.13355472683906555, "learning_rate": 6.131019695695702e-05, "loss": 0.0302, "step": 27360 }, { "epoch": 12.248825240545983, "grad_norm": 0.11929863691329956, "learning_rate": 6.12833517100818e-05, "loss": 0.0335, "step": 27370 }, { "epoch": 12.253300514656523, "grad_norm": 0.140811026096344, "learning_rate": 6.125650303562221e-05, "loss": 0.0347, "step": 27380 }, { "epoch": 12.257775788767063, "grad_norm": 0.13037587702274323, "learning_rate": 6.122965094173424e-05, "loss": 0.0348, "step": 27390 }, { "epoch": 12.262251062877601, "grad_norm": 0.153762549161911, "learning_rate": 6.12027954365748e-05, "loss": 0.0301, "step": 27400 }, { "epoch": 12.26672633698814, "grad_norm": 0.13698507845401764, "learning_rate": 6.117593652830191e-05, "loss": 0.0338, "step": 27410 }, { "epoch": 12.27120161109868, "grad_norm": 0.13692879676818848, "learning_rate": 6.11490742250746e-05, "loss": 0.0328, "step": 27420 }, { "epoch": 12.275676885209219, "grad_norm": 0.12920254468917847, "learning_rate": 6.112220853505288e-05, "loss": 0.0331, "step": 27430 }, { "epoch": 12.280152159319758, "grad_norm": 0.11602600663900375, "learning_rate": 6.10953394663979e-05, "loss": 0.0331, "step": 27440 }, { "epoch": 12.284627433430298, "grad_norm": 0.14413368701934814, "learning_rate": 6.106846702727172e-05, "loss": 0.0375, "step": 27450 }, { "epoch": 12.289102707540836, "grad_norm": 0.1504700481891632, "learning_rate": 6.104159122583752e-05, "loss": 0.0324, "step": 27460 }, { "epoch": 12.293577981651376, "grad_norm": 0.12728677690029144, "learning_rate": 6.101471207025945e-05, "loss": 0.0307, "step": 27470 }, { "epoch": 12.298053255761916, "grad_norm": 0.14551055431365967, "learning_rate": 6.0987829568702656e-05, "loss": 0.0354, "step": 27480 }, { "epoch": 12.302528529872454, "grad_norm": 0.13354197144508362, "learning_rate": 6.0960943729333374e-05, "loss": 0.0316, "step": 27490 }, { "epoch": 12.307003803982994, "grad_norm": 0.1125042587518692, "learning_rate": 6.09340545603188e-05, "loss": 0.034, "step": 27500 }, { "epoch": 12.311479078093534, "grad_norm": 0.14106528460979462, "learning_rate": 6.090716206982714e-05, "loss": 0.0337, "step": 27510 }, { "epoch": 12.315954352204072, "grad_norm": 0.16535574197769165, "learning_rate": 6.088026626602763e-05, "loss": 0.0342, "step": 27520 }, { "epoch": 12.320429626314612, "grad_norm": 0.14433090388774872, "learning_rate": 6.085336715709049e-05, "loss": 0.0312, "step": 27530 }, { "epoch": 12.324904900425151, "grad_norm": 0.1405676156282425, "learning_rate": 6.0826464751186994e-05, "loss": 0.0341, "step": 27540 }, { "epoch": 12.32938017453569, "grad_norm": 0.14199012517929077, "learning_rate": 6.079955905648934e-05, "loss": 0.0338, "step": 27550 }, { "epoch": 12.33385544864623, "grad_norm": 0.17010881006717682, "learning_rate": 6.077265008117081e-05, "loss": 0.0348, "step": 27560 }, { "epoch": 12.33833072275677, "grad_norm": 0.1454591006040573, "learning_rate": 6.074573783340562e-05, "loss": 0.032, "step": 27570 }, { "epoch": 12.342805996867309, "grad_norm": 0.10591168701648712, "learning_rate": 6.071882232136901e-05, "loss": 0.0323, "step": 27580 }, { "epoch": 12.347281270977847, "grad_norm": 0.14633867144584656, "learning_rate": 6.069190355323717e-05, "loss": 0.0358, "step": 27590 }, { "epoch": 12.351756545088387, "grad_norm": 0.14646165072917938, "learning_rate": 6.066498153718735e-05, "loss": 0.0361, "step": 27600 }, { "epoch": 12.356231819198927, "grad_norm": 0.12903325259685516, "learning_rate": 6.0638056281397726e-05, "loss": 0.0307, "step": 27610 }, { "epoch": 12.360707093309465, "grad_norm": 0.11483772099018097, "learning_rate": 6.0611127794047486e-05, "loss": 0.0306, "step": 27620 }, { "epoch": 12.365182367420005, "grad_norm": 0.1336284577846527, "learning_rate": 6.0584196083316794e-05, "loss": 0.0343, "step": 27630 }, { "epoch": 12.369657641530544, "grad_norm": 0.12865601480007172, "learning_rate": 6.055726115738678e-05, "loss": 0.0334, "step": 27640 }, { "epoch": 12.374132915641082, "grad_norm": 0.13068966567516327, "learning_rate": 6.053032302443959e-05, "loss": 0.0359, "step": 27650 }, { "epoch": 12.378608189751622, "grad_norm": 0.11288808286190033, "learning_rate": 6.05033816926583e-05, "loss": 0.0354, "step": 27660 }, { "epoch": 12.383083463862162, "grad_norm": 0.16374117136001587, "learning_rate": 6.0476437170226975e-05, "loss": 0.0332, "step": 27670 }, { "epoch": 12.3875587379727, "grad_norm": 0.15562547743320465, "learning_rate": 6.044948946533064e-05, "loss": 0.0315, "step": 27680 }, { "epoch": 12.39203401208324, "grad_norm": 0.16301068663597107, "learning_rate": 6.042253858615532e-05, "loss": 0.035, "step": 27690 }, { "epoch": 12.39650928619378, "grad_norm": 0.13873720169067383, "learning_rate": 6.0395584540887963e-05, "loss": 0.0366, "step": 27700 }, { "epoch": 12.400984560304318, "grad_norm": 0.15428349375724792, "learning_rate": 6.036862733771651e-05, "loss": 0.0349, "step": 27710 }, { "epoch": 12.405459834414858, "grad_norm": 0.15458786487579346, "learning_rate": 6.034166698482984e-05, "loss": 0.0302, "step": 27720 }, { "epoch": 12.409935108525397, "grad_norm": 0.1405366063117981, "learning_rate": 6.03147034904178e-05, "loss": 0.0327, "step": 27730 }, { "epoch": 12.414410382635936, "grad_norm": 0.14194057881832123, "learning_rate": 6.0287736862671175e-05, "loss": 0.0318, "step": 27740 }, { "epoch": 12.418885656746475, "grad_norm": 0.14416222274303436, "learning_rate": 6.026076710978171e-05, "loss": 0.0352, "step": 27750 }, { "epoch": 12.423360930857015, "grad_norm": 0.13408862054347992, "learning_rate": 6.023379423994214e-05, "loss": 0.0323, "step": 27760 }, { "epoch": 12.427836204967555, "grad_norm": 0.15217842161655426, "learning_rate": 6.020681826134609e-05, "loss": 0.0334, "step": 27770 }, { "epoch": 12.432311479078093, "grad_norm": 0.13179028034210205, "learning_rate": 6.017983918218812e-05, "loss": 0.031, "step": 27780 }, { "epoch": 12.436786753188633, "grad_norm": 0.1545323133468628, "learning_rate": 6.015285701066382e-05, "loss": 0.0336, "step": 27790 }, { "epoch": 12.441262027299173, "grad_norm": 0.1366642862558365, "learning_rate": 6.012587175496961e-05, "loss": 0.0343, "step": 27800 }, { "epoch": 12.44573730140971, "grad_norm": 0.14490632712841034, "learning_rate": 6.009888342330292e-05, "loss": 0.0386, "step": 27810 }, { "epoch": 12.45021257552025, "grad_norm": 0.11663785576820374, "learning_rate": 6.0071892023862105e-05, "loss": 0.032, "step": 27820 }, { "epoch": 12.45468784963079, "grad_norm": 0.12878049910068512, "learning_rate": 6.004489756484641e-05, "loss": 0.0317, "step": 27830 }, { "epoch": 12.459163123741329, "grad_norm": 0.1289111226797104, "learning_rate": 6.001790005445607e-05, "loss": 0.0351, "step": 27840 }, { "epoch": 12.463638397851868, "grad_norm": 0.1495627462863922, "learning_rate": 5.999089950089218e-05, "loss": 0.0312, "step": 27850 }, { "epoch": 12.468113671962408, "grad_norm": 0.14087776839733124, "learning_rate": 5.9963895912356836e-05, "loss": 0.0319, "step": 27860 }, { "epoch": 12.472588946072946, "grad_norm": 0.13821978867053986, "learning_rate": 5.9936889297052986e-05, "loss": 0.0347, "step": 27870 }, { "epoch": 12.477064220183486, "grad_norm": 0.13910362124443054, "learning_rate": 5.9909879663184544e-05, "loss": 0.0341, "step": 27880 }, { "epoch": 12.481539494294026, "grad_norm": 0.14160852134227753, "learning_rate": 5.988286701895631e-05, "loss": 0.0326, "step": 27890 }, { "epoch": 12.486014768404564, "grad_norm": 0.16832590103149414, "learning_rate": 5.985585137257401e-05, "loss": 0.0324, "step": 27900 }, { "epoch": 12.490490042515104, "grad_norm": 0.12562766671180725, "learning_rate": 5.982883273224431e-05, "loss": 0.03, "step": 27910 }, { "epoch": 12.494965316625644, "grad_norm": 0.1414230614900589, "learning_rate": 5.980181110617473e-05, "loss": 0.0334, "step": 27920 }, { "epoch": 12.499440590736183, "grad_norm": 0.12992385029792786, "learning_rate": 5.977478650257374e-05, "loss": 0.0331, "step": 27930 }, { "epoch": 12.503915864846721, "grad_norm": 0.12346511334180832, "learning_rate": 5.974775892965071e-05, "loss": 0.0322, "step": 27940 }, { "epoch": 12.508391138957261, "grad_norm": 0.13401558995246887, "learning_rate": 5.9720728395615875e-05, "loss": 0.0317, "step": 27950 }, { "epoch": 12.512866413067801, "grad_norm": 0.13001649081707, "learning_rate": 5.969369490868042e-05, "loss": 0.0344, "step": 27960 }, { "epoch": 12.51734168717834, "grad_norm": 0.150409534573555, "learning_rate": 5.966665847705639e-05, "loss": 0.0352, "step": 27970 }, { "epoch": 12.521816961288879, "grad_norm": 0.15381775796413422, "learning_rate": 5.963961910895676e-05, "loss": 0.0335, "step": 27980 }, { "epoch": 12.526292235399419, "grad_norm": 0.1511426717042923, "learning_rate": 5.961257681259535e-05, "loss": 0.0304, "step": 27990 }, { "epoch": 12.530767509509957, "grad_norm": 0.13259081542491913, "learning_rate": 5.958553159618693e-05, "loss": 0.0326, "step": 28000 }, { "epoch": 12.535242783620497, "grad_norm": 0.12402434647083282, "learning_rate": 5.955848346794708e-05, "loss": 0.0345, "step": 28010 }, { "epoch": 12.539718057731037, "grad_norm": 0.16819195449352264, "learning_rate": 5.953143243609235e-05, "loss": 0.0368, "step": 28020 }, { "epoch": 12.544193331841575, "grad_norm": 0.12120665609836578, "learning_rate": 5.95043785088401e-05, "loss": 0.0334, "step": 28030 }, { "epoch": 12.548668605952114, "grad_norm": 0.12400295585393906, "learning_rate": 5.9477321694408606e-05, "loss": 0.0331, "step": 28040 }, { "epoch": 12.553143880062654, "grad_norm": 0.15238764882087708, "learning_rate": 5.945026200101702e-05, "loss": 0.0351, "step": 28050 }, { "epoch": 12.557619154173192, "grad_norm": 0.15080447494983673, "learning_rate": 5.9423199436885345e-05, "loss": 0.0332, "step": 28060 }, { "epoch": 12.562094428283732, "grad_norm": 0.15582329034805298, "learning_rate": 5.939613401023451e-05, "loss": 0.0367, "step": 28070 }, { "epoch": 12.566569702394272, "grad_norm": 0.1537589728832245, "learning_rate": 5.9369065729286245e-05, "loss": 0.035, "step": 28080 }, { "epoch": 12.571044976504812, "grad_norm": 0.13240863382816315, "learning_rate": 5.934199460226317e-05, "loss": 0.0351, "step": 28090 }, { "epoch": 12.57552025061535, "grad_norm": 0.15393105149269104, "learning_rate": 5.9314920637388815e-05, "loss": 0.0322, "step": 28100 }, { "epoch": 12.57999552472589, "grad_norm": 0.17344340682029724, "learning_rate": 5.92878438428875e-05, "loss": 0.0369, "step": 28110 }, { "epoch": 12.58447079883643, "grad_norm": 0.1414402723312378, "learning_rate": 5.9260764226984476e-05, "loss": 0.0322, "step": 28120 }, { "epoch": 12.588946072946968, "grad_norm": 0.13963384926319122, "learning_rate": 5.9233681797905785e-05, "loss": 0.0342, "step": 28130 }, { "epoch": 12.593421347057507, "grad_norm": 0.15500792860984802, "learning_rate": 5.9206596563878357e-05, "loss": 0.0343, "step": 28140 }, { "epoch": 12.597896621168047, "grad_norm": 0.14789707958698273, "learning_rate": 5.9179508533130004e-05, "loss": 0.0323, "step": 28150 }, { "epoch": 12.602371895278585, "grad_norm": 0.14480319619178772, "learning_rate": 5.915241771388931e-05, "loss": 0.0383, "step": 28160 }, { "epoch": 12.606847169389125, "grad_norm": 0.13053004443645477, "learning_rate": 5.912532411438576e-05, "loss": 0.0325, "step": 28170 }, { "epoch": 12.611322443499665, "grad_norm": 0.1333998739719391, "learning_rate": 5.909822774284971e-05, "loss": 0.0384, "step": 28180 }, { "epoch": 12.615797717610203, "grad_norm": 0.14067727327346802, "learning_rate": 5.9071128607512285e-05, "loss": 0.0356, "step": 28190 }, { "epoch": 12.620272991720743, "grad_norm": 0.11945313215255737, "learning_rate": 5.90440267166055e-05, "loss": 0.036, "step": 28200 }, { "epoch": 12.624748265831283, "grad_norm": 0.11646659672260284, "learning_rate": 5.901692207836219e-05, "loss": 0.033, "step": 28210 }, { "epoch": 12.62922353994182, "grad_norm": 0.11471141874790192, "learning_rate": 5.8989814701016035e-05, "loss": 0.0352, "step": 28220 }, { "epoch": 12.63369881405236, "grad_norm": 0.16150972247123718, "learning_rate": 5.896270459280153e-05, "loss": 0.0318, "step": 28230 }, { "epoch": 12.6381740881629, "grad_norm": 0.12370206415653229, "learning_rate": 5.8935591761954025e-05, "loss": 0.0314, "step": 28240 }, { "epoch": 12.642649362273438, "grad_norm": 0.12148093432188034, "learning_rate": 5.890847621670966e-05, "loss": 0.032, "step": 28250 }, { "epoch": 12.647124636383978, "grad_norm": 0.16930708289146423, "learning_rate": 5.888135796530544e-05, "loss": 0.0348, "step": 28260 }, { "epoch": 12.651599910494518, "grad_norm": 0.15154284238815308, "learning_rate": 5.885423701597917e-05, "loss": 0.0343, "step": 28270 }, { "epoch": 12.656075184605058, "grad_norm": 0.1194644346833229, "learning_rate": 5.8827113376969465e-05, "loss": 0.0328, "step": 28280 }, { "epoch": 12.660550458715596, "grad_norm": 0.15143336355686188, "learning_rate": 5.8799987056515804e-05, "loss": 0.0317, "step": 28290 }, { "epoch": 12.665025732826136, "grad_norm": 0.13754035532474518, "learning_rate": 5.8772858062858416e-05, "loss": 0.0328, "step": 28300 }, { "epoch": 12.669501006936676, "grad_norm": 0.1415853202342987, "learning_rate": 5.874572640423839e-05, "loss": 0.0322, "step": 28310 }, { "epoch": 12.673976281047214, "grad_norm": 0.13312219083309174, "learning_rate": 5.871859208889759e-05, "loss": 0.0312, "step": 28320 }, { "epoch": 12.678451555157753, "grad_norm": 0.15009424090385437, "learning_rate": 5.869145512507872e-05, "loss": 0.0319, "step": 28330 }, { "epoch": 12.682926829268293, "grad_norm": 0.13367749750614166, "learning_rate": 5.866431552102526e-05, "loss": 0.0348, "step": 28340 }, { "epoch": 12.687402103378831, "grad_norm": 0.12104342132806778, "learning_rate": 5.8637173284981526e-05, "loss": 0.03, "step": 28350 }, { "epoch": 12.691877377489371, "grad_norm": 0.13067728281021118, "learning_rate": 5.861002842519259e-05, "loss": 0.0347, "step": 28360 }, { "epoch": 12.696352651599911, "grad_norm": 0.15220020711421967, "learning_rate": 5.858288094990436e-05, "loss": 0.0328, "step": 28370 }, { "epoch": 12.700827925710449, "grad_norm": 0.11737646907567978, "learning_rate": 5.85557308673635e-05, "loss": 0.033, "step": 28380 }, { "epoch": 12.705303199820989, "grad_norm": 0.11890550702810287, "learning_rate": 5.8528578185817514e-05, "loss": 0.032, "step": 28390 }, { "epoch": 12.709778473931529, "grad_norm": 0.12148312479257584, "learning_rate": 5.850142291351466e-05, "loss": 0.032, "step": 28400 }, { "epoch": 12.714253748042067, "grad_norm": 0.10231666266918182, "learning_rate": 5.847426505870399e-05, "loss": 0.0337, "step": 28410 }, { "epoch": 12.718729022152607, "grad_norm": 0.12330391258001328, "learning_rate": 5.8447104629635344e-05, "loss": 0.0358, "step": 28420 }, { "epoch": 12.723204296263146, "grad_norm": 0.14808860421180725, "learning_rate": 5.841994163455934e-05, "loss": 0.034, "step": 28430 }, { "epoch": 12.727679570373684, "grad_norm": 0.1336820423603058, "learning_rate": 5.8392776081727385e-05, "loss": 0.0347, "step": 28440 }, { "epoch": 12.732154844484224, "grad_norm": 0.1326117068529129, "learning_rate": 5.8365607979391666e-05, "loss": 0.0319, "step": 28450 }, { "epoch": 12.736630118594764, "grad_norm": 0.1585078239440918, "learning_rate": 5.833843733580512e-05, "loss": 0.0313, "step": 28460 }, { "epoch": 12.741105392705304, "grad_norm": 0.14396141469478607, "learning_rate": 5.831126415922148e-05, "loss": 0.033, "step": 28470 }, { "epoch": 12.745580666815842, "grad_norm": 0.13617925345897675, "learning_rate": 5.828408845789522e-05, "loss": 0.0337, "step": 28480 }, { "epoch": 12.750055940926382, "grad_norm": 0.14393801987171173, "learning_rate": 5.825691024008162e-05, "loss": 0.0351, "step": 28490 }, { "epoch": 12.754531215036922, "grad_norm": 0.13088369369506836, "learning_rate": 5.8229729514036705e-05, "loss": 0.0317, "step": 28500 }, { "epoch": 12.75900648914746, "grad_norm": 0.1443496197462082, "learning_rate": 5.820254628801726e-05, "loss": 0.0331, "step": 28510 }, { "epoch": 12.763481763258, "grad_norm": 0.11706359684467316, "learning_rate": 5.817536057028081e-05, "loss": 0.0347, "step": 28520 }, { "epoch": 12.76795703736854, "grad_norm": 0.14746452867984772, "learning_rate": 5.8148172369085686e-05, "loss": 0.0301, "step": 28530 }, { "epoch": 12.772432311479077, "grad_norm": 0.1445552259683609, "learning_rate": 5.812098169269094e-05, "loss": 0.0313, "step": 28540 }, { "epoch": 12.776907585589617, "grad_norm": 0.1561867594718933, "learning_rate": 5.809378854935639e-05, "loss": 0.033, "step": 28550 }, { "epoch": 12.781382859700157, "grad_norm": 0.13516157865524292, "learning_rate": 5.8066592947342555e-05, "loss": 0.0329, "step": 28560 }, { "epoch": 12.785858133810695, "grad_norm": 0.13041874766349792, "learning_rate": 5.8039394894910796e-05, "loss": 0.032, "step": 28570 }, { "epoch": 12.790333407921235, "grad_norm": 0.1367386430501938, "learning_rate": 5.8012194400323116e-05, "loss": 0.0342, "step": 28580 }, { "epoch": 12.794808682031775, "grad_norm": 0.14689943194389343, "learning_rate": 5.798499147184233e-05, "loss": 0.0325, "step": 28590 }, { "epoch": 12.799283956142315, "grad_norm": 0.11955304443836212, "learning_rate": 5.795778611773197e-05, "loss": 0.0345, "step": 28600 }, { "epoch": 12.803759230252853, "grad_norm": 0.12809526920318604, "learning_rate": 5.7930578346256305e-05, "loss": 0.0339, "step": 28610 }, { "epoch": 12.808234504363393, "grad_norm": 0.13213978707790375, "learning_rate": 5.7903368165680327e-05, "loss": 0.0322, "step": 28620 }, { "epoch": 12.812709778473932, "grad_norm": 0.1585305631160736, "learning_rate": 5.7876155584269785e-05, "loss": 0.0347, "step": 28630 }, { "epoch": 12.81718505258447, "grad_norm": 0.16318641602993011, "learning_rate": 5.7848940610291125e-05, "loss": 0.0367, "step": 28640 }, { "epoch": 12.82166032669501, "grad_norm": 0.14435282349586487, "learning_rate": 5.782172325201155e-05, "loss": 0.0298, "step": 28650 }, { "epoch": 12.82613560080555, "grad_norm": 0.1766786277294159, "learning_rate": 5.7794503517698974e-05, "loss": 0.0344, "step": 28660 }, { "epoch": 12.830610874916088, "grad_norm": 0.134751558303833, "learning_rate": 5.776728141562203e-05, "loss": 0.0327, "step": 28670 }, { "epoch": 12.835086149026628, "grad_norm": 0.1494368016719818, "learning_rate": 5.7740056954050084e-05, "loss": 0.0401, "step": 28680 }, { "epoch": 12.839561423137168, "grad_norm": 0.15930302441120148, "learning_rate": 5.771283014125317e-05, "loss": 0.0359, "step": 28690 }, { "epoch": 12.844036697247706, "grad_norm": 0.11956675350666046, "learning_rate": 5.768560098550213e-05, "loss": 0.0331, "step": 28700 }, { "epoch": 12.848511971358246, "grad_norm": 0.14312247931957245, "learning_rate": 5.765836949506843e-05, "loss": 0.0353, "step": 28710 }, { "epoch": 12.852987245468785, "grad_norm": 0.13837289810180664, "learning_rate": 5.763113567822429e-05, "loss": 0.0324, "step": 28720 }, { "epoch": 12.857462519579324, "grad_norm": 0.1333795189857483, "learning_rate": 5.760389954324261e-05, "loss": 0.033, "step": 28730 }, { "epoch": 12.861937793689863, "grad_norm": 0.14759472012519836, "learning_rate": 5.757666109839702e-05, "loss": 0.0349, "step": 28740 }, { "epoch": 12.866413067800403, "grad_norm": 0.13066726922988892, "learning_rate": 5.7549420351961844e-05, "loss": 0.0293, "step": 28750 }, { "epoch": 12.870888341910941, "grad_norm": 0.14807258546352386, "learning_rate": 5.75221773122121e-05, "loss": 0.0335, "step": 28760 }, { "epoch": 12.875363616021481, "grad_norm": 0.14455674588680267, "learning_rate": 5.74949319874235e-05, "loss": 0.0326, "step": 28770 }, { "epoch": 12.87983889013202, "grad_norm": 0.12674450874328613, "learning_rate": 5.746768438587245e-05, "loss": 0.0343, "step": 28780 }, { "epoch": 12.88431416424256, "grad_norm": 0.1456819474697113, "learning_rate": 5.7440434515836064e-05, "loss": 0.0305, "step": 28790 }, { "epoch": 12.888789438353099, "grad_norm": 0.16363844275474548, "learning_rate": 5.74131823855921e-05, "loss": 0.0305, "step": 28800 }, { "epoch": 12.893264712463639, "grad_norm": 0.14132054150104523, "learning_rate": 5.7385928003419085e-05, "loss": 0.0323, "step": 28810 }, { "epoch": 12.897739986574178, "grad_norm": 0.1205425038933754, "learning_rate": 5.735867137759615e-05, "loss": 0.0319, "step": 28820 }, { "epoch": 12.902215260684716, "grad_norm": 0.11455909162759781, "learning_rate": 5.733141251640315e-05, "loss": 0.031, "step": 28830 }, { "epoch": 12.906690534795256, "grad_norm": 0.12965752184391022, "learning_rate": 5.730415142812059e-05, "loss": 0.0327, "step": 28840 }, { "epoch": 12.911165808905796, "grad_norm": 0.14743860065937042, "learning_rate": 5.727688812102967e-05, "loss": 0.0352, "step": 28850 }, { "epoch": 12.915641083016334, "grad_norm": 0.12545014917850494, "learning_rate": 5.72496226034123e-05, "loss": 0.0311, "step": 28860 }, { "epoch": 12.920116357126874, "grad_norm": 0.13684684038162231, "learning_rate": 5.722235488355099e-05, "loss": 0.0321, "step": 28870 }, { "epoch": 12.924591631237414, "grad_norm": 0.13515116274356842, "learning_rate": 5.719508496972896e-05, "loss": 0.0325, "step": 28880 }, { "epoch": 12.929066905347952, "grad_norm": 0.1402328759431839, "learning_rate": 5.7167812870230094e-05, "loss": 0.0304, "step": 28890 }, { "epoch": 12.933542179458492, "grad_norm": 0.14326298236846924, "learning_rate": 5.714053859333893e-05, "loss": 0.0352, "step": 28900 }, { "epoch": 12.938017453569032, "grad_norm": 0.1253584772348404, "learning_rate": 5.711326214734069e-05, "loss": 0.0283, "step": 28910 }, { "epoch": 12.94249272767957, "grad_norm": 0.12615130841732025, "learning_rate": 5.7085983540521216e-05, "loss": 0.0317, "step": 28920 }, { "epoch": 12.94696800179011, "grad_norm": 0.1294141709804535, "learning_rate": 5.705870278116703e-05, "loss": 0.0334, "step": 28930 }, { "epoch": 12.95144327590065, "grad_norm": 0.11255909502506256, "learning_rate": 5.7031419877565317e-05, "loss": 0.0272, "step": 28940 }, { "epoch": 12.955918550011187, "grad_norm": 0.12663571536540985, "learning_rate": 5.70041348380039e-05, "loss": 0.0375, "step": 28950 }, { "epoch": 12.960393824121727, "grad_norm": 0.13780725002288818, "learning_rate": 5.697684767077125e-05, "loss": 0.032, "step": 28960 }, { "epoch": 12.964869098232267, "grad_norm": 0.10933590680360794, "learning_rate": 5.69495583841565e-05, "loss": 0.032, "step": 28970 }, { "epoch": 12.969344372342807, "grad_norm": 0.17601555585861206, "learning_rate": 5.692226698644938e-05, "loss": 0.0315, "step": 28980 }, { "epoch": 12.973819646453345, "grad_norm": 0.18741671741008759, "learning_rate": 5.689497348594035e-05, "loss": 0.0324, "step": 28990 }, { "epoch": 12.978294920563885, "grad_norm": 0.15901517868041992, "learning_rate": 5.686767789092041e-05, "loss": 0.0327, "step": 29000 }, { "epoch": 12.982770194674425, "grad_norm": 0.14170877635478973, "learning_rate": 5.6840380209681255e-05, "loss": 0.0301, "step": 29010 }, { "epoch": 12.987245468784963, "grad_norm": 0.16915041208267212, "learning_rate": 5.681308045051522e-05, "loss": 0.0325, "step": 29020 }, { "epoch": 12.991720742895502, "grad_norm": 0.1580161303281784, "learning_rate": 5.6785778621715225e-05, "loss": 0.0344, "step": 29030 }, { "epoch": 12.996196017006042, "grad_norm": 0.11788511276245117, "learning_rate": 5.675847473157485e-05, "loss": 0.033, "step": 29040 }, { "epoch": 13.00067129111658, "grad_norm": 0.1678008735179901, "learning_rate": 5.67311687883883e-05, "loss": 0.0345, "step": 29050 }, { "epoch": 13.00514656522712, "grad_norm": 0.150923490524292, "learning_rate": 5.670386080045039e-05, "loss": 0.032, "step": 29060 }, { "epoch": 13.00962183933766, "grad_norm": 0.14624273777008057, "learning_rate": 5.667655077605659e-05, "loss": 0.0329, "step": 29070 }, { "epoch": 13.014097113448198, "grad_norm": 0.09547947347164154, "learning_rate": 5.664923872350294e-05, "loss": 0.0335, "step": 29080 }, { "epoch": 13.018572387558738, "grad_norm": 0.14130528271198273, "learning_rate": 5.662192465108613e-05, "loss": 0.0306, "step": 29090 }, { "epoch": 13.023047661669278, "grad_norm": 0.1384781301021576, "learning_rate": 5.6594608567103456e-05, "loss": 0.0355, "step": 29100 }, { "epoch": 13.027522935779816, "grad_norm": 0.1324111521244049, "learning_rate": 5.656729047985281e-05, "loss": 0.0349, "step": 29110 }, { "epoch": 13.031998209890356, "grad_norm": 0.11588434875011444, "learning_rate": 5.653997039763273e-05, "loss": 0.0308, "step": 29120 }, { "epoch": 13.036473484000895, "grad_norm": 0.11578142642974854, "learning_rate": 5.65126483287423e-05, "loss": 0.0317, "step": 29130 }, { "epoch": 13.040948758111435, "grad_norm": 0.12072783708572388, "learning_rate": 5.648532428148128e-05, "loss": 0.0256, "step": 29140 }, { "epoch": 13.045424032221973, "grad_norm": 0.11105549335479736, "learning_rate": 5.645799826414997e-05, "loss": 0.0336, "step": 29150 }, { "epoch": 13.049899306332513, "grad_norm": 0.13619087636470795, "learning_rate": 5.6430670285049314e-05, "loss": 0.0334, "step": 29160 }, { "epoch": 13.054374580443053, "grad_norm": 0.13612055778503418, "learning_rate": 5.6403340352480806e-05, "loss": 0.0332, "step": 29170 }, { "epoch": 13.058849854553591, "grad_norm": 0.14264491200447083, "learning_rate": 5.637600847474656e-05, "loss": 0.0337, "step": 29180 }, { "epoch": 13.06332512866413, "grad_norm": 0.12419291585683823, "learning_rate": 5.634867466014932e-05, "loss": 0.0323, "step": 29190 }, { "epoch": 13.06780040277467, "grad_norm": 0.14296609163284302, "learning_rate": 5.6321338916992315e-05, "loss": 0.0311, "step": 29200 }, { "epoch": 13.072275676885209, "grad_norm": 0.13931237161159515, "learning_rate": 5.6294001253579475e-05, "loss": 0.0298, "step": 29210 }, { "epoch": 13.076750950995748, "grad_norm": 0.1508978009223938, "learning_rate": 5.6266661678215216e-05, "loss": 0.0337, "step": 29220 }, { "epoch": 13.081226225106288, "grad_norm": 0.11879527568817139, "learning_rate": 5.6239320199204616e-05, "loss": 0.0342, "step": 29230 }, { "epoch": 13.085701499216826, "grad_norm": 0.12876127660274506, "learning_rate": 5.621197682485327e-05, "loss": 0.0314, "step": 29240 }, { "epoch": 13.090176773327366, "grad_norm": 0.14699330925941467, "learning_rate": 5.618463156346739e-05, "loss": 0.0325, "step": 29250 }, { "epoch": 13.094652047437906, "grad_norm": 0.09953813999891281, "learning_rate": 5.615728442335373e-05, "loss": 0.0309, "step": 29260 }, { "epoch": 13.099127321548444, "grad_norm": 0.12592361867427826, "learning_rate": 5.6129935412819635e-05, "loss": 0.0343, "step": 29270 }, { "epoch": 13.103602595658984, "grad_norm": 0.1432393342256546, "learning_rate": 5.6102584540173006e-05, "loss": 0.0345, "step": 29280 }, { "epoch": 13.108077869769524, "grad_norm": 0.12392891198396683, "learning_rate": 5.607523181372234e-05, "loss": 0.0339, "step": 29290 }, { "epoch": 13.112553143880064, "grad_norm": 0.12356960773468018, "learning_rate": 5.604787724177666e-05, "loss": 0.0303, "step": 29300 }, { "epoch": 13.117028417990602, "grad_norm": 0.13504093885421753, "learning_rate": 5.602052083264555e-05, "loss": 0.0319, "step": 29310 }, { "epoch": 13.121503692101141, "grad_norm": 0.11014268547296524, "learning_rate": 5.599316259463916e-05, "loss": 0.0299, "step": 29320 }, { "epoch": 13.125978966211681, "grad_norm": 0.17760136723518372, "learning_rate": 5.596580253606824e-05, "loss": 0.0314, "step": 29330 }, { "epoch": 13.13045424032222, "grad_norm": 0.15460306406021118, "learning_rate": 5.5938440665244006e-05, "loss": 0.0383, "step": 29340 }, { "epoch": 13.13492951443276, "grad_norm": 0.13816586136817932, "learning_rate": 5.5911076990478304e-05, "loss": 0.0319, "step": 29350 }, { "epoch": 13.139404788543299, "grad_norm": 0.14552748203277588, "learning_rate": 5.588371152008349e-05, "loss": 0.0336, "step": 29360 }, { "epoch": 13.143880062653837, "grad_norm": 0.11390821635723114, "learning_rate": 5.585634426237246e-05, "loss": 0.0301, "step": 29370 }, { "epoch": 13.148355336764377, "grad_norm": 0.14427359402179718, "learning_rate": 5.5828975225658666e-05, "loss": 0.0333, "step": 29380 }, { "epoch": 13.152830610874917, "grad_norm": 0.1238352581858635, "learning_rate": 5.5801604418256117e-05, "loss": 0.0334, "step": 29390 }, { "epoch": 13.157305884985455, "grad_norm": 0.15447455644607544, "learning_rate": 5.577423184847932e-05, "loss": 0.0347, "step": 29400 }, { "epoch": 13.161781159095995, "grad_norm": 0.1518227756023407, "learning_rate": 5.574685752464334e-05, "loss": 0.0339, "step": 29410 }, { "epoch": 13.166256433206534, "grad_norm": 0.11140147596597672, "learning_rate": 5.5719481455063784e-05, "loss": 0.0327, "step": 29420 }, { "epoch": 13.170731707317072, "grad_norm": 0.12223586440086365, "learning_rate": 5.569210364805677e-05, "loss": 0.0302, "step": 29430 }, { "epoch": 13.175206981427612, "grad_norm": 0.13598231971263885, "learning_rate": 5.566472411193897e-05, "loss": 0.0295, "step": 29440 }, { "epoch": 13.179682255538152, "grad_norm": 0.13273850083351135, "learning_rate": 5.5637342855027554e-05, "loss": 0.0351, "step": 29450 }, { "epoch": 13.18415752964869, "grad_norm": 0.13307665288448334, "learning_rate": 5.560995988564023e-05, "loss": 0.0322, "step": 29460 }, { "epoch": 13.18863280375923, "grad_norm": 0.13147813081741333, "learning_rate": 5.558257521209521e-05, "loss": 0.036, "step": 29470 }, { "epoch": 13.19310807786977, "grad_norm": 0.13874870538711548, "learning_rate": 5.555518884271122e-05, "loss": 0.0328, "step": 29480 }, { "epoch": 13.19758335198031, "grad_norm": 0.12127004563808441, "learning_rate": 5.552780078580756e-05, "loss": 0.0308, "step": 29490 }, { "epoch": 13.202058626090848, "grad_norm": 0.11536136269569397, "learning_rate": 5.550041104970397e-05, "loss": 0.0287, "step": 29500 }, { "epoch": 13.206533900201388, "grad_norm": 0.13774096965789795, "learning_rate": 5.5473019642720734e-05, "loss": 0.031, "step": 29510 }, { "epoch": 13.211009174311927, "grad_norm": 0.13604530692100525, "learning_rate": 5.544562657317863e-05, "loss": 0.0344, "step": 29520 }, { "epoch": 13.215484448422465, "grad_norm": 0.16842885315418243, "learning_rate": 5.541823184939896e-05, "loss": 0.0336, "step": 29530 }, { "epoch": 13.219959722533005, "grad_norm": 0.15972889959812164, "learning_rate": 5.539083547970352e-05, "loss": 0.0344, "step": 29540 }, { "epoch": 13.224434996643545, "grad_norm": 0.15928727388381958, "learning_rate": 5.5363437472414595e-05, "loss": 0.0289, "step": 29550 }, { "epoch": 13.228910270754083, "grad_norm": 0.12857243418693542, "learning_rate": 5.533603783585496e-05, "loss": 0.0286, "step": 29560 }, { "epoch": 13.233385544864623, "grad_norm": 0.1795995980501175, "learning_rate": 5.5308636578347936e-05, "loss": 0.0324, "step": 29570 }, { "epoch": 13.237860818975163, "grad_norm": 0.1298300325870514, "learning_rate": 5.52812337082173e-05, "loss": 0.0316, "step": 29580 }, { "epoch": 13.2423360930857, "grad_norm": 0.13869334757328033, "learning_rate": 5.525382923378728e-05, "loss": 0.0344, "step": 29590 }, { "epoch": 13.24681136719624, "grad_norm": 0.14266221225261688, "learning_rate": 5.522642316338268e-05, "loss": 0.0341, "step": 29600 }, { "epoch": 13.25128664130678, "grad_norm": 0.13473469018936157, "learning_rate": 5.519901550532871e-05, "loss": 0.0308, "step": 29610 }, { "epoch": 13.255761915417319, "grad_norm": 0.10128007084131241, "learning_rate": 5.51716062679511e-05, "loss": 0.0306, "step": 29620 }, { "epoch": 13.260237189527858, "grad_norm": 0.11718986928462982, "learning_rate": 5.514419545957606e-05, "loss": 0.031, "step": 29630 }, { "epoch": 13.264712463638398, "grad_norm": 0.11036083847284317, "learning_rate": 5.511678308853026e-05, "loss": 0.0314, "step": 29640 }, { "epoch": 13.269187737748936, "grad_norm": 0.12076466530561447, "learning_rate": 5.508936916314086e-05, "loss": 0.032, "step": 29650 }, { "epoch": 13.273663011859476, "grad_norm": 0.15841896831989288, "learning_rate": 5.506195369173548e-05, "loss": 0.0314, "step": 29660 }, { "epoch": 13.278138285970016, "grad_norm": 0.12585927546024323, "learning_rate": 5.5034536682642224e-05, "loss": 0.0322, "step": 29670 }, { "epoch": 13.282613560080556, "grad_norm": 0.15128961205482483, "learning_rate": 5.500711814418966e-05, "loss": 0.0323, "step": 29680 }, { "epoch": 13.287088834191094, "grad_norm": 0.1065284013748169, "learning_rate": 5.49796980847068e-05, "loss": 0.0297, "step": 29690 }, { "epoch": 13.291564108301634, "grad_norm": 0.1213453933596611, "learning_rate": 5.495227651252315e-05, "loss": 0.0303, "step": 29700 }, { "epoch": 13.296039382412173, "grad_norm": 0.1330939382314682, "learning_rate": 5.4924853435968635e-05, "loss": 0.0356, "step": 29710 }, { "epoch": 13.300514656522711, "grad_norm": 0.14479555189609528, "learning_rate": 5.48974288633737e-05, "loss": 0.0314, "step": 29720 }, { "epoch": 13.304989930633251, "grad_norm": 0.12235034257173538, "learning_rate": 5.487000280306917e-05, "loss": 0.0302, "step": 29730 }, { "epoch": 13.309465204743791, "grad_norm": 0.1268160194158554, "learning_rate": 5.4842575263386386e-05, "loss": 0.0343, "step": 29740 }, { "epoch": 13.31394047885433, "grad_norm": 0.15193456411361694, "learning_rate": 5.481514625265709e-05, "loss": 0.0346, "step": 29750 }, { "epoch": 13.318415752964869, "grad_norm": 0.1207757368683815, "learning_rate": 5.478771577921351e-05, "loss": 0.0328, "step": 29760 }, { "epoch": 13.322891027075409, "grad_norm": 0.13809150457382202, "learning_rate": 5.476028385138828e-05, "loss": 0.0344, "step": 29770 }, { "epoch": 13.327366301185947, "grad_norm": 0.12526008486747742, "learning_rate": 5.473285047751451e-05, "loss": 0.0346, "step": 29780 }, { "epoch": 13.331841575296487, "grad_norm": 0.1179049164056778, "learning_rate": 5.470541566592573e-05, "loss": 0.0327, "step": 29790 }, { "epoch": 13.336316849407027, "grad_norm": 0.15459372103214264, "learning_rate": 5.467797942495589e-05, "loss": 0.0329, "step": 29800 }, { "epoch": 13.340792123517565, "grad_norm": 0.1581123322248459, "learning_rate": 5.4650541762939435e-05, "loss": 0.033, "step": 29810 }, { "epoch": 13.345267397628104, "grad_norm": 0.1335754543542862, "learning_rate": 5.462310268821118e-05, "loss": 0.0314, "step": 29820 }, { "epoch": 13.349742671738644, "grad_norm": 0.13598516583442688, "learning_rate": 5.45956622091064e-05, "loss": 0.0311, "step": 29830 }, { "epoch": 13.354217945849184, "grad_norm": 0.10244589298963547, "learning_rate": 5.456822033396076e-05, "loss": 0.0301, "step": 29840 }, { "epoch": 13.358693219959722, "grad_norm": 0.13814866542816162, "learning_rate": 5.454077707111042e-05, "loss": 0.0296, "step": 29850 }, { "epoch": 13.363168494070262, "grad_norm": 0.11847434192895889, "learning_rate": 5.4513332428891887e-05, "loss": 0.0302, "step": 29860 }, { "epoch": 13.367643768180802, "grad_norm": 0.13137011229991913, "learning_rate": 5.448588641564213e-05, "loss": 0.0304, "step": 29870 }, { "epoch": 13.37211904229134, "grad_norm": 0.11431223899126053, "learning_rate": 5.445843903969854e-05, "loss": 0.0308, "step": 29880 }, { "epoch": 13.37659431640188, "grad_norm": 0.12492846697568893, "learning_rate": 5.443099030939887e-05, "loss": 0.0355, "step": 29890 }, { "epoch": 13.38106959051242, "grad_norm": 0.12130162119865417, "learning_rate": 5.440354023308134e-05, "loss": 0.0298, "step": 29900 }, { "epoch": 13.385544864622958, "grad_norm": 0.11931376904249191, "learning_rate": 5.4376088819084556e-05, "loss": 0.0322, "step": 29910 }, { "epoch": 13.390020138733497, "grad_norm": 0.1292925477027893, "learning_rate": 5.4348636075747536e-05, "loss": 0.0352, "step": 29920 }, { "epoch": 13.394495412844037, "grad_norm": 0.12486175447702408, "learning_rate": 5.43211820114097e-05, "loss": 0.0311, "step": 29930 }, { "epoch": 13.398970686954575, "grad_norm": 0.148141011595726, "learning_rate": 5.4293726634410855e-05, "loss": 0.0331, "step": 29940 }, { "epoch": 13.403445961065115, "grad_norm": 0.13366000354290009, "learning_rate": 5.426626995309123e-05, "loss": 0.0322, "step": 29950 }, { "epoch": 13.407921235175655, "grad_norm": 0.1507595032453537, "learning_rate": 5.423881197579144e-05, "loss": 0.0332, "step": 29960 }, { "epoch": 13.412396509286193, "grad_norm": 0.12081130594015121, "learning_rate": 5.4211352710852495e-05, "loss": 0.0305, "step": 29970 }, { "epoch": 13.416871783396733, "grad_norm": 0.14344638586044312, "learning_rate": 5.418389216661579e-05, "loss": 0.0317, "step": 29980 }, { "epoch": 13.421347057507273, "grad_norm": 0.13010284304618835, "learning_rate": 5.415643035142309e-05, "loss": 0.0321, "step": 29990 }, { "epoch": 13.425822331617812, "grad_norm": 0.1173761859536171, "learning_rate": 5.4128967273616625e-05, "loss": 0.0321, "step": 30000 }, { "epoch": 13.43029760572835, "grad_norm": 0.13104677200317383, "learning_rate": 5.4101502941538896e-05, "loss": 0.0322, "step": 30010 }, { "epoch": 13.43477287983889, "grad_norm": 0.13568729162216187, "learning_rate": 5.407403736353288e-05, "loss": 0.0315, "step": 30020 }, { "epoch": 13.43924815394943, "grad_norm": 0.12030728906393051, "learning_rate": 5.404657054794189e-05, "loss": 0.0279, "step": 30030 }, { "epoch": 13.443723428059968, "grad_norm": 0.13155007362365723, "learning_rate": 5.401910250310961e-05, "loss": 0.0354, "step": 30040 }, { "epoch": 13.448198702170508, "grad_norm": 0.12753742933273315, "learning_rate": 5.39916332373801e-05, "loss": 0.0293, "step": 30050 }, { "epoch": 13.452673976281048, "grad_norm": 0.12386578321456909, "learning_rate": 5.396416275909779e-05, "loss": 0.0316, "step": 30060 }, { "epoch": 13.457149250391586, "grad_norm": 0.13188521564006805, "learning_rate": 5.393669107660753e-05, "loss": 0.03, "step": 30070 }, { "epoch": 13.461624524502126, "grad_norm": 0.13726018369197845, "learning_rate": 5.390921819825445e-05, "loss": 0.029, "step": 30080 }, { "epoch": 13.466099798612666, "grad_norm": 0.13086214661598206, "learning_rate": 5.3881744132384104e-05, "loss": 0.0295, "step": 30090 }, { "epoch": 13.470575072723204, "grad_norm": 0.14233984053134918, "learning_rate": 5.3854268887342374e-05, "loss": 0.0291, "step": 30100 }, { "epoch": 13.475050346833743, "grad_norm": 0.12856899201869965, "learning_rate": 5.3826792471475516e-05, "loss": 0.0332, "step": 30110 }, { "epoch": 13.479525620944283, "grad_norm": 0.13138075172901154, "learning_rate": 5.379931489313016e-05, "loss": 0.0317, "step": 30120 }, { "epoch": 13.484000895054821, "grad_norm": 0.11429773271083832, "learning_rate": 5.3771836160653254e-05, "loss": 0.0347, "step": 30130 }, { "epoch": 13.488476169165361, "grad_norm": 0.13217096030712128, "learning_rate": 5.3744356282392104e-05, "loss": 0.0303, "step": 30140 }, { "epoch": 13.492951443275901, "grad_norm": 0.13640196621418, "learning_rate": 5.371687526669439e-05, "loss": 0.0297, "step": 30150 }, { "epoch": 13.497426717386439, "grad_norm": 0.12516124546527863, "learning_rate": 5.368939312190808e-05, "loss": 0.0296, "step": 30160 }, { "epoch": 13.501901991496979, "grad_norm": 0.12700645625591278, "learning_rate": 5.366190985638159e-05, "loss": 0.0335, "step": 30170 }, { "epoch": 13.506377265607519, "grad_norm": 0.12719671428203583, "learning_rate": 5.363442547846356e-05, "loss": 0.0338, "step": 30180 }, { "epoch": 13.510852539718059, "grad_norm": 0.12611007690429688, "learning_rate": 5.360693999650303e-05, "loss": 0.0343, "step": 30190 }, { "epoch": 13.515327813828597, "grad_norm": 0.12228722870349884, "learning_rate": 5.357945341884936e-05, "loss": 0.0306, "step": 30200 }, { "epoch": 13.519803087939136, "grad_norm": 0.12007497996091843, "learning_rate": 5.355196575385225e-05, "loss": 0.0319, "step": 30210 }, { "epoch": 13.524278362049676, "grad_norm": 0.12763945758342743, "learning_rate": 5.352447700986173e-05, "loss": 0.0362, "step": 30220 }, { "epoch": 13.528753636160214, "grad_norm": 0.13290199637413025, "learning_rate": 5.3496987195228156e-05, "loss": 0.0324, "step": 30230 }, { "epoch": 13.533228910270754, "grad_norm": 0.12047987431287766, "learning_rate": 5.3469496318302204e-05, "loss": 0.0305, "step": 30240 }, { "epoch": 13.537704184381294, "grad_norm": 0.1240907534956932, "learning_rate": 5.344200438743489e-05, "loss": 0.031, "step": 30250 }, { "epoch": 13.542179458491832, "grad_norm": 0.13734768331050873, "learning_rate": 5.341451141097751e-05, "loss": 0.0379, "step": 30260 }, { "epoch": 13.546654732602372, "grad_norm": 0.12878063321113586, "learning_rate": 5.3387017397281704e-05, "loss": 0.0304, "step": 30270 }, { "epoch": 13.551130006712912, "grad_norm": 0.14352194964885712, "learning_rate": 5.335952235469947e-05, "loss": 0.0294, "step": 30280 }, { "epoch": 13.55560528082345, "grad_norm": 0.14930449426174164, "learning_rate": 5.3332026291583016e-05, "loss": 0.0349, "step": 30290 }, { "epoch": 13.56008055493399, "grad_norm": 0.13285823166370392, "learning_rate": 5.330452921628497e-05, "loss": 0.0304, "step": 30300 }, { "epoch": 13.56455582904453, "grad_norm": 0.11428272724151611, "learning_rate": 5.3277031137158205e-05, "loss": 0.0334, "step": 30310 }, { "epoch": 13.569031103155067, "grad_norm": 0.14818106591701508, "learning_rate": 5.32495320625559e-05, "loss": 0.031, "step": 30320 }, { "epoch": 13.573506377265607, "grad_norm": 0.115068219602108, "learning_rate": 5.322203200083154e-05, "loss": 0.0313, "step": 30330 }, { "epoch": 13.577981651376147, "grad_norm": 0.12211815267801285, "learning_rate": 5.319453096033896e-05, "loss": 0.0317, "step": 30340 }, { "epoch": 13.582456925486685, "grad_norm": 0.12742716073989868, "learning_rate": 5.316702894943221e-05, "loss": 0.0291, "step": 30350 }, { "epoch": 13.586932199597225, "grad_norm": 0.11133644729852676, "learning_rate": 5.313952597646568e-05, "loss": 0.033, "step": 30360 }, { "epoch": 13.591407473707765, "grad_norm": 0.14326819777488708, "learning_rate": 5.311202204979404e-05, "loss": 0.0325, "step": 30370 }, { "epoch": 13.595882747818305, "grad_norm": 0.13947199285030365, "learning_rate": 5.308451717777228e-05, "loss": 0.0338, "step": 30380 }, { "epoch": 13.600358021928843, "grad_norm": 0.14546211063861847, "learning_rate": 5.305701136875566e-05, "loss": 0.0308, "step": 30390 }, { "epoch": 13.604833296039383, "grad_norm": 0.1351669877767563, "learning_rate": 5.3029504631099694e-05, "loss": 0.0354, "step": 30400 }, { "epoch": 13.609308570149922, "grad_norm": 0.11052409559488297, "learning_rate": 5.3001996973160206e-05, "loss": 0.0339, "step": 30410 }, { "epoch": 13.61378384426046, "grad_norm": 0.09739786386489868, "learning_rate": 5.297448840329329e-05, "loss": 0.032, "step": 30420 }, { "epoch": 13.618259118371, "grad_norm": 0.09074174612760544, "learning_rate": 5.294697892985534e-05, "loss": 0.033, "step": 30430 }, { "epoch": 13.62273439248154, "grad_norm": 0.11761877685785294, "learning_rate": 5.2919468561203e-05, "loss": 0.0306, "step": 30440 }, { "epoch": 13.627209666592078, "grad_norm": 0.12211631238460541, "learning_rate": 5.2891957305693205e-05, "loss": 0.0329, "step": 30450 }, { "epoch": 13.631684940702618, "grad_norm": 0.14701806008815765, "learning_rate": 5.286444517168313e-05, "loss": 0.0323, "step": 30460 }, { "epoch": 13.636160214813158, "grad_norm": 0.12269391864538193, "learning_rate": 5.283693216753025e-05, "loss": 0.0316, "step": 30470 }, { "epoch": 13.640635488923696, "grad_norm": 0.11716341227293015, "learning_rate": 5.280941830159227e-05, "loss": 0.0344, "step": 30480 }, { "epoch": 13.645110763034236, "grad_norm": 0.1466369479894638, "learning_rate": 5.278190358222721e-05, "loss": 0.0318, "step": 30490 }, { "epoch": 13.649586037144775, "grad_norm": 0.10867336392402649, "learning_rate": 5.2754388017793274e-05, "loss": 0.0283, "step": 30500 }, { "epoch": 13.654061311255315, "grad_norm": 0.1593461036682129, "learning_rate": 5.2726871616649e-05, "loss": 0.0349, "step": 30510 }, { "epoch": 13.658536585365853, "grad_norm": 0.13165190815925598, "learning_rate": 5.269935438715312e-05, "loss": 0.0351, "step": 30520 }, { "epoch": 13.663011859476393, "grad_norm": 0.13079673051834106, "learning_rate": 5.2671836337664634e-05, "loss": 0.0329, "step": 30530 }, { "epoch": 13.667487133586933, "grad_norm": 0.14295710623264313, "learning_rate": 5.264431747654284e-05, "loss": 0.0312, "step": 30540 }, { "epoch": 13.671962407697471, "grad_norm": 0.11045211553573608, "learning_rate": 5.26167978121472e-05, "loss": 0.0329, "step": 30550 }, { "epoch": 13.676437681808011, "grad_norm": 0.1185019239783287, "learning_rate": 5.258927735283748e-05, "loss": 0.0285, "step": 30560 }, { "epoch": 13.68091295591855, "grad_norm": 0.14228685200214386, "learning_rate": 5.2561756106973656e-05, "loss": 0.0311, "step": 30570 }, { "epoch": 13.685388230029089, "grad_norm": 0.12576939165592194, "learning_rate": 5.2534234082915976e-05, "loss": 0.0334, "step": 30580 }, { "epoch": 13.689863504139629, "grad_norm": 0.13182410597801208, "learning_rate": 5.25067112890249e-05, "loss": 0.0329, "step": 30590 }, { "epoch": 13.694338778250168, "grad_norm": 0.14511901140213013, "learning_rate": 5.247918773366112e-05, "loss": 0.0288, "step": 30600 }, { "epoch": 13.698814052360706, "grad_norm": 0.12834177911281586, "learning_rate": 5.245166342518556e-05, "loss": 0.0354, "step": 30610 }, { "epoch": 13.703289326471246, "grad_norm": 0.12259993702173233, "learning_rate": 5.242413837195938e-05, "loss": 0.0322, "step": 30620 }, { "epoch": 13.707764600581786, "grad_norm": 0.15608076751232147, "learning_rate": 5.2396612582343986e-05, "loss": 0.0334, "step": 30630 }, { "epoch": 13.712239874692324, "grad_norm": 0.12725412845611572, "learning_rate": 5.2369086064700945e-05, "loss": 0.0325, "step": 30640 }, { "epoch": 13.716715148802864, "grad_norm": 0.11936045438051224, "learning_rate": 5.234155882739212e-05, "loss": 0.0338, "step": 30650 }, { "epoch": 13.721190422913404, "grad_norm": 0.13925597071647644, "learning_rate": 5.231403087877955e-05, "loss": 0.0286, "step": 30660 }, { "epoch": 13.725665697023942, "grad_norm": 0.12936291098594666, "learning_rate": 5.2286502227225506e-05, "loss": 0.0342, "step": 30670 }, { "epoch": 13.730140971134482, "grad_norm": 0.121385857462883, "learning_rate": 5.225897288109245e-05, "loss": 0.0305, "step": 30680 }, { "epoch": 13.734616245245022, "grad_norm": 0.14951276779174805, "learning_rate": 5.2231442848743064e-05, "loss": 0.0301, "step": 30690 }, { "epoch": 13.739091519355561, "grad_norm": 0.12693019211292267, "learning_rate": 5.220391213854028e-05, "loss": 0.0329, "step": 30700 }, { "epoch": 13.7435667934661, "grad_norm": 0.1228460818529129, "learning_rate": 5.217638075884716e-05, "loss": 0.03, "step": 30710 }, { "epoch": 13.74804206757664, "grad_norm": 0.15006065368652344, "learning_rate": 5.214884871802703e-05, "loss": 0.0326, "step": 30720 }, { "epoch": 13.752517341687179, "grad_norm": 0.11059547960758209, "learning_rate": 5.2121316024443415e-05, "loss": 0.0323, "step": 30730 }, { "epoch": 13.756992615797717, "grad_norm": 0.11481555551290512, "learning_rate": 5.209378268645998e-05, "loss": 0.031, "step": 30740 }, { "epoch": 13.761467889908257, "grad_norm": 0.11420293152332306, "learning_rate": 5.2066248712440656e-05, "loss": 0.0321, "step": 30750 }, { "epoch": 13.765943164018797, "grad_norm": 0.12824778258800507, "learning_rate": 5.203871411074954e-05, "loss": 0.0297, "step": 30760 }, { "epoch": 13.770418438129335, "grad_norm": 0.10991160571575165, "learning_rate": 5.20111788897509e-05, "loss": 0.0344, "step": 30770 }, { "epoch": 13.774893712239875, "grad_norm": 0.12222378700971603, "learning_rate": 5.198364305780922e-05, "loss": 0.0317, "step": 30780 }, { "epoch": 13.779368986350415, "grad_norm": 0.13036805391311646, "learning_rate": 5.1956106623289145e-05, "loss": 0.0305, "step": 30790 }, { "epoch": 13.783844260460953, "grad_norm": 0.11242999136447906, "learning_rate": 5.1928569594555524e-05, "loss": 0.0281, "step": 30800 }, { "epoch": 13.788319534571492, "grad_norm": 0.131133571267128, "learning_rate": 5.1901031979973394e-05, "loss": 0.031, "step": 30810 }, { "epoch": 13.792794808682032, "grad_norm": 0.11457362025976181, "learning_rate": 5.187349378790793e-05, "loss": 0.0305, "step": 30820 }, { "epoch": 13.79727008279257, "grad_norm": 0.13131816685199738, "learning_rate": 5.1845955026724535e-05, "loss": 0.0295, "step": 30830 }, { "epoch": 13.80174535690311, "grad_norm": 0.1389068067073822, "learning_rate": 5.1818415704788725e-05, "loss": 0.0329, "step": 30840 }, { "epoch": 13.80622063101365, "grad_norm": 0.1225694790482521, "learning_rate": 5.179087583046625e-05, "loss": 0.0283, "step": 30850 }, { "epoch": 13.810695905124188, "grad_norm": 0.13307297229766846, "learning_rate": 5.176333541212296e-05, "loss": 0.0299, "step": 30860 }, { "epoch": 13.815171179234728, "grad_norm": 0.11041750758886337, "learning_rate": 5.1735794458124956e-05, "loss": 0.031, "step": 30870 }, { "epoch": 13.819646453345268, "grad_norm": 0.12050902098417282, "learning_rate": 5.170825297683841e-05, "loss": 0.0358, "step": 30880 }, { "epoch": 13.824121727455807, "grad_norm": 0.12176880985498428, "learning_rate": 5.168071097662972e-05, "loss": 0.0316, "step": 30890 }, { "epoch": 13.828597001566346, "grad_norm": 0.11309217661619186, "learning_rate": 5.165316846586541e-05, "loss": 0.0303, "step": 30900 }, { "epoch": 13.833072275676885, "grad_norm": 0.13581761717796326, "learning_rate": 5.1625625452912174e-05, "loss": 0.0366, "step": 30910 }, { "epoch": 13.837547549787425, "grad_norm": 0.1133120134472847, "learning_rate": 5.1598081946136843e-05, "loss": 0.0334, "step": 30920 }, { "epoch": 13.842022823897963, "grad_norm": 0.11661548912525177, "learning_rate": 5.157053795390642e-05, "loss": 0.0313, "step": 30930 }, { "epoch": 13.846498098008503, "grad_norm": 0.09593261778354645, "learning_rate": 5.154299348458803e-05, "loss": 0.0335, "step": 30940 }, { "epoch": 13.850973372119043, "grad_norm": 0.12037575244903564, "learning_rate": 5.151544854654895e-05, "loss": 0.0308, "step": 30950 }, { "epoch": 13.855448646229581, "grad_norm": 0.1451137512922287, "learning_rate": 5.148790314815663e-05, "loss": 0.0301, "step": 30960 }, { "epoch": 13.85992392034012, "grad_norm": 0.13126228749752045, "learning_rate": 5.1460357297778615e-05, "loss": 0.0319, "step": 30970 }, { "epoch": 13.86439919445066, "grad_norm": 0.1353387087583542, "learning_rate": 5.143281100378261e-05, "loss": 0.0335, "step": 30980 }, { "epoch": 13.868874468561199, "grad_norm": 0.13162504136562347, "learning_rate": 5.1405264274536445e-05, "loss": 0.0289, "step": 30990 }, { "epoch": 13.873349742671738, "grad_norm": 0.1606297492980957, "learning_rate": 5.1377717118408105e-05, "loss": 0.033, "step": 31000 }, { "epoch": 13.877825016782278, "grad_norm": 0.0936846137046814, "learning_rate": 5.135016954376567e-05, "loss": 0.0304, "step": 31010 }, { "epoch": 13.882300290892816, "grad_norm": 0.1256527304649353, "learning_rate": 5.132262155897739e-05, "loss": 0.0295, "step": 31020 }, { "epoch": 13.886775565003356, "grad_norm": 0.10972697287797928, "learning_rate": 5.12950731724116e-05, "loss": 0.0301, "step": 31030 }, { "epoch": 13.891250839113896, "grad_norm": 0.12571163475513458, "learning_rate": 5.1267524392436784e-05, "loss": 0.0322, "step": 31040 }, { "epoch": 13.895726113224434, "grad_norm": 0.10742966085672379, "learning_rate": 5.123997522742151e-05, "loss": 0.03, "step": 31050 }, { "epoch": 13.900201387334974, "grad_norm": 0.1314852088689804, "learning_rate": 5.12124256857345e-05, "loss": 0.0291, "step": 31060 }, { "epoch": 13.904676661445514, "grad_norm": 0.12451211363077164, "learning_rate": 5.118487577574459e-05, "loss": 0.0296, "step": 31070 }, { "epoch": 13.909151935556054, "grad_norm": 0.1497722715139389, "learning_rate": 5.1157325505820694e-05, "loss": 0.0312, "step": 31080 }, { "epoch": 13.913627209666592, "grad_norm": 0.152971088886261, "learning_rate": 5.112977488433188e-05, "loss": 0.0312, "step": 31090 }, { "epoch": 13.918102483777131, "grad_norm": 0.1573476493358612, "learning_rate": 5.110222391964728e-05, "loss": 0.0345, "step": 31100 }, { "epoch": 13.922577757887671, "grad_norm": 0.13260331749916077, "learning_rate": 5.107467262013614e-05, "loss": 0.0314, "step": 31110 }, { "epoch": 13.92705303199821, "grad_norm": 0.13938508927822113, "learning_rate": 5.104712099416785e-05, "loss": 0.0323, "step": 31120 }, { "epoch": 13.93152830610875, "grad_norm": 0.1340818852186203, "learning_rate": 5.101956905011185e-05, "loss": 0.0309, "step": 31130 }, { "epoch": 13.936003580219289, "grad_norm": 0.11996565014123917, "learning_rate": 5.0992016796337686e-05, "loss": 0.0341, "step": 31140 }, { "epoch": 13.940478854329827, "grad_norm": 0.13631349802017212, "learning_rate": 5.096446424121502e-05, "loss": 0.0337, "step": 31150 }, { "epoch": 13.944954128440367, "grad_norm": 0.1112804040312767, "learning_rate": 5.093691139311356e-05, "loss": 0.0309, "step": 31160 }, { "epoch": 13.949429402550907, "grad_norm": 0.10118910670280457, "learning_rate": 5.0909358260403186e-05, "loss": 0.0331, "step": 31170 }, { "epoch": 13.953904676661445, "grad_norm": 0.12334148585796356, "learning_rate": 5.088180485145378e-05, "loss": 0.0333, "step": 31180 }, { "epoch": 13.958379950771985, "grad_norm": 0.10503482818603516, "learning_rate": 5.085425117463533e-05, "loss": 0.0322, "step": 31190 }, { "epoch": 13.962855224882524, "grad_norm": 0.13542021811008453, "learning_rate": 5.0826697238317935e-05, "loss": 0.0312, "step": 31200 }, { "epoch": 13.967330498993064, "grad_norm": 0.12279608100652695, "learning_rate": 5.0799143050871746e-05, "loss": 0.0296, "step": 31210 }, { "epoch": 13.971805773103602, "grad_norm": 0.12470947951078415, "learning_rate": 5.077158862066699e-05, "loss": 0.0332, "step": 31220 }, { "epoch": 13.976281047214142, "grad_norm": 0.10537810623645782, "learning_rate": 5.074403395607399e-05, "loss": 0.0335, "step": 31230 }, { "epoch": 13.980756321324682, "grad_norm": 0.1326228529214859, "learning_rate": 5.071647906546312e-05, "loss": 0.0291, "step": 31240 }, { "epoch": 13.98523159543522, "grad_norm": 0.1257198303937912, "learning_rate": 5.068892395720483e-05, "loss": 0.0319, "step": 31250 }, { "epoch": 13.98970686954576, "grad_norm": 0.09099692106246948, "learning_rate": 5.066136863966963e-05, "loss": 0.0313, "step": 31260 }, { "epoch": 13.9941821436563, "grad_norm": 0.13224878907203674, "learning_rate": 5.063381312122809e-05, "loss": 0.0309, "step": 31270 }, { "epoch": 13.998657417766838, "grad_norm": 0.1358027309179306, "learning_rate": 5.0606257410250866e-05, "loss": 0.0356, "step": 31280 }, { "epoch": 14.003132691877378, "grad_norm": 0.1327366679906845, "learning_rate": 5.057870151510864e-05, "loss": 0.03, "step": 31290 }, { "epoch": 14.007607965987917, "grad_norm": 0.11036446690559387, "learning_rate": 5.0551145444172186e-05, "loss": 0.029, "step": 31300 }, { "epoch": 14.012083240098455, "grad_norm": 0.15579086542129517, "learning_rate": 5.052358920581229e-05, "loss": 0.0338, "step": 31310 }, { "epoch": 14.016558514208995, "grad_norm": 0.12303535640239716, "learning_rate": 5.0496032808399815e-05, "loss": 0.0329, "step": 31320 }, { "epoch": 14.021033788319535, "grad_norm": 0.13010956346988678, "learning_rate": 5.046847626030569e-05, "loss": 0.03, "step": 31330 }, { "epoch": 14.025509062430073, "grad_norm": 0.14484697580337524, "learning_rate": 5.0440919569900835e-05, "loss": 0.0318, "step": 31340 }, { "epoch": 14.029984336540613, "grad_norm": 0.13377319276332855, "learning_rate": 5.041336274555625e-05, "loss": 0.0308, "step": 31350 }, { "epoch": 14.034459610651153, "grad_norm": 0.13517095148563385, "learning_rate": 5.038580579564298e-05, "loss": 0.0355, "step": 31360 }, { "epoch": 14.03893488476169, "grad_norm": 0.14167922735214233, "learning_rate": 5.0358248728532096e-05, "loss": 0.032, "step": 31370 }, { "epoch": 14.04341015887223, "grad_norm": 0.12823939323425293, "learning_rate": 5.033069155259471e-05, "loss": 0.0338, "step": 31380 }, { "epoch": 14.04788543298277, "grad_norm": 0.16445623338222504, "learning_rate": 5.030313427620197e-05, "loss": 0.0333, "step": 31390 }, { "epoch": 14.05236070709331, "grad_norm": 0.17246906459331512, "learning_rate": 5.027557690772503e-05, "loss": 0.0324, "step": 31400 }, { "epoch": 14.056835981203848, "grad_norm": 0.12994414567947388, "learning_rate": 5.02480194555351e-05, "loss": 0.0298, "step": 31410 }, { "epoch": 14.061311255314388, "grad_norm": 0.16869299113750458, "learning_rate": 5.0220461928003406e-05, "loss": 0.0315, "step": 31420 }, { "epoch": 14.065786529424928, "grad_norm": 0.13431929051876068, "learning_rate": 5.0192904333501214e-05, "loss": 0.0288, "step": 31430 }, { "epoch": 14.070261803535466, "grad_norm": 0.11996549367904663, "learning_rate": 5.016534668039976e-05, "loss": 0.032, "step": 31440 }, { "epoch": 14.074737077646006, "grad_norm": 0.12651178240776062, "learning_rate": 5.0137788977070353e-05, "loss": 0.0299, "step": 31450 }, { "epoch": 14.079212351756546, "grad_norm": 0.11782991886138916, "learning_rate": 5.011023123188431e-05, "loss": 0.0319, "step": 31460 }, { "epoch": 14.083687625867084, "grad_norm": 0.10619687288999557, "learning_rate": 5.0082673453212914e-05, "loss": 0.0284, "step": 31470 }, { "epoch": 14.088162899977624, "grad_norm": 0.1258830428123474, "learning_rate": 5.005511564942751e-05, "loss": 0.032, "step": 31480 }, { "epoch": 14.092638174088163, "grad_norm": 0.16015779972076416, "learning_rate": 5.0027557828899426e-05, "loss": 0.032, "step": 31490 }, { "epoch": 14.097113448198701, "grad_norm": 0.14840373396873474, "learning_rate": 5e-05, "loss": 0.0346, "step": 31500 }, { "epoch": 14.101588722309241, "grad_norm": 0.1410653293132782, "learning_rate": 4.997244217110058e-05, "loss": 0.0332, "step": 31510 }, { "epoch": 14.106063996419781, "grad_norm": 0.13705404102802277, "learning_rate": 4.994488435057251e-05, "loss": 0.0334, "step": 31520 }, { "epoch": 14.11053927053032, "grad_norm": 0.13360488414764404, "learning_rate": 4.991732654678709e-05, "loss": 0.0306, "step": 31530 }, { "epoch": 14.115014544640859, "grad_norm": 0.13160264492034912, "learning_rate": 4.988976876811571e-05, "loss": 0.03, "step": 31540 }, { "epoch": 14.119489818751399, "grad_norm": 0.15852802991867065, "learning_rate": 4.986221102292965e-05, "loss": 0.0324, "step": 31550 }, { "epoch": 14.123965092861937, "grad_norm": 0.1407870054244995, "learning_rate": 4.9834653319600246e-05, "loss": 0.031, "step": 31560 }, { "epoch": 14.128440366972477, "grad_norm": 0.11955949664115906, "learning_rate": 4.98070956664988e-05, "loss": 0.0324, "step": 31570 }, { "epoch": 14.132915641083017, "grad_norm": 0.12219425290822983, "learning_rate": 4.97795380719966e-05, "loss": 0.0328, "step": 31580 }, { "epoch": 14.137390915193556, "grad_norm": 0.13518451154232025, "learning_rate": 4.975198054446492e-05, "loss": 0.0327, "step": 31590 }, { "epoch": 14.141866189304094, "grad_norm": 0.10912986099720001, "learning_rate": 4.972442309227498e-05, "loss": 0.0305, "step": 31600 }, { "epoch": 14.146341463414634, "grad_norm": 0.1317816525697708, "learning_rate": 4.9696865723798035e-05, "loss": 0.0299, "step": 31610 }, { "epoch": 14.150816737525174, "grad_norm": 0.1391362100839615, "learning_rate": 4.96693084474053e-05, "loss": 0.0348, "step": 31620 }, { "epoch": 14.155292011635712, "grad_norm": 0.14707893133163452, "learning_rate": 4.96417512714679e-05, "loss": 0.032, "step": 31630 }, { "epoch": 14.159767285746252, "grad_norm": 0.1378260999917984, "learning_rate": 4.961419420435703e-05, "loss": 0.0302, "step": 31640 }, { "epoch": 14.164242559856792, "grad_norm": 0.16727516055107117, "learning_rate": 4.9586637254443756e-05, "loss": 0.0321, "step": 31650 }, { "epoch": 14.16871783396733, "grad_norm": 0.13376499712467194, "learning_rate": 4.955908043009917e-05, "loss": 0.0334, "step": 31660 }, { "epoch": 14.17319310807787, "grad_norm": 0.13006842136383057, "learning_rate": 4.9531523739694325e-05, "loss": 0.0276, "step": 31670 }, { "epoch": 14.17766838218841, "grad_norm": 0.12356937676668167, "learning_rate": 4.950396719160018e-05, "loss": 0.0324, "step": 31680 }, { "epoch": 14.182143656298948, "grad_norm": 0.10567120462656021, "learning_rate": 4.947641079418773e-05, "loss": 0.0334, "step": 31690 }, { "epoch": 14.186618930409487, "grad_norm": 0.12214721739292145, "learning_rate": 4.9448854555827825e-05, "loss": 0.033, "step": 31700 }, { "epoch": 14.191094204520027, "grad_norm": 0.10224590450525284, "learning_rate": 4.942129848489137e-05, "loss": 0.0307, "step": 31710 }, { "epoch": 14.195569478630565, "grad_norm": 0.11990765482187271, "learning_rate": 4.9393742589749145e-05, "loss": 0.0282, "step": 31720 }, { "epoch": 14.200044752741105, "grad_norm": 0.11781120300292969, "learning_rate": 4.9366186878771926e-05, "loss": 0.0316, "step": 31730 }, { "epoch": 14.204520026851645, "grad_norm": 0.1313457489013672, "learning_rate": 4.93386313603304e-05, "loss": 0.0287, "step": 31740 }, { "epoch": 14.208995300962185, "grad_norm": 0.15767353773117065, "learning_rate": 4.9311076042795185e-05, "loss": 0.0309, "step": 31750 }, { "epoch": 14.213470575072723, "grad_norm": 0.13133835792541504, "learning_rate": 4.9283520934536904e-05, "loss": 0.0312, "step": 31760 }, { "epoch": 14.217945849183263, "grad_norm": 0.10753316432237625, "learning_rate": 4.925596604392603e-05, "loss": 0.029, "step": 31770 }, { "epoch": 14.222421123293802, "grad_norm": 0.11250606924295425, "learning_rate": 4.9228411379333014e-05, "loss": 0.0314, "step": 31780 }, { "epoch": 14.22689639740434, "grad_norm": 0.10699483007192612, "learning_rate": 4.920085694912828e-05, "loss": 0.0294, "step": 31790 }, { "epoch": 14.23137167151488, "grad_norm": 0.13622616231441498, "learning_rate": 4.917330276168208e-05, "loss": 0.0308, "step": 31800 }, { "epoch": 14.23584694562542, "grad_norm": 0.13992980122566223, "learning_rate": 4.91457488253647e-05, "loss": 0.0304, "step": 31810 }, { "epoch": 14.240322219735958, "grad_norm": 0.14489486813545227, "learning_rate": 4.911819514854624e-05, "loss": 0.0306, "step": 31820 }, { "epoch": 14.244797493846498, "grad_norm": 0.12649431824684143, "learning_rate": 4.909064173959681e-05, "loss": 0.0337, "step": 31830 }, { "epoch": 14.249272767957038, "grad_norm": 0.14469313621520996, "learning_rate": 4.906308860688645e-05, "loss": 0.0314, "step": 31840 }, { "epoch": 14.253748042067576, "grad_norm": 0.13449855148792267, "learning_rate": 4.903553575878499e-05, "loss": 0.0349, "step": 31850 }, { "epoch": 14.258223316178116, "grad_norm": 0.1242353618144989, "learning_rate": 4.9007983203662326e-05, "loss": 0.0327, "step": 31860 }, { "epoch": 14.262698590288656, "grad_norm": 0.12639164924621582, "learning_rate": 4.8980430949888156e-05, "loss": 0.0334, "step": 31870 }, { "epoch": 14.267173864399194, "grad_norm": 0.1209057942032814, "learning_rate": 4.895287900583216e-05, "loss": 0.0336, "step": 31880 }, { "epoch": 14.271649138509733, "grad_norm": 0.13829714059829712, "learning_rate": 4.892532737986387e-05, "loss": 0.0302, "step": 31890 }, { "epoch": 14.276124412620273, "grad_norm": 0.13672955334186554, "learning_rate": 4.889777608035273e-05, "loss": 0.0319, "step": 31900 }, { "epoch": 14.280599686730813, "grad_norm": 0.1243743747472763, "learning_rate": 4.887022511566814e-05, "loss": 0.0322, "step": 31910 }, { "epoch": 14.285074960841351, "grad_norm": 0.14596043527126312, "learning_rate": 4.884267449417931e-05, "loss": 0.0305, "step": 31920 }, { "epoch": 14.289550234951891, "grad_norm": 0.12117727100849152, "learning_rate": 4.881512422425541e-05, "loss": 0.0328, "step": 31930 }, { "epoch": 14.29402550906243, "grad_norm": 0.10290488600730896, "learning_rate": 4.878757431426551e-05, "loss": 0.0299, "step": 31940 }, { "epoch": 14.298500783172969, "grad_norm": 0.1146087646484375, "learning_rate": 4.87600247725785e-05, "loss": 0.0311, "step": 31950 }, { "epoch": 14.302976057283509, "grad_norm": 0.11357443034648895, "learning_rate": 4.873247560756324e-05, "loss": 0.0349, "step": 31960 }, { "epoch": 14.307451331394049, "grad_norm": 0.15128080546855927, "learning_rate": 4.870492682758841e-05, "loss": 0.0301, "step": 31970 }, { "epoch": 14.311926605504587, "grad_norm": 0.1255815327167511, "learning_rate": 4.867737844102261e-05, "loss": 0.0301, "step": 31980 }, { "epoch": 14.316401879615126, "grad_norm": 0.15673905611038208, "learning_rate": 4.864983045623434e-05, "loss": 0.03, "step": 31990 }, { "epoch": 14.320877153725666, "grad_norm": 0.11141011118888855, "learning_rate": 4.8622282881591906e-05, "loss": 0.0335, "step": 32000 }, { "epoch": 14.325352427836204, "grad_norm": 0.11469302326440811, "learning_rate": 4.8594735725463567e-05, "loss": 0.0299, "step": 32010 }, { "epoch": 14.329827701946744, "grad_norm": 0.1116335317492485, "learning_rate": 4.85671889962174e-05, "loss": 0.03, "step": 32020 }, { "epoch": 14.334302976057284, "grad_norm": 0.12906479835510254, "learning_rate": 4.85396427022214e-05, "loss": 0.0338, "step": 32030 }, { "epoch": 14.338778250167822, "grad_norm": 0.11063835769891739, "learning_rate": 4.851209685184338e-05, "loss": 0.0305, "step": 32040 }, { "epoch": 14.343253524278362, "grad_norm": 0.13330627977848053, "learning_rate": 4.848455145345105e-05, "loss": 0.0299, "step": 32050 }, { "epoch": 14.347728798388902, "grad_norm": 0.1495882272720337, "learning_rate": 4.845700651541199e-05, "loss": 0.033, "step": 32060 }, { "epoch": 14.35220407249944, "grad_norm": 0.1175476685166359, "learning_rate": 4.8429462046093585e-05, "loss": 0.0297, "step": 32070 }, { "epoch": 14.35667934660998, "grad_norm": 0.13035406172275543, "learning_rate": 4.840191805386315e-05, "loss": 0.033, "step": 32080 }, { "epoch": 14.36115462072052, "grad_norm": 0.12800446152687073, "learning_rate": 4.837437454708784e-05, "loss": 0.0328, "step": 32090 }, { "epoch": 14.36562989483106, "grad_norm": 0.14750936627388, "learning_rate": 4.834683153413459e-05, "loss": 0.0311, "step": 32100 }, { "epoch": 14.370105168941597, "grad_norm": 0.1449756622314453, "learning_rate": 4.8319289023370296e-05, "loss": 0.0291, "step": 32110 }, { "epoch": 14.374580443052137, "grad_norm": 0.1357119381427765, "learning_rate": 4.82917470231616e-05, "loss": 0.0307, "step": 32120 }, { "epoch": 14.379055717162677, "grad_norm": 0.14254286885261536, "learning_rate": 4.826420554187506e-05, "loss": 0.0283, "step": 32130 }, { "epoch": 14.383530991273215, "grad_norm": 0.11331890523433685, "learning_rate": 4.823666458787705e-05, "loss": 0.029, "step": 32140 }, { "epoch": 14.388006265383755, "grad_norm": 0.1209830567240715, "learning_rate": 4.820912416953377e-05, "loss": 0.0327, "step": 32150 }, { "epoch": 14.392481539494295, "grad_norm": 0.10751473903656006, "learning_rate": 4.818158429521129e-05, "loss": 0.032, "step": 32160 }, { "epoch": 14.396956813604833, "grad_norm": 0.1049581915140152, "learning_rate": 4.8154044973275484e-05, "loss": 0.0307, "step": 32170 }, { "epoch": 14.401432087715373, "grad_norm": 0.1268339902162552, "learning_rate": 4.812650621209209e-05, "loss": 0.0328, "step": 32180 }, { "epoch": 14.405907361825912, "grad_norm": 0.1130162701010704, "learning_rate": 4.809896802002662e-05, "loss": 0.0308, "step": 32190 }, { "epoch": 14.41038263593645, "grad_norm": 0.13817046582698822, "learning_rate": 4.8071430405444474e-05, "loss": 0.0302, "step": 32200 }, { "epoch": 14.41485791004699, "grad_norm": 0.11351260542869568, "learning_rate": 4.804389337671087e-05, "loss": 0.0317, "step": 32210 }, { "epoch": 14.41933318415753, "grad_norm": 0.12498049437999725, "learning_rate": 4.801635694219079e-05, "loss": 0.0276, "step": 32220 }, { "epoch": 14.423808458268068, "grad_norm": 0.13237562775611877, "learning_rate": 4.798882111024912e-05, "loss": 0.0276, "step": 32230 }, { "epoch": 14.428283732378608, "grad_norm": 0.12295904010534286, "learning_rate": 4.7961285889250475e-05, "loss": 0.0293, "step": 32240 }, { "epoch": 14.432759006489148, "grad_norm": 0.11380310356616974, "learning_rate": 4.7933751287559335e-05, "loss": 0.0295, "step": 32250 }, { "epoch": 14.437234280599686, "grad_norm": 0.14228765666484833, "learning_rate": 4.790621731354003e-05, "loss": 0.0286, "step": 32260 }, { "epoch": 14.441709554710226, "grad_norm": 0.10858723521232605, "learning_rate": 4.78786839755566e-05, "loss": 0.0298, "step": 32270 }, { "epoch": 14.446184828820765, "grad_norm": 0.11803919821977615, "learning_rate": 4.785115128197298e-05, "loss": 0.0292, "step": 32280 }, { "epoch": 14.450660102931305, "grad_norm": 0.12369152158498764, "learning_rate": 4.7823619241152854e-05, "loss": 0.0286, "step": 32290 }, { "epoch": 14.455135377041843, "grad_norm": 0.11459558457136154, "learning_rate": 4.779608786145974e-05, "loss": 0.0294, "step": 32300 }, { "epoch": 14.459610651152383, "grad_norm": 0.14332158863544464, "learning_rate": 4.776855715125694e-05, "loss": 0.03, "step": 32310 }, { "epoch": 14.464085925262923, "grad_norm": 0.12062323838472366, "learning_rate": 4.774102711890756e-05, "loss": 0.0295, "step": 32320 }, { "epoch": 14.468561199373461, "grad_norm": 0.1253206580877304, "learning_rate": 4.771349777277452e-05, "loss": 0.0291, "step": 32330 }, { "epoch": 14.473036473484001, "grad_norm": 0.13548044860363007, "learning_rate": 4.7685969121220456e-05, "loss": 0.0316, "step": 32340 }, { "epoch": 14.47751174759454, "grad_norm": 0.10651445388793945, "learning_rate": 4.7658441172607876e-05, "loss": 0.0315, "step": 32350 }, { "epoch": 14.481987021705079, "grad_norm": 0.11210950464010239, "learning_rate": 4.7630913935299066e-05, "loss": 0.0298, "step": 32360 }, { "epoch": 14.486462295815619, "grad_norm": 0.11848791688680649, "learning_rate": 4.7603387417656026e-05, "loss": 0.0275, "step": 32370 }, { "epoch": 14.490937569926158, "grad_norm": 0.13750965893268585, "learning_rate": 4.7575861628040635e-05, "loss": 0.0312, "step": 32380 }, { "epoch": 14.495412844036696, "grad_norm": 0.13563555479049683, "learning_rate": 4.754833657481445e-05, "loss": 0.0312, "step": 32390 }, { "epoch": 14.499888118147236, "grad_norm": 0.16382996737957, "learning_rate": 4.7520812266338885e-05, "loss": 0.0329, "step": 32400 }, { "epoch": 14.504363392257776, "grad_norm": 0.18432340025901794, "learning_rate": 4.749328871097512e-05, "loss": 0.0332, "step": 32410 }, { "epoch": 14.508838666368316, "grad_norm": 0.12504731118679047, "learning_rate": 4.746576591708403e-05, "loss": 0.0344, "step": 32420 }, { "epoch": 14.513313940478854, "grad_norm": 0.12269795686006546, "learning_rate": 4.743824389302635e-05, "loss": 0.0335, "step": 32430 }, { "epoch": 14.517789214589394, "grad_norm": 0.12577089667320251, "learning_rate": 4.741072264716252e-05, "loss": 0.0277, "step": 32440 }, { "epoch": 14.522264488699934, "grad_norm": 0.12183735519647598, "learning_rate": 4.738320218785281e-05, "loss": 0.028, "step": 32450 }, { "epoch": 14.526739762810472, "grad_norm": 0.1452815681695938, "learning_rate": 4.735568252345718e-05, "loss": 0.0264, "step": 32460 }, { "epoch": 14.531215036921012, "grad_norm": 0.14250698685646057, "learning_rate": 4.7328163662335364e-05, "loss": 0.0302, "step": 32470 }, { "epoch": 14.535690311031551, "grad_norm": 0.13644102215766907, "learning_rate": 4.7300645612846907e-05, "loss": 0.0276, "step": 32480 }, { "epoch": 14.54016558514209, "grad_norm": 0.15207436680793762, "learning_rate": 4.7273128383351015e-05, "loss": 0.0316, "step": 32490 }, { "epoch": 14.54464085925263, "grad_norm": 0.147593691945076, "learning_rate": 4.7245611982206724e-05, "loss": 0.0336, "step": 32500 }, { "epoch": 14.54911613336317, "grad_norm": 0.131923109292984, "learning_rate": 4.721809641777281e-05, "loss": 0.0309, "step": 32510 }, { "epoch": 14.553591407473707, "grad_norm": 0.13170352578163147, "learning_rate": 4.7190581698407725e-05, "loss": 0.0337, "step": 32520 }, { "epoch": 14.558066681584247, "grad_norm": 0.15180641412734985, "learning_rate": 4.716306783246977e-05, "loss": 0.028, "step": 32530 }, { "epoch": 14.562541955694787, "grad_norm": 0.13998201489448547, "learning_rate": 4.713555482831688e-05, "loss": 0.0315, "step": 32540 }, { "epoch": 14.567017229805325, "grad_norm": 0.14166703820228577, "learning_rate": 4.710804269430681e-05, "loss": 0.0293, "step": 32550 }, { "epoch": 14.571492503915865, "grad_norm": 0.09385187923908234, "learning_rate": 4.708053143879701e-05, "loss": 0.0255, "step": 32560 }, { "epoch": 14.575967778026405, "grad_norm": 0.15853294730186462, "learning_rate": 4.7053021070144664e-05, "loss": 0.0317, "step": 32570 }, { "epoch": 14.580443052136943, "grad_norm": 0.13322177529335022, "learning_rate": 4.702551159670672e-05, "loss": 0.0276, "step": 32580 }, { "epoch": 14.584918326247482, "grad_norm": 0.12798243761062622, "learning_rate": 4.699800302683981e-05, "loss": 0.0306, "step": 32590 }, { "epoch": 14.589393600358022, "grad_norm": 0.12000150233507156, "learning_rate": 4.697049536890033e-05, "loss": 0.0323, "step": 32600 }, { "epoch": 14.593868874468562, "grad_norm": 0.11592309176921844, "learning_rate": 4.694298863124435e-05, "loss": 0.0323, "step": 32610 }, { "epoch": 14.5983441485791, "grad_norm": 0.12926514446735382, "learning_rate": 4.691548282222771e-05, "loss": 0.0322, "step": 32620 }, { "epoch": 14.60281942268964, "grad_norm": 0.11726488173007965, "learning_rate": 4.688797795020597e-05, "loss": 0.0324, "step": 32630 }, { "epoch": 14.60729469680018, "grad_norm": 0.12735868990421295, "learning_rate": 4.6860474023534335e-05, "loss": 0.0309, "step": 32640 }, { "epoch": 14.611769970910718, "grad_norm": 0.12895458936691284, "learning_rate": 4.683297105056782e-05, "loss": 0.0319, "step": 32650 }, { "epoch": 14.616245245021258, "grad_norm": 0.14839546382427216, "learning_rate": 4.680546903966106e-05, "loss": 0.0315, "step": 32660 }, { "epoch": 14.620720519131797, "grad_norm": 0.14199529588222504, "learning_rate": 4.677796799916845e-05, "loss": 0.0292, "step": 32670 }, { "epoch": 14.625195793242336, "grad_norm": 0.11606661230325699, "learning_rate": 4.6750467937444115e-05, "loss": 0.031, "step": 32680 }, { "epoch": 14.629671067352875, "grad_norm": 0.1324678212404251, "learning_rate": 4.6722968862841806e-05, "loss": 0.0315, "step": 32690 }, { "epoch": 14.634146341463415, "grad_norm": 0.11977871507406235, "learning_rate": 4.669547078371504e-05, "loss": 0.0311, "step": 32700 }, { "epoch": 14.638621615573953, "grad_norm": 0.11287666857242584, "learning_rate": 4.666797370841699e-05, "loss": 0.0333, "step": 32710 }, { "epoch": 14.643096889684493, "grad_norm": 0.11241611838340759, "learning_rate": 4.664047764530055e-05, "loss": 0.0281, "step": 32720 }, { "epoch": 14.647572163795033, "grad_norm": 0.14163479208946228, "learning_rate": 4.66129826027183e-05, "loss": 0.0333, "step": 32730 }, { "epoch": 14.652047437905571, "grad_norm": 0.13394586741924286, "learning_rate": 4.65854885890225e-05, "loss": 0.0303, "step": 32740 }, { "epoch": 14.65652271201611, "grad_norm": 0.12339895963668823, "learning_rate": 4.6557995612565144e-05, "loss": 0.0323, "step": 32750 }, { "epoch": 14.66099798612665, "grad_norm": 0.11070023477077484, "learning_rate": 4.65305036816978e-05, "loss": 0.0282, "step": 32760 }, { "epoch": 14.665473260237189, "grad_norm": 0.12498302757740021, "learning_rate": 4.650301280477184e-05, "loss": 0.0307, "step": 32770 }, { "epoch": 14.669948534347728, "grad_norm": 0.11857137829065323, "learning_rate": 4.647552299013828e-05, "loss": 0.0253, "step": 32780 }, { "epoch": 14.674423808458268, "grad_norm": 0.11728523671627045, "learning_rate": 4.6448034246147754e-05, "loss": 0.0299, "step": 32790 }, { "epoch": 14.678899082568808, "grad_norm": 0.13904239237308502, "learning_rate": 4.642054658115067e-05, "loss": 0.0298, "step": 32800 }, { "epoch": 14.683374356679346, "grad_norm": 0.1447119116783142, "learning_rate": 4.6393060003496984e-05, "loss": 0.029, "step": 32810 }, { "epoch": 14.687849630789886, "grad_norm": 0.11610020697116852, "learning_rate": 4.6365574521536445e-05, "loss": 0.0318, "step": 32820 }, { "epoch": 14.692324904900426, "grad_norm": 0.125487819314003, "learning_rate": 4.633809014361843e-05, "loss": 0.0294, "step": 32830 }, { "epoch": 14.696800179010964, "grad_norm": 0.12563854455947876, "learning_rate": 4.631060687809191e-05, "loss": 0.0307, "step": 32840 }, { "epoch": 14.701275453121504, "grad_norm": 0.12682169675827026, "learning_rate": 4.6283124733305624e-05, "loss": 0.031, "step": 32850 }, { "epoch": 14.705750727232044, "grad_norm": 0.12426036596298218, "learning_rate": 4.625564371760791e-05, "loss": 0.0309, "step": 32860 }, { "epoch": 14.710226001342582, "grad_norm": 0.13727645576000214, "learning_rate": 4.622816383934676e-05, "loss": 0.029, "step": 32870 }, { "epoch": 14.714701275453121, "grad_norm": 0.12812651693820953, "learning_rate": 4.620068510686985e-05, "loss": 0.0346, "step": 32880 }, { "epoch": 14.719176549563661, "grad_norm": 0.12856920063495636, "learning_rate": 4.6173207528524476e-05, "loss": 0.033, "step": 32890 }, { "epoch": 14.7236518236742, "grad_norm": 0.1329096406698227, "learning_rate": 4.6145731112657644e-05, "loss": 0.031, "step": 32900 }, { "epoch": 14.72812709778474, "grad_norm": 0.12359577417373657, "learning_rate": 4.611825586761591e-05, "loss": 0.0286, "step": 32910 }, { "epoch": 14.732602371895279, "grad_norm": 0.1315600723028183, "learning_rate": 4.609078180174555e-05, "loss": 0.0303, "step": 32920 }, { "epoch": 14.737077646005817, "grad_norm": 0.10717107355594635, "learning_rate": 4.6063308923392485e-05, "loss": 0.0284, "step": 32930 }, { "epoch": 14.741552920116357, "grad_norm": 0.11980253458023071, "learning_rate": 4.60358372409022e-05, "loss": 0.0303, "step": 32940 }, { "epoch": 14.746028194226897, "grad_norm": 0.12122227996587753, "learning_rate": 4.6008366762619926e-05, "loss": 0.0323, "step": 32950 }, { "epoch": 14.750503468337435, "grad_norm": 0.11414924263954163, "learning_rate": 4.598089749689041e-05, "loss": 0.0315, "step": 32960 }, { "epoch": 14.754978742447975, "grad_norm": 0.1373867690563202, "learning_rate": 4.5953429452058135e-05, "loss": 0.0291, "step": 32970 }, { "epoch": 14.759454016558514, "grad_norm": 0.1523166447877884, "learning_rate": 4.5925962636467126e-05, "loss": 0.0278, "step": 32980 }, { "epoch": 14.763929290669054, "grad_norm": 0.11222619563341141, "learning_rate": 4.58984970584611e-05, "loss": 0.0325, "step": 32990 }, { "epoch": 14.768404564779592, "grad_norm": 0.138473242521286, "learning_rate": 4.5871032726383386e-05, "loss": 0.0312, "step": 33000 }, { "epoch": 14.772879838890132, "grad_norm": 0.11273611336946487, "learning_rate": 4.5843569648576913e-05, "loss": 0.0277, "step": 33010 }, { "epoch": 14.777355113000672, "grad_norm": 0.10784223675727844, "learning_rate": 4.5816107833384234e-05, "loss": 0.031, "step": 33020 }, { "epoch": 14.78183038711121, "grad_norm": 0.1283324956893921, "learning_rate": 4.5788647289147516e-05, "loss": 0.0301, "step": 33030 }, { "epoch": 14.78630566122175, "grad_norm": 0.12388399988412857, "learning_rate": 4.576118802420856e-05, "loss": 0.0303, "step": 33040 }, { "epoch": 14.79078093533229, "grad_norm": 0.11389532685279846, "learning_rate": 4.573373004690878e-05, "loss": 0.0288, "step": 33050 }, { "epoch": 14.795256209442828, "grad_norm": 0.13880857825279236, "learning_rate": 4.570627336558915e-05, "loss": 0.0311, "step": 33060 }, { "epoch": 14.799731483553368, "grad_norm": 0.12674002349376678, "learning_rate": 4.567881798859032e-05, "loss": 0.0322, "step": 33070 }, { "epoch": 14.804206757663907, "grad_norm": 0.11833342164754868, "learning_rate": 4.565136392425247e-05, "loss": 0.0288, "step": 33080 }, { "epoch": 14.808682031774445, "grad_norm": 0.13420405983924866, "learning_rate": 4.562391118091544e-05, "loss": 0.0307, "step": 33090 }, { "epoch": 14.813157305884985, "grad_norm": 0.13333390653133392, "learning_rate": 4.559645976691868e-05, "loss": 0.0292, "step": 33100 }, { "epoch": 14.817632579995525, "grad_norm": 0.1485365331172943, "learning_rate": 4.5569009690601136e-05, "loss": 0.0318, "step": 33110 }, { "epoch": 14.822107854106065, "grad_norm": 0.1541544646024704, "learning_rate": 4.554156096030149e-05, "loss": 0.0312, "step": 33120 }, { "epoch": 14.826583128216603, "grad_norm": 0.1377989798784256, "learning_rate": 4.5514113584357873e-05, "loss": 0.0318, "step": 33130 }, { "epoch": 14.831058402327143, "grad_norm": 0.12795640528202057, "learning_rate": 4.548666757110812e-05, "loss": 0.0324, "step": 33140 }, { "epoch": 14.835533676437683, "grad_norm": 0.1303573101758957, "learning_rate": 4.545922292888959e-05, "loss": 0.0313, "step": 33150 }, { "epoch": 14.84000895054822, "grad_norm": 0.10543419420719147, "learning_rate": 4.543177966603925e-05, "loss": 0.0308, "step": 33160 }, { "epoch": 14.84448422465876, "grad_norm": 0.15393780171871185, "learning_rate": 4.5404337790893626e-05, "loss": 0.0322, "step": 33170 }, { "epoch": 14.8489594987693, "grad_norm": 0.12940755486488342, "learning_rate": 4.537689731178883e-05, "loss": 0.0297, "step": 33180 }, { "epoch": 14.853434772879838, "grad_norm": 0.14517442882061005, "learning_rate": 4.534945823706056e-05, "loss": 0.0266, "step": 33190 }, { "epoch": 14.857910046990378, "grad_norm": 0.11749600619077682, "learning_rate": 4.5322020575044114e-05, "loss": 0.0349, "step": 33200 }, { "epoch": 14.862385321100918, "grad_norm": 0.12712909281253815, "learning_rate": 4.529458433407429e-05, "loss": 0.0285, "step": 33210 }, { "epoch": 14.866860595211456, "grad_norm": 0.11094397306442261, "learning_rate": 4.526714952248551e-05, "loss": 0.0285, "step": 33220 }, { "epoch": 14.871335869321996, "grad_norm": 0.12098449468612671, "learning_rate": 4.5239716148611724e-05, "loss": 0.0292, "step": 33230 }, { "epoch": 14.875811143432536, "grad_norm": 0.12406046688556671, "learning_rate": 4.5212284220786494e-05, "loss": 0.0324, "step": 33240 }, { "epoch": 14.880286417543074, "grad_norm": 0.09281608462333679, "learning_rate": 4.518485374734292e-05, "loss": 0.028, "step": 33250 }, { "epoch": 14.884761691653614, "grad_norm": 0.12825612723827362, "learning_rate": 4.515742473661362e-05, "loss": 0.0308, "step": 33260 }, { "epoch": 14.889236965764153, "grad_norm": 0.1347646266222, "learning_rate": 4.5129997196930845e-05, "loss": 0.0316, "step": 33270 }, { "epoch": 14.893712239874692, "grad_norm": 0.13335227966308594, "learning_rate": 4.510257113662632e-05, "loss": 0.036, "step": 33280 }, { "epoch": 14.898187513985231, "grad_norm": 0.1530974954366684, "learning_rate": 4.507514656403137e-05, "loss": 0.0297, "step": 33290 }, { "epoch": 14.902662788095771, "grad_norm": 0.10594990104436874, "learning_rate": 4.504772348747687e-05, "loss": 0.0282, "step": 33300 }, { "epoch": 14.907138062206311, "grad_norm": 0.12245457619428635, "learning_rate": 4.5020301915293214e-05, "loss": 0.0275, "step": 33310 }, { "epoch": 14.911613336316849, "grad_norm": 0.12244008481502533, "learning_rate": 4.4992881855810366e-05, "loss": 0.0337, "step": 33320 }, { "epoch": 14.916088610427389, "grad_norm": 0.13025809824466705, "learning_rate": 4.496546331735778e-05, "loss": 0.0318, "step": 33330 }, { "epoch": 14.920563884537929, "grad_norm": 0.11410422623157501, "learning_rate": 4.4938046308264544e-05, "loss": 0.0276, "step": 33340 }, { "epoch": 14.925039158648467, "grad_norm": 0.11061383038759232, "learning_rate": 4.491063083685916e-05, "loss": 0.0289, "step": 33350 }, { "epoch": 14.929514432759007, "grad_norm": 0.12102396786212921, "learning_rate": 4.488321691146975e-05, "loss": 0.0311, "step": 33360 }, { "epoch": 14.933989706869546, "grad_norm": 0.1155838891863823, "learning_rate": 4.4855804540423964e-05, "loss": 0.0318, "step": 33370 }, { "epoch": 14.938464980980084, "grad_norm": 0.15328601002693176, "learning_rate": 4.482839373204891e-05, "loss": 0.0291, "step": 33380 }, { "epoch": 14.942940255090624, "grad_norm": 0.1184406504034996, "learning_rate": 4.480098449467132e-05, "loss": 0.0287, "step": 33390 }, { "epoch": 14.947415529201164, "grad_norm": 0.13852618634700775, "learning_rate": 4.477357683661734e-05, "loss": 0.0305, "step": 33400 }, { "epoch": 14.951890803311702, "grad_norm": 0.11070691049098969, "learning_rate": 4.474617076621272e-05, "loss": 0.0288, "step": 33410 }, { "epoch": 14.956366077422242, "grad_norm": 0.12366168200969696, "learning_rate": 4.471876629178273e-05, "loss": 0.0312, "step": 33420 }, { "epoch": 14.960841351532782, "grad_norm": 0.16598984599113464, "learning_rate": 4.469136342165207e-05, "loss": 0.0313, "step": 33430 }, { "epoch": 14.96531662564332, "grad_norm": 0.14061176776885986, "learning_rate": 4.4663962164145045e-05, "loss": 0.028, "step": 33440 }, { "epoch": 14.96979189975386, "grad_norm": 0.11137966811656952, "learning_rate": 4.463656252758542e-05, "loss": 0.0308, "step": 33450 }, { "epoch": 14.9742671738644, "grad_norm": 0.10475951433181763, "learning_rate": 4.46091645202965e-05, "loss": 0.0317, "step": 33460 }, { "epoch": 14.978742447974938, "grad_norm": 0.12989367544651031, "learning_rate": 4.4581768150601055e-05, "loss": 0.0331, "step": 33470 }, { "epoch": 14.983217722085477, "grad_norm": 0.1188843846321106, "learning_rate": 4.4554373426821374e-05, "loss": 0.0274, "step": 33480 }, { "epoch": 14.987692996196017, "grad_norm": 0.11549718677997589, "learning_rate": 4.452698035727929e-05, "loss": 0.0322, "step": 33490 }, { "epoch": 14.992168270306557, "grad_norm": 0.11638665944337845, "learning_rate": 4.449958895029604e-05, "loss": 0.0306, "step": 33500 }, { "epoch": 14.996643544417095, "grad_norm": 0.13507023453712463, "learning_rate": 4.447219921419244e-05, "loss": 0.029, "step": 33510 }, { "epoch": 15.001118818527635, "grad_norm": 0.13740850985050201, "learning_rate": 4.444481115728878e-05, "loss": 0.0263, "step": 33520 }, { "epoch": 15.005594092638175, "grad_norm": 0.15614794194698334, "learning_rate": 4.441742478790481e-05, "loss": 0.0282, "step": 33530 }, { "epoch": 15.010069366748713, "grad_norm": 0.129995197057724, "learning_rate": 4.439004011435979e-05, "loss": 0.0281, "step": 33540 }, { "epoch": 15.014544640859253, "grad_norm": 0.12246780097484589, "learning_rate": 4.436265714497245e-05, "loss": 0.0276, "step": 33550 }, { "epoch": 15.019019914969792, "grad_norm": 0.11661151796579361, "learning_rate": 4.433527588806103e-05, "loss": 0.0293, "step": 33560 }, { "epoch": 15.02349518908033, "grad_norm": 0.12105749547481537, "learning_rate": 4.430789635194324e-05, "loss": 0.0311, "step": 33570 }, { "epoch": 15.02797046319087, "grad_norm": 0.12071439623832703, "learning_rate": 4.428051854493623e-05, "loss": 0.0309, "step": 33580 }, { "epoch": 15.03244573730141, "grad_norm": 0.12862898409366608, "learning_rate": 4.425314247535668e-05, "loss": 0.0312, "step": 33590 }, { "epoch": 15.036921011411948, "grad_norm": 0.11985460668802261, "learning_rate": 4.4225768151520694e-05, "loss": 0.0311, "step": 33600 }, { "epoch": 15.041396285522488, "grad_norm": 0.12248274683952332, "learning_rate": 4.41983955817439e-05, "loss": 0.0321, "step": 33610 }, { "epoch": 15.045871559633028, "grad_norm": 0.12388491630554199, "learning_rate": 4.4171024774341346e-05, "loss": 0.0278, "step": 33620 }, { "epoch": 15.050346833743566, "grad_norm": 0.11845362186431885, "learning_rate": 4.414365573762755e-05, "loss": 0.0305, "step": 33630 }, { "epoch": 15.054822107854106, "grad_norm": 0.11820126324892044, "learning_rate": 4.411628847991653e-05, "loss": 0.0304, "step": 33640 }, { "epoch": 15.059297381964646, "grad_norm": 0.13109911978244781, "learning_rate": 4.408892300952171e-05, "loss": 0.0276, "step": 33650 }, { "epoch": 15.063772656075185, "grad_norm": 0.10942456126213074, "learning_rate": 4.406155933475599e-05, "loss": 0.0303, "step": 33660 }, { "epoch": 15.068247930185724, "grad_norm": 0.13067620992660522, "learning_rate": 4.4034197463931774e-05, "loss": 0.0266, "step": 33670 }, { "epoch": 15.072723204296263, "grad_norm": 0.12353359907865524, "learning_rate": 4.400683740536083e-05, "loss": 0.0296, "step": 33680 }, { "epoch": 15.077198478406803, "grad_norm": 0.11504416167736053, "learning_rate": 4.3979479167354477e-05, "loss": 0.0294, "step": 33690 }, { "epoch": 15.081673752517341, "grad_norm": 0.12461525201797485, "learning_rate": 4.3952122758223354e-05, "loss": 0.0272, "step": 33700 }, { "epoch": 15.086149026627881, "grad_norm": 0.10198252648115158, "learning_rate": 4.392476818627765e-05, "loss": 0.0302, "step": 33710 }, { "epoch": 15.09062430073842, "grad_norm": 0.14661996066570282, "learning_rate": 4.3897415459827e-05, "loss": 0.0338, "step": 33720 }, { "epoch": 15.095099574848959, "grad_norm": 0.1422874480485916, "learning_rate": 4.387006458718037e-05, "loss": 0.0317, "step": 33730 }, { "epoch": 15.099574848959499, "grad_norm": 0.1334601789712906, "learning_rate": 4.384271557664628e-05, "loss": 0.0279, "step": 33740 }, { "epoch": 15.104050123070039, "grad_norm": 0.10640275478363037, "learning_rate": 4.381536843653262e-05, "loss": 0.0302, "step": 33750 }, { "epoch": 15.108525397180577, "grad_norm": 0.12948548793792725, "learning_rate": 4.3788023175146747e-05, "loss": 0.0286, "step": 33760 }, { "epoch": 15.113000671291116, "grad_norm": 0.11232514679431915, "learning_rate": 4.3760679800795396e-05, "loss": 0.0319, "step": 33770 }, { "epoch": 15.117475945401656, "grad_norm": 0.11743997782468796, "learning_rate": 4.373333832178478e-05, "loss": 0.0281, "step": 33780 }, { "epoch": 15.121951219512194, "grad_norm": 0.11485754698514938, "learning_rate": 4.370599874642055e-05, "loss": 0.0311, "step": 33790 }, { "epoch": 15.126426493622734, "grad_norm": 0.14061212539672852, "learning_rate": 4.367866108300769e-05, "loss": 0.0281, "step": 33800 }, { "epoch": 15.130901767733274, "grad_norm": 0.10281524807214737, "learning_rate": 4.365132533985071e-05, "loss": 0.0343, "step": 33810 }, { "epoch": 15.135377041843814, "grad_norm": 0.12237539887428284, "learning_rate": 4.362399152525344e-05, "loss": 0.0286, "step": 33820 }, { "epoch": 15.139852315954352, "grad_norm": 0.13618189096450806, "learning_rate": 4.35966596475192e-05, "loss": 0.0292, "step": 33830 }, { "epoch": 15.144327590064892, "grad_norm": 0.1269778311252594, "learning_rate": 4.3569329714950704e-05, "loss": 0.0316, "step": 33840 }, { "epoch": 15.148802864175432, "grad_norm": 0.13572001457214355, "learning_rate": 4.3542001735850034e-05, "loss": 0.0285, "step": 33850 }, { "epoch": 15.15327813828597, "grad_norm": 0.10910475254058838, "learning_rate": 4.3514675718518734e-05, "loss": 0.0294, "step": 33860 }, { "epoch": 15.15775341239651, "grad_norm": 0.12275724112987518, "learning_rate": 4.348735167125771e-05, "loss": 0.0325, "step": 33870 }, { "epoch": 15.16222868650705, "grad_norm": 0.11117519438266754, "learning_rate": 4.3460029602367284e-05, "loss": 0.031, "step": 33880 }, { "epoch": 15.166703960617587, "grad_norm": 0.14833082258701324, "learning_rate": 4.3432709520147205e-05, "loss": 0.0295, "step": 33890 }, { "epoch": 15.171179234728127, "grad_norm": 0.11448965966701508, "learning_rate": 4.3405391432896555e-05, "loss": 0.0286, "step": 33900 }, { "epoch": 15.175654508838667, "grad_norm": 0.12539628148078918, "learning_rate": 4.3378075348913886e-05, "loss": 0.0306, "step": 33910 }, { "epoch": 15.180129782949205, "grad_norm": 0.12908929586410522, "learning_rate": 4.335076127649707e-05, "loss": 0.0302, "step": 33920 }, { "epoch": 15.184605057059745, "grad_norm": 0.12374712526798248, "learning_rate": 4.3323449223943416e-05, "loss": 0.0283, "step": 33930 }, { "epoch": 15.189080331170285, "grad_norm": 0.14808256924152374, "learning_rate": 4.329613919954962e-05, "loss": 0.0284, "step": 33940 }, { "epoch": 15.193555605280823, "grad_norm": 0.14131830632686615, "learning_rate": 4.3268831211611707e-05, "loss": 0.032, "step": 33950 }, { "epoch": 15.198030879391363, "grad_norm": 0.13159038126468658, "learning_rate": 4.324152526842517e-05, "loss": 0.0316, "step": 33960 }, { "epoch": 15.202506153501902, "grad_norm": 0.1252039521932602, "learning_rate": 4.321422137828479e-05, "loss": 0.032, "step": 33970 }, { "epoch": 15.20698142761244, "grad_norm": 0.13225272297859192, "learning_rate": 4.3186919549484784e-05, "loss": 0.0324, "step": 33980 }, { "epoch": 15.21145670172298, "grad_norm": 0.11962137371301651, "learning_rate": 4.315961979031875e-05, "loss": 0.0294, "step": 33990 }, { "epoch": 15.21593197583352, "grad_norm": 0.11754762381315231, "learning_rate": 4.3132322109079596e-05, "loss": 0.0295, "step": 34000 }, { "epoch": 15.22040724994406, "grad_norm": 0.12378687411546707, "learning_rate": 4.310502651405967e-05, "loss": 0.0297, "step": 34010 }, { "epoch": 15.224882524054598, "grad_norm": 0.09647126495838165, "learning_rate": 4.307773301355062e-05, "loss": 0.0278, "step": 34020 }, { "epoch": 15.229357798165138, "grad_norm": 0.1292782723903656, "learning_rate": 4.305044161584352e-05, "loss": 0.0314, "step": 34030 }, { "epoch": 15.233833072275678, "grad_norm": 0.10593372583389282, "learning_rate": 4.302315232922876e-05, "loss": 0.0286, "step": 34040 }, { "epoch": 15.238308346386216, "grad_norm": 0.10966765880584717, "learning_rate": 4.2995865161996105e-05, "loss": 0.0318, "step": 34050 }, { "epoch": 15.242783620496756, "grad_norm": 0.12288285046815872, "learning_rate": 4.29685801224347e-05, "loss": 0.0308, "step": 34060 }, { "epoch": 15.247258894607295, "grad_norm": 0.07946906238794327, "learning_rate": 4.294129721883298e-05, "loss": 0.0274, "step": 34070 }, { "epoch": 15.251734168717833, "grad_norm": 0.12748223543167114, "learning_rate": 4.291401645947879e-05, "loss": 0.0306, "step": 34080 }, { "epoch": 15.256209442828373, "grad_norm": 0.12754347920417786, "learning_rate": 4.2886737852659325e-05, "loss": 0.0337, "step": 34090 }, { "epoch": 15.260684716938913, "grad_norm": 0.12921297550201416, "learning_rate": 4.2859461406661065e-05, "loss": 0.034, "step": 34100 }, { "epoch": 15.265159991049451, "grad_norm": 0.13133518397808075, "learning_rate": 4.283218712976992e-05, "loss": 0.0314, "step": 34110 }, { "epoch": 15.269635265159991, "grad_norm": 0.12021429091691971, "learning_rate": 4.280491503027104e-05, "loss": 0.0282, "step": 34120 }, { "epoch": 15.27411053927053, "grad_norm": 0.13338862359523773, "learning_rate": 4.2777645116449004e-05, "loss": 0.0259, "step": 34130 }, { "epoch": 15.278585813381069, "grad_norm": 0.12844784557819366, "learning_rate": 4.275037739658771e-05, "loss": 0.03, "step": 34140 }, { "epoch": 15.283061087491609, "grad_norm": 0.12235454469919205, "learning_rate": 4.2723111878970326e-05, "loss": 0.0282, "step": 34150 }, { "epoch": 15.287536361602148, "grad_norm": 0.10605985671281815, "learning_rate": 4.269584857187943e-05, "loss": 0.0315, "step": 34160 }, { "epoch": 15.292011635712687, "grad_norm": 0.12198268622159958, "learning_rate": 4.2668587483596864e-05, "loss": 0.029, "step": 34170 }, { "epoch": 15.296486909823226, "grad_norm": 0.11223436892032623, "learning_rate": 4.264132862240387e-05, "loss": 0.0293, "step": 34180 }, { "epoch": 15.300962183933766, "grad_norm": 0.11842066049575806, "learning_rate": 4.261407199658093e-05, "loss": 0.0289, "step": 34190 }, { "epoch": 15.305437458044306, "grad_norm": 0.11673494428396225, "learning_rate": 4.2586817614407895e-05, "loss": 0.0277, "step": 34200 }, { "epoch": 15.309912732154844, "grad_norm": 0.14125730097293854, "learning_rate": 4.255956548416397e-05, "loss": 0.0346, "step": 34210 }, { "epoch": 15.314388006265384, "grad_norm": 0.09566212445497513, "learning_rate": 4.253231561412756e-05, "loss": 0.031, "step": 34220 }, { "epoch": 15.318863280375924, "grad_norm": 0.12486616522073746, "learning_rate": 4.250506801257653e-05, "loss": 0.0315, "step": 34230 }, { "epoch": 15.323338554486462, "grad_norm": 0.11415784806013107, "learning_rate": 4.247782268778791e-05, "loss": 0.0271, "step": 34240 }, { "epoch": 15.327813828597002, "grad_norm": 0.1510656177997589, "learning_rate": 4.2450579648038154e-05, "loss": 0.0297, "step": 34250 }, { "epoch": 15.332289102707541, "grad_norm": 0.13823893666267395, "learning_rate": 4.2423338901602985e-05, "loss": 0.0316, "step": 34260 }, { "epoch": 15.33676437681808, "grad_norm": 0.12835584580898285, "learning_rate": 4.239610045675739e-05, "loss": 0.0278, "step": 34270 }, { "epoch": 15.34123965092862, "grad_norm": 0.12108340114355087, "learning_rate": 4.236886432177572e-05, "loss": 0.0272, "step": 34280 }, { "epoch": 15.34571492503916, "grad_norm": 0.11946593970060349, "learning_rate": 4.234163050493158e-05, "loss": 0.0275, "step": 34290 }, { "epoch": 15.350190199149697, "grad_norm": 0.11551414430141449, "learning_rate": 4.231439901449788e-05, "loss": 0.0317, "step": 34300 }, { "epoch": 15.354665473260237, "grad_norm": 0.11573553830385208, "learning_rate": 4.2287169858746836e-05, "loss": 0.0307, "step": 34310 }, { "epoch": 15.359140747370777, "grad_norm": 0.09707608819007874, "learning_rate": 4.2259943045949934e-05, "loss": 0.0288, "step": 34320 }, { "epoch": 15.363616021481315, "grad_norm": 0.13951323926448822, "learning_rate": 4.223271858437799e-05, "loss": 0.0294, "step": 34330 }, { "epoch": 15.368091295591855, "grad_norm": 0.10796412080526352, "learning_rate": 4.220549648230104e-05, "loss": 0.0276, "step": 34340 }, { "epoch": 15.372566569702395, "grad_norm": 0.10531069338321686, "learning_rate": 4.2178276747988446e-05, "loss": 0.0287, "step": 34350 }, { "epoch": 15.377041843812934, "grad_norm": 0.15002453327178955, "learning_rate": 4.215105938970889e-05, "loss": 0.0293, "step": 34360 }, { "epoch": 15.381517117923472, "grad_norm": 0.13020622730255127, "learning_rate": 4.212384441573023e-05, "loss": 0.0306, "step": 34370 }, { "epoch": 15.385992392034012, "grad_norm": 0.1159132868051529, "learning_rate": 4.209663183431969e-05, "loss": 0.03, "step": 34380 }, { "epoch": 15.390467666144552, "grad_norm": 0.11981093883514404, "learning_rate": 4.2069421653743706e-05, "loss": 0.0289, "step": 34390 }, { "epoch": 15.39494294025509, "grad_norm": 0.12263787537813187, "learning_rate": 4.2042213882268025e-05, "loss": 0.033, "step": 34400 }, { "epoch": 15.39941821436563, "grad_norm": 0.10852167010307312, "learning_rate": 4.201500852815768e-05, "loss": 0.0291, "step": 34410 }, { "epoch": 15.40389348847617, "grad_norm": 0.1420011818408966, "learning_rate": 4.1987805599676896e-05, "loss": 0.0291, "step": 34420 }, { "epoch": 15.408368762586708, "grad_norm": 0.13816672563552856, "learning_rate": 4.196060510508922e-05, "loss": 0.0301, "step": 34430 }, { "epoch": 15.412844036697248, "grad_norm": 0.15476442873477936, "learning_rate": 4.1933407052657456e-05, "loss": 0.0329, "step": 34440 }, { "epoch": 15.417319310807788, "grad_norm": 0.15688054263591766, "learning_rate": 4.190621145064363e-05, "loss": 0.0279, "step": 34450 }, { "epoch": 15.421794584918326, "grad_norm": 0.11421308666467667, "learning_rate": 4.187901830730906e-05, "loss": 0.0308, "step": 34460 }, { "epoch": 15.426269859028865, "grad_norm": 0.12730462849140167, "learning_rate": 4.1851827630914305e-05, "loss": 0.0299, "step": 34470 }, { "epoch": 15.430745133139405, "grad_norm": 0.1456589698791504, "learning_rate": 4.18246394297192e-05, "loss": 0.0293, "step": 34480 }, { "epoch": 15.435220407249943, "grad_norm": 0.12228720635175705, "learning_rate": 4.179745371198276e-05, "loss": 0.0282, "step": 34490 }, { "epoch": 15.439695681360483, "grad_norm": 0.11615891754627228, "learning_rate": 4.17702704859633e-05, "loss": 0.0339, "step": 34500 }, { "epoch": 15.444170955471023, "grad_norm": 0.13306130468845367, "learning_rate": 4.1743089759918394e-05, "loss": 0.0315, "step": 34510 }, { "epoch": 15.448646229581563, "grad_norm": 0.11769947409629822, "learning_rate": 4.171591154210479e-05, "loss": 0.0271, "step": 34520 }, { "epoch": 15.4531215036921, "grad_norm": 0.10950897634029388, "learning_rate": 4.1688735840778546e-05, "loss": 0.0328, "step": 34530 }, { "epoch": 15.45759677780264, "grad_norm": 0.11970655620098114, "learning_rate": 4.166156266419489e-05, "loss": 0.0287, "step": 34540 }, { "epoch": 15.46207205191318, "grad_norm": 0.1207582950592041, "learning_rate": 4.163439202060833e-05, "loss": 0.0274, "step": 34550 }, { "epoch": 15.466547326023719, "grad_norm": 0.10850560665130615, "learning_rate": 4.160722391827262e-05, "loss": 0.0278, "step": 34560 }, { "epoch": 15.471022600134258, "grad_norm": 0.13854126632213593, "learning_rate": 4.158005836544066e-05, "loss": 0.0345, "step": 34570 }, { "epoch": 15.475497874244798, "grad_norm": 0.13773860037326813, "learning_rate": 4.155289537036466e-05, "loss": 0.0315, "step": 34580 }, { "epoch": 15.479973148355336, "grad_norm": 0.1091095581650734, "learning_rate": 4.1525734941296026e-05, "loss": 0.033, "step": 34590 }, { "epoch": 15.484448422465876, "grad_norm": 0.11441115289926529, "learning_rate": 4.149857708648536e-05, "loss": 0.0267, "step": 34600 }, { "epoch": 15.488923696576416, "grad_norm": 0.1261981576681137, "learning_rate": 4.147142181418249e-05, "loss": 0.0311, "step": 34610 }, { "epoch": 15.493398970686954, "grad_norm": 0.12656733393669128, "learning_rate": 4.14442691326365e-05, "loss": 0.0313, "step": 34620 }, { "epoch": 15.497874244797494, "grad_norm": 0.12071989476680756, "learning_rate": 4.141711905009566e-05, "loss": 0.0312, "step": 34630 }, { "epoch": 15.502349518908034, "grad_norm": 0.15146178007125854, "learning_rate": 4.1389971574807416e-05, "loss": 0.0287, "step": 34640 }, { "epoch": 15.506824793018572, "grad_norm": 0.14288444817066193, "learning_rate": 4.13628267150185e-05, "loss": 0.0298, "step": 34650 }, { "epoch": 15.511300067129111, "grad_norm": 0.13101017475128174, "learning_rate": 4.1335684478974744e-05, "loss": 0.0261, "step": 34660 }, { "epoch": 15.515775341239651, "grad_norm": 0.124989815056324, "learning_rate": 4.130854487492128e-05, "loss": 0.0348, "step": 34670 }, { "epoch": 15.52025061535019, "grad_norm": 0.11102715134620667, "learning_rate": 4.1281407911102425e-05, "loss": 0.0263, "step": 34680 }, { "epoch": 15.52472588946073, "grad_norm": 0.10988423973321915, "learning_rate": 4.125427359576162e-05, "loss": 0.0311, "step": 34690 }, { "epoch": 15.529201163571269, "grad_norm": 0.09577330946922302, "learning_rate": 4.12271419371416e-05, "loss": 0.0311, "step": 34700 }, { "epoch": 15.533676437681809, "grad_norm": 0.13184860348701477, "learning_rate": 4.120001294348421e-05, "loss": 0.0291, "step": 34710 }, { "epoch": 15.538151711792347, "grad_norm": 0.13182798027992249, "learning_rate": 4.1172886623030526e-05, "loss": 0.0327, "step": 34720 }, { "epoch": 15.542626985902887, "grad_norm": 0.11751872301101685, "learning_rate": 4.114576298402084e-05, "loss": 0.0303, "step": 34730 }, { "epoch": 15.547102260013427, "grad_norm": 0.11740916222333908, "learning_rate": 4.111864203469457e-05, "loss": 0.0307, "step": 34740 }, { "epoch": 15.551577534123965, "grad_norm": 0.1256125420331955, "learning_rate": 4.109152378329036e-05, "loss": 0.0288, "step": 34750 }, { "epoch": 15.556052808234504, "grad_norm": 0.1188087910413742, "learning_rate": 4.1064408238045994e-05, "loss": 0.0328, "step": 34760 }, { "epoch": 15.560528082345044, "grad_norm": 0.11635129153728485, "learning_rate": 4.103729540719847e-05, "loss": 0.0329, "step": 34770 }, { "epoch": 15.565003356455582, "grad_norm": 0.1130930706858635, "learning_rate": 4.1010185298983984e-05, "loss": 0.0321, "step": 34780 }, { "epoch": 15.569478630566122, "grad_norm": 0.13788451254367828, "learning_rate": 4.0983077921637815e-05, "loss": 0.031, "step": 34790 }, { "epoch": 15.573953904676662, "grad_norm": 0.13662856817245483, "learning_rate": 4.095597328339452e-05, "loss": 0.0289, "step": 34800 }, { "epoch": 15.5784291787872, "grad_norm": 0.14358358085155487, "learning_rate": 4.092887139248772e-05, "loss": 0.034, "step": 34810 }, { "epoch": 15.58290445289774, "grad_norm": 0.13497814536094666, "learning_rate": 4.09017722571503e-05, "loss": 0.0289, "step": 34820 }, { "epoch": 15.58737972700828, "grad_norm": 0.12831814587116241, "learning_rate": 4.087467588561424e-05, "loss": 0.0278, "step": 34830 }, { "epoch": 15.591855001118818, "grad_norm": 0.13425730168819427, "learning_rate": 4.08475822861107e-05, "loss": 0.0258, "step": 34840 }, { "epoch": 15.596330275229358, "grad_norm": 0.1484915167093277, "learning_rate": 4.082049146687003e-05, "loss": 0.0279, "step": 34850 }, { "epoch": 15.600805549339897, "grad_norm": 0.1300646811723709, "learning_rate": 4.079340343612165e-05, "loss": 0.0331, "step": 34860 }, { "epoch": 15.605280823450435, "grad_norm": 0.125468447804451, "learning_rate": 4.076631820209422e-05, "loss": 0.0327, "step": 34870 }, { "epoch": 15.609756097560975, "grad_norm": 0.11826711148023605, "learning_rate": 4.0739235773015536e-05, "loss": 0.0302, "step": 34880 }, { "epoch": 15.614231371671515, "grad_norm": 0.12440413236618042, "learning_rate": 4.07121561571125e-05, "loss": 0.0277, "step": 34890 }, { "epoch": 15.618706645782055, "grad_norm": 0.14381590485572815, "learning_rate": 4.0685079362611204e-05, "loss": 0.0313, "step": 34900 }, { "epoch": 15.623181919892593, "grad_norm": 0.11574332416057587, "learning_rate": 4.065800539773683e-05, "loss": 0.0294, "step": 34910 }, { "epoch": 15.627657194003133, "grad_norm": 0.12022630870342255, "learning_rate": 4.063093427071376e-05, "loss": 0.0308, "step": 34920 }, { "epoch": 15.632132468113673, "grad_norm": 0.15203703939914703, "learning_rate": 4.0603865989765504e-05, "loss": 0.0325, "step": 34930 }, { "epoch": 15.63660774222421, "grad_norm": 0.1409400999546051, "learning_rate": 4.0576800563114646e-05, "loss": 0.0341, "step": 34940 }, { "epoch": 15.64108301633475, "grad_norm": 0.14429333806037903, "learning_rate": 4.0549737998983e-05, "loss": 0.0329, "step": 34950 }, { "epoch": 15.64555829044529, "grad_norm": 0.1479385793209076, "learning_rate": 4.05226783055914e-05, "loss": 0.0307, "step": 34960 }, { "epoch": 15.650033564555828, "grad_norm": 0.10798654705286026, "learning_rate": 4.049562149115992e-05, "loss": 0.0288, "step": 34970 }, { "epoch": 15.654508838666368, "grad_norm": 0.1282823234796524, "learning_rate": 4.046856756390767e-05, "loss": 0.03, "step": 34980 }, { "epoch": 15.658984112776908, "grad_norm": 0.1269093155860901, "learning_rate": 4.044151653205292e-05, "loss": 0.0297, "step": 34990 }, { "epoch": 15.663459386887446, "grad_norm": 0.10640934109687805, "learning_rate": 4.0414468403813095e-05, "loss": 0.0316, "step": 35000 }, { "epoch": 15.667934660997986, "grad_norm": 0.11840309947729111, "learning_rate": 4.038742318740465e-05, "loss": 0.0273, "step": 35010 }, { "epoch": 15.672409935108526, "grad_norm": 0.11777601391077042, "learning_rate": 4.036038089104326e-05, "loss": 0.0298, "step": 35020 }, { "epoch": 15.676885209219066, "grad_norm": 0.12459491938352585, "learning_rate": 4.0333341522943614e-05, "loss": 0.0273, "step": 35030 }, { "epoch": 15.681360483329604, "grad_norm": 0.13421577215194702, "learning_rate": 4.0306305091319595e-05, "loss": 0.0266, "step": 35040 }, { "epoch": 15.685835757440143, "grad_norm": 0.12426677346229553, "learning_rate": 4.0279271604384144e-05, "loss": 0.0291, "step": 35050 }, { "epoch": 15.690311031550683, "grad_norm": 0.10267273336648941, "learning_rate": 4.0252241070349304e-05, "loss": 0.0267, "step": 35060 }, { "epoch": 15.694786305661221, "grad_norm": 0.1498933881521225, "learning_rate": 4.0225213497426276e-05, "loss": 0.0286, "step": 35070 }, { "epoch": 15.699261579771761, "grad_norm": 0.1372346580028534, "learning_rate": 4.019818889382528e-05, "loss": 0.0306, "step": 35080 }, { "epoch": 15.703736853882301, "grad_norm": 0.1045532301068306, "learning_rate": 4.0171167267755696e-05, "loss": 0.029, "step": 35090 }, { "epoch": 15.708212127992839, "grad_norm": 0.11310985684394836, "learning_rate": 4.0144148627425993e-05, "loss": 0.026, "step": 35100 }, { "epoch": 15.712687402103379, "grad_norm": 0.10115356743335724, "learning_rate": 4.0117132981043693e-05, "loss": 0.0324, "step": 35110 }, { "epoch": 15.717162676213919, "grad_norm": 0.11376688629388809, "learning_rate": 4.0090120336815474e-05, "loss": 0.0329, "step": 35120 }, { "epoch": 15.721637950324457, "grad_norm": 0.13223955035209656, "learning_rate": 4.006311070294702e-05, "loss": 0.0282, "step": 35130 }, { "epoch": 15.726113224434997, "grad_norm": 0.11750921607017517, "learning_rate": 4.003610408764317e-05, "loss": 0.0336, "step": 35140 }, { "epoch": 15.730588498545536, "grad_norm": 0.1286448836326599, "learning_rate": 4.0009100499107824e-05, "loss": 0.0272, "step": 35150 }, { "epoch": 15.735063772656074, "grad_norm": 0.11419714987277985, "learning_rate": 3.9982099945543945e-05, "loss": 0.0288, "step": 35160 }, { "epoch": 15.739539046766614, "grad_norm": 0.11059560626745224, "learning_rate": 3.99551024351536e-05, "loss": 0.0279, "step": 35170 }, { "epoch": 15.744014320877154, "grad_norm": 0.12814708054065704, "learning_rate": 3.9928107976137906e-05, "loss": 0.0313, "step": 35180 }, { "epoch": 15.748489594987692, "grad_norm": 0.12098269909620285, "learning_rate": 3.9901116576697083e-05, "loss": 0.028, "step": 35190 }, { "epoch": 15.752964869098232, "grad_norm": 0.11606735736131668, "learning_rate": 3.9874128245030404e-05, "loss": 0.0309, "step": 35200 }, { "epoch": 15.757440143208772, "grad_norm": 0.09415601193904877, "learning_rate": 3.984714298933619e-05, "loss": 0.027, "step": 35210 }, { "epoch": 15.761915417319312, "grad_norm": 0.11746934056282043, "learning_rate": 3.982016081781189e-05, "loss": 0.0333, "step": 35220 }, { "epoch": 15.76639069142985, "grad_norm": 0.11381088197231293, "learning_rate": 3.979318173865393e-05, "loss": 0.0308, "step": 35230 }, { "epoch": 15.77086596554039, "grad_norm": 0.12760710716247559, "learning_rate": 3.976620576005786e-05, "loss": 0.0327, "step": 35240 }, { "epoch": 15.77534123965093, "grad_norm": 0.12005200237035751, "learning_rate": 3.973923289021829e-05, "loss": 0.0268, "step": 35250 }, { "epoch": 15.779816513761467, "grad_norm": 0.11406591534614563, "learning_rate": 3.9712263137328836e-05, "loss": 0.0259, "step": 35260 }, { "epoch": 15.784291787872007, "grad_norm": 0.149457186460495, "learning_rate": 3.9685296509582224e-05, "loss": 0.0309, "step": 35270 }, { "epoch": 15.788767061982547, "grad_norm": 0.1198110282421112, "learning_rate": 3.965833301517017e-05, "loss": 0.0313, "step": 35280 }, { "epoch": 15.793242336093085, "grad_norm": 0.10512330383062363, "learning_rate": 3.963137266228349e-05, "loss": 0.0303, "step": 35290 }, { "epoch": 15.797717610203625, "grad_norm": 0.11016476154327393, "learning_rate": 3.960441545911204e-05, "loss": 0.0322, "step": 35300 }, { "epoch": 15.802192884314165, "grad_norm": 0.13493680953979492, "learning_rate": 3.9577461413844684e-05, "loss": 0.0319, "step": 35310 }, { "epoch": 15.806668158424703, "grad_norm": 0.11974453926086426, "learning_rate": 3.955051053466937e-05, "loss": 0.0292, "step": 35320 }, { "epoch": 15.811143432535243, "grad_norm": 0.13259069621562958, "learning_rate": 3.9523562829773036e-05, "loss": 0.0307, "step": 35330 }, { "epoch": 15.815618706645783, "grad_norm": 0.15214063227176666, "learning_rate": 3.949661830734172e-05, "loss": 0.0299, "step": 35340 }, { "epoch": 15.82009398075632, "grad_norm": 0.11019638180732727, "learning_rate": 3.946967697556042e-05, "loss": 0.028, "step": 35350 }, { "epoch": 15.82456925486686, "grad_norm": 0.13689260184764862, "learning_rate": 3.944273884261322e-05, "loss": 0.0335, "step": 35360 }, { "epoch": 15.8290445289774, "grad_norm": 0.14189884066581726, "learning_rate": 3.9415803916683224e-05, "loss": 0.0293, "step": 35370 }, { "epoch": 15.833519803087938, "grad_norm": 0.15181511640548706, "learning_rate": 3.9388872205952526e-05, "loss": 0.0272, "step": 35380 }, { "epoch": 15.837995077198478, "grad_norm": 0.10840349644422531, "learning_rate": 3.93619437186023e-05, "loss": 0.029, "step": 35390 }, { "epoch": 15.842470351309018, "grad_norm": 0.11795758455991745, "learning_rate": 3.933501846281267e-05, "loss": 0.0276, "step": 35400 }, { "epoch": 15.846945625419558, "grad_norm": 0.12330693751573563, "learning_rate": 3.930809644676283e-05, "loss": 0.0282, "step": 35410 }, { "epoch": 15.851420899530096, "grad_norm": 0.12176439166069031, "learning_rate": 3.928117767863102e-05, "loss": 0.0298, "step": 35420 }, { "epoch": 15.855896173640636, "grad_norm": 0.09381072968244553, "learning_rate": 3.925426216659438e-05, "loss": 0.028, "step": 35430 }, { "epoch": 15.860371447751175, "grad_norm": 0.13560688495635986, "learning_rate": 3.92273499188292e-05, "loss": 0.0316, "step": 35440 }, { "epoch": 15.864846721861714, "grad_norm": 0.10850423574447632, "learning_rate": 3.9200440943510665e-05, "loss": 0.0272, "step": 35450 }, { "epoch": 15.869321995972253, "grad_norm": 0.12954455614089966, "learning_rate": 3.917353524881302e-05, "loss": 0.0317, "step": 35460 }, { "epoch": 15.873797270082793, "grad_norm": 0.11720463633537292, "learning_rate": 3.914663284290952e-05, "loss": 0.0286, "step": 35470 }, { "epoch": 15.878272544193331, "grad_norm": 0.12070638686418533, "learning_rate": 3.9119733733972387e-05, "loss": 0.029, "step": 35480 }, { "epoch": 15.882747818303871, "grad_norm": 0.11484009027481079, "learning_rate": 3.9092837930172884e-05, "loss": 0.0289, "step": 35490 }, { "epoch": 15.88722309241441, "grad_norm": 0.10381428897380829, "learning_rate": 3.9065945439681214e-05, "loss": 0.027, "step": 35500 }, { "epoch": 15.891698366524949, "grad_norm": 0.14204084873199463, "learning_rate": 3.903905627066662e-05, "loss": 0.0273, "step": 35510 }, { "epoch": 15.896173640635489, "grad_norm": 0.1236523985862732, "learning_rate": 3.901217043129735e-05, "loss": 0.0285, "step": 35520 }, { "epoch": 15.900648914746029, "grad_norm": 0.11895988881587982, "learning_rate": 3.898528792974056e-05, "loss": 0.0292, "step": 35530 }, { "epoch": 15.905124188856567, "grad_norm": 0.1613478660583496, "learning_rate": 3.895840877416249e-05, "loss": 0.0302, "step": 35540 }, { "epoch": 15.909599462967106, "grad_norm": 0.1310071349143982, "learning_rate": 3.8931532972728285e-05, "loss": 0.03, "step": 35550 }, { "epoch": 15.914074737077646, "grad_norm": 0.12804530560970306, "learning_rate": 3.890466053360211e-05, "loss": 0.0288, "step": 35560 }, { "epoch": 15.918550011188184, "grad_norm": 0.1129092276096344, "learning_rate": 3.8877791464947136e-05, "loss": 0.0323, "step": 35570 }, { "epoch": 15.923025285298724, "grad_norm": 0.14106561243534088, "learning_rate": 3.8850925774925425e-05, "loss": 0.0269, "step": 35580 }, { "epoch": 15.927500559409264, "grad_norm": 0.14848747849464417, "learning_rate": 3.8824063471698105e-05, "loss": 0.0323, "step": 35590 }, { "epoch": 15.931975833519804, "grad_norm": 0.17037728428840637, "learning_rate": 3.879720456342521e-05, "loss": 0.0282, "step": 35600 }, { "epoch": 15.936451107630342, "grad_norm": 0.10837133973836899, "learning_rate": 3.877034905826577e-05, "loss": 0.0275, "step": 35610 }, { "epoch": 15.940926381740882, "grad_norm": 0.11542152613401413, "learning_rate": 3.87434969643778e-05, "loss": 0.0307, "step": 35620 }, { "epoch": 15.945401655851422, "grad_norm": 0.1193246990442276, "learning_rate": 3.871664828991822e-05, "loss": 0.0313, "step": 35630 }, { "epoch": 15.94987692996196, "grad_norm": 0.12618400156497955, "learning_rate": 3.8689803043043e-05, "loss": 0.027, "step": 35640 }, { "epoch": 15.9543522040725, "grad_norm": 0.11170677095651627, "learning_rate": 3.866296123190696e-05, "loss": 0.0268, "step": 35650 }, { "epoch": 15.95882747818304, "grad_norm": 0.10933439433574677, "learning_rate": 3.863612286466396e-05, "loss": 0.0284, "step": 35660 }, { "epoch": 15.963302752293577, "grad_norm": 0.10662899166345596, "learning_rate": 3.860928794946682e-05, "loss": 0.0268, "step": 35670 }, { "epoch": 15.967778026404117, "grad_norm": 0.1024302989244461, "learning_rate": 3.858245649446721e-05, "loss": 0.0274, "step": 35680 }, { "epoch": 15.972253300514657, "grad_norm": 0.11154831945896149, "learning_rate": 3.855562850781589e-05, "loss": 0.0315, "step": 35690 }, { "epoch": 15.976728574625195, "grad_norm": 0.12937642633914948, "learning_rate": 3.852880399766243e-05, "loss": 0.0295, "step": 35700 }, { "epoch": 15.981203848735735, "grad_norm": 0.11134210973978043, "learning_rate": 3.850198297215543e-05, "loss": 0.0308, "step": 35710 }, { "epoch": 15.985679122846275, "grad_norm": 0.1226847693324089, "learning_rate": 3.8475165439442446e-05, "loss": 0.0275, "step": 35720 }, { "epoch": 15.990154396956815, "grad_norm": 0.14144796133041382, "learning_rate": 3.844835140766988e-05, "loss": 0.0312, "step": 35730 }, { "epoch": 15.994629671067353, "grad_norm": 0.11845991760492325, "learning_rate": 3.842154088498316e-05, "loss": 0.0299, "step": 35740 }, { "epoch": 15.999104945177892, "grad_norm": 0.10662637650966644, "learning_rate": 3.839473387952662e-05, "loss": 0.0264, "step": 35750 }, { "epoch": 16.00358021928843, "grad_norm": 0.12288357317447662, "learning_rate": 3.836793039944349e-05, "loss": 0.0284, "step": 35760 }, { "epoch": 16.00805549339897, "grad_norm": 0.12464229762554169, "learning_rate": 3.834113045287599e-05, "loss": 0.0294, "step": 35770 }, { "epoch": 16.01253076750951, "grad_norm": 0.12403640151023865, "learning_rate": 3.831433404796521e-05, "loss": 0.0285, "step": 35780 }, { "epoch": 16.01700604162005, "grad_norm": 0.146604984998703, "learning_rate": 3.828754119285123e-05, "loss": 0.029, "step": 35790 }, { "epoch": 16.02148131573059, "grad_norm": 0.12214037030935287, "learning_rate": 3.826075189567296e-05, "loss": 0.0246, "step": 35800 }, { "epoch": 16.025956589841126, "grad_norm": 0.1147671639919281, "learning_rate": 3.823396616456833e-05, "loss": 0.0259, "step": 35810 }, { "epoch": 16.030431863951666, "grad_norm": 0.10071864724159241, "learning_rate": 3.820718400767409e-05, "loss": 0.0275, "step": 35820 }, { "epoch": 16.034907138062206, "grad_norm": 0.09052669256925583, "learning_rate": 3.818040543312598e-05, "loss": 0.0273, "step": 35830 }, { "epoch": 16.039382412172746, "grad_norm": 0.09816714376211166, "learning_rate": 3.8153630449058646e-05, "loss": 0.0271, "step": 35840 }, { "epoch": 16.043857686283285, "grad_norm": 0.10085497051477432, "learning_rate": 3.812685906360557e-05, "loss": 0.027, "step": 35850 }, { "epoch": 16.048332960393825, "grad_norm": 0.10589668899774551, "learning_rate": 3.810009128489925e-05, "loss": 0.0306, "step": 35860 }, { "epoch": 16.052808234504365, "grad_norm": 0.1127466931939125, "learning_rate": 3.807332712107097e-05, "loss": 0.0305, "step": 35870 }, { "epoch": 16.0572835086149, "grad_norm": 0.1445375680923462, "learning_rate": 3.8046566580251e-05, "loss": 0.0272, "step": 35880 }, { "epoch": 16.06175878272544, "grad_norm": 0.13234594464302063, "learning_rate": 3.801980967056851e-05, "loss": 0.0299, "step": 35890 }, { "epoch": 16.06623405683598, "grad_norm": 0.11460193246603012, "learning_rate": 3.799305640015152e-05, "loss": 0.025, "step": 35900 }, { "epoch": 16.07070933094652, "grad_norm": 0.10751451551914215, "learning_rate": 3.796630677712697e-05, "loss": 0.0276, "step": 35910 }, { "epoch": 16.07518460505706, "grad_norm": 0.13994240760803223, "learning_rate": 3.793956080962068e-05, "loss": 0.0295, "step": 35920 }, { "epoch": 16.0796598791676, "grad_norm": 0.1051056832075119, "learning_rate": 3.791281850575737e-05, "loss": 0.0289, "step": 35930 }, { "epoch": 16.084135153278137, "grad_norm": 0.11111237853765488, "learning_rate": 3.788607987366069e-05, "loss": 0.028, "step": 35940 }, { "epoch": 16.088610427388677, "grad_norm": 0.12830011546611786, "learning_rate": 3.7859344921453064e-05, "loss": 0.0277, "step": 35950 }, { "epoch": 16.093085701499216, "grad_norm": 0.12916171550750732, "learning_rate": 3.783261365725592e-05, "loss": 0.028, "step": 35960 }, { "epoch": 16.097560975609756, "grad_norm": 0.13066773116588593, "learning_rate": 3.780588608918947e-05, "loss": 0.027, "step": 35970 }, { "epoch": 16.102036249720296, "grad_norm": 0.1313597410917282, "learning_rate": 3.777916222537285e-05, "loss": 0.0309, "step": 35980 }, { "epoch": 16.106511523830836, "grad_norm": 0.1241750419139862, "learning_rate": 3.77524420739241e-05, "loss": 0.0281, "step": 35990 }, { "epoch": 16.110986797941376, "grad_norm": 0.09918716549873352, "learning_rate": 3.772572564296005e-05, "loss": 0.0294, "step": 36000 }, { "epoch": 16.115462072051912, "grad_norm": 0.10692169517278671, "learning_rate": 3.7699012940596477e-05, "loss": 0.0334, "step": 36010 }, { "epoch": 16.11993734616245, "grad_norm": 0.12140136957168579, "learning_rate": 3.767230397494798e-05, "loss": 0.0282, "step": 36020 }, { "epoch": 16.12441262027299, "grad_norm": 0.12761487066745758, "learning_rate": 3.764559875412803e-05, "loss": 0.0292, "step": 36030 }, { "epoch": 16.12888789438353, "grad_norm": 0.134360209107399, "learning_rate": 3.761889728624899e-05, "loss": 0.0275, "step": 36040 }, { "epoch": 16.13336316849407, "grad_norm": 0.11307501792907715, "learning_rate": 3.7592199579422035e-05, "loss": 0.0282, "step": 36050 }, { "epoch": 16.13783844260461, "grad_norm": 0.09561816602945328, "learning_rate": 3.756550564175727e-05, "loss": 0.0318, "step": 36060 }, { "epoch": 16.142313716715147, "grad_norm": 0.1252911537885666, "learning_rate": 3.7538815481363554e-05, "loss": 0.0257, "step": 36070 }, { "epoch": 16.146788990825687, "grad_norm": 0.12241494655609131, "learning_rate": 3.751212910634867e-05, "loss": 0.0289, "step": 36080 }, { "epoch": 16.151264264936227, "grad_norm": 0.10184754431247711, "learning_rate": 3.748544652481927e-05, "loss": 0.027, "step": 36090 }, { "epoch": 16.155739539046767, "grad_norm": 0.12102202326059341, "learning_rate": 3.7458767744880765e-05, "loss": 0.0269, "step": 36100 }, { "epoch": 16.160214813157307, "grad_norm": 0.1082586720585823, "learning_rate": 3.7432092774637504e-05, "loss": 0.0279, "step": 36110 }, { "epoch": 16.164690087267847, "grad_norm": 0.12078297883272171, "learning_rate": 3.74054216221926e-05, "loss": 0.03, "step": 36120 }, { "epoch": 16.169165361378383, "grad_norm": 0.11991987377405167, "learning_rate": 3.737875429564807e-05, "loss": 0.0311, "step": 36130 }, { "epoch": 16.173640635488923, "grad_norm": 0.12828829884529114, "learning_rate": 3.7352090803104765e-05, "loss": 0.0286, "step": 36140 }, { "epoch": 16.178115909599462, "grad_norm": 0.09586569666862488, "learning_rate": 3.73254311526623e-05, "loss": 0.0246, "step": 36150 }, { "epoch": 16.182591183710002, "grad_norm": 0.114891916513443, "learning_rate": 3.7298775352419206e-05, "loss": 0.032, "step": 36160 }, { "epoch": 16.187066457820542, "grad_norm": 0.11297760903835297, "learning_rate": 3.727212341047281e-05, "loss": 0.0263, "step": 36170 }, { "epoch": 16.191541731931082, "grad_norm": 0.09981515258550644, "learning_rate": 3.7245475334919246e-05, "loss": 0.0247, "step": 36180 }, { "epoch": 16.19601700604162, "grad_norm": 0.0971793681383133, "learning_rate": 3.721883113385353e-05, "loss": 0.0293, "step": 36190 }, { "epoch": 16.200492280152158, "grad_norm": 0.09850508719682693, "learning_rate": 3.719219081536942e-05, "loss": 0.0297, "step": 36200 }, { "epoch": 16.204967554262698, "grad_norm": 0.09853540360927582, "learning_rate": 3.716555438755961e-05, "loss": 0.0294, "step": 36210 }, { "epoch": 16.209442828373238, "grad_norm": 0.12143515795469284, "learning_rate": 3.713892185851548e-05, "loss": 0.0325, "step": 36220 }, { "epoch": 16.213918102483778, "grad_norm": 0.1320570558309555, "learning_rate": 3.711229323632732e-05, "loss": 0.0298, "step": 36230 }, { "epoch": 16.218393376594317, "grad_norm": 0.1809171438217163, "learning_rate": 3.7085668529084184e-05, "loss": 0.0285, "step": 36240 }, { "epoch": 16.222868650704857, "grad_norm": 0.12209392338991165, "learning_rate": 3.705904774487396e-05, "loss": 0.0293, "step": 36250 }, { "epoch": 16.227343924815393, "grad_norm": 0.126169815659523, "learning_rate": 3.703243089178337e-05, "loss": 0.0292, "step": 36260 }, { "epoch": 16.231819198925933, "grad_norm": 0.10708770155906677, "learning_rate": 3.700581797789786e-05, "loss": 0.0267, "step": 36270 }, { "epoch": 16.236294473036473, "grad_norm": 0.11154186725616455, "learning_rate": 3.697920901130178e-05, "loss": 0.0299, "step": 36280 }, { "epoch": 16.240769747147013, "grad_norm": 0.11074917763471603, "learning_rate": 3.695260400007819e-05, "loss": 0.0271, "step": 36290 }, { "epoch": 16.245245021257553, "grad_norm": 0.13067054748535156, "learning_rate": 3.6926002952309016e-05, "loss": 0.0322, "step": 36300 }, { "epoch": 16.249720295368093, "grad_norm": 0.13534215092658997, "learning_rate": 3.6899405876074944e-05, "loss": 0.0287, "step": 36310 }, { "epoch": 16.25419556947863, "grad_norm": 0.132888063788414, "learning_rate": 3.687281277945547e-05, "loss": 0.03, "step": 36320 }, { "epoch": 16.25867084358917, "grad_norm": 0.12586714327335358, "learning_rate": 3.684622367052887e-05, "loss": 0.0297, "step": 36330 }, { "epoch": 16.26314611769971, "grad_norm": 0.11870598047971725, "learning_rate": 3.68196385573722e-05, "loss": 0.0279, "step": 36340 }, { "epoch": 16.26762139181025, "grad_norm": 0.10647913813591003, "learning_rate": 3.679305744806134e-05, "loss": 0.0274, "step": 36350 }, { "epoch": 16.272096665920788, "grad_norm": 0.13626894354820251, "learning_rate": 3.676648035067093e-05, "loss": 0.0272, "step": 36360 }, { "epoch": 16.276571940031328, "grad_norm": 0.10295109450817108, "learning_rate": 3.673990727327435e-05, "loss": 0.0274, "step": 36370 }, { "epoch": 16.281047214141868, "grad_norm": 0.10023539513349533, "learning_rate": 3.6713338223943867e-05, "loss": 0.0264, "step": 36380 }, { "epoch": 16.285522488252404, "grad_norm": 0.150278702378273, "learning_rate": 3.6686773210750385e-05, "loss": 0.0331, "step": 36390 }, { "epoch": 16.289997762362944, "grad_norm": 0.12414079159498215, "learning_rate": 3.666021224176369e-05, "loss": 0.0314, "step": 36400 }, { "epoch": 16.294473036473484, "grad_norm": 0.1500927060842514, "learning_rate": 3.663365532505232e-05, "loss": 0.0278, "step": 36410 }, { "epoch": 16.298948310584024, "grad_norm": 0.12025880068540573, "learning_rate": 3.6607102468683526e-05, "loss": 0.0265, "step": 36420 }, { "epoch": 16.303423584694563, "grad_norm": 0.11825131624937057, "learning_rate": 3.658055368072339e-05, "loss": 0.028, "step": 36430 }, { "epoch": 16.307898858805103, "grad_norm": 0.10616494715213776, "learning_rate": 3.655400896923672e-05, "loss": 0.029, "step": 36440 }, { "epoch": 16.31237413291564, "grad_norm": 0.09486425668001175, "learning_rate": 3.65274683422871e-05, "loss": 0.0295, "step": 36450 }, { "epoch": 16.31684940702618, "grad_norm": 0.13141772150993347, "learning_rate": 3.650093180793689e-05, "loss": 0.0301, "step": 36460 }, { "epoch": 16.32132468113672, "grad_norm": 0.14622598886489868, "learning_rate": 3.647439937424717e-05, "loss": 0.0276, "step": 36470 }, { "epoch": 16.32579995524726, "grad_norm": 0.10138409584760666, "learning_rate": 3.6447871049277796e-05, "loss": 0.0285, "step": 36480 }, { "epoch": 16.3302752293578, "grad_norm": 0.11991895735263824, "learning_rate": 3.642134684108737e-05, "loss": 0.0296, "step": 36490 }, { "epoch": 16.33475050346834, "grad_norm": 0.12466270476579666, "learning_rate": 3.639482675773324e-05, "loss": 0.0293, "step": 36500 }, { "epoch": 16.339225777578875, "grad_norm": 0.1216922253370285, "learning_rate": 3.636831080727154e-05, "loss": 0.0283, "step": 36510 }, { "epoch": 16.343701051689415, "grad_norm": 0.11328652501106262, "learning_rate": 3.634179899775708e-05, "loss": 0.0273, "step": 36520 }, { "epoch": 16.348176325799955, "grad_norm": 0.1123642846941948, "learning_rate": 3.631529133724348e-05, "loss": 0.0291, "step": 36530 }, { "epoch": 16.352651599910494, "grad_norm": 0.13414429128170013, "learning_rate": 3.628878783378302e-05, "loss": 0.0293, "step": 36540 }, { "epoch": 16.357126874021034, "grad_norm": 0.11246030032634735, "learning_rate": 3.62622884954268e-05, "loss": 0.0283, "step": 36550 }, { "epoch": 16.361602148131574, "grad_norm": 0.10399256646633148, "learning_rate": 3.6235793330224635e-05, "loss": 0.033, "step": 36560 }, { "epoch": 16.366077422242114, "grad_norm": 0.11634261161088943, "learning_rate": 3.6209302346225006e-05, "loss": 0.0296, "step": 36570 }, { "epoch": 16.37055269635265, "grad_norm": 0.1486562043428421, "learning_rate": 3.618281555147522e-05, "loss": 0.0286, "step": 36580 }, { "epoch": 16.37502797046319, "grad_norm": 0.1093926876783371, "learning_rate": 3.615633295402123e-05, "loss": 0.0298, "step": 36590 }, { "epoch": 16.37950324457373, "grad_norm": 0.10709859430789948, "learning_rate": 3.612985456190778e-05, "loss": 0.0262, "step": 36600 }, { "epoch": 16.38397851868427, "grad_norm": 0.11277598142623901, "learning_rate": 3.610338038317828e-05, "loss": 0.0317, "step": 36610 }, { "epoch": 16.38845379279481, "grad_norm": 0.10346702486276627, "learning_rate": 3.607691042587492e-05, "loss": 0.0268, "step": 36620 }, { "epoch": 16.39292906690535, "grad_norm": 0.10163304954767227, "learning_rate": 3.605044469803854e-05, "loss": 0.0255, "step": 36630 }, { "epoch": 16.397404341015886, "grad_norm": 0.12036854028701782, "learning_rate": 3.602398320770875e-05, "loss": 0.0338, "step": 36640 }, { "epoch": 16.401879615126425, "grad_norm": 0.11368953436613083, "learning_rate": 3.599752596292386e-05, "loss": 0.025, "step": 36650 }, { "epoch": 16.406354889236965, "grad_norm": 0.13927844166755676, "learning_rate": 3.597107297172084e-05, "loss": 0.0279, "step": 36660 }, { "epoch": 16.410830163347505, "grad_norm": 0.13418474793434143, "learning_rate": 3.594462424213545e-05, "loss": 0.0279, "step": 36670 }, { "epoch": 16.415305437458045, "grad_norm": 0.112474225461483, "learning_rate": 3.591817978220212e-05, "loss": 0.0274, "step": 36680 }, { "epoch": 16.419780711568585, "grad_norm": 0.13478009402751923, "learning_rate": 3.5891739599953945e-05, "loss": 0.0285, "step": 36690 }, { "epoch": 16.424255985679125, "grad_norm": 0.10949638485908508, "learning_rate": 3.586530370342279e-05, "loss": 0.0295, "step": 36700 }, { "epoch": 16.42873125978966, "grad_norm": 0.11431214958429337, "learning_rate": 3.5838872100639146e-05, "loss": 0.0278, "step": 36710 }, { "epoch": 16.4332065339002, "grad_norm": 0.142310231924057, "learning_rate": 3.581244479963225e-05, "loss": 0.0312, "step": 36720 }, { "epoch": 16.43768180801074, "grad_norm": 0.1361013501882553, "learning_rate": 3.5786021808430054e-05, "loss": 0.0291, "step": 36730 }, { "epoch": 16.44215708212128, "grad_norm": 0.1308324635028839, "learning_rate": 3.57596031350591e-05, "loss": 0.0266, "step": 36740 }, { "epoch": 16.44663235623182, "grad_norm": 0.11957643181085587, "learning_rate": 3.5733188787544745e-05, "loss": 0.0301, "step": 36750 }, { "epoch": 16.45110763034236, "grad_norm": 0.1243235319852829, "learning_rate": 3.570677877391092e-05, "loss": 0.0296, "step": 36760 }, { "epoch": 16.455582904452896, "grad_norm": 0.10522710531949997, "learning_rate": 3.568037310218033e-05, "loss": 0.0279, "step": 36770 }, { "epoch": 16.460058178563436, "grad_norm": 0.13408392667770386, "learning_rate": 3.5653971780374295e-05, "loss": 0.0269, "step": 36780 }, { "epoch": 16.464533452673976, "grad_norm": 0.12760722637176514, "learning_rate": 3.5627574816512846e-05, "loss": 0.0282, "step": 36790 }, { "epoch": 16.469008726784516, "grad_norm": 0.11744362115859985, "learning_rate": 3.56011822186147e-05, "loss": 0.0303, "step": 36800 }, { "epoch": 16.473484000895056, "grad_norm": 0.10067330300807953, "learning_rate": 3.557479399469721e-05, "loss": 0.0285, "step": 36810 }, { "epoch": 16.477959275005595, "grad_norm": 0.14779295027256012, "learning_rate": 3.554841015277641e-05, "loss": 0.0342, "step": 36820 }, { "epoch": 16.48243454911613, "grad_norm": 0.10188426077365875, "learning_rate": 3.552203070086707e-05, "loss": 0.0297, "step": 36830 }, { "epoch": 16.48690982322667, "grad_norm": 0.12341033667325974, "learning_rate": 3.5495655646982505e-05, "loss": 0.031, "step": 36840 }, { "epoch": 16.49138509733721, "grad_norm": 0.12931331992149353, "learning_rate": 3.5469284999134815e-05, "loss": 0.0286, "step": 36850 }, { "epoch": 16.49586037144775, "grad_norm": 0.11508669704198837, "learning_rate": 3.544291876533466e-05, "loss": 0.0285, "step": 36860 }, { "epoch": 16.50033564555829, "grad_norm": 0.1311425268650055, "learning_rate": 3.541655695359142e-05, "loss": 0.0298, "step": 36870 }, { "epoch": 16.50481091966883, "grad_norm": 0.11485176533460617, "learning_rate": 3.539019957191315e-05, "loss": 0.023, "step": 36880 }, { "epoch": 16.50928619377937, "grad_norm": 0.10720150917768478, "learning_rate": 3.536384662830648e-05, "loss": 0.029, "step": 36890 }, { "epoch": 16.513761467889907, "grad_norm": 0.10013247281312943, "learning_rate": 3.533749813077677e-05, "loss": 0.0311, "step": 36900 }, { "epoch": 16.518236742000447, "grad_norm": 0.11165263503789902, "learning_rate": 3.5311154087327975e-05, "loss": 0.0258, "step": 36910 }, { "epoch": 16.522712016110987, "grad_norm": 0.1103915274143219, "learning_rate": 3.528481450596274e-05, "loss": 0.0281, "step": 36920 }, { "epoch": 16.527187290221526, "grad_norm": 0.09702851623296738, "learning_rate": 3.525847939468233e-05, "loss": 0.0228, "step": 36930 }, { "epoch": 16.531662564332066, "grad_norm": 0.09571859985589981, "learning_rate": 3.523214876148664e-05, "loss": 0.025, "step": 36940 }, { "epoch": 16.536137838442606, "grad_norm": 0.1344972848892212, "learning_rate": 3.5205822614374255e-05, "loss": 0.0305, "step": 36950 }, { "epoch": 16.540613112553142, "grad_norm": 0.11694079637527466, "learning_rate": 3.517950096134232e-05, "loss": 0.0248, "step": 36960 }, { "epoch": 16.545088386663682, "grad_norm": 0.09371545165777206, "learning_rate": 3.515318381038668e-05, "loss": 0.0234, "step": 36970 }, { "epoch": 16.549563660774222, "grad_norm": 0.10675708949565887, "learning_rate": 3.512687116950182e-05, "loss": 0.0298, "step": 36980 }, { "epoch": 16.554038934884762, "grad_norm": 0.13382457196712494, "learning_rate": 3.5100563046680764e-05, "loss": 0.0273, "step": 36990 }, { "epoch": 16.5585142089953, "grad_norm": 0.1241687685251236, "learning_rate": 3.5074259449915284e-05, "loss": 0.0277, "step": 37000 }, { "epoch": 16.56298948310584, "grad_norm": 0.12030500173568726, "learning_rate": 3.504796038719567e-05, "loss": 0.0297, "step": 37010 }, { "epoch": 16.567464757216378, "grad_norm": 0.10767023265361786, "learning_rate": 3.5021665866510925e-05, "loss": 0.0257, "step": 37020 }, { "epoch": 16.571940031326918, "grad_norm": 0.11844927072525024, "learning_rate": 3.499537589584859e-05, "loss": 0.033, "step": 37030 }, { "epoch": 16.576415305437457, "grad_norm": 0.12920306622982025, "learning_rate": 3.496909048319489e-05, "loss": 0.0298, "step": 37040 }, { "epoch": 16.580890579547997, "grad_norm": 0.1400853991508484, "learning_rate": 3.494280963653463e-05, "loss": 0.0275, "step": 37050 }, { "epoch": 16.585365853658537, "grad_norm": 0.11911115050315857, "learning_rate": 3.491653336385124e-05, "loss": 0.0273, "step": 37060 }, { "epoch": 16.589841127769077, "grad_norm": 0.1158897653222084, "learning_rate": 3.489026167312678e-05, "loss": 0.0264, "step": 37070 }, { "epoch": 16.594316401879617, "grad_norm": 0.11474811285734177, "learning_rate": 3.4863994572341843e-05, "loss": 0.0279, "step": 37080 }, { "epoch": 16.598791675990153, "grad_norm": 0.12277276813983917, "learning_rate": 3.483773206947572e-05, "loss": 0.0295, "step": 37090 }, { "epoch": 16.603266950100693, "grad_norm": 0.10958962887525558, "learning_rate": 3.4811474172506275e-05, "loss": 0.0294, "step": 37100 }, { "epoch": 16.607742224211233, "grad_norm": 0.13766519725322723, "learning_rate": 3.478522088940993e-05, "loss": 0.0274, "step": 37110 }, { "epoch": 16.612217498321773, "grad_norm": 0.1312839388847351, "learning_rate": 3.475897222816178e-05, "loss": 0.0307, "step": 37120 }, { "epoch": 16.616692772432312, "grad_norm": 0.13531824946403503, "learning_rate": 3.473272819673542e-05, "loss": 0.0244, "step": 37130 }, { "epoch": 16.621168046542852, "grad_norm": 0.11884219944477081, "learning_rate": 3.470648880310313e-05, "loss": 0.031, "step": 37140 }, { "epoch": 16.62564332065339, "grad_norm": 0.12120656669139862, "learning_rate": 3.468025405523576e-05, "loss": 0.0276, "step": 37150 }, { "epoch": 16.63011859476393, "grad_norm": 0.13882024586200714, "learning_rate": 3.465402396110269e-05, "loss": 0.0294, "step": 37160 }, { "epoch": 16.634593868874468, "grad_norm": 0.1338890790939331, "learning_rate": 3.462779852867197e-05, "loss": 0.0265, "step": 37170 }, { "epoch": 16.639069142985008, "grad_norm": 0.13130781054496765, "learning_rate": 3.460157776591018e-05, "loss": 0.0291, "step": 37180 }, { "epoch": 16.643544417095548, "grad_norm": 0.10723642259836197, "learning_rate": 3.457536168078247e-05, "loss": 0.0292, "step": 37190 }, { "epoch": 16.648019691206088, "grad_norm": 0.10171042382717133, "learning_rate": 3.4549150281252636e-05, "loss": 0.0307, "step": 37200 }, { "epoch": 16.652494965316627, "grad_norm": 0.10492350906133652, "learning_rate": 3.452294357528297e-05, "loss": 0.0286, "step": 37210 }, { "epoch": 16.656970239427164, "grad_norm": 0.11192900687456131, "learning_rate": 3.449674157083443e-05, "loss": 0.028, "step": 37220 }, { "epoch": 16.661445513537704, "grad_norm": 0.11932540684938431, "learning_rate": 3.447054427586644e-05, "loss": 0.0274, "step": 37230 }, { "epoch": 16.665920787648243, "grad_norm": 0.1169833391904831, "learning_rate": 3.444435169833706e-05, "loss": 0.0252, "step": 37240 }, { "epoch": 16.670396061758783, "grad_norm": 0.1008252203464508, "learning_rate": 3.4418163846202944e-05, "loss": 0.0269, "step": 37250 }, { "epoch": 16.674871335869323, "grad_norm": 0.10270050168037415, "learning_rate": 3.439198072741921e-05, "loss": 0.0308, "step": 37260 }, { "epoch": 16.679346609979863, "grad_norm": 0.13072055578231812, "learning_rate": 3.436580234993965e-05, "loss": 0.0297, "step": 37270 }, { "epoch": 16.6838218840904, "grad_norm": 0.12618114054203033, "learning_rate": 3.4339628721716505e-05, "loss": 0.0294, "step": 37280 }, { "epoch": 16.68829715820094, "grad_norm": 0.11230115592479706, "learning_rate": 3.431345985070067e-05, "loss": 0.0294, "step": 37290 }, { "epoch": 16.69277243231148, "grad_norm": 0.09165281057357788, "learning_rate": 3.4287295744841586e-05, "loss": 0.0299, "step": 37300 }, { "epoch": 16.69724770642202, "grad_norm": 0.12507502734661102, "learning_rate": 3.4261136412087155e-05, "loss": 0.028, "step": 37310 }, { "epoch": 16.70172298053256, "grad_norm": 0.11701013892889023, "learning_rate": 3.423498186038393e-05, "loss": 0.0265, "step": 37320 }, { "epoch": 16.7061982546431, "grad_norm": 0.11322382837533951, "learning_rate": 3.420883209767697e-05, "loss": 0.0319, "step": 37330 }, { "epoch": 16.710673528753635, "grad_norm": 0.12204425781965256, "learning_rate": 3.418268713190986e-05, "loss": 0.0279, "step": 37340 }, { "epoch": 16.715148802864174, "grad_norm": 0.09810697287321091, "learning_rate": 3.4156546971024784e-05, "loss": 0.0295, "step": 37350 }, { "epoch": 16.719624076974714, "grad_norm": 0.13618606328964233, "learning_rate": 3.413041162296241e-05, "loss": 0.0303, "step": 37360 }, { "epoch": 16.724099351085254, "grad_norm": 0.12427922338247299, "learning_rate": 3.410428109566198e-05, "loss": 0.0283, "step": 37370 }, { "epoch": 16.728574625195794, "grad_norm": 0.1498093605041504, "learning_rate": 3.407815539706124e-05, "loss": 0.0255, "step": 37380 }, { "epoch": 16.733049899306334, "grad_norm": 0.10313691943883896, "learning_rate": 3.40520345350965e-05, "loss": 0.0275, "step": 37390 }, { "epoch": 16.73752517341687, "grad_norm": 0.11117719113826752, "learning_rate": 3.40259185177026e-05, "loss": 0.0262, "step": 37400 }, { "epoch": 16.74200044752741, "grad_norm": 0.11073888838291168, "learning_rate": 3.399980735281286e-05, "loss": 0.0252, "step": 37410 }, { "epoch": 16.74647572163795, "grad_norm": 0.12477529793977737, "learning_rate": 3.397370104835922e-05, "loss": 0.0304, "step": 37420 }, { "epoch": 16.75095099574849, "grad_norm": 0.11304154992103577, "learning_rate": 3.394759961227202e-05, "loss": 0.0274, "step": 37430 }, { "epoch": 16.75542626985903, "grad_norm": 0.11466948688030243, "learning_rate": 3.392150305248024e-05, "loss": 0.0255, "step": 37440 }, { "epoch": 16.75990154396957, "grad_norm": 0.10938846319913864, "learning_rate": 3.389541137691129e-05, "loss": 0.0285, "step": 37450 }, { "epoch": 16.76437681808011, "grad_norm": 0.12363659590482712, "learning_rate": 3.386932459349114e-05, "loss": 0.0287, "step": 37460 }, { "epoch": 16.768852092190645, "grad_norm": 0.11495689302682877, "learning_rate": 3.384324271014429e-05, "loss": 0.0266, "step": 37470 }, { "epoch": 16.773327366301185, "grad_norm": 0.12313652783632278, "learning_rate": 3.3817165734793705e-05, "loss": 0.0271, "step": 37480 }, { "epoch": 16.777802640411725, "grad_norm": 0.09508269280195236, "learning_rate": 3.379109367536089e-05, "loss": 0.0261, "step": 37490 }, { "epoch": 16.782277914522265, "grad_norm": 0.10206080973148346, "learning_rate": 3.3765026539765834e-05, "loss": 0.0276, "step": 37500 }, { "epoch": 16.786753188632805, "grad_norm": 0.11330022662878036, "learning_rate": 3.373896433592705e-05, "loss": 0.0264, "step": 37510 }, { "epoch": 16.791228462743344, "grad_norm": 0.1113836020231247, "learning_rate": 3.371290707176158e-05, "loss": 0.0282, "step": 37520 }, { "epoch": 16.79570373685388, "grad_norm": 0.11587876081466675, "learning_rate": 3.368685475518488e-05, "loss": 0.0259, "step": 37530 }, { "epoch": 16.80017901096442, "grad_norm": 0.10130292177200317, "learning_rate": 3.366080739411101e-05, "loss": 0.0267, "step": 37540 }, { "epoch": 16.80465428507496, "grad_norm": 0.10408546775579453, "learning_rate": 3.363476499645241e-05, "loss": 0.0319, "step": 37550 }, { "epoch": 16.8091295591855, "grad_norm": 0.1128779947757721, "learning_rate": 3.360872757012011e-05, "loss": 0.0294, "step": 37560 }, { "epoch": 16.81360483329604, "grad_norm": 0.09926128387451172, "learning_rate": 3.358269512302361e-05, "loss": 0.0249, "step": 37570 }, { "epoch": 16.81808010740658, "grad_norm": 0.12033192068338394, "learning_rate": 3.355666766307084e-05, "loss": 0.0275, "step": 37580 }, { "epoch": 16.82255538151712, "grad_norm": 0.10222283750772476, "learning_rate": 3.3530645198168295e-05, "loss": 0.0259, "step": 37590 }, { "epoch": 16.827030655627656, "grad_norm": 0.11162515729665756, "learning_rate": 3.350462773622086e-05, "loss": 0.0266, "step": 37600 }, { "epoch": 16.831505929738196, "grad_norm": 0.13603736460208893, "learning_rate": 3.3478615285131994e-05, "loss": 0.0259, "step": 37610 }, { "epoch": 16.835981203848736, "grad_norm": 0.12141349166631699, "learning_rate": 3.3452607852803584e-05, "loss": 0.0283, "step": 37620 }, { "epoch": 16.840456477959275, "grad_norm": 0.10436423122882843, "learning_rate": 3.3426605447136004e-05, "loss": 0.0276, "step": 37630 }, { "epoch": 16.844931752069815, "grad_norm": 0.09157493710517883, "learning_rate": 3.3400608076028094e-05, "loss": 0.027, "step": 37640 }, { "epoch": 16.849407026180355, "grad_norm": 0.12361376732587814, "learning_rate": 3.337461574737716e-05, "loss": 0.028, "step": 37650 }, { "epoch": 16.85388230029089, "grad_norm": 0.11201279610395432, "learning_rate": 3.3348628469079e-05, "loss": 0.0286, "step": 37660 }, { "epoch": 16.85835757440143, "grad_norm": 0.08892941474914551, "learning_rate": 3.332264624902787e-05, "loss": 0.0237, "step": 37670 }, { "epoch": 16.86283284851197, "grad_norm": 0.10786428302526474, "learning_rate": 3.329666909511645e-05, "loss": 0.0257, "step": 37680 }, { "epoch": 16.86730812262251, "grad_norm": 0.10265939682722092, "learning_rate": 3.327069701523595e-05, "loss": 0.0273, "step": 37690 }, { "epoch": 16.87178339673305, "grad_norm": 0.12330908328294754, "learning_rate": 3.324473001727597e-05, "loss": 0.0254, "step": 37700 }, { "epoch": 16.87625867084359, "grad_norm": 0.13853390514850616, "learning_rate": 3.321876810912461e-05, "loss": 0.0303, "step": 37710 }, { "epoch": 16.88073394495413, "grad_norm": 0.11704840511083603, "learning_rate": 3.3192811298668434e-05, "loss": 0.0303, "step": 37720 }, { "epoch": 16.885209219064667, "grad_norm": 0.10327155888080597, "learning_rate": 3.316685959379241e-05, "loss": 0.0292, "step": 37730 }, { "epoch": 16.889684493175206, "grad_norm": 0.10627672076225281, "learning_rate": 3.3140913002379995e-05, "loss": 0.0268, "step": 37740 }, { "epoch": 16.894159767285746, "grad_norm": 0.11708403378725052, "learning_rate": 3.3114971532313056e-05, "loss": 0.0304, "step": 37750 }, { "epoch": 16.898635041396286, "grad_norm": 0.10928094387054443, "learning_rate": 3.308903519147194e-05, "loss": 0.0274, "step": 37760 }, { "epoch": 16.903110315506826, "grad_norm": 0.11370643228292465, "learning_rate": 3.3063103987735433e-05, "loss": 0.0293, "step": 37770 }, { "epoch": 16.907585589617366, "grad_norm": 0.1272449940443039, "learning_rate": 3.3037177928980735e-05, "loss": 0.0241, "step": 37780 }, { "epoch": 16.912060863727902, "grad_norm": 0.1109994575381279, "learning_rate": 3.301125702308353e-05, "loss": 0.0275, "step": 37790 }, { "epoch": 16.916536137838442, "grad_norm": 0.11855780333280563, "learning_rate": 3.298534127791785e-05, "loss": 0.0283, "step": 37800 }, { "epoch": 16.92101141194898, "grad_norm": 0.11367309838533401, "learning_rate": 3.295943070135625e-05, "loss": 0.0285, "step": 37810 }, { "epoch": 16.92548668605952, "grad_norm": 0.09504936635494232, "learning_rate": 3.2933525301269684e-05, "loss": 0.0269, "step": 37820 }, { "epoch": 16.92996196017006, "grad_norm": 0.08959267288446426, "learning_rate": 3.2907625085527503e-05, "loss": 0.0298, "step": 37830 }, { "epoch": 16.9344372342806, "grad_norm": 0.11422551423311234, "learning_rate": 3.288173006199755e-05, "loss": 0.0318, "step": 37840 }, { "epoch": 16.938912508391137, "grad_norm": 0.10880929231643677, "learning_rate": 3.2855840238546e-05, "loss": 0.0274, "step": 37850 }, { "epoch": 16.943387782501677, "grad_norm": 0.1418597400188446, "learning_rate": 3.282995562303754e-05, "loss": 0.0302, "step": 37860 }, { "epoch": 16.947863056612217, "grad_norm": 0.1302928626537323, "learning_rate": 3.280407622333518e-05, "loss": 0.0289, "step": 37870 }, { "epoch": 16.952338330722757, "grad_norm": 0.10970053821802139, "learning_rate": 3.2778202047300444e-05, "loss": 0.0257, "step": 37880 }, { "epoch": 16.956813604833297, "grad_norm": 0.0934176817536354, "learning_rate": 3.275233310279321e-05, "loss": 0.0284, "step": 37890 }, { "epoch": 16.961288878943837, "grad_norm": 0.11517870426177979, "learning_rate": 3.272646939767179e-05, "loss": 0.0278, "step": 37900 }, { "epoch": 16.965764153054373, "grad_norm": 0.10606295615434647, "learning_rate": 3.2700610939792885e-05, "loss": 0.028, "step": 37910 }, { "epoch": 16.970239427164913, "grad_norm": 0.12468689680099487, "learning_rate": 3.267475773701161e-05, "loss": 0.0287, "step": 37920 }, { "epoch": 16.974714701275452, "grad_norm": 0.11335453391075134, "learning_rate": 3.264890979718147e-05, "loss": 0.0267, "step": 37930 }, { "epoch": 16.979189975385992, "grad_norm": 0.1240297481417656, "learning_rate": 3.262306712815444e-05, "loss": 0.0264, "step": 37940 }, { "epoch": 16.983665249496532, "grad_norm": 0.12743216753005981, "learning_rate": 3.2597229737780774e-05, "loss": 0.0297, "step": 37950 }, { "epoch": 16.988140523607072, "grad_norm": 0.11449532955884933, "learning_rate": 3.257139763390925e-05, "loss": 0.0267, "step": 37960 }, { "epoch": 16.99261579771761, "grad_norm": 0.1021808609366417, "learning_rate": 3.2545570824386925e-05, "loss": 0.0279, "step": 37970 }, { "epoch": 16.997091071828148, "grad_norm": 0.09115379303693771, "learning_rate": 3.251974931705933e-05, "loss": 0.0289, "step": 37980 }, { "epoch": 17.001566345938688, "grad_norm": 0.10842156410217285, "learning_rate": 3.249393311977037e-05, "loss": 0.0295, "step": 37990 }, { "epoch": 17.006041620049228, "grad_norm": 0.09444145113229752, "learning_rate": 3.2468122240362284e-05, "loss": 0.0274, "step": 38000 }, { "epoch": 17.010516894159768, "grad_norm": 0.10303789377212524, "learning_rate": 3.244231668667578e-05, "loss": 0.0281, "step": 38010 }, { "epoch": 17.014992168270307, "grad_norm": 0.11140690743923187, "learning_rate": 3.241651646654986e-05, "loss": 0.0279, "step": 38020 }, { "epoch": 17.019467442380847, "grad_norm": 0.10050667822360992, "learning_rate": 3.239072158782198e-05, "loss": 0.0305, "step": 38030 }, { "epoch": 17.023942716491383, "grad_norm": 0.11334235221147537, "learning_rate": 3.236493205832795e-05, "loss": 0.0268, "step": 38040 }, { "epoch": 17.028417990601923, "grad_norm": 0.09533849358558655, "learning_rate": 3.233914788590192e-05, "loss": 0.0305, "step": 38050 }, { "epoch": 17.032893264712463, "grad_norm": 0.11817830801010132, "learning_rate": 3.231336907837646e-05, "loss": 0.0295, "step": 38060 }, { "epoch": 17.037368538823003, "grad_norm": 0.11168840527534485, "learning_rate": 3.228759564358248e-05, "loss": 0.0325, "step": 38070 }, { "epoch": 17.041843812933543, "grad_norm": 0.11449489742517471, "learning_rate": 3.226182758934927e-05, "loss": 0.0308, "step": 38080 }, { "epoch": 17.046319087044083, "grad_norm": 0.11377935856580734, "learning_rate": 3.223606492350451e-05, "loss": 0.0279, "step": 38090 }, { "epoch": 17.050794361154622, "grad_norm": 0.12651319801807404, "learning_rate": 3.221030765387417e-05, "loss": 0.0274, "step": 38100 }, { "epoch": 17.05526963526516, "grad_norm": 0.11622317880392075, "learning_rate": 3.218455578828269e-05, "loss": 0.0277, "step": 38110 }, { "epoch": 17.0597449093757, "grad_norm": 0.11708182841539383, "learning_rate": 3.2158809334552745e-05, "loss": 0.0283, "step": 38120 }, { "epoch": 17.06422018348624, "grad_norm": 0.14126279950141907, "learning_rate": 3.2133068300505455e-05, "loss": 0.0293, "step": 38130 }, { "epoch": 17.068695457596778, "grad_norm": 0.13035550713539124, "learning_rate": 3.210733269396028e-05, "loss": 0.0282, "step": 38140 }, { "epoch": 17.073170731707318, "grad_norm": 0.16123491525650024, "learning_rate": 3.2081602522734986e-05, "loss": 0.028, "step": 38150 }, { "epoch": 17.077646005817858, "grad_norm": 0.13191990554332733, "learning_rate": 3.205587779464576e-05, "loss": 0.0266, "step": 38160 }, { "epoch": 17.082121279928394, "grad_norm": 0.10683462768793106, "learning_rate": 3.203015851750706e-05, "loss": 0.0275, "step": 38170 }, { "epoch": 17.086596554038934, "grad_norm": 0.1103854700922966, "learning_rate": 3.2004444699131727e-05, "loss": 0.0307, "step": 38180 }, { "epoch": 17.091071828149474, "grad_norm": 0.11193264275789261, "learning_rate": 3.197873634733096e-05, "loss": 0.0241, "step": 38190 }, { "epoch": 17.095547102260014, "grad_norm": 0.12261912226676941, "learning_rate": 3.1953033469914276e-05, "loss": 0.0283, "step": 38200 }, { "epoch": 17.100022376370553, "grad_norm": 0.09153813868761063, "learning_rate": 3.192733607468951e-05, "loss": 0.025, "step": 38210 }, { "epoch": 17.104497650481093, "grad_norm": 0.13988031446933746, "learning_rate": 3.190164416946285e-05, "loss": 0.0291, "step": 38220 }, { "epoch": 17.10897292459163, "grad_norm": 0.10385967791080475, "learning_rate": 3.187595776203886e-05, "loss": 0.0256, "step": 38230 }, { "epoch": 17.11344819870217, "grad_norm": 0.09942251443862915, "learning_rate": 3.1850276860220346e-05, "loss": 0.0254, "step": 38240 }, { "epoch": 17.11792347281271, "grad_norm": 0.10803288966417313, "learning_rate": 3.18246014718085e-05, "loss": 0.0314, "step": 38250 }, { "epoch": 17.12239874692325, "grad_norm": 0.12442849576473236, "learning_rate": 3.1798931604602864e-05, "loss": 0.029, "step": 38260 }, { "epoch": 17.12687402103379, "grad_norm": 0.10084778070449829, "learning_rate": 3.1773267266401206e-05, "loss": 0.029, "step": 38270 }, { "epoch": 17.13134929514433, "grad_norm": 0.14097923040390015, "learning_rate": 3.1747608464999725e-05, "loss": 0.0259, "step": 38280 }, { "epoch": 17.13582456925487, "grad_norm": 0.11999552696943283, "learning_rate": 3.172195520819285e-05, "loss": 0.026, "step": 38290 }, { "epoch": 17.140299843365405, "grad_norm": 0.08972629904747009, "learning_rate": 3.169630750377337e-05, "loss": 0.0275, "step": 38300 }, { "epoch": 17.144775117475945, "grad_norm": 0.10673673450946808, "learning_rate": 3.167066535953242e-05, "loss": 0.0275, "step": 38310 }, { "epoch": 17.149250391586484, "grad_norm": 0.12458302080631256, "learning_rate": 3.1645028783259345e-05, "loss": 0.0278, "step": 38320 }, { "epoch": 17.153725665697024, "grad_norm": 0.09749539196491241, "learning_rate": 3.161939778274191e-05, "loss": 0.0257, "step": 38330 }, { "epoch": 17.158200939807564, "grad_norm": 0.11654997617006302, "learning_rate": 3.1593772365766105e-05, "loss": 0.0267, "step": 38340 }, { "epoch": 17.162676213918104, "grad_norm": 0.0978444516658783, "learning_rate": 3.156815254011627e-05, "loss": 0.0251, "step": 38350 }, { "epoch": 17.16715148802864, "grad_norm": 0.09024540334939957, "learning_rate": 3.1542538313575035e-05, "loss": 0.0295, "step": 38360 }, { "epoch": 17.17162676213918, "grad_norm": 0.10423298180103302, "learning_rate": 3.1516929693923315e-05, "loss": 0.0301, "step": 38370 }, { "epoch": 17.17610203624972, "grad_norm": 0.1281515657901764, "learning_rate": 3.1491326688940345e-05, "loss": 0.0243, "step": 38380 }, { "epoch": 17.18057731036026, "grad_norm": 0.1312241405248642, "learning_rate": 3.146572930640362e-05, "loss": 0.028, "step": 38390 }, { "epoch": 17.1850525844708, "grad_norm": 0.12553247809410095, "learning_rate": 3.144013755408895e-05, "loss": 0.0274, "step": 38400 }, { "epoch": 17.18952785858134, "grad_norm": 0.10392063856124878, "learning_rate": 3.141455143977049e-05, "loss": 0.0251, "step": 38410 }, { "epoch": 17.194003132691876, "grad_norm": 0.11410681158304214, "learning_rate": 3.1388970971220546e-05, "loss": 0.027, "step": 38420 }, { "epoch": 17.198478406802415, "grad_norm": 0.10973562300205231, "learning_rate": 3.136339615620985e-05, "loss": 0.0273, "step": 38430 }, { "epoch": 17.202953680912955, "grad_norm": 0.08757303655147552, "learning_rate": 3.133782700250731e-05, "loss": 0.026, "step": 38440 }, { "epoch": 17.207428955023495, "grad_norm": 0.10540883243083954, "learning_rate": 3.131226351788018e-05, "loss": 0.0272, "step": 38450 }, { "epoch": 17.211904229134035, "grad_norm": 0.11927720159292221, "learning_rate": 3.128670571009399e-05, "loss": 0.0315, "step": 38460 }, { "epoch": 17.216379503244575, "grad_norm": 0.14013174176216125, "learning_rate": 3.126115358691249e-05, "loss": 0.0264, "step": 38470 }, { "epoch": 17.220854777355115, "grad_norm": 0.10050756484270096, "learning_rate": 3.123560715609777e-05, "loss": 0.0252, "step": 38480 }, { "epoch": 17.22533005146565, "grad_norm": 0.1447315365076065, "learning_rate": 3.121006642541014e-05, "loss": 0.031, "step": 38490 }, { "epoch": 17.22980532557619, "grad_norm": 0.09356110543012619, "learning_rate": 3.118453140260823e-05, "loss": 0.0284, "step": 38500 }, { "epoch": 17.23428059968673, "grad_norm": 0.15020741522312164, "learning_rate": 3.115900209544887e-05, "loss": 0.0306, "step": 38510 }, { "epoch": 17.23875587379727, "grad_norm": 0.11209181696176529, "learning_rate": 3.113347851168721e-05, "loss": 0.0307, "step": 38520 }, { "epoch": 17.24323114790781, "grad_norm": 0.10218106210231781, "learning_rate": 3.110796065907665e-05, "loss": 0.0248, "step": 38530 }, { "epoch": 17.24770642201835, "grad_norm": 0.1045261025428772, "learning_rate": 3.1082448545368814e-05, "loss": 0.0309, "step": 38540 }, { "epoch": 17.252181696128886, "grad_norm": 0.10972704738378525, "learning_rate": 3.105694217831361e-05, "loss": 0.0271, "step": 38550 }, { "epoch": 17.256656970239426, "grad_norm": 0.1264219880104065, "learning_rate": 3.1031441565659235e-05, "loss": 0.0274, "step": 38560 }, { "epoch": 17.261132244349966, "grad_norm": 0.11750900000333786, "learning_rate": 3.100594671515206e-05, "loss": 0.0293, "step": 38570 }, { "epoch": 17.265607518460506, "grad_norm": 0.123197540640831, "learning_rate": 3.098045763453678e-05, "loss": 0.027, "step": 38580 }, { "epoch": 17.270082792571046, "grad_norm": 0.13871927559375763, "learning_rate": 3.095497433155626e-05, "loss": 0.0261, "step": 38590 }, { "epoch": 17.274558066681585, "grad_norm": 0.11607930064201355, "learning_rate": 3.0929496813951694e-05, "loss": 0.028, "step": 38600 }, { "epoch": 17.279033340792125, "grad_norm": 0.13209325075149536, "learning_rate": 3.090402508946249e-05, "loss": 0.0305, "step": 38610 }, { "epoch": 17.28350861490266, "grad_norm": 0.11027669161558151, "learning_rate": 3.0878559165826236e-05, "loss": 0.0288, "step": 38620 }, { "epoch": 17.2879838890132, "grad_norm": 0.13301758468151093, "learning_rate": 3.0853099050778854e-05, "loss": 0.0251, "step": 38630 }, { "epoch": 17.29245916312374, "grad_norm": 0.11408563703298569, "learning_rate": 3.082764475205442e-05, "loss": 0.0284, "step": 38640 }, { "epoch": 17.29693443723428, "grad_norm": 0.10992127656936646, "learning_rate": 3.0802196277385317e-05, "loss": 0.0271, "step": 38650 }, { "epoch": 17.30140971134482, "grad_norm": 0.101523756980896, "learning_rate": 3.077675363450207e-05, "loss": 0.0266, "step": 38660 }, { "epoch": 17.30588498545536, "grad_norm": 0.11509256064891815, "learning_rate": 3.075131683113352e-05, "loss": 0.0238, "step": 38670 }, { "epoch": 17.310360259565897, "grad_norm": 0.12729695439338684, "learning_rate": 3.072588587500669e-05, "loss": 0.0267, "step": 38680 }, { "epoch": 17.314835533676437, "grad_norm": 0.11561334133148193, "learning_rate": 3.070046077384682e-05, "loss": 0.0286, "step": 38690 }, { "epoch": 17.319310807786977, "grad_norm": 0.09631238877773285, "learning_rate": 3.0675041535377405e-05, "loss": 0.0272, "step": 38700 }, { "epoch": 17.323786081897516, "grad_norm": 0.12263011932373047, "learning_rate": 3.06496281673201e-05, "loss": 0.0272, "step": 38710 }, { "epoch": 17.328261356008056, "grad_norm": 0.11035686731338501, "learning_rate": 3.062422067739485e-05, "loss": 0.0268, "step": 38720 }, { "epoch": 17.332736630118596, "grad_norm": 0.09220956265926361, "learning_rate": 3.059881907331979e-05, "loss": 0.0269, "step": 38730 }, { "epoch": 17.337211904229132, "grad_norm": 0.1053856909275055, "learning_rate": 3.057342336281122e-05, "loss": 0.0249, "step": 38740 }, { "epoch": 17.341687178339672, "grad_norm": 0.09804161638021469, "learning_rate": 3.0548033553583705e-05, "loss": 0.0259, "step": 38750 }, { "epoch": 17.346162452450212, "grad_norm": 0.12914246320724487, "learning_rate": 3.052264965335e-05, "loss": 0.0272, "step": 38760 }, { "epoch": 17.350637726560752, "grad_norm": 0.11268316209316254, "learning_rate": 3.049727166982105e-05, "loss": 0.0291, "step": 38770 }, { "epoch": 17.35511300067129, "grad_norm": 0.11731106787919998, "learning_rate": 3.0471899610706038e-05, "loss": 0.0312, "step": 38780 }, { "epoch": 17.35958827478183, "grad_norm": 0.10193055868148804, "learning_rate": 3.0446533483712304e-05, "loss": 0.0284, "step": 38790 }, { "epoch": 17.36406354889237, "grad_norm": 0.12462897598743439, "learning_rate": 3.042117329654544e-05, "loss": 0.0252, "step": 38800 }, { "epoch": 17.368538823002908, "grad_norm": 0.09543940424919128, "learning_rate": 3.039581905690916e-05, "loss": 0.0251, "step": 38810 }, { "epoch": 17.373014097113447, "grad_norm": 0.10112626850605011, "learning_rate": 3.0370470772505433e-05, "loss": 0.0249, "step": 38820 }, { "epoch": 17.377489371223987, "grad_norm": 0.10265851765871048, "learning_rate": 3.034512845103441e-05, "loss": 0.029, "step": 38830 }, { "epoch": 17.381964645334527, "grad_norm": 0.1312079280614853, "learning_rate": 3.03197921001944e-05, "loss": 0.0279, "step": 38840 }, { "epoch": 17.386439919445067, "grad_norm": 0.10635077953338623, "learning_rate": 3.0294461727681932e-05, "loss": 0.0257, "step": 38850 }, { "epoch": 17.390915193555607, "grad_norm": 0.10174386203289032, "learning_rate": 3.0269137341191677e-05, "loss": 0.0272, "step": 38860 }, { "epoch": 17.395390467666143, "grad_norm": 0.11089413613080978, "learning_rate": 3.0243818948416543e-05, "loss": 0.0233, "step": 38870 }, { "epoch": 17.399865741776683, "grad_norm": 0.11324204504489899, "learning_rate": 3.0218506557047598e-05, "loss": 0.0252, "step": 38880 }, { "epoch": 17.404341015887223, "grad_norm": 0.10335860401391983, "learning_rate": 3.0193200174774038e-05, "loss": 0.0252, "step": 38890 }, { "epoch": 17.408816289997763, "grad_norm": 0.09542369097471237, "learning_rate": 3.0167899809283308e-05, "loss": 0.0288, "step": 38900 }, { "epoch": 17.413291564108302, "grad_norm": 0.10876217484474182, "learning_rate": 3.0142605468260978e-05, "loss": 0.0284, "step": 38910 }, { "epoch": 17.417766838218842, "grad_norm": 0.10189259797334671, "learning_rate": 3.0117317159390794e-05, "loss": 0.0279, "step": 38920 }, { "epoch": 17.42224211232938, "grad_norm": 0.13037694990634918, "learning_rate": 3.0092034890354694e-05, "loss": 0.0279, "step": 38930 }, { "epoch": 17.42671738643992, "grad_norm": 0.10639213770627975, "learning_rate": 3.006675866883275e-05, "loss": 0.0252, "step": 38940 }, { "epoch": 17.431192660550458, "grad_norm": 0.08334638923406601, "learning_rate": 3.004148850250323e-05, "loss": 0.0231, "step": 38950 }, { "epoch": 17.435667934660998, "grad_norm": 0.11481092870235443, "learning_rate": 3.0016224399042515e-05, "loss": 0.027, "step": 38960 }, { "epoch": 17.440143208771538, "grad_norm": 0.09579560905694962, "learning_rate": 2.999096636612518e-05, "loss": 0.028, "step": 38970 }, { "epoch": 17.444618482882078, "grad_norm": 0.10967721045017242, "learning_rate": 2.9965714411423972e-05, "loss": 0.03, "step": 38980 }, { "epoch": 17.449093756992617, "grad_norm": 0.125322163105011, "learning_rate": 2.994046854260974e-05, "loss": 0.0307, "step": 38990 }, { "epoch": 17.453569031103154, "grad_norm": 0.10735398530960083, "learning_rate": 2.991522876735154e-05, "loss": 0.0287, "step": 39000 }, { "epoch": 17.458044305213694, "grad_norm": 0.11774599552154541, "learning_rate": 2.9889995093316515e-05, "loss": 0.031, "step": 39010 }, { "epoch": 17.462519579324233, "grad_norm": 0.10450869798660278, "learning_rate": 2.9864767528170002e-05, "loss": 0.0248, "step": 39020 }, { "epoch": 17.466994853434773, "grad_norm": 0.10548854619264603, "learning_rate": 2.9839546079575497e-05, "loss": 0.028, "step": 39030 }, { "epoch": 17.471470127545313, "grad_norm": 0.11060821264982224, "learning_rate": 2.9814330755194564e-05, "loss": 0.0275, "step": 39040 }, { "epoch": 17.475945401655853, "grad_norm": 0.12738829851150513, "learning_rate": 2.978912156268699e-05, "loss": 0.0216, "step": 39050 }, { "epoch": 17.48042067576639, "grad_norm": 0.10970912128686905, "learning_rate": 2.976391850971065e-05, "loss": 0.0278, "step": 39060 }, { "epoch": 17.48489594987693, "grad_norm": 0.09312223643064499, "learning_rate": 2.973872160392156e-05, "loss": 0.0282, "step": 39070 }, { "epoch": 17.48937122398747, "grad_norm": 0.10858701169490814, "learning_rate": 2.971353085297387e-05, "loss": 0.0287, "step": 39080 }, { "epoch": 17.49384649809801, "grad_norm": 0.14434170722961426, "learning_rate": 2.9688346264519866e-05, "loss": 0.0287, "step": 39090 }, { "epoch": 17.49832177220855, "grad_norm": 0.10270176827907562, "learning_rate": 2.9663167846209998e-05, "loss": 0.0306, "step": 39100 }, { "epoch": 17.50279704631909, "grad_norm": 0.12677603960037231, "learning_rate": 2.963799560569275e-05, "loss": 0.0274, "step": 39110 }, { "epoch": 17.507272320429628, "grad_norm": 0.08802841603755951, "learning_rate": 2.9612829550614836e-05, "loss": 0.0292, "step": 39120 }, { "epoch": 17.511747594540164, "grad_norm": 0.13264067471027374, "learning_rate": 2.9587669688620988e-05, "loss": 0.0269, "step": 39130 }, { "epoch": 17.516222868650704, "grad_norm": 0.09286723285913467, "learning_rate": 2.956251602735413e-05, "loss": 0.0279, "step": 39140 }, { "epoch": 17.520698142761244, "grad_norm": 0.08609084784984589, "learning_rate": 2.9537368574455304e-05, "loss": 0.0266, "step": 39150 }, { "epoch": 17.525173416871784, "grad_norm": 0.11096392571926117, "learning_rate": 2.9512227337563604e-05, "loss": 0.0291, "step": 39160 }, { "epoch": 17.529648690982324, "grad_norm": 0.13065481185913086, "learning_rate": 2.948709232431631e-05, "loss": 0.0266, "step": 39170 }, { "epoch": 17.534123965092864, "grad_norm": 0.08403553813695908, "learning_rate": 2.9461963542348737e-05, "loss": 0.0274, "step": 39180 }, { "epoch": 17.5385992392034, "grad_norm": 0.12827034294605255, "learning_rate": 2.943684099929436e-05, "loss": 0.0253, "step": 39190 }, { "epoch": 17.54307451331394, "grad_norm": 0.10747060924768448, "learning_rate": 2.9411724702784758e-05, "loss": 0.029, "step": 39200 }, { "epoch": 17.54754978742448, "grad_norm": 0.10623320937156677, "learning_rate": 2.9386614660449596e-05, "loss": 0.0252, "step": 39210 }, { "epoch": 17.55202506153502, "grad_norm": 0.08786159753799438, "learning_rate": 2.936151087991663e-05, "loss": 0.0251, "step": 39220 }, { "epoch": 17.55650033564556, "grad_norm": 0.11099749058485031, "learning_rate": 2.9336413368811723e-05, "loss": 0.0282, "step": 39230 }, { "epoch": 17.5609756097561, "grad_norm": 0.1189601719379425, "learning_rate": 2.931132213475884e-05, "loss": 0.0248, "step": 39240 }, { "epoch": 17.565450883866635, "grad_norm": 0.10888567566871643, "learning_rate": 2.928623718538006e-05, "loss": 0.0263, "step": 39250 }, { "epoch": 17.569926157977175, "grad_norm": 0.09850136935710907, "learning_rate": 2.9261158528295495e-05, "loss": 0.0255, "step": 39260 }, { "epoch": 17.574401432087715, "grad_norm": 0.11641981452703476, "learning_rate": 2.9236086171123404e-05, "loss": 0.0286, "step": 39270 }, { "epoch": 17.578876706198255, "grad_norm": 0.11552740633487701, "learning_rate": 2.9211020121480083e-05, "loss": 0.027, "step": 39280 }, { "epoch": 17.583351980308795, "grad_norm": 0.10112637281417847, "learning_rate": 2.918596038697995e-05, "loss": 0.0298, "step": 39290 }, { "epoch": 17.587827254419334, "grad_norm": 0.09089617431163788, "learning_rate": 2.916090697523549e-05, "loss": 0.0311, "step": 39300 }, { "epoch": 17.59230252852987, "grad_norm": 0.11039931327104568, "learning_rate": 2.9135859893857248e-05, "loss": 0.025, "step": 39310 }, { "epoch": 17.59677780264041, "grad_norm": 0.1299215704202652, "learning_rate": 2.9110819150453927e-05, "loss": 0.0258, "step": 39320 }, { "epoch": 17.60125307675095, "grad_norm": 0.10537048429250717, "learning_rate": 2.9085784752632157e-05, "loss": 0.0249, "step": 39330 }, { "epoch": 17.60572835086149, "grad_norm": 0.11879709362983704, "learning_rate": 2.9060756707996796e-05, "loss": 0.0262, "step": 39340 }, { "epoch": 17.61020362497203, "grad_norm": 0.14044317603111267, "learning_rate": 2.9035735024150674e-05, "loss": 0.0281, "step": 39350 }, { "epoch": 17.61467889908257, "grad_norm": 0.10544421523809433, "learning_rate": 2.9010719708694722e-05, "loss": 0.0263, "step": 39360 }, { "epoch": 17.61915417319311, "grad_norm": 0.10714434087276459, "learning_rate": 2.8985710769227936e-05, "loss": 0.027, "step": 39370 }, { "epoch": 17.623629447303646, "grad_norm": 0.10808310657739639, "learning_rate": 2.8960708213347366e-05, "loss": 0.0295, "step": 39380 }, { "epoch": 17.628104721414186, "grad_norm": 0.09665873646736145, "learning_rate": 2.8935712048648112e-05, "loss": 0.0263, "step": 39390 }, { "epoch": 17.632579995524726, "grad_norm": 0.11012893170118332, "learning_rate": 2.89107222827234e-05, "loss": 0.0265, "step": 39400 }, { "epoch": 17.637055269635265, "grad_norm": 0.13068968057632446, "learning_rate": 2.8885738923164395e-05, "loss": 0.0275, "step": 39410 }, { "epoch": 17.641530543745805, "grad_norm": 0.10468827933073044, "learning_rate": 2.8860761977560436e-05, "loss": 0.027, "step": 39420 }, { "epoch": 17.646005817856345, "grad_norm": 0.118044413626194, "learning_rate": 2.883579145349884e-05, "loss": 0.0288, "step": 39430 }, { "epoch": 17.65048109196688, "grad_norm": 0.1420523077249527, "learning_rate": 2.881082735856499e-05, "loss": 0.0287, "step": 39440 }, { "epoch": 17.65495636607742, "grad_norm": 0.11969068646430969, "learning_rate": 2.878586970034232e-05, "loss": 0.0274, "step": 39450 }, { "epoch": 17.65943164018796, "grad_norm": 0.12898734211921692, "learning_rate": 2.8760918486412292e-05, "loss": 0.0251, "step": 39460 }, { "epoch": 17.6639069142985, "grad_norm": 0.1167379766702652, "learning_rate": 2.8735973724354482e-05, "loss": 0.0248, "step": 39470 }, { "epoch": 17.66838218840904, "grad_norm": 0.14109882712364197, "learning_rate": 2.8711035421746367e-05, "loss": 0.0289, "step": 39480 }, { "epoch": 17.67285746251958, "grad_norm": 0.09090426564216614, "learning_rate": 2.8686103586163626e-05, "loss": 0.0249, "step": 39490 }, { "epoch": 17.67733273663012, "grad_norm": 0.09161577373743057, "learning_rate": 2.866117822517982e-05, "loss": 0.0325, "step": 39500 }, { "epoch": 17.681808010740657, "grad_norm": 0.11463805288076401, "learning_rate": 2.8636259346366666e-05, "loss": 0.0278, "step": 39510 }, { "epoch": 17.686283284851196, "grad_norm": 0.12743209302425385, "learning_rate": 2.861134695729385e-05, "loss": 0.024, "step": 39520 }, { "epoch": 17.690758558961736, "grad_norm": 0.10281714051961899, "learning_rate": 2.858644106552909e-05, "loss": 0.0282, "step": 39530 }, { "epoch": 17.695233833072276, "grad_norm": 0.10574746131896973, "learning_rate": 2.8561541678638142e-05, "loss": 0.0266, "step": 39540 }, { "epoch": 17.699709107182816, "grad_norm": 0.11343108117580414, "learning_rate": 2.8536648804184785e-05, "loss": 0.0263, "step": 39550 }, { "epoch": 17.704184381293356, "grad_norm": 0.11022262275218964, "learning_rate": 2.8511762449730795e-05, "loss": 0.0239, "step": 39560 }, { "epoch": 17.708659655403892, "grad_norm": 0.08955880254507065, "learning_rate": 2.8486882622836026e-05, "loss": 0.0241, "step": 39570 }, { "epoch": 17.713134929514432, "grad_norm": 0.11874634772539139, "learning_rate": 2.846200933105829e-05, "loss": 0.028, "step": 39580 }, { "epoch": 17.71761020362497, "grad_norm": 0.09417340904474258, "learning_rate": 2.843714258195346e-05, "loss": 0.0275, "step": 39590 }, { "epoch": 17.72208547773551, "grad_norm": 0.12863171100616455, "learning_rate": 2.8412282383075363e-05, "loss": 0.0288, "step": 39600 }, { "epoch": 17.72656075184605, "grad_norm": 0.09576547145843506, "learning_rate": 2.838742874197587e-05, "loss": 0.0237, "step": 39610 }, { "epoch": 17.73103602595659, "grad_norm": 0.11083522439002991, "learning_rate": 2.8362581666204918e-05, "loss": 0.026, "step": 39620 }, { "epoch": 17.735511300067127, "grad_norm": 0.10525250434875488, "learning_rate": 2.8337741163310317e-05, "loss": 0.0279, "step": 39630 }, { "epoch": 17.739986574177667, "grad_norm": 0.11445559561252594, "learning_rate": 2.8312907240838027e-05, "loss": 0.029, "step": 39640 }, { "epoch": 17.744461848288207, "grad_norm": 0.12586311995983124, "learning_rate": 2.8288079906331864e-05, "loss": 0.0289, "step": 39650 }, { "epoch": 17.748937122398747, "grad_norm": 0.12269816547632217, "learning_rate": 2.8263259167333777e-05, "loss": 0.0285, "step": 39660 }, { "epoch": 17.753412396509287, "grad_norm": 0.08573880046606064, "learning_rate": 2.823844503138363e-05, "loss": 0.0275, "step": 39670 }, { "epoch": 17.757887670619827, "grad_norm": 0.09791209548711777, "learning_rate": 2.8213637506019304e-05, "loss": 0.0278, "step": 39680 }, { "epoch": 17.762362944730366, "grad_norm": 0.12036086618900299, "learning_rate": 2.8188836598776662e-05, "loss": 0.0309, "step": 39690 }, { "epoch": 17.766838218840903, "grad_norm": 0.1195632740855217, "learning_rate": 2.8164042317189575e-05, "loss": 0.0278, "step": 39700 }, { "epoch": 17.771313492951442, "grad_norm": 0.12125252932310104, "learning_rate": 2.8139254668789867e-05, "loss": 0.0256, "step": 39710 }, { "epoch": 17.775788767061982, "grad_norm": 0.08245177567005157, "learning_rate": 2.811447366110741e-05, "loss": 0.0292, "step": 39720 }, { "epoch": 17.780264041172522, "grad_norm": 0.11123598366975784, "learning_rate": 2.8089699301670002e-05, "loss": 0.0264, "step": 39730 }, { "epoch": 17.784739315283062, "grad_norm": 0.09250421077013016, "learning_rate": 2.8064931598003436e-05, "loss": 0.0261, "step": 39740 }, { "epoch": 17.789214589393602, "grad_norm": 0.09876759350299835, "learning_rate": 2.804017055763149e-05, "loss": 0.0292, "step": 39750 }, { "epoch": 17.793689863504138, "grad_norm": 0.11493329703807831, "learning_rate": 2.8015416188075893e-05, "loss": 0.0268, "step": 39760 }, { "epoch": 17.798165137614678, "grad_norm": 0.11152395606040955, "learning_rate": 2.7990668496856427e-05, "loss": 0.0267, "step": 39770 }, { "epoch": 17.802640411725218, "grad_norm": 0.09378352016210556, "learning_rate": 2.7965927491490705e-05, "loss": 0.029, "step": 39780 }, { "epoch": 17.807115685835758, "grad_norm": 0.10940643399953842, "learning_rate": 2.7941193179494484e-05, "loss": 0.0306, "step": 39790 }, { "epoch": 17.811590959946297, "grad_norm": 0.11046629399061203, "learning_rate": 2.79164655683813e-05, "loss": 0.0275, "step": 39800 }, { "epoch": 17.816066234056837, "grad_norm": 0.11447036266326904, "learning_rate": 2.7891744665662823e-05, "loss": 0.0306, "step": 39810 }, { "epoch": 17.820541508167373, "grad_norm": 0.09201160818338394, "learning_rate": 2.7867030478848577e-05, "loss": 0.0304, "step": 39820 }, { "epoch": 17.825016782277913, "grad_norm": 0.14558112621307373, "learning_rate": 2.7842323015446082e-05, "loss": 0.028, "step": 39830 }, { "epoch": 17.829492056388453, "grad_norm": 0.11771298199892044, "learning_rate": 2.7817622282960815e-05, "loss": 0.028, "step": 39840 }, { "epoch": 17.833967330498993, "grad_norm": 0.09947410970926285, "learning_rate": 2.7792928288896202e-05, "loss": 0.028, "step": 39850 }, { "epoch": 17.838442604609533, "grad_norm": 0.11104267090559006, "learning_rate": 2.776824104075364e-05, "loss": 0.025, "step": 39860 }, { "epoch": 17.842917878720073, "grad_norm": 0.09303706139326096, "learning_rate": 2.774356054603243e-05, "loss": 0.0289, "step": 39870 }, { "epoch": 17.847393152830612, "grad_norm": 0.10437455773353577, "learning_rate": 2.7718886812229907e-05, "loss": 0.0287, "step": 39880 }, { "epoch": 17.85186842694115, "grad_norm": 0.10037052631378174, "learning_rate": 2.7694219846841262e-05, "loss": 0.027, "step": 39890 }, { "epoch": 17.85634370105169, "grad_norm": 0.1195313110947609, "learning_rate": 2.766955965735968e-05, "loss": 0.0265, "step": 39900 }, { "epoch": 17.86081897516223, "grad_norm": 0.10418152064085007, "learning_rate": 2.764490625127627e-05, "loss": 0.027, "step": 39910 }, { "epoch": 17.865294249272768, "grad_norm": 0.10883905738592148, "learning_rate": 2.762025963608009e-05, "loss": 0.0242, "step": 39920 }, { "epoch": 17.869769523383308, "grad_norm": 0.10148809850215912, "learning_rate": 2.7595619819258116e-05, "loss": 0.0329, "step": 39930 }, { "epoch": 17.874244797493848, "grad_norm": 0.11369045823812485, "learning_rate": 2.7570986808295324e-05, "loss": 0.0276, "step": 39940 }, { "epoch": 17.878720071604384, "grad_norm": 0.11056964099407196, "learning_rate": 2.7546360610674493e-05, "loss": 0.0288, "step": 39950 }, { "epoch": 17.883195345714924, "grad_norm": 0.10854403674602509, "learning_rate": 2.7521741233876496e-05, "loss": 0.0288, "step": 39960 }, { "epoch": 17.887670619825464, "grad_norm": 0.0992741659283638, "learning_rate": 2.749712868537997e-05, "loss": 0.0286, "step": 39970 }, { "epoch": 17.892145893936004, "grad_norm": 0.1182321235537529, "learning_rate": 2.747252297266162e-05, "loss": 0.0222, "step": 39980 }, { "epoch": 17.896621168046543, "grad_norm": 0.08304061740636826, "learning_rate": 2.7447924103195976e-05, "loss": 0.0276, "step": 39990 }, { "epoch": 17.901096442157083, "grad_norm": 0.10887594521045685, "learning_rate": 2.7423332084455544e-05, "loss": 0.0284, "step": 40000 }, { "epoch": 17.905571716267623, "grad_norm": 0.11864496767520905, "learning_rate": 2.7398746923910723e-05, "loss": 0.0286, "step": 40010 }, { "epoch": 17.91004699037816, "grad_norm": 0.11049585044384003, "learning_rate": 2.7374168629029813e-05, "loss": 0.027, "step": 40020 }, { "epoch": 17.9145222644887, "grad_norm": 0.11417008191347122, "learning_rate": 2.7349597207279088e-05, "loss": 0.0272, "step": 40030 }, { "epoch": 17.91899753859924, "grad_norm": 0.10104408115148544, "learning_rate": 2.7325032666122686e-05, "loss": 0.0252, "step": 40040 }, { "epoch": 17.92347281270978, "grad_norm": 0.08881418406963348, "learning_rate": 2.7300475013022663e-05, "loss": 0.0242, "step": 40050 }, { "epoch": 17.92794808682032, "grad_norm": 0.07891953736543655, "learning_rate": 2.727592425543899e-05, "loss": 0.0254, "step": 40060 }, { "epoch": 17.93242336093086, "grad_norm": 0.08881057798862457, "learning_rate": 2.725138040082953e-05, "loss": 0.0242, "step": 40070 }, { "epoch": 17.936898635041395, "grad_norm": 0.088539719581604, "learning_rate": 2.7226843456650037e-05, "loss": 0.0273, "step": 40080 }, { "epoch": 17.941373909151935, "grad_norm": 0.09989922493696213, "learning_rate": 2.7202313430354253e-05, "loss": 0.0236, "step": 40090 }, { "epoch": 17.945849183262474, "grad_norm": 0.11996471136808395, "learning_rate": 2.717779032939367e-05, "loss": 0.0246, "step": 40100 }, { "epoch": 17.950324457373014, "grad_norm": 0.11931006610393524, "learning_rate": 2.7153274161217846e-05, "loss": 0.0274, "step": 40110 }, { "epoch": 17.954799731483554, "grad_norm": 0.11529998481273651, "learning_rate": 2.7128764933274052e-05, "loss": 0.0263, "step": 40120 }, { "epoch": 17.959275005594094, "grad_norm": 0.09310784190893173, "learning_rate": 2.7104262653007616e-05, "loss": 0.0247, "step": 40130 }, { "epoch": 17.96375027970463, "grad_norm": 0.10926765203475952, "learning_rate": 2.707976732786166e-05, "loss": 0.0256, "step": 40140 }, { "epoch": 17.96822555381517, "grad_norm": 0.10828964412212372, "learning_rate": 2.705527896527721e-05, "loss": 0.0259, "step": 40150 }, { "epoch": 17.97270082792571, "grad_norm": 0.1060025617480278, "learning_rate": 2.703079757269319e-05, "loss": 0.0255, "step": 40160 }, { "epoch": 17.97717610203625, "grad_norm": 0.11310405284166336, "learning_rate": 2.7006323157546386e-05, "loss": 0.0271, "step": 40170 }, { "epoch": 17.98165137614679, "grad_norm": 0.12149262428283691, "learning_rate": 2.698185572727151e-05, "loss": 0.0273, "step": 40180 }, { "epoch": 17.98612665025733, "grad_norm": 0.14854450523853302, "learning_rate": 2.695739528930111e-05, "loss": 0.0289, "step": 40190 }, { "epoch": 17.99060192436787, "grad_norm": 0.1143186017870903, "learning_rate": 2.693294185106562e-05, "loss": 0.0274, "step": 40200 }, { "epoch": 17.995077198478405, "grad_norm": 0.1182013750076294, "learning_rate": 2.690849541999333e-05, "loss": 0.0289, "step": 40210 }, { "epoch": 17.999552472588945, "grad_norm": 0.15934893488883972, "learning_rate": 2.688405600351045e-05, "loss": 0.0299, "step": 40220 }, { "epoch": 18.004027746699485, "grad_norm": 0.10786369442939758, "learning_rate": 2.6859623609040984e-05, "loss": 0.0297, "step": 40230 }, { "epoch": 18.008503020810025, "grad_norm": 0.10486860573291779, "learning_rate": 2.6835198244006927e-05, "loss": 0.0289, "step": 40240 }, { "epoch": 18.012978294920565, "grad_norm": 0.11533516645431519, "learning_rate": 2.681077991582797e-05, "loss": 0.0271, "step": 40250 }, { "epoch": 18.017453569031105, "grad_norm": 0.09775801748037338, "learning_rate": 2.6786368631921836e-05, "loss": 0.0262, "step": 40260 }, { "epoch": 18.02192884314164, "grad_norm": 0.11448591202497482, "learning_rate": 2.6761964399703955e-05, "loss": 0.0244, "step": 40270 }, { "epoch": 18.02640411725218, "grad_norm": 0.1070185974240303, "learning_rate": 2.6737567226587747e-05, "loss": 0.0246, "step": 40280 }, { "epoch": 18.03087939136272, "grad_norm": 0.1051558330655098, "learning_rate": 2.67131771199844e-05, "loss": 0.0264, "step": 40290 }, { "epoch": 18.03535466547326, "grad_norm": 0.11260990798473358, "learning_rate": 2.6688794087302993e-05, "loss": 0.0298, "step": 40300 }, { "epoch": 18.0398299395838, "grad_norm": 0.10500326752662659, "learning_rate": 2.6664418135950453e-05, "loss": 0.0257, "step": 40310 }, { "epoch": 18.04430521369434, "grad_norm": 0.1036401242017746, "learning_rate": 2.6640049273331515e-05, "loss": 0.0284, "step": 40320 }, { "epoch": 18.048780487804876, "grad_norm": 0.10181750357151031, "learning_rate": 2.6615687506848864e-05, "loss": 0.0262, "step": 40330 }, { "epoch": 18.053255761915416, "grad_norm": 0.11634695529937744, "learning_rate": 2.6591332843902884e-05, "loss": 0.0241, "step": 40340 }, { "epoch": 18.057731036025956, "grad_norm": 0.12667925655841827, "learning_rate": 2.656698529189193e-05, "loss": 0.0298, "step": 40350 }, { "epoch": 18.062206310136496, "grad_norm": 0.10401729494333267, "learning_rate": 2.654264485821214e-05, "loss": 0.0257, "step": 40360 }, { "epoch": 18.066681584247036, "grad_norm": 0.1220976710319519, "learning_rate": 2.6518311550257478e-05, "loss": 0.0262, "step": 40370 }, { "epoch": 18.071156858357575, "grad_norm": 0.10220004618167877, "learning_rate": 2.6493985375419778e-05, "loss": 0.025, "step": 40380 }, { "epoch": 18.075632132468115, "grad_norm": 0.089901864528656, "learning_rate": 2.6469666341088677e-05, "loss": 0.0252, "step": 40390 }, { "epoch": 18.08010740657865, "grad_norm": 0.09511516243219376, "learning_rate": 2.644535445465164e-05, "loss": 0.0249, "step": 40400 }, { "epoch": 18.08458268068919, "grad_norm": 0.10121265798807144, "learning_rate": 2.642104972349403e-05, "loss": 0.028, "step": 40410 }, { "epoch": 18.08905795479973, "grad_norm": 0.10552633553743362, "learning_rate": 2.6396752154998915e-05, "loss": 0.0281, "step": 40420 }, { "epoch": 18.09353322891027, "grad_norm": 0.11233502626419067, "learning_rate": 2.6372461756547306e-05, "loss": 0.0311, "step": 40430 }, { "epoch": 18.09800850302081, "grad_norm": 0.13596956431865692, "learning_rate": 2.6348178535517966e-05, "loss": 0.0278, "step": 40440 }, { "epoch": 18.10248377713135, "grad_norm": 0.09534571319818497, "learning_rate": 2.6323902499287473e-05, "loss": 0.026, "step": 40450 }, { "epoch": 18.106959051241887, "grad_norm": 0.12807290256023407, "learning_rate": 2.629963365523031e-05, "loss": 0.029, "step": 40460 }, { "epoch": 18.111434325352427, "grad_norm": 0.10003679990768433, "learning_rate": 2.6275372010718635e-05, "loss": 0.0284, "step": 40470 }, { "epoch": 18.115909599462967, "grad_norm": 0.09679880738258362, "learning_rate": 2.6251117573122563e-05, "loss": 0.0236, "step": 40480 }, { "epoch": 18.120384873573506, "grad_norm": 0.09431570768356323, "learning_rate": 2.6226870349809885e-05, "loss": 0.0246, "step": 40490 }, { "epoch": 18.124860147684046, "grad_norm": 0.0988299697637558, "learning_rate": 2.6202630348146324e-05, "loss": 0.0253, "step": 40500 }, { "epoch": 18.129335421794586, "grad_norm": 0.11003097146749496, "learning_rate": 2.6178397575495328e-05, "loss": 0.0261, "step": 40510 }, { "epoch": 18.133810695905126, "grad_norm": 0.0971628725528717, "learning_rate": 2.6154172039218172e-05, "loss": 0.0243, "step": 40520 }, { "epoch": 18.138285970015662, "grad_norm": 0.10408058762550354, "learning_rate": 2.612995374667394e-05, "loss": 0.0289, "step": 40530 }, { "epoch": 18.142761244126202, "grad_norm": 0.10565894842147827, "learning_rate": 2.6105742705219515e-05, "loss": 0.028, "step": 40540 }, { "epoch": 18.147236518236742, "grad_norm": 0.08722098171710968, "learning_rate": 2.6081538922209535e-05, "loss": 0.0262, "step": 40550 }, { "epoch": 18.15171179234728, "grad_norm": 0.10710829496383667, "learning_rate": 2.6057342404996522e-05, "loss": 0.0274, "step": 40560 }, { "epoch": 18.15618706645782, "grad_norm": 0.10792136192321777, "learning_rate": 2.6033153160930722e-05, "loss": 0.0276, "step": 40570 }, { "epoch": 18.16066234056836, "grad_norm": 0.1130899116396904, "learning_rate": 2.6008971197360176e-05, "loss": 0.0273, "step": 40580 }, { "epoch": 18.165137614678898, "grad_norm": 0.11300644278526306, "learning_rate": 2.5984796521630737e-05, "loss": 0.0244, "step": 40590 }, { "epoch": 18.169612888789437, "grad_norm": 0.10063886642456055, "learning_rate": 2.5960629141086012e-05, "loss": 0.0263, "step": 40600 }, { "epoch": 18.174088162899977, "grad_norm": 0.1034046933054924, "learning_rate": 2.593646906306747e-05, "loss": 0.0255, "step": 40610 }, { "epoch": 18.178563437010517, "grad_norm": 0.09230414777994156, "learning_rate": 2.591231629491423e-05, "loss": 0.0301, "step": 40620 }, { "epoch": 18.183038711121057, "grad_norm": 0.08651192486286163, "learning_rate": 2.5888170843963332e-05, "loss": 0.026, "step": 40630 }, { "epoch": 18.187513985231597, "grad_norm": 0.08527199923992157, "learning_rate": 2.586403271754947e-05, "loss": 0.0266, "step": 40640 }, { "epoch": 18.191989259342133, "grad_norm": 0.08489000052213669, "learning_rate": 2.5839901923005205e-05, "loss": 0.0268, "step": 40650 }, { "epoch": 18.196464533452673, "grad_norm": 0.12551850080490112, "learning_rate": 2.5815778467660823e-05, "loss": 0.0248, "step": 40660 }, { "epoch": 18.200939807563213, "grad_norm": 0.08936256915330887, "learning_rate": 2.57916623588444e-05, "loss": 0.028, "step": 40670 }, { "epoch": 18.205415081673753, "grad_norm": 0.09106092900037766, "learning_rate": 2.5767553603881767e-05, "loss": 0.0228, "step": 40680 }, { "epoch": 18.209890355784292, "grad_norm": 0.10985039174556732, "learning_rate": 2.574345221009653e-05, "loss": 0.0279, "step": 40690 }, { "epoch": 18.214365629894832, "grad_norm": 0.09043742716312408, "learning_rate": 2.571935818481005e-05, "loss": 0.0294, "step": 40700 }, { "epoch": 18.218840904005372, "grad_norm": 0.1080927774310112, "learning_rate": 2.5695271535341443e-05, "loss": 0.0269, "step": 40710 }, { "epoch": 18.22331617811591, "grad_norm": 0.09976181387901306, "learning_rate": 2.567119226900764e-05, "loss": 0.0256, "step": 40720 }, { "epoch": 18.227791452226448, "grad_norm": 0.10087592154741287, "learning_rate": 2.5647120393123246e-05, "loss": 0.0252, "step": 40730 }, { "epoch": 18.232266726336988, "grad_norm": 0.09989523142576218, "learning_rate": 2.562305591500069e-05, "loss": 0.0274, "step": 40740 }, { "epoch": 18.236742000447528, "grad_norm": 0.12048488110303879, "learning_rate": 2.5598998841950107e-05, "loss": 0.0269, "step": 40750 }, { "epoch": 18.241217274558068, "grad_norm": 0.1242198646068573, "learning_rate": 2.55749491812794e-05, "loss": 0.0245, "step": 40760 }, { "epoch": 18.245692548668607, "grad_norm": 0.09615784138441086, "learning_rate": 2.555090694029421e-05, "loss": 0.0268, "step": 40770 }, { "epoch": 18.250167822779144, "grad_norm": 0.09110631048679352, "learning_rate": 2.552687212629799e-05, "loss": 0.0251, "step": 40780 }, { "epoch": 18.254643096889684, "grad_norm": 0.11241819709539413, "learning_rate": 2.5502844746591804e-05, "loss": 0.0274, "step": 40790 }, { "epoch": 18.259118371000223, "grad_norm": 0.11339984089136124, "learning_rate": 2.547882480847461e-05, "loss": 0.0257, "step": 40800 }, { "epoch": 18.263593645110763, "grad_norm": 0.09169421344995499, "learning_rate": 2.545481231924296e-05, "loss": 0.0258, "step": 40810 }, { "epoch": 18.268068919221303, "grad_norm": 0.10276671499013901, "learning_rate": 2.543080728619127e-05, "loss": 0.0247, "step": 40820 }, { "epoch": 18.272544193331843, "grad_norm": 0.11379365622997284, "learning_rate": 2.540680971661161e-05, "loss": 0.0269, "step": 40830 }, { "epoch": 18.27701946744238, "grad_norm": 0.10253484547138214, "learning_rate": 2.5382819617793813e-05, "loss": 0.0238, "step": 40840 }, { "epoch": 18.28149474155292, "grad_norm": 0.08288746327161789, "learning_rate": 2.5358836997025437e-05, "loss": 0.0269, "step": 40850 }, { "epoch": 18.28597001566346, "grad_norm": 0.12124630063772202, "learning_rate": 2.5334861861591753e-05, "loss": 0.0307, "step": 40860 }, { "epoch": 18.290445289774, "grad_norm": 0.089940644800663, "learning_rate": 2.5310894218775805e-05, "loss": 0.0247, "step": 40870 }, { "epoch": 18.29492056388454, "grad_norm": 0.11872800439596176, "learning_rate": 2.528693407585832e-05, "loss": 0.0264, "step": 40880 }, { "epoch": 18.29939583799508, "grad_norm": 0.1183207631111145, "learning_rate": 2.526298144011775e-05, "loss": 0.0297, "step": 40890 }, { "epoch": 18.303871112105618, "grad_norm": 0.11884105950593948, "learning_rate": 2.5239036318830278e-05, "loss": 0.0288, "step": 40900 }, { "epoch": 18.308346386216154, "grad_norm": 0.10016050934791565, "learning_rate": 2.521509871926979e-05, "loss": 0.0271, "step": 40910 }, { "epoch": 18.312821660326694, "grad_norm": 0.10618429630994797, "learning_rate": 2.5191168648707887e-05, "loss": 0.0272, "step": 40920 }, { "epoch": 18.317296934437234, "grad_norm": 0.11775879561901093, "learning_rate": 2.5167246114413956e-05, "loss": 0.0288, "step": 40930 }, { "epoch": 18.321772208547774, "grad_norm": 0.10570640116930008, "learning_rate": 2.5143331123654933e-05, "loss": 0.0268, "step": 40940 }, { "epoch": 18.326247482658314, "grad_norm": 0.1085716187953949, "learning_rate": 2.511942368369566e-05, "loss": 0.0258, "step": 40950 }, { "epoch": 18.330722756768854, "grad_norm": 0.16391734778881073, "learning_rate": 2.5095523801798495e-05, "loss": 0.0256, "step": 40960 }, { "epoch": 18.33519803087939, "grad_norm": 0.10291681438684464, "learning_rate": 2.5071631485223658e-05, "loss": 0.0282, "step": 40970 }, { "epoch": 18.33967330498993, "grad_norm": 0.12458930164575577, "learning_rate": 2.5047746741228978e-05, "loss": 0.0264, "step": 40980 }, { "epoch": 18.34414857910047, "grad_norm": 0.10890492051839828, "learning_rate": 2.5023869577070013e-05, "loss": 0.0278, "step": 40990 }, { "epoch": 18.34862385321101, "grad_norm": 0.13402670621871948, "learning_rate": 2.500000000000001e-05, "loss": 0.0315, "step": 41000 }, { "epoch": 18.35309912732155, "grad_norm": 0.11953707784414291, "learning_rate": 2.4976138017269908e-05, "loss": 0.0285, "step": 41010 }, { "epoch": 18.35757440143209, "grad_norm": 0.10560168325901031, "learning_rate": 2.4952283636128372e-05, "loss": 0.0264, "step": 41020 }, { "epoch": 18.362049675542625, "grad_norm": 0.10450264066457748, "learning_rate": 2.4928436863821725e-05, "loss": 0.0274, "step": 41030 }, { "epoch": 18.366524949653165, "grad_norm": 0.10194500535726547, "learning_rate": 2.490459770759398e-05, "loss": 0.0275, "step": 41040 }, { "epoch": 18.371000223763705, "grad_norm": 0.10323040187358856, "learning_rate": 2.4880766174686848e-05, "loss": 0.0248, "step": 41050 }, { "epoch": 18.375475497874245, "grad_norm": 0.09424642473459244, "learning_rate": 2.485694227233971e-05, "loss": 0.0287, "step": 41060 }, { "epoch": 18.379950771984785, "grad_norm": 0.09249241650104523, "learning_rate": 2.4833126007789653e-05, "loss": 0.027, "step": 41070 }, { "epoch": 18.384426046095324, "grad_norm": 0.11273205280303955, "learning_rate": 2.4809317388271426e-05, "loss": 0.0238, "step": 41080 }, { "epoch": 18.388901320205864, "grad_norm": 0.09139535576105118, "learning_rate": 2.478551642101743e-05, "loss": 0.0258, "step": 41090 }, { "epoch": 18.3933765943164, "grad_norm": 0.12815262377262115, "learning_rate": 2.476172311325783e-05, "loss": 0.0266, "step": 41100 }, { "epoch": 18.39785186842694, "grad_norm": 0.1054467186331749, "learning_rate": 2.4737937472220336e-05, "loss": 0.0253, "step": 41110 }, { "epoch": 18.40232714253748, "grad_norm": 0.10787903517484665, "learning_rate": 2.4714159505130452e-05, "loss": 0.0254, "step": 41120 }, { "epoch": 18.40680241664802, "grad_norm": 0.1239837259054184, "learning_rate": 2.4690389219211273e-05, "loss": 0.0275, "step": 41130 }, { "epoch": 18.41127769075856, "grad_norm": 0.11382511258125305, "learning_rate": 2.4666626621683592e-05, "loss": 0.0247, "step": 41140 }, { "epoch": 18.4157529648691, "grad_norm": 0.11288341879844666, "learning_rate": 2.4642871719765852e-05, "loss": 0.0284, "step": 41150 }, { "epoch": 18.420228238979636, "grad_norm": 0.10892853885889053, "learning_rate": 2.4619124520674146e-05, "loss": 0.0254, "step": 41160 }, { "epoch": 18.424703513090176, "grad_norm": 0.10715874284505844, "learning_rate": 2.459538503162231e-05, "loss": 0.0287, "step": 41170 }, { "epoch": 18.429178787200716, "grad_norm": 0.12336155027151108, "learning_rate": 2.4571653259821694e-05, "loss": 0.0265, "step": 41180 }, { "epoch": 18.433654061311255, "grad_norm": 0.10335179418325424, "learning_rate": 2.4547929212481435e-05, "loss": 0.0281, "step": 41190 }, { "epoch": 18.438129335421795, "grad_norm": 0.11257285624742508, "learning_rate": 2.4524212896808263e-05, "loss": 0.024, "step": 41200 }, { "epoch": 18.442604609532335, "grad_norm": 0.1034289300441742, "learning_rate": 2.4500504320006562e-05, "loss": 0.0285, "step": 41210 }, { "epoch": 18.44707988364287, "grad_norm": 0.12373567372560501, "learning_rate": 2.447680348927837e-05, "loss": 0.0263, "step": 41220 }, { "epoch": 18.45155515775341, "grad_norm": 0.07855936139822006, "learning_rate": 2.4453110411823382e-05, "loss": 0.0255, "step": 41230 }, { "epoch": 18.45603043186395, "grad_norm": 0.08476711809635162, "learning_rate": 2.4429425094838903e-05, "loss": 0.0254, "step": 41240 }, { "epoch": 18.46050570597449, "grad_norm": 0.10228963941335678, "learning_rate": 2.4405747545519963e-05, "loss": 0.0281, "step": 41250 }, { "epoch": 18.46498098008503, "grad_norm": 0.08626151829957962, "learning_rate": 2.438207777105911e-05, "loss": 0.0253, "step": 41260 }, { "epoch": 18.46945625419557, "grad_norm": 0.11029870808124542, "learning_rate": 2.4358415778646643e-05, "loss": 0.0247, "step": 41270 }, { "epoch": 18.47393152830611, "grad_norm": 0.10470389574766159, "learning_rate": 2.433476157547044e-05, "loss": 0.0259, "step": 41280 }, { "epoch": 18.478406802416647, "grad_norm": 0.09558732807636261, "learning_rate": 2.4311115168716013e-05, "loss": 0.0244, "step": 41290 }, { "epoch": 18.482882076527186, "grad_norm": 0.11527890712022781, "learning_rate": 2.4287476565566527e-05, "loss": 0.029, "step": 41300 }, { "epoch": 18.487357350637726, "grad_norm": 0.11236262321472168, "learning_rate": 2.4263845773202736e-05, "loss": 0.0261, "step": 41310 }, { "epoch": 18.491832624748266, "grad_norm": 0.09150092303752899, "learning_rate": 2.424022279880312e-05, "loss": 0.0266, "step": 41320 }, { "epoch": 18.496307898858806, "grad_norm": 0.10240788012742996, "learning_rate": 2.4216607649543628e-05, "loss": 0.0241, "step": 41330 }, { "epoch": 18.500783172969346, "grad_norm": 0.1050696074962616, "learning_rate": 2.419300033259798e-05, "loss": 0.0256, "step": 41340 }, { "epoch": 18.505258447079882, "grad_norm": 0.10295847058296204, "learning_rate": 2.4169400855137436e-05, "loss": 0.0294, "step": 41350 }, { "epoch": 18.509733721190422, "grad_norm": 0.0955745130777359, "learning_rate": 2.4145809224330896e-05, "loss": 0.0243, "step": 41360 }, { "epoch": 18.51420899530096, "grad_norm": 0.0928441509604454, "learning_rate": 2.4122225447344875e-05, "loss": 0.024, "step": 41370 }, { "epoch": 18.5186842694115, "grad_norm": 0.08643052726984024, "learning_rate": 2.4098649531343497e-05, "loss": 0.0283, "step": 41380 }, { "epoch": 18.52315954352204, "grad_norm": 0.09498447179794312, "learning_rate": 2.4075081483488494e-05, "loss": 0.0228, "step": 41390 }, { "epoch": 18.52763481763258, "grad_norm": 0.11252176761627197, "learning_rate": 2.405152131093926e-05, "loss": 0.0263, "step": 41400 }, { "epoch": 18.53211009174312, "grad_norm": 0.09955187886953354, "learning_rate": 2.402796902085268e-05, "loss": 0.0311, "step": 41410 }, { "epoch": 18.536585365853657, "grad_norm": 0.1125381663441658, "learning_rate": 2.4004424620383386e-05, "loss": 0.0263, "step": 41420 }, { "epoch": 18.541060639964197, "grad_norm": 0.11458365619182587, "learning_rate": 2.3980888116683515e-05, "loss": 0.0242, "step": 41430 }, { "epoch": 18.545535914074737, "grad_norm": 0.12929819524288177, "learning_rate": 2.3957359516902845e-05, "loss": 0.0257, "step": 41440 }, { "epoch": 18.550011188185277, "grad_norm": 0.0972658097743988, "learning_rate": 2.3933838828188733e-05, "loss": 0.031, "step": 41450 }, { "epoch": 18.554486462295817, "grad_norm": 0.11463576555252075, "learning_rate": 2.3910326057686127e-05, "loss": 0.0264, "step": 41460 }, { "epoch": 18.558961736406356, "grad_norm": 0.09947403520345688, "learning_rate": 2.3886821212537647e-05, "loss": 0.027, "step": 41470 }, { "epoch": 18.563437010516893, "grad_norm": 0.09305677562952042, "learning_rate": 2.3863324299883366e-05, "loss": 0.0276, "step": 41480 }, { "epoch": 18.567912284627432, "grad_norm": 0.1007392629981041, "learning_rate": 2.3839835326861104e-05, "loss": 0.0258, "step": 41490 }, { "epoch": 18.572387558737972, "grad_norm": 0.11260421574115753, "learning_rate": 2.381635430060611e-05, "loss": 0.0288, "step": 41500 }, { "epoch": 18.576862832848512, "grad_norm": 0.0922384113073349, "learning_rate": 2.3792881228251356e-05, "loss": 0.0252, "step": 41510 }, { "epoch": 18.581338106959052, "grad_norm": 0.09548863023519516, "learning_rate": 2.3769416116927335e-05, "loss": 0.0225, "step": 41520 }, { "epoch": 18.585813381069592, "grad_norm": 0.11019735038280487, "learning_rate": 2.374595897376211e-05, "loss": 0.0244, "step": 41530 }, { "epoch": 18.590288655180128, "grad_norm": 0.11400657147169113, "learning_rate": 2.3722509805881356e-05, "loss": 0.0249, "step": 41540 }, { "epoch": 18.594763929290668, "grad_norm": 0.09743454307317734, "learning_rate": 2.3699068620408304e-05, "loss": 0.0291, "step": 41550 }, { "epoch": 18.599239203401208, "grad_norm": 0.09259656816720963, "learning_rate": 2.3675635424463754e-05, "loss": 0.0288, "step": 41560 }, { "epoch": 18.603714477511748, "grad_norm": 0.10165227204561234, "learning_rate": 2.3652210225166122e-05, "loss": 0.0258, "step": 41570 }, { "epoch": 18.608189751622287, "grad_norm": 0.11902398616075516, "learning_rate": 2.362879302963135e-05, "loss": 0.0246, "step": 41580 }, { "epoch": 18.612665025732827, "grad_norm": 0.09915472567081451, "learning_rate": 2.3605383844972966e-05, "loss": 0.027, "step": 41590 }, { "epoch": 18.617140299843367, "grad_norm": 0.1041216105222702, "learning_rate": 2.3581982678302063e-05, "loss": 0.0254, "step": 41600 }, { "epoch": 18.621615573953903, "grad_norm": 0.09934891015291214, "learning_rate": 2.3558589536727277e-05, "loss": 0.0267, "step": 41610 }, { "epoch": 18.626090848064443, "grad_norm": 0.10428817570209503, "learning_rate": 2.353520442735488e-05, "loss": 0.0262, "step": 41620 }, { "epoch": 18.630566122174983, "grad_norm": 0.10479407757520676, "learning_rate": 2.3511827357288575e-05, "loss": 0.026, "step": 41630 }, { "epoch": 18.635041396285523, "grad_norm": 0.10610247403383255, "learning_rate": 2.3488458333629777e-05, "loss": 0.0241, "step": 41640 }, { "epoch": 18.639516670396063, "grad_norm": 0.09419596195220947, "learning_rate": 2.3465097363477306e-05, "loss": 0.0285, "step": 41650 }, { "epoch": 18.643991944506602, "grad_norm": 0.11028274893760681, "learning_rate": 2.344174445392766e-05, "loss": 0.0266, "step": 41660 }, { "epoch": 18.64846721861714, "grad_norm": 0.1078493520617485, "learning_rate": 2.341839961207482e-05, "loss": 0.0269, "step": 41670 }, { "epoch": 18.65294249272768, "grad_norm": 0.12006353586912155, "learning_rate": 2.339506284501033e-05, "loss": 0.028, "step": 41680 }, { "epoch": 18.65741776683822, "grad_norm": 0.09423677623271942, "learning_rate": 2.3371734159823284e-05, "loss": 0.0264, "step": 41690 }, { "epoch": 18.66189304094876, "grad_norm": 0.08965729922056198, "learning_rate": 2.3348413563600325e-05, "loss": 0.0275, "step": 41700 }, { "epoch": 18.666368315059298, "grad_norm": 0.12014691531658173, "learning_rate": 2.3325101063425608e-05, "loss": 0.0252, "step": 41710 }, { "epoch": 18.670843589169838, "grad_norm": 0.09071977436542511, "learning_rate": 2.3301796666380898e-05, "loss": 0.0246, "step": 41720 }, { "epoch": 18.675318863280374, "grad_norm": 0.0982757955789566, "learning_rate": 2.3278500379545436e-05, "loss": 0.0261, "step": 41730 }, { "epoch": 18.679794137390914, "grad_norm": 0.10947185009717941, "learning_rate": 2.3255212209996025e-05, "loss": 0.0253, "step": 41740 }, { "epoch": 18.684269411501454, "grad_norm": 0.08686758577823639, "learning_rate": 2.323193216480698e-05, "loss": 0.0258, "step": 41750 }, { "epoch": 18.688744685611994, "grad_norm": 0.09266096353530884, "learning_rate": 2.3208660251050158e-05, "loss": 0.0261, "step": 41760 }, { "epoch": 18.693219959722533, "grad_norm": 0.10226021707057953, "learning_rate": 2.3185396475795007e-05, "loss": 0.0268, "step": 41770 }, { "epoch": 18.697695233833073, "grad_norm": 0.11018811911344528, "learning_rate": 2.3162140846108366e-05, "loss": 0.0262, "step": 41780 }, { "epoch": 18.702170507943613, "grad_norm": 0.09715130925178528, "learning_rate": 2.3138893369054766e-05, "loss": 0.024, "step": 41790 }, { "epoch": 18.70664578205415, "grad_norm": 0.11009696125984192, "learning_rate": 2.3115654051696095e-05, "loss": 0.0271, "step": 41800 }, { "epoch": 18.71112105616469, "grad_norm": 0.09806596487760544, "learning_rate": 2.3092422901091897e-05, "loss": 0.0239, "step": 41810 }, { "epoch": 18.71559633027523, "grad_norm": 0.11044594645500183, "learning_rate": 2.3069199924299174e-05, "loss": 0.0244, "step": 41820 }, { "epoch": 18.72007160438577, "grad_norm": 0.10783880203962326, "learning_rate": 2.3045985128372442e-05, "loss": 0.0245, "step": 41830 }, { "epoch": 18.72454687849631, "grad_norm": 0.14464139938354492, "learning_rate": 2.3022778520363753e-05, "loss": 0.0283, "step": 41840 }, { "epoch": 18.72902215260685, "grad_norm": 0.10104833543300629, "learning_rate": 2.2999580107322653e-05, "loss": 0.026, "step": 41850 }, { "epoch": 18.733497426717385, "grad_norm": 0.09913673996925354, "learning_rate": 2.2976389896296203e-05, "loss": 0.0272, "step": 41860 }, { "epoch": 18.737972700827925, "grad_norm": 0.1074201837182045, "learning_rate": 2.2953207894328992e-05, "loss": 0.0235, "step": 41870 }, { "epoch": 18.742447974938464, "grad_norm": 0.09927832335233688, "learning_rate": 2.29300341084631e-05, "loss": 0.0243, "step": 41880 }, { "epoch": 18.746923249049004, "grad_norm": 0.11839217692613602, "learning_rate": 2.2906868545738102e-05, "loss": 0.0272, "step": 41890 }, { "epoch": 18.751398523159544, "grad_norm": 0.10216718912124634, "learning_rate": 2.288371121319109e-05, "loss": 0.0252, "step": 41900 }, { "epoch": 18.755873797270084, "grad_norm": 0.11618883907794952, "learning_rate": 2.2860562117856647e-05, "loss": 0.0246, "step": 41910 }, { "epoch": 18.760349071380624, "grad_norm": 0.09433958679437637, "learning_rate": 2.2837421266766857e-05, "loss": 0.0239, "step": 41920 }, { "epoch": 18.76482434549116, "grad_norm": 0.09428665786981583, "learning_rate": 2.281428866695128e-05, "loss": 0.0266, "step": 41930 }, { "epoch": 18.7692996196017, "grad_norm": 0.10757685452699661, "learning_rate": 2.279116432543705e-05, "loss": 0.0273, "step": 41940 }, { "epoch": 18.77377489371224, "grad_norm": 0.13262413442134857, "learning_rate": 2.2768048249248648e-05, "loss": 0.0272, "step": 41950 }, { "epoch": 18.77825016782278, "grad_norm": 0.10291531682014465, "learning_rate": 2.2744940445408202e-05, "loss": 0.0233, "step": 41960 }, { "epoch": 18.78272544193332, "grad_norm": 0.11892333626747131, "learning_rate": 2.2721840920935196e-05, "loss": 0.0254, "step": 41970 }, { "epoch": 18.78720071604386, "grad_norm": 0.09914720803499222, "learning_rate": 2.2698749682846687e-05, "loss": 0.0283, "step": 41980 }, { "epoch": 18.791675990154395, "grad_norm": 0.11024247854948044, "learning_rate": 2.2675666738157186e-05, "loss": 0.025, "step": 41990 }, { "epoch": 18.796151264264935, "grad_norm": 0.07860592752695084, "learning_rate": 2.2652592093878666e-05, "loss": 0.0264, "step": 42000 }, { "epoch": 18.800626538375475, "grad_norm": 0.11989726126194, "learning_rate": 2.26295257570206e-05, "loss": 0.0275, "step": 42010 }, { "epoch": 18.805101812486015, "grad_norm": 0.0895085334777832, "learning_rate": 2.2606467734589924e-05, "loss": 0.026, "step": 42020 }, { "epoch": 18.809577086596555, "grad_norm": 0.10768409818410873, "learning_rate": 2.258341803359108e-05, "loss": 0.0295, "step": 42030 }, { "epoch": 18.814052360707095, "grad_norm": 0.13516457378864288, "learning_rate": 2.2560376661025945e-05, "loss": 0.0236, "step": 42040 }, { "epoch": 18.81852763481763, "grad_norm": 0.10237962752580643, "learning_rate": 2.2537343623893874e-05, "loss": 0.0255, "step": 42050 }, { "epoch": 18.82300290892817, "grad_norm": 0.11831894516944885, "learning_rate": 2.251431892919171e-05, "loss": 0.0253, "step": 42060 }, { "epoch": 18.82747818303871, "grad_norm": 0.09747354686260223, "learning_rate": 2.249130258391373e-05, "loss": 0.0252, "step": 42070 }, { "epoch": 18.83195345714925, "grad_norm": 0.11818135529756546, "learning_rate": 2.2468294595051687e-05, "loss": 0.0255, "step": 42080 }, { "epoch": 18.83642873125979, "grad_norm": 0.08894097059965134, "learning_rate": 2.2445294969594844e-05, "loss": 0.0276, "step": 42090 }, { "epoch": 18.84090400537033, "grad_norm": 0.0997655987739563, "learning_rate": 2.242230371452982e-05, "loss": 0.0283, "step": 42100 }, { "epoch": 18.84537927948087, "grad_norm": 0.11598735302686691, "learning_rate": 2.239932083684081e-05, "loss": 0.0261, "step": 42110 }, { "epoch": 18.849854553591406, "grad_norm": 0.09545058012008667, "learning_rate": 2.237634634350934e-05, "loss": 0.0239, "step": 42120 }, { "epoch": 18.854329827701946, "grad_norm": 0.10056865960359573, "learning_rate": 2.2353380241514515e-05, "loss": 0.0271, "step": 42130 }, { "epoch": 18.858805101812486, "grad_norm": 0.10265415161848068, "learning_rate": 2.23304225378328e-05, "loss": 0.0222, "step": 42140 }, { "epoch": 18.863280375923026, "grad_norm": 0.08597219735383987, "learning_rate": 2.2307473239438154e-05, "loss": 0.0233, "step": 42150 }, { "epoch": 18.867755650033565, "grad_norm": 0.11170630156993866, "learning_rate": 2.2284532353301953e-05, "loss": 0.0281, "step": 42160 }, { "epoch": 18.872230924144105, "grad_norm": 0.10501236468553543, "learning_rate": 2.2261599886393014e-05, "loss": 0.0248, "step": 42170 }, { "epoch": 18.87670619825464, "grad_norm": 0.11515551060438156, "learning_rate": 2.2238675845677663e-05, "loss": 0.0241, "step": 42180 }, { "epoch": 18.88118147236518, "grad_norm": 0.08168746531009674, "learning_rate": 2.22157602381196e-05, "loss": 0.0267, "step": 42190 }, { "epoch": 18.88565674647572, "grad_norm": 0.11320456862449646, "learning_rate": 2.219285307067997e-05, "loss": 0.023, "step": 42200 }, { "epoch": 18.89013202058626, "grad_norm": 0.08965571224689484, "learning_rate": 2.2169954350317374e-05, "loss": 0.0203, "step": 42210 }, { "epoch": 18.8946072946968, "grad_norm": 0.11564560979604721, "learning_rate": 2.2147064083987838e-05, "loss": 0.026, "step": 42220 }, { "epoch": 18.89908256880734, "grad_norm": 0.10051178187131882, "learning_rate": 2.21241822786448e-05, "loss": 0.0287, "step": 42230 }, { "epoch": 18.903557842917877, "grad_norm": 0.08708908408880234, "learning_rate": 2.2101308941239203e-05, "loss": 0.025, "step": 42240 }, { "epoch": 18.908033117028417, "grad_norm": 0.08095596730709076, "learning_rate": 2.207844407871929e-05, "loss": 0.0277, "step": 42250 }, { "epoch": 18.912508391138957, "grad_norm": 0.08221080154180527, "learning_rate": 2.2055587698030877e-05, "loss": 0.0244, "step": 42260 }, { "epoch": 18.916983665249496, "grad_norm": 0.11123578995466232, "learning_rate": 2.2032739806117058e-05, "loss": 0.0287, "step": 42270 }, { "epoch": 18.921458939360036, "grad_norm": 0.1134650781750679, "learning_rate": 2.2009900409918465e-05, "loss": 0.0251, "step": 42280 }, { "epoch": 18.925934213470576, "grad_norm": 0.11513236910104752, "learning_rate": 2.1987069516373098e-05, "loss": 0.025, "step": 42290 }, { "epoch": 18.930409487581116, "grad_norm": 0.10151014477014542, "learning_rate": 2.196424713241637e-05, "loss": 0.0272, "step": 42300 }, { "epoch": 18.934884761691652, "grad_norm": 0.10495670139789581, "learning_rate": 2.1941433264981125e-05, "loss": 0.0275, "step": 42310 }, { "epoch": 18.939360035802192, "grad_norm": 0.10784729570150375, "learning_rate": 2.1918627920997593e-05, "loss": 0.026, "step": 42320 }, { "epoch": 18.943835309912732, "grad_norm": 0.10229761898517609, "learning_rate": 2.1895831107393484e-05, "loss": 0.0288, "step": 42330 }, { "epoch": 18.94831058402327, "grad_norm": 0.10470611602067947, "learning_rate": 2.1873042831093803e-05, "loss": 0.0256, "step": 42340 }, { "epoch": 18.95278585813381, "grad_norm": 0.11287857592105865, "learning_rate": 2.1850263099021077e-05, "loss": 0.0277, "step": 42350 }, { "epoch": 18.95726113224435, "grad_norm": 0.11469049006700516, "learning_rate": 2.182749191809518e-05, "loss": 0.028, "step": 42360 }, { "epoch": 18.961736406354888, "grad_norm": 0.09632209688425064, "learning_rate": 2.180472929523338e-05, "loss": 0.0271, "step": 42370 }, { "epoch": 18.966211680465427, "grad_norm": 0.08329709619283676, "learning_rate": 2.1781975237350366e-05, "loss": 0.0242, "step": 42380 }, { "epoch": 18.970686954575967, "grad_norm": 0.12254087626934052, "learning_rate": 2.1759229751358217e-05, "loss": 0.0261, "step": 42390 }, { "epoch": 18.975162228686507, "grad_norm": 0.10663367062807083, "learning_rate": 2.1736492844166407e-05, "loss": 0.0273, "step": 42400 }, { "epoch": 18.979637502797047, "grad_norm": 0.11630713939666748, "learning_rate": 2.1713764522681846e-05, "loss": 0.0258, "step": 42410 }, { "epoch": 18.984112776907587, "grad_norm": 0.09251351654529572, "learning_rate": 2.1691044793808734e-05, "loss": 0.0254, "step": 42420 }, { "epoch": 18.988588051018127, "grad_norm": 0.07894586026668549, "learning_rate": 2.1668333664448776e-05, "loss": 0.0238, "step": 42430 }, { "epoch": 18.993063325128663, "grad_norm": 0.09753181785345078, "learning_rate": 2.1645631141500994e-05, "loss": 0.0239, "step": 42440 }, { "epoch": 18.997538599239203, "grad_norm": 0.16706036031246185, "learning_rate": 2.1622937231861822e-05, "loss": 0.0255, "step": 42450 }, { "epoch": 19.002013873349743, "grad_norm": 0.09109700471162796, "learning_rate": 2.1600251942425066e-05, "loss": 0.0255, "step": 42460 }, { "epoch": 19.006489147460282, "grad_norm": 0.0926537811756134, "learning_rate": 2.1577575280081896e-05, "loss": 0.0243, "step": 42470 }, { "epoch": 19.010964421570822, "grad_norm": 0.11471781879663467, "learning_rate": 2.1554907251720945e-05, "loss": 0.0242, "step": 42480 }, { "epoch": 19.015439695681362, "grad_norm": 0.11010495573282242, "learning_rate": 2.1532247864228084e-05, "loss": 0.0239, "step": 42490 }, { "epoch": 19.0199149697919, "grad_norm": 0.09352383762598038, "learning_rate": 2.150959712448669e-05, "loss": 0.0249, "step": 42500 }, { "epoch": 19.024390243902438, "grad_norm": 0.09770387411117554, "learning_rate": 2.148695503937745e-05, "loss": 0.0248, "step": 42510 }, { "epoch": 19.028865518012978, "grad_norm": 0.08757480978965759, "learning_rate": 2.1464321615778422e-05, "loss": 0.029, "step": 42520 }, { "epoch": 19.033340792123518, "grad_norm": 0.09746949374675751, "learning_rate": 2.1441696860565048e-05, "loss": 0.0262, "step": 42530 }, { "epoch": 19.037816066234058, "grad_norm": 0.07725771516561508, "learning_rate": 2.1419080780610123e-05, "loss": 0.0271, "step": 42540 }, { "epoch": 19.042291340344597, "grad_norm": 0.13394732773303986, "learning_rate": 2.139647338278381e-05, "loss": 0.0242, "step": 42550 }, { "epoch": 19.046766614455134, "grad_norm": 0.10532264411449432, "learning_rate": 2.1373874673953685e-05, "loss": 0.0236, "step": 42560 }, { "epoch": 19.051241888565674, "grad_norm": 0.08417601883411407, "learning_rate": 2.1351284660984572e-05, "loss": 0.0238, "step": 42570 }, { "epoch": 19.055717162676213, "grad_norm": 0.09688309580087662, "learning_rate": 2.1328703350738765e-05, "loss": 0.0258, "step": 42580 }, { "epoch": 19.060192436786753, "grad_norm": 0.09950326383113861, "learning_rate": 2.1306130750075865e-05, "loss": 0.022, "step": 42590 }, { "epoch": 19.064667710897293, "grad_norm": 0.09690635651350021, "learning_rate": 2.128356686585282e-05, "loss": 0.0265, "step": 42600 }, { "epoch": 19.069142985007833, "grad_norm": 0.09644177556037903, "learning_rate": 2.1261011704923955e-05, "loss": 0.0244, "step": 42610 }, { "epoch": 19.073618259118373, "grad_norm": 0.1048724427819252, "learning_rate": 2.12384652741409e-05, "loss": 0.0238, "step": 42620 }, { "epoch": 19.07809353322891, "grad_norm": 0.12032526731491089, "learning_rate": 2.121592758035273e-05, "loss": 0.0269, "step": 42630 }, { "epoch": 19.08256880733945, "grad_norm": 0.08956959843635559, "learning_rate": 2.1193398630405725e-05, "loss": 0.0257, "step": 42640 }, { "epoch": 19.08704408144999, "grad_norm": 0.09766323864459991, "learning_rate": 2.1170878431143637e-05, "loss": 0.0269, "step": 42650 }, { "epoch": 19.09151935556053, "grad_norm": 0.09732164442539215, "learning_rate": 2.1148366989407496e-05, "loss": 0.0282, "step": 42660 }, { "epoch": 19.09599462967107, "grad_norm": 0.1328122466802597, "learning_rate": 2.112586431203568e-05, "loss": 0.0279, "step": 42670 }, { "epoch": 19.100469903781608, "grad_norm": 0.10893654823303223, "learning_rate": 2.110337040586391e-05, "loss": 0.0244, "step": 42680 }, { "epoch": 19.104945177892144, "grad_norm": 0.09433279931545258, "learning_rate": 2.1080885277725236e-05, "loss": 0.0245, "step": 42690 }, { "epoch": 19.109420452002684, "grad_norm": 0.10756467282772064, "learning_rate": 2.105840893445005e-05, "loss": 0.0242, "step": 42700 }, { "epoch": 19.113895726113224, "grad_norm": 0.0966048613190651, "learning_rate": 2.103594138286607e-05, "loss": 0.024, "step": 42710 }, { "epoch": 19.118371000223764, "grad_norm": 0.10515527427196503, "learning_rate": 2.1013482629798333e-05, "loss": 0.0262, "step": 42720 }, { "epoch": 19.122846274334304, "grad_norm": 0.10400571674108505, "learning_rate": 2.0991032682069246e-05, "loss": 0.0292, "step": 42730 }, { "epoch": 19.127321548444844, "grad_norm": 0.11050355434417725, "learning_rate": 2.0968591546498488e-05, "loss": 0.0256, "step": 42740 }, { "epoch": 19.13179682255538, "grad_norm": 0.11088448762893677, "learning_rate": 2.094615922990309e-05, "loss": 0.0254, "step": 42750 }, { "epoch": 19.13627209666592, "grad_norm": 0.11177422851324081, "learning_rate": 2.09237357390974e-05, "loss": 0.0251, "step": 42760 }, { "epoch": 19.14074737077646, "grad_norm": 0.10123983025550842, "learning_rate": 2.0901321080893056e-05, "loss": 0.026, "step": 42770 }, { "epoch": 19.145222644887, "grad_norm": 0.10081936419010162, "learning_rate": 2.0878915262099098e-05, "loss": 0.0244, "step": 42780 }, { "epoch": 19.14969791899754, "grad_norm": 0.08873939514160156, "learning_rate": 2.085651828952175e-05, "loss": 0.0278, "step": 42790 }, { "epoch": 19.15417319310808, "grad_norm": 0.09278330206871033, "learning_rate": 2.0834130169964692e-05, "loss": 0.0242, "step": 42800 }, { "epoch": 19.15864846721862, "grad_norm": 0.09024201333522797, "learning_rate": 2.0811750910228774e-05, "loss": 0.0257, "step": 42810 }, { "epoch": 19.163123741329155, "grad_norm": 0.09791358560323715, "learning_rate": 2.0789380517112272e-05, "loss": 0.0265, "step": 42820 }, { "epoch": 19.167599015439695, "grad_norm": 0.09215427935123444, "learning_rate": 2.0767018997410713e-05, "loss": 0.0269, "step": 42830 }, { "epoch": 19.172074289550235, "grad_norm": 0.10973268747329712, "learning_rate": 2.0744666357916925e-05, "loss": 0.0241, "step": 42840 }, { "epoch": 19.176549563660775, "grad_norm": 0.09497436881065369, "learning_rate": 2.072232260542106e-05, "loss": 0.0212, "step": 42850 }, { "epoch": 19.181024837771314, "grad_norm": 0.10638841986656189, "learning_rate": 2.0699987746710554e-05, "loss": 0.0234, "step": 42860 }, { "epoch": 19.185500111881854, "grad_norm": 0.08283191174268723, "learning_rate": 2.067766178857013e-05, "loss": 0.0264, "step": 42870 }, { "epoch": 19.18997538599239, "grad_norm": 0.11476048827171326, "learning_rate": 2.065534473778186e-05, "loss": 0.0262, "step": 42880 }, { "epoch": 19.19445066010293, "grad_norm": 0.09669831395149231, "learning_rate": 2.063303660112506e-05, "loss": 0.0244, "step": 42890 }, { "epoch": 19.19892593421347, "grad_norm": 0.08843382447957993, "learning_rate": 2.061073738537635e-05, "loss": 0.0267, "step": 42900 }, { "epoch": 19.20340120832401, "grad_norm": 0.09081729501485825, "learning_rate": 2.0588447097309645e-05, "loss": 0.0285, "step": 42910 }, { "epoch": 19.20787648243455, "grad_norm": 0.1282443255186081, "learning_rate": 2.056616574369612e-05, "loss": 0.0284, "step": 42920 }, { "epoch": 19.21235175654509, "grad_norm": 0.1036304384469986, "learning_rate": 2.0543893331304333e-05, "loss": 0.0231, "step": 42930 }, { "epoch": 19.216827030655626, "grad_norm": 0.0918002650141716, "learning_rate": 2.0521629866899966e-05, "loss": 0.0229, "step": 42940 }, { "epoch": 19.221302304766166, "grad_norm": 0.0959216058254242, "learning_rate": 2.0499375357246148e-05, "loss": 0.0255, "step": 42950 }, { "epoch": 19.225777578876706, "grad_norm": 0.10566829144954681, "learning_rate": 2.0477129809103147e-05, "loss": 0.0236, "step": 42960 }, { "epoch": 19.230252852987245, "grad_norm": 0.10810498148202896, "learning_rate": 2.0454893229228617e-05, "loss": 0.026, "step": 42970 }, { "epoch": 19.234728127097785, "grad_norm": 0.0938263088464737, "learning_rate": 2.0432665624377434e-05, "loss": 0.0256, "step": 42980 }, { "epoch": 19.239203401208325, "grad_norm": 0.08573058247566223, "learning_rate": 2.0410447001301753e-05, "loss": 0.0229, "step": 42990 }, { "epoch": 19.243678675318865, "grad_norm": 0.12742294371128082, "learning_rate": 2.0388237366751006e-05, "loss": 0.0268, "step": 43000 }, { "epoch": 19.2481539494294, "grad_norm": 0.09953684359788895, "learning_rate": 2.03660367274719e-05, "loss": 0.0236, "step": 43010 }, { "epoch": 19.25262922353994, "grad_norm": 0.10041037946939468, "learning_rate": 2.0343845090208368e-05, "loss": 0.023, "step": 43020 }, { "epoch": 19.25710449765048, "grad_norm": 0.08606907725334167, "learning_rate": 2.0321662461701696e-05, "loss": 0.0248, "step": 43030 }, { "epoch": 19.26157977176102, "grad_norm": 0.0721895843744278, "learning_rate": 2.0299488848690355e-05, "loss": 0.026, "step": 43040 }, { "epoch": 19.26605504587156, "grad_norm": 0.08075106143951416, "learning_rate": 2.0277324257910106e-05, "loss": 0.0295, "step": 43050 }, { "epoch": 19.2705303199821, "grad_norm": 0.10670875757932663, "learning_rate": 2.0255168696093968e-05, "loss": 0.0271, "step": 43060 }, { "epoch": 19.275005594092637, "grad_norm": 0.08975598961114883, "learning_rate": 2.0233022169972192e-05, "loss": 0.0238, "step": 43070 }, { "epoch": 19.279480868203176, "grad_norm": 0.1018344908952713, "learning_rate": 2.0210884686272368e-05, "loss": 0.025, "step": 43080 }, { "epoch": 19.283956142313716, "grad_norm": 0.0952652096748352, "learning_rate": 2.0188756251719203e-05, "loss": 0.0278, "step": 43090 }, { "epoch": 19.288431416424256, "grad_norm": 0.0977315604686737, "learning_rate": 2.0166636873034805e-05, "loss": 0.0226, "step": 43100 }, { "epoch": 19.292906690534796, "grad_norm": 0.07868881523609161, "learning_rate": 2.0144526556938387e-05, "loss": 0.0248, "step": 43110 }, { "epoch": 19.297381964645336, "grad_norm": 0.09130308777093887, "learning_rate": 2.0122425310146542e-05, "loss": 0.0264, "step": 43120 }, { "epoch": 19.301857238755872, "grad_norm": 0.11168379336595535, "learning_rate": 2.0100333139372985e-05, "loss": 0.0266, "step": 43130 }, { "epoch": 19.306332512866412, "grad_norm": 0.09305288642644882, "learning_rate": 2.0078250051328784e-05, "loss": 0.0268, "step": 43140 }, { "epoch": 19.31080778697695, "grad_norm": 0.10903023183345795, "learning_rate": 2.0056176052722174e-05, "loss": 0.0232, "step": 43150 }, { "epoch": 19.31528306108749, "grad_norm": 0.08903105556964874, "learning_rate": 2.0034111150258666e-05, "loss": 0.0252, "step": 43160 }, { "epoch": 19.31975833519803, "grad_norm": 0.07635834813117981, "learning_rate": 2.0012055350640986e-05, "loss": 0.0279, "step": 43170 }, { "epoch": 19.32423360930857, "grad_norm": 0.08418940007686615, "learning_rate": 1.999000866056908e-05, "loss": 0.0278, "step": 43180 }, { "epoch": 19.32870888341911, "grad_norm": 0.11500602215528488, "learning_rate": 1.9967971086740195e-05, "loss": 0.0244, "step": 43190 }, { "epoch": 19.333184157529647, "grad_norm": 0.10653950273990631, "learning_rate": 1.9945942635848748e-05, "loss": 0.0235, "step": 43200 }, { "epoch": 19.337659431640187, "grad_norm": 0.12274147570133209, "learning_rate": 1.9923923314586397e-05, "loss": 0.0266, "step": 43210 }, { "epoch": 19.342134705750727, "grad_norm": 0.09834884852170944, "learning_rate": 1.9901913129642024e-05, "loss": 0.026, "step": 43220 }, { "epoch": 19.346609979861267, "grad_norm": 0.07499609887599945, "learning_rate": 1.9879912087701753e-05, "loss": 0.0232, "step": 43230 }, { "epoch": 19.351085253971807, "grad_norm": 0.10019140690565109, "learning_rate": 1.98579201954489e-05, "loss": 0.0314, "step": 43240 }, { "epoch": 19.355560528082346, "grad_norm": 0.09200822561979294, "learning_rate": 1.9835937459564064e-05, "loss": 0.0253, "step": 43250 }, { "epoch": 19.360035802192883, "grad_norm": 0.0850878432393074, "learning_rate": 1.981396388672496e-05, "loss": 0.0287, "step": 43260 }, { "epoch": 19.364511076303422, "grad_norm": 0.09865380823612213, "learning_rate": 1.9791999483606648e-05, "loss": 0.0264, "step": 43270 }, { "epoch": 19.368986350413962, "grad_norm": 0.0850420668721199, "learning_rate": 1.977004425688126e-05, "loss": 0.0268, "step": 43280 }, { "epoch": 19.373461624524502, "grad_norm": 0.09177303314208984, "learning_rate": 1.974809821321827e-05, "loss": 0.0274, "step": 43290 }, { "epoch": 19.377936898635042, "grad_norm": 0.09028152376413345, "learning_rate": 1.9726161359284286e-05, "loss": 0.0262, "step": 43300 }, { "epoch": 19.382412172745582, "grad_norm": 0.10327286273241043, "learning_rate": 1.9704233701743148e-05, "loss": 0.026, "step": 43310 }, { "epoch": 19.38688744685612, "grad_norm": 0.09204687923192978, "learning_rate": 1.9682315247255894e-05, "loss": 0.025, "step": 43320 }, { "epoch": 19.391362720966658, "grad_norm": 0.1095535010099411, "learning_rate": 1.9660406002480765e-05, "loss": 0.0255, "step": 43330 }, { "epoch": 19.395837995077198, "grad_norm": 0.10410057008266449, "learning_rate": 1.9638505974073234e-05, "loss": 0.0244, "step": 43340 }, { "epoch": 19.400313269187738, "grad_norm": 0.08429017663002014, "learning_rate": 1.9616615168685943e-05, "loss": 0.0271, "step": 43350 }, { "epoch": 19.404788543298277, "grad_norm": 0.1006828024983406, "learning_rate": 1.9594733592968733e-05, "loss": 0.0227, "step": 43360 }, { "epoch": 19.409263817408817, "grad_norm": 0.09153228253126144, "learning_rate": 1.9572861253568657e-05, "loss": 0.0249, "step": 43370 }, { "epoch": 19.413739091519357, "grad_norm": 0.11383718997240067, "learning_rate": 1.9550998157129946e-05, "loss": 0.0244, "step": 43380 }, { "epoch": 19.418214365629893, "grad_norm": 0.08728727698326111, "learning_rate": 1.9529144310294023e-05, "loss": 0.0277, "step": 43390 }, { "epoch": 19.422689639740433, "grad_norm": 0.09735829383134842, "learning_rate": 1.950729971969955e-05, "loss": 0.024, "step": 43400 }, { "epoch": 19.427164913850973, "grad_norm": 0.0989517942070961, "learning_rate": 1.9485464391982284e-05, "loss": 0.0254, "step": 43410 }, { "epoch": 19.431640187961513, "grad_norm": 0.085866779088974, "learning_rate": 1.9463638333775276e-05, "loss": 0.0261, "step": 43420 }, { "epoch": 19.436115462072053, "grad_norm": 0.0941055417060852, "learning_rate": 1.944182155170864e-05, "loss": 0.0258, "step": 43430 }, { "epoch": 19.440590736182592, "grad_norm": 0.10506486892700195, "learning_rate": 1.942001405240979e-05, "loss": 0.0234, "step": 43440 }, { "epoch": 19.44506601029313, "grad_norm": 0.07749630510807037, "learning_rate": 1.939821584250326e-05, "loss": 0.0237, "step": 43450 }, { "epoch": 19.44954128440367, "grad_norm": 0.1079121008515358, "learning_rate": 1.937642692861076e-05, "loss": 0.0271, "step": 43460 }, { "epoch": 19.45401655851421, "grad_norm": 0.08546680957078934, "learning_rate": 1.9354647317351188e-05, "loss": 0.0245, "step": 43470 }, { "epoch": 19.45849183262475, "grad_norm": 0.1175324097275734, "learning_rate": 1.93328770153406e-05, "loss": 0.0236, "step": 43480 }, { "epoch": 19.462967106735288, "grad_norm": 0.09493392705917358, "learning_rate": 1.9311116029192278e-05, "loss": 0.0251, "step": 43490 }, { "epoch": 19.467442380845828, "grad_norm": 0.10112158954143524, "learning_rate": 1.928936436551661e-05, "loss": 0.0279, "step": 43500 }, { "epoch": 19.471917654956368, "grad_norm": 0.09575123339891434, "learning_rate": 1.9267622030921184e-05, "loss": 0.0259, "step": 43510 }, { "epoch": 19.476392929066904, "grad_norm": 0.08910831809043884, "learning_rate": 1.924588903201074e-05, "loss": 0.0278, "step": 43520 }, { "epoch": 19.480868203177444, "grad_norm": 0.09281788021326065, "learning_rate": 1.9224165375387193e-05, "loss": 0.0266, "step": 43530 }, { "epoch": 19.485343477287984, "grad_norm": 0.1008223220705986, "learning_rate": 1.920245106764962e-05, "loss": 0.0281, "step": 43540 }, { "epoch": 19.489818751398523, "grad_norm": 0.11716561764478683, "learning_rate": 1.9180746115394243e-05, "loss": 0.0269, "step": 43550 }, { "epoch": 19.494294025509063, "grad_norm": 0.10360610485076904, "learning_rate": 1.9159050525214452e-05, "loss": 0.0276, "step": 43560 }, { "epoch": 19.498769299619603, "grad_norm": 0.09970337152481079, "learning_rate": 1.9137364303700833e-05, "loss": 0.022, "step": 43570 }, { "epoch": 19.50324457373014, "grad_norm": 0.08366693556308746, "learning_rate": 1.9115687457441022e-05, "loss": 0.0248, "step": 43580 }, { "epoch": 19.50771984784068, "grad_norm": 0.08319178223609924, "learning_rate": 1.909401999301993e-05, "loss": 0.0254, "step": 43590 }, { "epoch": 19.51219512195122, "grad_norm": 0.10893997550010681, "learning_rate": 1.9072361917019536e-05, "loss": 0.0268, "step": 43600 }, { "epoch": 19.51667039606176, "grad_norm": 0.1022329330444336, "learning_rate": 1.9050713236018998e-05, "loss": 0.0231, "step": 43610 }, { "epoch": 19.5211456701723, "grad_norm": 0.11150937527418137, "learning_rate": 1.9029073956594606e-05, "loss": 0.0274, "step": 43620 }, { "epoch": 19.52562094428284, "grad_norm": 0.11181092262268066, "learning_rate": 1.9007444085319786e-05, "loss": 0.0281, "step": 43630 }, { "epoch": 19.530096218393375, "grad_norm": 0.1165379136800766, "learning_rate": 1.8985823628765188e-05, "loss": 0.0288, "step": 43640 }, { "epoch": 19.534571492503915, "grad_norm": 0.10671112686395645, "learning_rate": 1.8964212593498442e-05, "loss": 0.0247, "step": 43650 }, { "epoch": 19.539046766614454, "grad_norm": 0.11104617267847061, "learning_rate": 1.8942610986084486e-05, "loss": 0.026, "step": 43660 }, { "epoch": 19.543522040724994, "grad_norm": 0.0852707251906395, "learning_rate": 1.8921018813085283e-05, "loss": 0.0276, "step": 43670 }, { "epoch": 19.547997314835534, "grad_norm": 0.12202014774084091, "learning_rate": 1.8899436081059975e-05, "loss": 0.028, "step": 43680 }, { "epoch": 19.552472588946074, "grad_norm": 0.09011222422122955, "learning_rate": 1.887786279656482e-05, "loss": 0.0254, "step": 43690 }, { "epoch": 19.556947863056614, "grad_norm": 0.11647108197212219, "learning_rate": 1.8856298966153212e-05, "loss": 0.0293, "step": 43700 }, { "epoch": 19.56142313716715, "grad_norm": 0.07753046602010727, "learning_rate": 1.8834744596375666e-05, "loss": 0.0254, "step": 43710 }, { "epoch": 19.56589841127769, "grad_norm": 0.09894897788763046, "learning_rate": 1.881319969377987e-05, "loss": 0.0248, "step": 43720 }, { "epoch": 19.57037368538823, "grad_norm": 0.09107312560081482, "learning_rate": 1.8791664264910537e-05, "loss": 0.0321, "step": 43730 }, { "epoch": 19.57484895949877, "grad_norm": 0.11665727943181992, "learning_rate": 1.877013831630961e-05, "loss": 0.0246, "step": 43740 }, { "epoch": 19.57932423360931, "grad_norm": 0.09700016677379608, "learning_rate": 1.874862185451608e-05, "loss": 0.0226, "step": 43750 }, { "epoch": 19.58379950771985, "grad_norm": 0.10766100138425827, "learning_rate": 1.872711488606609e-05, "loss": 0.0283, "step": 43760 }, { "epoch": 19.588274781830386, "grad_norm": 0.11112651973962784, "learning_rate": 1.8705617417492883e-05, "loss": 0.0247, "step": 43770 }, { "epoch": 19.592750055940925, "grad_norm": 0.08885182440280914, "learning_rate": 1.868412945532681e-05, "loss": 0.0258, "step": 43780 }, { "epoch": 19.597225330051465, "grad_norm": 0.11814051121473312, "learning_rate": 1.8662651006095387e-05, "loss": 0.0236, "step": 43790 }, { "epoch": 19.601700604162005, "grad_norm": 0.12529921531677246, "learning_rate": 1.8641182076323148e-05, "loss": 0.0275, "step": 43800 }, { "epoch": 19.606175878272545, "grad_norm": 0.09791426360607147, "learning_rate": 1.8619722672531825e-05, "loss": 0.0247, "step": 43810 }, { "epoch": 19.610651152383085, "grad_norm": 0.10012223571538925, "learning_rate": 1.8598272801240213e-05, "loss": 0.0282, "step": 43820 }, { "epoch": 19.615126426493624, "grad_norm": 0.11190581321716309, "learning_rate": 1.85768324689642e-05, "loss": 0.027, "step": 43830 }, { "epoch": 19.61960170060416, "grad_norm": 0.0972650796175003, "learning_rate": 1.855540168221681e-05, "loss": 0.0229, "step": 43840 }, { "epoch": 19.6240769747147, "grad_norm": 0.09134052693843842, "learning_rate": 1.8533980447508137e-05, "loss": 0.0248, "step": 43850 }, { "epoch": 19.62855224882524, "grad_norm": 0.08420588076114655, "learning_rate": 1.851256877134538e-05, "loss": 0.023, "step": 43860 }, { "epoch": 19.63302752293578, "grad_norm": 0.0808253288269043, "learning_rate": 1.849116666023288e-05, "loss": 0.0254, "step": 43870 }, { "epoch": 19.63750279704632, "grad_norm": 0.1011141687631607, "learning_rate": 1.846977412067198e-05, "loss": 0.0225, "step": 43880 }, { "epoch": 19.64197807115686, "grad_norm": 0.09628481417894363, "learning_rate": 1.8448391159161204e-05, "loss": 0.0252, "step": 43890 }, { "epoch": 19.646453345267396, "grad_norm": 0.10189926624298096, "learning_rate": 1.8427017782196127e-05, "loss": 0.0254, "step": 43900 }, { "epoch": 19.650928619377936, "grad_norm": 0.08753615617752075, "learning_rate": 1.840565399626939e-05, "loss": 0.0242, "step": 43910 }, { "epoch": 19.655403893488476, "grad_norm": 0.08176752179861069, "learning_rate": 1.838429980787081e-05, "loss": 0.0234, "step": 43920 }, { "epoch": 19.659879167599016, "grad_norm": 0.0995369553565979, "learning_rate": 1.8362955223487143e-05, "loss": 0.0254, "step": 43930 }, { "epoch": 19.664354441709555, "grad_norm": 0.08822035789489746, "learning_rate": 1.8341620249602387e-05, "loss": 0.0255, "step": 43940 }, { "epoch": 19.668829715820095, "grad_norm": 0.13426855206489563, "learning_rate": 1.8320294892697478e-05, "loss": 0.0253, "step": 43950 }, { "epoch": 19.67330498993063, "grad_norm": 0.12039274722337723, "learning_rate": 1.8298979159250557e-05, "loss": 0.0226, "step": 43960 }, { "epoch": 19.67778026404117, "grad_norm": 0.10992763936519623, "learning_rate": 1.8277673055736715e-05, "loss": 0.0245, "step": 43970 }, { "epoch": 19.68225553815171, "grad_norm": 0.09549752622842789, "learning_rate": 1.8256376588628238e-05, "loss": 0.0247, "step": 43980 }, { "epoch": 19.68673081226225, "grad_norm": 0.0971386507153511, "learning_rate": 1.8235089764394408e-05, "loss": 0.0249, "step": 43990 }, { "epoch": 19.69120608637279, "grad_norm": 0.09002842009067535, "learning_rate": 1.821381258950161e-05, "loss": 0.0225, "step": 44000 }, { "epoch": 19.69568136048333, "grad_norm": 0.10324123501777649, "learning_rate": 1.8192545070413282e-05, "loss": 0.027, "step": 44010 }, { "epoch": 19.70015663459387, "grad_norm": 0.0942336842417717, "learning_rate": 1.817128721358991e-05, "loss": 0.0234, "step": 44020 }, { "epoch": 19.704631908704407, "grad_norm": 0.10344292968511581, "learning_rate": 1.8150039025489113e-05, "loss": 0.026, "step": 44030 }, { "epoch": 19.709107182814947, "grad_norm": 0.11041392385959625, "learning_rate": 1.8128800512565513e-05, "loss": 0.027, "step": 44040 }, { "epoch": 19.713582456925486, "grad_norm": 0.09221185743808746, "learning_rate": 1.810757168127081e-05, "loss": 0.0257, "step": 44050 }, { "epoch": 19.718057731036026, "grad_norm": 0.09878844022750854, "learning_rate": 1.808635253805376e-05, "loss": 0.0275, "step": 44060 }, { "epoch": 19.722533005146566, "grad_norm": 0.09864797443151474, "learning_rate": 1.8065143089360172e-05, "loss": 0.0249, "step": 44070 }, { "epoch": 19.727008279257106, "grad_norm": 0.09896019846200943, "learning_rate": 1.8043943341632907e-05, "loss": 0.0245, "step": 44080 }, { "epoch": 19.731483553367642, "grad_norm": 0.10441941767930984, "learning_rate": 1.8022753301311935e-05, "loss": 0.0239, "step": 44090 }, { "epoch": 19.735958827478182, "grad_norm": 0.10640055686235428, "learning_rate": 1.800157297483417e-05, "loss": 0.0241, "step": 44100 }, { "epoch": 19.740434101588722, "grad_norm": 0.0932871624827385, "learning_rate": 1.7980402368633703e-05, "loss": 0.0238, "step": 44110 }, { "epoch": 19.74490937569926, "grad_norm": 0.09116657078266144, "learning_rate": 1.7959241489141525e-05, "loss": 0.0251, "step": 44120 }, { "epoch": 19.7493846498098, "grad_norm": 0.07138507068157196, "learning_rate": 1.7938090342785817e-05, "loss": 0.0246, "step": 44130 }, { "epoch": 19.75385992392034, "grad_norm": 0.08356033265590668, "learning_rate": 1.7916948935991718e-05, "loss": 0.0234, "step": 44140 }, { "epoch": 19.758335198030878, "grad_norm": 0.11001971364021301, "learning_rate": 1.789581727518143e-05, "loss": 0.0279, "step": 44150 }, { "epoch": 19.762810472141418, "grad_norm": 0.08963297307491302, "learning_rate": 1.787469536677419e-05, "loss": 0.0259, "step": 44160 }, { "epoch": 19.767285746251957, "grad_norm": 0.11708585917949677, "learning_rate": 1.785358321718626e-05, "loss": 0.0261, "step": 44170 }, { "epoch": 19.771761020362497, "grad_norm": 0.08763671666383743, "learning_rate": 1.7832480832830987e-05, "loss": 0.0272, "step": 44180 }, { "epoch": 19.776236294473037, "grad_norm": 0.10132744908332825, "learning_rate": 1.7811388220118707e-05, "loss": 0.0247, "step": 44190 }, { "epoch": 19.780711568583577, "grad_norm": 0.1044347807765007, "learning_rate": 1.7790305385456795e-05, "loss": 0.0276, "step": 44200 }, { "epoch": 19.785186842694117, "grad_norm": 0.08039877563714981, "learning_rate": 1.7769232335249657e-05, "loss": 0.0229, "step": 44210 }, { "epoch": 19.789662116804653, "grad_norm": 0.11399157345294952, "learning_rate": 1.774816907589873e-05, "loss": 0.0259, "step": 44220 }, { "epoch": 19.794137390915193, "grad_norm": 0.1031387597322464, "learning_rate": 1.7727115613802465e-05, "loss": 0.0243, "step": 44230 }, { "epoch": 19.798612665025733, "grad_norm": 0.10426655411720276, "learning_rate": 1.770607195535639e-05, "loss": 0.0263, "step": 44240 }, { "epoch": 19.803087939136272, "grad_norm": 0.09889999032020569, "learning_rate": 1.768503810695295e-05, "loss": 0.0234, "step": 44250 }, { "epoch": 19.807563213246812, "grad_norm": 0.09754256159067154, "learning_rate": 1.7664014074981742e-05, "loss": 0.0272, "step": 44260 }, { "epoch": 19.812038487357352, "grad_norm": 0.09321747720241547, "learning_rate": 1.764299986582924e-05, "loss": 0.0269, "step": 44270 }, { "epoch": 19.81651376146789, "grad_norm": 0.10501128435134888, "learning_rate": 1.7621995485879062e-05, "loss": 0.0253, "step": 44280 }, { "epoch": 19.820989035578428, "grad_norm": 0.11199890822172165, "learning_rate": 1.7601000941511757e-05, "loss": 0.0262, "step": 44290 }, { "epoch": 19.825464309688968, "grad_norm": 0.10929009318351746, "learning_rate": 1.7580016239104924e-05, "loss": 0.0233, "step": 44300 }, { "epoch": 19.829939583799508, "grad_norm": 0.13320523500442505, "learning_rate": 1.755904138503316e-05, "loss": 0.0291, "step": 44310 }, { "epoch": 19.834414857910048, "grad_norm": 0.10093508660793304, "learning_rate": 1.753807638566805e-05, "loss": 0.0238, "step": 44320 }, { "epoch": 19.838890132020587, "grad_norm": 0.09030492603778839, "learning_rate": 1.751712124737826e-05, "loss": 0.0211, "step": 44330 }, { "epoch": 19.843365406131127, "grad_norm": 0.09752381592988968, "learning_rate": 1.749617597652934e-05, "loss": 0.0239, "step": 44340 }, { "epoch": 19.847840680241664, "grad_norm": 0.12640294432640076, "learning_rate": 1.7475240579483958e-05, "loss": 0.0256, "step": 44350 }, { "epoch": 19.852315954352203, "grad_norm": 0.09587319195270538, "learning_rate": 1.745431506260173e-05, "loss": 0.0254, "step": 44360 }, { "epoch": 19.856791228462743, "grad_norm": 0.11091336607933044, "learning_rate": 1.743339943223926e-05, "loss": 0.0276, "step": 44370 }, { "epoch": 19.861266502573283, "grad_norm": 0.0823993980884552, "learning_rate": 1.7412493694750176e-05, "loss": 0.0269, "step": 44380 }, { "epoch": 19.865741776683823, "grad_norm": 0.09758573770523071, "learning_rate": 1.7391597856485083e-05, "loss": 0.0256, "step": 44390 }, { "epoch": 19.870217050794363, "grad_norm": 0.09708986431360245, "learning_rate": 1.7370711923791567e-05, "loss": 0.0258, "step": 44400 }, { "epoch": 19.8746923249049, "grad_norm": 0.11638449132442474, "learning_rate": 1.7349835903014277e-05, "loss": 0.024, "step": 44410 }, { "epoch": 19.87916759901544, "grad_norm": 0.08145073056221008, "learning_rate": 1.7328969800494726e-05, "loss": 0.0236, "step": 44420 }, { "epoch": 19.88364287312598, "grad_norm": 0.11242931336164474, "learning_rate": 1.7308113622571544e-05, "loss": 0.028, "step": 44430 }, { "epoch": 19.88811814723652, "grad_norm": 0.09214776009321213, "learning_rate": 1.7287267375580256e-05, "loss": 0.0279, "step": 44440 }, { "epoch": 19.89259342134706, "grad_norm": 0.11314426362514496, "learning_rate": 1.7266431065853416e-05, "loss": 0.0251, "step": 44450 }, { "epoch": 19.897068695457598, "grad_norm": 0.09857405722141266, "learning_rate": 1.7245604699720535e-05, "loss": 0.0283, "step": 44460 }, { "epoch": 19.901543969568134, "grad_norm": 0.1136384904384613, "learning_rate": 1.7224788283508103e-05, "loss": 0.0258, "step": 44470 }, { "epoch": 19.906019243678674, "grad_norm": 0.09902313351631165, "learning_rate": 1.7203981823539643e-05, "loss": 0.026, "step": 44480 }, { "epoch": 19.910494517789214, "grad_norm": 0.088041752576828, "learning_rate": 1.7183185326135543e-05, "loss": 0.0232, "step": 44490 }, { "epoch": 19.914969791899754, "grad_norm": 0.09009792655706406, "learning_rate": 1.7162398797613282e-05, "loss": 0.0226, "step": 44500 }, { "epoch": 19.919445066010294, "grad_norm": 0.0983462929725647, "learning_rate": 1.7141622244287235e-05, "loss": 0.027, "step": 44510 }, { "epoch": 19.923920340120834, "grad_norm": 0.09254274517297745, "learning_rate": 1.712085567246878e-05, "loss": 0.0281, "step": 44520 }, { "epoch": 19.92839561423137, "grad_norm": 0.10182631760835648, "learning_rate": 1.7100099088466242e-05, "loss": 0.0266, "step": 44530 }, { "epoch": 19.93287088834191, "grad_norm": 0.09368887543678284, "learning_rate": 1.7079352498584934e-05, "loss": 0.0258, "step": 44540 }, { "epoch": 19.93734616245245, "grad_norm": 0.10117658972740173, "learning_rate": 1.70586159091271e-05, "loss": 0.0286, "step": 44550 }, { "epoch": 19.94182143656299, "grad_norm": 0.09119237959384918, "learning_rate": 1.703788932639202e-05, "loss": 0.0273, "step": 44560 }, { "epoch": 19.94629671067353, "grad_norm": 0.06838057935237885, "learning_rate": 1.7017172756675813e-05, "loss": 0.0248, "step": 44570 }, { "epoch": 19.95077198478407, "grad_norm": 0.08926523476839066, "learning_rate": 1.699646620627168e-05, "loss": 0.0219, "step": 44580 }, { "epoch": 19.95524725889461, "grad_norm": 0.0939595103263855, "learning_rate": 1.6975769681469705e-05, "loss": 0.0266, "step": 44590 }, { "epoch": 19.959722533005145, "grad_norm": 0.09806382656097412, "learning_rate": 1.6955083188556947e-05, "loss": 0.0249, "step": 44600 }, { "epoch": 19.964197807115685, "grad_norm": 0.0920901969075203, "learning_rate": 1.6934406733817414e-05, "loss": 0.025, "step": 44610 }, { "epoch": 19.968673081226225, "grad_norm": 0.10171191394329071, "learning_rate": 1.691374032353205e-05, "loss": 0.026, "step": 44620 }, { "epoch": 19.973148355336765, "grad_norm": 0.09154309332370758, "learning_rate": 1.689308396397882e-05, "loss": 0.0275, "step": 44630 }, { "epoch": 19.977623629447304, "grad_norm": 0.08745722472667694, "learning_rate": 1.6872437661432517e-05, "loss": 0.0265, "step": 44640 }, { "epoch": 19.982098903557844, "grad_norm": 0.08804555237293243, "learning_rate": 1.685180142216498e-05, "loss": 0.0239, "step": 44650 }, { "epoch": 19.98657417766838, "grad_norm": 0.09693148732185364, "learning_rate": 1.6831175252444943e-05, "loss": 0.0225, "step": 44660 }, { "epoch": 19.99104945177892, "grad_norm": 0.0855623334646225, "learning_rate": 1.6810559158538092e-05, "loss": 0.0205, "step": 44670 }, { "epoch": 19.99552472588946, "grad_norm": 0.07334919273853302, "learning_rate": 1.6789953146707053e-05, "loss": 0.0209, "step": 44680 }, { "epoch": 20.0, "grad_norm": 0.09589879959821701, "learning_rate": 1.676935722321139e-05, "loss": 0.0266, "step": 44690 }, { "epoch": 20.00447527411054, "grad_norm": 0.1178017258644104, "learning_rate": 1.6748771394307585e-05, "loss": 0.0248, "step": 44700 }, { "epoch": 20.00895054822108, "grad_norm": 0.10480465739965439, "learning_rate": 1.672819566624911e-05, "loss": 0.0253, "step": 44710 }, { "epoch": 20.01342582233162, "grad_norm": 0.07882021367549896, "learning_rate": 1.6707630045286265e-05, "loss": 0.028, "step": 44720 }, { "epoch": 20.017901096442156, "grad_norm": 0.08482827246189117, "learning_rate": 1.6687074537666398e-05, "loss": 0.0224, "step": 44730 }, { "epoch": 20.022376370552696, "grad_norm": 0.11232418566942215, "learning_rate": 1.666652914963371e-05, "loss": 0.0218, "step": 44740 }, { "epoch": 20.026851644663235, "grad_norm": 0.10406429320573807, "learning_rate": 1.6645993887429345e-05, "loss": 0.0247, "step": 44750 }, { "epoch": 20.031326918773775, "grad_norm": 0.08407416939735413, "learning_rate": 1.662546875729138e-05, "loss": 0.0215, "step": 44760 }, { "epoch": 20.035802192884315, "grad_norm": 0.08137864619493484, "learning_rate": 1.660495376545478e-05, "loss": 0.0266, "step": 44770 }, { "epoch": 20.040277466994855, "grad_norm": 0.10174823552370071, "learning_rate": 1.658444891815152e-05, "loss": 0.0238, "step": 44780 }, { "epoch": 20.04475274110539, "grad_norm": 0.09414715319871902, "learning_rate": 1.6563954221610355e-05, "loss": 0.0264, "step": 44790 }, { "epoch": 20.04922801521593, "grad_norm": 0.09673429280519485, "learning_rate": 1.6543469682057106e-05, "loss": 0.026, "step": 44800 }, { "epoch": 20.05370328932647, "grad_norm": 0.0881957858800888, "learning_rate": 1.652299530571436e-05, "loss": 0.0236, "step": 44810 }, { "epoch": 20.05817856343701, "grad_norm": 0.0908500999212265, "learning_rate": 1.6502531098801753e-05, "loss": 0.0243, "step": 44820 }, { "epoch": 20.06265383754755, "grad_norm": 0.08644580096006393, "learning_rate": 1.648207706753575e-05, "loss": 0.0249, "step": 44830 }, { "epoch": 20.06712911165809, "grad_norm": 0.09879783540964127, "learning_rate": 1.646163321812974e-05, "loss": 0.0249, "step": 44840 }, { "epoch": 20.071604385768627, "grad_norm": 0.08003653585910797, "learning_rate": 1.6441199556794033e-05, "loss": 0.0211, "step": 44850 }, { "epoch": 20.076079659879166, "grad_norm": 0.10389537364244461, "learning_rate": 1.6420776089735827e-05, "loss": 0.0252, "step": 44860 }, { "epoch": 20.080554933989706, "grad_norm": 0.09553767740726471, "learning_rate": 1.6400362823159228e-05, "loss": 0.0251, "step": 44870 }, { "epoch": 20.085030208100246, "grad_norm": 0.07855000346899033, "learning_rate": 1.637995976326527e-05, "loss": 0.0234, "step": 44880 }, { "epoch": 20.089505482210786, "grad_norm": 0.0901799276471138, "learning_rate": 1.6359566916251845e-05, "loss": 0.0279, "step": 44890 }, { "epoch": 20.093980756321326, "grad_norm": 0.11533284187316895, "learning_rate": 1.633918428831377e-05, "loss": 0.0258, "step": 44900 }, { "epoch": 20.098456030431866, "grad_norm": 0.09338999539613724, "learning_rate": 1.631881188564275e-05, "loss": 0.025, "step": 44910 }, { "epoch": 20.102931304542402, "grad_norm": 0.08693401515483856, "learning_rate": 1.6298449714427355e-05, "loss": 0.0234, "step": 44920 }, { "epoch": 20.10740657865294, "grad_norm": 0.09104791283607483, "learning_rate": 1.6278097780853136e-05, "loss": 0.0227, "step": 44930 }, { "epoch": 20.11188185276348, "grad_norm": 0.09849260002374649, "learning_rate": 1.62577560911024e-05, "loss": 0.0261, "step": 44940 }, { "epoch": 20.11635712687402, "grad_norm": 0.09894324094057083, "learning_rate": 1.623742465135449e-05, "loss": 0.0221, "step": 44950 }, { "epoch": 20.12083240098456, "grad_norm": 0.08460716158151627, "learning_rate": 1.6217103467785484e-05, "loss": 0.0241, "step": 44960 }, { "epoch": 20.1253076750951, "grad_norm": 0.08375568687915802, "learning_rate": 1.6196792546568472e-05, "loss": 0.0237, "step": 44970 }, { "epoch": 20.129782949205637, "grad_norm": 0.1245795264840126, "learning_rate": 1.617649189387337e-05, "loss": 0.0254, "step": 44980 }, { "epoch": 20.134258223316177, "grad_norm": 0.06946686655282974, "learning_rate": 1.615620151586697e-05, "loss": 0.0219, "step": 44990 }, { "epoch": 20.138733497426717, "grad_norm": 0.08520293235778809, "learning_rate": 1.6135921418712956e-05, "loss": 0.0259, "step": 45000 }, { "epoch": 20.143208771537257, "grad_norm": 0.11213190853595734, "learning_rate": 1.6115651608571887e-05, "loss": 0.0235, "step": 45010 }, { "epoch": 20.147684045647797, "grad_norm": 0.08326393365859985, "learning_rate": 1.6095392091601175e-05, "loss": 0.0239, "step": 45020 }, { "epoch": 20.152159319758336, "grad_norm": 0.10926520824432373, "learning_rate": 1.6075142873955164e-05, "loss": 0.0228, "step": 45030 }, { "epoch": 20.156634593868873, "grad_norm": 0.08452285826206207, "learning_rate": 1.6054903961785013e-05, "loss": 0.0214, "step": 45040 }, { "epoch": 20.161109867979413, "grad_norm": 0.09559773653745651, "learning_rate": 1.603467536123877e-05, "loss": 0.024, "step": 45050 }, { "epoch": 20.165585142089952, "grad_norm": 0.11725886911153793, "learning_rate": 1.6014457078461353e-05, "loss": 0.0305, "step": 45060 }, { "epoch": 20.170060416200492, "grad_norm": 0.08731025457382202, "learning_rate": 1.5994249119594517e-05, "loss": 0.0228, "step": 45070 }, { "epoch": 20.174535690311032, "grad_norm": 0.09887498617172241, "learning_rate": 1.597405149077697e-05, "loss": 0.0278, "step": 45080 }, { "epoch": 20.179010964421572, "grad_norm": 0.10007632523775101, "learning_rate": 1.5953864198144135e-05, "loss": 0.0219, "step": 45090 }, { "epoch": 20.18348623853211, "grad_norm": 0.08141699433326721, "learning_rate": 1.593368724782846e-05, "loss": 0.0253, "step": 45100 }, { "epoch": 20.187961512642648, "grad_norm": 0.09817419201135635, "learning_rate": 1.5913520645959097e-05, "loss": 0.0247, "step": 45110 }, { "epoch": 20.192436786753188, "grad_norm": 0.08454658091068268, "learning_rate": 1.5893364398662176e-05, "loss": 0.0247, "step": 45120 }, { "epoch": 20.196912060863728, "grad_norm": 0.07555490732192993, "learning_rate": 1.587321851206061e-05, "loss": 0.0256, "step": 45130 }, { "epoch": 20.201387334974267, "grad_norm": 0.1047835499048233, "learning_rate": 1.5853082992274205e-05, "loss": 0.0248, "step": 45140 }, { "epoch": 20.205862609084807, "grad_norm": 0.09787869453430176, "learning_rate": 1.583295784541958e-05, "loss": 0.025, "step": 45150 }, { "epoch": 20.210337883195347, "grad_norm": 0.11782798916101456, "learning_rate": 1.581284307761024e-05, "loss": 0.0257, "step": 45160 }, { "epoch": 20.214813157305883, "grad_norm": 0.0975295752286911, "learning_rate": 1.5792738694956516e-05, "loss": 0.0262, "step": 45170 }, { "epoch": 20.219288431416423, "grad_norm": 0.08448583632707596, "learning_rate": 1.5772644703565565e-05, "loss": 0.0243, "step": 45180 }, { "epoch": 20.223763705526963, "grad_norm": 0.0938783586025238, "learning_rate": 1.5752561109541447e-05, "loss": 0.0246, "step": 45190 }, { "epoch": 20.228238979637503, "grad_norm": 0.08005109429359436, "learning_rate": 1.5732487918985018e-05, "loss": 0.0229, "step": 45200 }, { "epoch": 20.232714253748043, "grad_norm": 0.09277331829071045, "learning_rate": 1.5712425137993973e-05, "loss": 0.028, "step": 45210 }, { "epoch": 20.237189527858582, "grad_norm": 0.10060814023017883, "learning_rate": 1.569237277266286e-05, "loss": 0.0225, "step": 45220 }, { "epoch": 20.241664801969122, "grad_norm": 0.10814248025417328, "learning_rate": 1.567233082908306e-05, "loss": 0.0258, "step": 45230 }, { "epoch": 20.24614007607966, "grad_norm": 0.11481980234384537, "learning_rate": 1.5652299313342773e-05, "loss": 0.0277, "step": 45240 }, { "epoch": 20.2506153501902, "grad_norm": 0.10928242653608322, "learning_rate": 1.563227823152708e-05, "loss": 0.0216, "step": 45250 }, { "epoch": 20.25509062430074, "grad_norm": 0.08914796262979507, "learning_rate": 1.5612267589717805e-05, "loss": 0.0243, "step": 45260 }, { "epoch": 20.259565898411278, "grad_norm": 0.09631980955600739, "learning_rate": 1.5592267393993716e-05, "loss": 0.0274, "step": 45270 }, { "epoch": 20.264041172521818, "grad_norm": 0.09378641098737717, "learning_rate": 1.557227765043027e-05, "loss": 0.0292, "step": 45280 }, { "epoch": 20.268516446632358, "grad_norm": 0.09988990426063538, "learning_rate": 1.5552298365099882e-05, "loss": 0.0208, "step": 45290 }, { "epoch": 20.272991720742894, "grad_norm": 0.11273593455553055, "learning_rate": 1.553232954407171e-05, "loss": 0.0275, "step": 45300 }, { "epoch": 20.277466994853434, "grad_norm": 0.09701873362064362, "learning_rate": 1.5512371193411762e-05, "loss": 0.0257, "step": 45310 }, { "epoch": 20.281942268963974, "grad_norm": 0.11817958205938339, "learning_rate": 1.549242331918285e-05, "loss": 0.0247, "step": 45320 }, { "epoch": 20.286417543074514, "grad_norm": 0.10564102232456207, "learning_rate": 1.5472485927444597e-05, "loss": 0.0264, "step": 45330 }, { "epoch": 20.290892817185053, "grad_norm": 0.09891942888498306, "learning_rate": 1.5452559024253487e-05, "loss": 0.0238, "step": 45340 }, { "epoch": 20.295368091295593, "grad_norm": 0.12181631475687027, "learning_rate": 1.5432642615662774e-05, "loss": 0.0272, "step": 45350 }, { "epoch": 20.29984336540613, "grad_norm": 0.1282103806734085, "learning_rate": 1.5412736707722537e-05, "loss": 0.0226, "step": 45360 }, { "epoch": 20.30431863951667, "grad_norm": 0.11611621081829071, "learning_rate": 1.5392841306479666e-05, "loss": 0.0255, "step": 45370 }, { "epoch": 20.30879391362721, "grad_norm": 0.08563504368066788, "learning_rate": 1.537295641797785e-05, "loss": 0.0233, "step": 45380 }, { "epoch": 20.31326918773775, "grad_norm": 0.08086259663105011, "learning_rate": 1.5353082048257596e-05, "loss": 0.022, "step": 45390 }, { "epoch": 20.31774446184829, "grad_norm": 0.09203419834375381, "learning_rate": 1.5333218203356243e-05, "loss": 0.024, "step": 45400 }, { "epoch": 20.32221973595883, "grad_norm": 0.08316654711961746, "learning_rate": 1.531336488930784e-05, "loss": 0.0232, "step": 45410 }, { "epoch": 20.32669501006937, "grad_norm": 0.10317549854516983, "learning_rate": 1.5293522112143373e-05, "loss": 0.0266, "step": 45420 }, { "epoch": 20.331170284179905, "grad_norm": 0.116429403424263, "learning_rate": 1.5273689877890485e-05, "loss": 0.0226, "step": 45430 }, { "epoch": 20.335645558290445, "grad_norm": 0.11432287096977234, "learning_rate": 1.5253868192573729e-05, "loss": 0.024, "step": 45440 }, { "epoch": 20.340120832400984, "grad_norm": 0.089174784719944, "learning_rate": 1.5234057062214402e-05, "loss": 0.0231, "step": 45450 }, { "epoch": 20.344596106511524, "grad_norm": 0.07851685583591461, "learning_rate": 1.5214256492830598e-05, "loss": 0.0233, "step": 45460 }, { "epoch": 20.349071380622064, "grad_norm": 0.10208455473184586, "learning_rate": 1.5194466490437203e-05, "loss": 0.0264, "step": 45470 }, { "epoch": 20.353546654732604, "grad_norm": 0.07750819623470306, "learning_rate": 1.517468706104589e-05, "loss": 0.0297, "step": 45480 }, { "epoch": 20.35802192884314, "grad_norm": 0.09051728248596191, "learning_rate": 1.5154918210665148e-05, "loss": 0.0268, "step": 45490 }, { "epoch": 20.36249720295368, "grad_norm": 0.11987569183111191, "learning_rate": 1.5135159945300231e-05, "loss": 0.0248, "step": 45500 }, { "epoch": 20.36697247706422, "grad_norm": 0.0858587846159935, "learning_rate": 1.5115412270953167e-05, "loss": 0.0251, "step": 45510 }, { "epoch": 20.37144775117476, "grad_norm": 0.08460421115159988, "learning_rate": 1.5095675193622777e-05, "loss": 0.0272, "step": 45520 }, { "epoch": 20.3759230252853, "grad_norm": 0.10670498013496399, "learning_rate": 1.5075948719304672e-05, "loss": 0.0268, "step": 45530 }, { "epoch": 20.38039829939584, "grad_norm": 0.11815550923347473, "learning_rate": 1.5056232853991209e-05, "loss": 0.0257, "step": 45540 }, { "epoch": 20.384873573506376, "grad_norm": 0.08015327155590057, "learning_rate": 1.5036527603671608e-05, "loss": 0.0246, "step": 45550 }, { "epoch": 20.389348847616915, "grad_norm": 0.07469115406274796, "learning_rate": 1.5016832974331724e-05, "loss": 0.0202, "step": 45560 }, { "epoch": 20.393824121727455, "grad_norm": 0.08507704734802246, "learning_rate": 1.4997148971954344e-05, "loss": 0.0236, "step": 45570 }, { "epoch": 20.398299395837995, "grad_norm": 0.10122325271368027, "learning_rate": 1.4977475602518876e-05, "loss": 0.0252, "step": 45580 }, { "epoch": 20.402774669948535, "grad_norm": 0.09801167249679565, "learning_rate": 1.4957812872001614e-05, "loss": 0.0243, "step": 45590 }, { "epoch": 20.407249944059075, "grad_norm": 0.08360510319471359, "learning_rate": 1.4938160786375572e-05, "loss": 0.0229, "step": 45600 }, { "epoch": 20.411725218169614, "grad_norm": 0.0953599363565445, "learning_rate": 1.4918519351610527e-05, "loss": 0.0227, "step": 45610 }, { "epoch": 20.41620049228015, "grad_norm": 0.11185566335916519, "learning_rate": 1.4898888573673031e-05, "loss": 0.0252, "step": 45620 }, { "epoch": 20.42067576639069, "grad_norm": 0.0826985239982605, "learning_rate": 1.4879268458526379e-05, "loss": 0.0225, "step": 45630 }, { "epoch": 20.42515104050123, "grad_norm": 0.08204559236764908, "learning_rate": 1.4859659012130695e-05, "loss": 0.0238, "step": 45640 }, { "epoch": 20.42962631461177, "grad_norm": 0.09902003407478333, "learning_rate": 1.4840060240442738e-05, "loss": 0.0266, "step": 45650 }, { "epoch": 20.43410158872231, "grad_norm": 0.09702104330062866, "learning_rate": 1.4820472149416154e-05, "loss": 0.024, "step": 45660 }, { "epoch": 20.43857686283285, "grad_norm": 0.1018274798989296, "learning_rate": 1.480089474500127e-05, "loss": 0.0224, "step": 45670 }, { "epoch": 20.443052136943386, "grad_norm": 0.09139332920312881, "learning_rate": 1.4781328033145187e-05, "loss": 0.0255, "step": 45680 }, { "epoch": 20.447527411053926, "grad_norm": 0.09715580195188522, "learning_rate": 1.4761772019791748e-05, "loss": 0.022, "step": 45690 }, { "epoch": 20.452002685164466, "grad_norm": 0.10340341180562973, "learning_rate": 1.4742226710881558e-05, "loss": 0.0262, "step": 45700 }, { "epoch": 20.456477959275006, "grad_norm": 0.08495616167783737, "learning_rate": 1.472269211235195e-05, "loss": 0.0251, "step": 45710 }, { "epoch": 20.460953233385546, "grad_norm": 0.10530070960521698, "learning_rate": 1.470316823013707e-05, "loss": 0.0282, "step": 45720 }, { "epoch": 20.465428507496085, "grad_norm": 0.08264901489019394, "learning_rate": 1.468365507016769e-05, "loss": 0.0222, "step": 45730 }, { "epoch": 20.469903781606625, "grad_norm": 0.09264574944972992, "learning_rate": 1.4664152638371437e-05, "loss": 0.0245, "step": 45740 }, { "epoch": 20.47437905571716, "grad_norm": 0.08869677037000656, "learning_rate": 1.4644660940672627e-05, "loss": 0.0258, "step": 45750 }, { "epoch": 20.4788543298277, "grad_norm": 0.09717968851327896, "learning_rate": 1.4625179982992321e-05, "loss": 0.0293, "step": 45760 }, { "epoch": 20.48332960393824, "grad_norm": 0.07704336196184158, "learning_rate": 1.4605709771248316e-05, "loss": 0.024, "step": 45770 }, { "epoch": 20.48780487804878, "grad_norm": 0.0999828577041626, "learning_rate": 1.4586250311355132e-05, "loss": 0.023, "step": 45780 }, { "epoch": 20.49228015215932, "grad_norm": 0.1044500544667244, "learning_rate": 1.4566801609224096e-05, "loss": 0.0229, "step": 45790 }, { "epoch": 20.49675542626986, "grad_norm": 0.07742497324943542, "learning_rate": 1.4547363670763137e-05, "loss": 0.0237, "step": 45800 }, { "epoch": 20.501230700380397, "grad_norm": 0.09686195850372314, "learning_rate": 1.4527936501877032e-05, "loss": 0.023, "step": 45810 }, { "epoch": 20.505705974490937, "grad_norm": 0.09599136561155319, "learning_rate": 1.450852010846724e-05, "loss": 0.0228, "step": 45820 }, { "epoch": 20.510181248601477, "grad_norm": 0.11150364577770233, "learning_rate": 1.4489114496431938e-05, "loss": 0.0263, "step": 45830 }, { "epoch": 20.514656522712016, "grad_norm": 0.08381577581167221, "learning_rate": 1.4469719671666043e-05, "loss": 0.0239, "step": 45840 }, { "epoch": 20.519131796822556, "grad_norm": 0.08553172647953033, "learning_rate": 1.4450335640061185e-05, "loss": 0.0231, "step": 45850 }, { "epoch": 20.523607070933096, "grad_norm": 0.101848304271698, "learning_rate": 1.443096240750571e-05, "loss": 0.0261, "step": 45860 }, { "epoch": 20.528082345043632, "grad_norm": 0.08707963675260544, "learning_rate": 1.4411599979884744e-05, "loss": 0.029, "step": 45870 }, { "epoch": 20.532557619154172, "grad_norm": 0.09169992059469223, "learning_rate": 1.439224836308002e-05, "loss": 0.0231, "step": 45880 }, { "epoch": 20.537032893264712, "grad_norm": 0.08055855333805084, "learning_rate": 1.4372907562970079e-05, "loss": 0.024, "step": 45890 }, { "epoch": 20.54150816737525, "grad_norm": 0.09983475506305695, "learning_rate": 1.435357758543015e-05, "loss": 0.0275, "step": 45900 }, { "epoch": 20.54598344148579, "grad_norm": 0.09052024781703949, "learning_rate": 1.433425843633216e-05, "loss": 0.0285, "step": 45910 }, { "epoch": 20.55045871559633, "grad_norm": 0.1080080047249794, "learning_rate": 1.4314950121544756e-05, "loss": 0.0223, "step": 45920 }, { "epoch": 20.55493398970687, "grad_norm": 0.12255092710256577, "learning_rate": 1.4295652646933277e-05, "loss": 0.0252, "step": 45930 }, { "epoch": 20.559409263817408, "grad_norm": 0.10226912796497345, "learning_rate": 1.4276366018359844e-05, "loss": 0.024, "step": 45940 }, { "epoch": 20.563884537927947, "grad_norm": 0.08817465603351593, "learning_rate": 1.4257090241683152e-05, "loss": 0.023, "step": 45950 }, { "epoch": 20.568359812038487, "grad_norm": 0.11458159238100052, "learning_rate": 1.4237825322758736e-05, "loss": 0.0236, "step": 45960 }, { "epoch": 20.572835086149027, "grad_norm": 0.11784665286540985, "learning_rate": 1.4218571267438712e-05, "loss": 0.0241, "step": 45970 }, { "epoch": 20.577310360259567, "grad_norm": 0.12449165433645248, "learning_rate": 1.4199328081572e-05, "loss": 0.024, "step": 45980 }, { "epoch": 20.581785634370107, "grad_norm": 0.10318472236394882, "learning_rate": 1.4180095771004154e-05, "loss": 0.0301, "step": 45990 }, { "epoch": 20.586260908480643, "grad_norm": 0.10051344335079193, "learning_rate": 1.4160874341577446e-05, "loss": 0.0249, "step": 46000 }, { "epoch": 20.590736182591183, "grad_norm": 0.07425229996442795, "learning_rate": 1.4141663799130833e-05, "loss": 0.0302, "step": 46010 }, { "epoch": 20.595211456701723, "grad_norm": 0.09004132449626923, "learning_rate": 1.412246414949997e-05, "loss": 0.0248, "step": 46020 }, { "epoch": 20.599686730812262, "grad_norm": 0.09411828219890594, "learning_rate": 1.4103275398517197e-05, "loss": 0.026, "step": 46030 }, { "epoch": 20.604162004922802, "grad_norm": 0.10052505135536194, "learning_rate": 1.4084097552011571e-05, "loss": 0.0272, "step": 46040 }, { "epoch": 20.608637279033342, "grad_norm": 0.08457311987876892, "learning_rate": 1.4064930615808808e-05, "loss": 0.0274, "step": 46050 }, { "epoch": 20.61311255314388, "grad_norm": 0.09912703186273575, "learning_rate": 1.4045774595731315e-05, "loss": 0.024, "step": 46060 }, { "epoch": 20.617587827254418, "grad_norm": 0.08885177224874496, "learning_rate": 1.4026629497598177e-05, "loss": 0.0261, "step": 46070 }, { "epoch": 20.622063101364958, "grad_norm": 0.09177926182746887, "learning_rate": 1.4007495327225162e-05, "loss": 0.0263, "step": 46080 }, { "epoch": 20.626538375475498, "grad_norm": 0.10078011453151703, "learning_rate": 1.3988372090424773e-05, "loss": 0.0288, "step": 46090 }, { "epoch": 20.631013649586038, "grad_norm": 0.1024729385972023, "learning_rate": 1.3969259793006079e-05, "loss": 0.0281, "step": 46100 }, { "epoch": 20.635488923696578, "grad_norm": 0.0858585461974144, "learning_rate": 1.3950158440774957e-05, "loss": 0.0254, "step": 46110 }, { "epoch": 20.639964197807117, "grad_norm": 0.09280335158109665, "learning_rate": 1.3931068039533823e-05, "loss": 0.0214, "step": 46120 }, { "epoch": 20.644439471917654, "grad_norm": 0.1025930866599083, "learning_rate": 1.3911988595081893e-05, "loss": 0.0248, "step": 46130 }, { "epoch": 20.648914746028193, "grad_norm": 0.08938366174697876, "learning_rate": 1.389292011321498e-05, "loss": 0.0228, "step": 46140 }, { "epoch": 20.653390020138733, "grad_norm": 0.1156732589006424, "learning_rate": 1.3873862599725578e-05, "loss": 0.0272, "step": 46150 }, { "epoch": 20.657865294249273, "grad_norm": 0.08630461990833282, "learning_rate": 1.385481606040287e-05, "loss": 0.023, "step": 46160 }, { "epoch": 20.662340568359813, "grad_norm": 0.09755116701126099, "learning_rate": 1.383578050103268e-05, "loss": 0.0236, "step": 46170 }, { "epoch": 20.666815842470353, "grad_norm": 0.09463081508874893, "learning_rate": 1.3816755927397502e-05, "loss": 0.0255, "step": 46180 }, { "epoch": 20.67129111658089, "grad_norm": 0.08256977796554565, "learning_rate": 1.3797742345276521e-05, "loss": 0.0236, "step": 46190 }, { "epoch": 20.67576639069143, "grad_norm": 0.10329052060842514, "learning_rate": 1.3778739760445552e-05, "loss": 0.0247, "step": 46200 }, { "epoch": 20.68024166480197, "grad_norm": 0.08502804487943649, "learning_rate": 1.3759748178677078e-05, "loss": 0.027, "step": 46210 }, { "epoch": 20.68471693891251, "grad_norm": 0.1079077422618866, "learning_rate": 1.374076760574024e-05, "loss": 0.0276, "step": 46220 }, { "epoch": 20.68919221302305, "grad_norm": 0.0808422639966011, "learning_rate": 1.3721798047400813e-05, "loss": 0.0227, "step": 46230 }, { "epoch": 20.693667487133588, "grad_norm": 0.09798799455165863, "learning_rate": 1.3702839509421305e-05, "loss": 0.0256, "step": 46240 }, { "epoch": 20.698142761244128, "grad_norm": 0.09463116526603699, "learning_rate": 1.368389199756075e-05, "loss": 0.0229, "step": 46250 }, { "epoch": 20.702618035354664, "grad_norm": 0.11442147940397263, "learning_rate": 1.3664955517574968e-05, "loss": 0.0225, "step": 46260 }, { "epoch": 20.707093309465204, "grad_norm": 0.10990161448717117, "learning_rate": 1.3646030075216293e-05, "loss": 0.0262, "step": 46270 }, { "epoch": 20.711568583575744, "grad_norm": 0.09599136561155319, "learning_rate": 1.3627115676233832e-05, "loss": 0.0252, "step": 46280 }, { "epoch": 20.716043857686284, "grad_norm": 0.08091891556978226, "learning_rate": 1.3608212326373249e-05, "loss": 0.023, "step": 46290 }, { "epoch": 20.720519131796824, "grad_norm": 0.07486991584300995, "learning_rate": 1.3589320031376901e-05, "loss": 0.0223, "step": 46300 }, { "epoch": 20.724994405907363, "grad_norm": 0.09711326658725739, "learning_rate": 1.3570438796983753e-05, "loss": 0.0237, "step": 46310 }, { "epoch": 20.7294696800179, "grad_norm": 0.08242142200469971, "learning_rate": 1.3551568628929434e-05, "loss": 0.0236, "step": 46320 }, { "epoch": 20.73394495412844, "grad_norm": 0.08576373010873795, "learning_rate": 1.3532709532946186e-05, "loss": 0.025, "step": 46330 }, { "epoch": 20.73842022823898, "grad_norm": 0.08901677280664444, "learning_rate": 1.3513861514762933e-05, "loss": 0.0278, "step": 46340 }, { "epoch": 20.74289550234952, "grad_norm": 0.11002212762832642, "learning_rate": 1.3495024580105192e-05, "loss": 0.024, "step": 46350 }, { "epoch": 20.74737077646006, "grad_norm": 0.07787686586380005, "learning_rate": 1.3476198734695122e-05, "loss": 0.0283, "step": 46360 }, { "epoch": 20.7518460505706, "grad_norm": 0.09830832481384277, "learning_rate": 1.3457383984251525e-05, "loss": 0.0253, "step": 46370 }, { "epoch": 20.756321324681135, "grad_norm": 0.11042024195194244, "learning_rate": 1.343858033448982e-05, "loss": 0.0254, "step": 46380 }, { "epoch": 20.760796598791675, "grad_norm": 0.09932022541761398, "learning_rate": 1.3419787791122062e-05, "loss": 0.0241, "step": 46390 }, { "epoch": 20.765271872902215, "grad_norm": 0.10871254652738571, "learning_rate": 1.3401006359856915e-05, "loss": 0.0265, "step": 46400 }, { "epoch": 20.769747147012755, "grad_norm": 0.09379199892282486, "learning_rate": 1.3382236046399722e-05, "loss": 0.025, "step": 46410 }, { "epoch": 20.774222421123294, "grad_norm": 0.08572835475206375, "learning_rate": 1.3363476856452356e-05, "loss": 0.0242, "step": 46420 }, { "epoch": 20.778697695233834, "grad_norm": 0.104039266705513, "learning_rate": 1.3344728795713413e-05, "loss": 0.0268, "step": 46430 }, { "epoch": 20.78317296934437, "grad_norm": 0.09968183189630508, "learning_rate": 1.3325991869878013e-05, "loss": 0.0227, "step": 46440 }, { "epoch": 20.78764824345491, "grad_norm": 0.10832429677248001, "learning_rate": 1.330726608463797e-05, "loss": 0.0246, "step": 46450 }, { "epoch": 20.79212351756545, "grad_norm": 0.09090448170900345, "learning_rate": 1.328855144568168e-05, "loss": 0.0295, "step": 46460 }, { "epoch": 20.79659879167599, "grad_norm": 0.11420460045337677, "learning_rate": 1.3269847958694148e-05, "loss": 0.0246, "step": 46470 }, { "epoch": 20.80107406578653, "grad_norm": 0.07775325328111649, "learning_rate": 1.325115562935701e-05, "loss": 0.0251, "step": 46480 }, { "epoch": 20.80554933989707, "grad_norm": 0.09363845735788345, "learning_rate": 1.323247446334847e-05, "loss": 0.0225, "step": 46490 }, { "epoch": 20.81002461400761, "grad_norm": 0.0891273245215416, "learning_rate": 1.3213804466343421e-05, "loss": 0.026, "step": 46500 }, { "epoch": 20.814499888118146, "grad_norm": 0.0831855833530426, "learning_rate": 1.3195145644013285e-05, "loss": 0.0233, "step": 46510 }, { "epoch": 20.818975162228686, "grad_norm": 0.12167749553918839, "learning_rate": 1.3176498002026128e-05, "loss": 0.0254, "step": 46520 }, { "epoch": 20.823450436339225, "grad_norm": 0.08736425638198853, "learning_rate": 1.3157861546046613e-05, "loss": 0.0238, "step": 46530 }, { "epoch": 20.827925710449765, "grad_norm": 0.10269371420145035, "learning_rate": 1.3139236281736e-05, "loss": 0.0241, "step": 46540 }, { "epoch": 20.832400984560305, "grad_norm": 0.07023721933364868, "learning_rate": 1.3120622214752132e-05, "loss": 0.0241, "step": 46550 }, { "epoch": 20.836876258670845, "grad_norm": 0.10603215545415878, "learning_rate": 1.3102019350749528e-05, "loss": 0.0282, "step": 46560 }, { "epoch": 20.84135153278138, "grad_norm": 0.09744709730148315, "learning_rate": 1.3083427695379175e-05, "loss": 0.0238, "step": 46570 }, { "epoch": 20.84582680689192, "grad_norm": 0.10361593216657639, "learning_rate": 1.3064847254288797e-05, "loss": 0.0236, "step": 46580 }, { "epoch": 20.85030208100246, "grad_norm": 0.10591905564069748, "learning_rate": 1.3046278033122577e-05, "loss": 0.0229, "step": 46590 }, { "epoch": 20.854777355113, "grad_norm": 0.07778507471084595, "learning_rate": 1.3027720037521397e-05, "loss": 0.0226, "step": 46600 }, { "epoch": 20.85925262922354, "grad_norm": 0.12068331986665726, "learning_rate": 1.300917327312267e-05, "loss": 0.023, "step": 46610 }, { "epoch": 20.86372790333408, "grad_norm": 0.08414651453495026, "learning_rate": 1.299063774556042e-05, "loss": 0.0229, "step": 46620 }, { "epoch": 20.86820317744462, "grad_norm": 0.07614418119192123, "learning_rate": 1.2972113460465246e-05, "loss": 0.0228, "step": 46630 }, { "epoch": 20.872678451555156, "grad_norm": 0.0750223696231842, "learning_rate": 1.2953600423464324e-05, "loss": 0.0249, "step": 46640 }, { "epoch": 20.877153725665696, "grad_norm": 0.07812948524951935, "learning_rate": 1.293509864018146e-05, "loss": 0.024, "step": 46650 }, { "epoch": 20.881628999776236, "grad_norm": 0.06567299365997314, "learning_rate": 1.2916608116236977e-05, "loss": 0.0264, "step": 46660 }, { "epoch": 20.886104273886776, "grad_norm": 0.10082142055034637, "learning_rate": 1.2898128857247827e-05, "loss": 0.0275, "step": 46670 }, { "epoch": 20.890579547997316, "grad_norm": 0.08143319934606552, "learning_rate": 1.2879660868827508e-05, "loss": 0.0286, "step": 46680 }, { "epoch": 20.895054822107856, "grad_norm": 0.09190678596496582, "learning_rate": 1.286120415658611e-05, "loss": 0.0216, "step": 46690 }, { "epoch": 20.899530096218392, "grad_norm": 0.10722075402736664, "learning_rate": 1.2842758726130283e-05, "loss": 0.0225, "step": 46700 }, { "epoch": 20.90400537032893, "grad_norm": 0.10459008067846298, "learning_rate": 1.2824324583063302e-05, "loss": 0.0264, "step": 46710 }, { "epoch": 20.90848064443947, "grad_norm": 0.08508118242025375, "learning_rate": 1.2805901732984921e-05, "loss": 0.0273, "step": 46720 }, { "epoch": 20.91295591855001, "grad_norm": 0.12178273499011993, "learning_rate": 1.2787490181491568e-05, "loss": 0.0273, "step": 46730 }, { "epoch": 20.91743119266055, "grad_norm": 0.10831298679113388, "learning_rate": 1.2769089934176126e-05, "loss": 0.0261, "step": 46740 }, { "epoch": 20.92190646677109, "grad_norm": 0.07541395723819733, "learning_rate": 1.275070099662815e-05, "loss": 0.0228, "step": 46750 }, { "epoch": 20.926381740881627, "grad_norm": 0.10253925621509552, "learning_rate": 1.2732323374433707e-05, "loss": 0.0231, "step": 46760 }, { "epoch": 20.930857014992167, "grad_norm": 0.07377245277166367, "learning_rate": 1.2713957073175425e-05, "loss": 0.024, "step": 46770 }, { "epoch": 20.935332289102707, "grad_norm": 0.0806640163064003, "learning_rate": 1.2695602098432502e-05, "loss": 0.0264, "step": 46780 }, { "epoch": 20.939807563213247, "grad_norm": 0.10175926238298416, "learning_rate": 1.2677258455780683e-05, "loss": 0.0254, "step": 46790 }, { "epoch": 20.944282837323787, "grad_norm": 0.09195506572723389, "learning_rate": 1.2658926150792322e-05, "loss": 0.0244, "step": 46800 }, { "epoch": 20.948758111434326, "grad_norm": 0.104121632874012, "learning_rate": 1.2640605189036232e-05, "loss": 0.0222, "step": 46810 }, { "epoch": 20.953233385544866, "grad_norm": 0.0870138630270958, "learning_rate": 1.2622295576077886e-05, "loss": 0.0258, "step": 46820 }, { "epoch": 20.957708659655403, "grad_norm": 0.08756020665168762, "learning_rate": 1.2603997317479238e-05, "loss": 0.0227, "step": 46830 }, { "epoch": 20.962183933765942, "grad_norm": 0.07421047240495682, "learning_rate": 1.2585710418798823e-05, "loss": 0.0238, "step": 46840 }, { "epoch": 20.966659207876482, "grad_norm": 0.11286778748035431, "learning_rate": 1.2567434885591722e-05, "loss": 0.0258, "step": 46850 }, { "epoch": 20.971134481987022, "grad_norm": 0.067361980676651, "learning_rate": 1.2549170723409549e-05, "loss": 0.0233, "step": 46860 }, { "epoch": 20.975609756097562, "grad_norm": 0.09240313619375229, "learning_rate": 1.2530917937800463e-05, "loss": 0.024, "step": 46870 }, { "epoch": 20.9800850302081, "grad_norm": 0.09164366126060486, "learning_rate": 1.251267653430922e-05, "loss": 0.0274, "step": 46880 }, { "epoch": 20.984560304318638, "grad_norm": 0.07954057306051254, "learning_rate": 1.2494446518477022e-05, "loss": 0.0243, "step": 46890 }, { "epoch": 20.989035578429178, "grad_norm": 0.09322148561477661, "learning_rate": 1.2476227895841713e-05, "loss": 0.0237, "step": 46900 }, { "epoch": 20.993510852539718, "grad_norm": 0.09156402200460434, "learning_rate": 1.245802067193761e-05, "loss": 0.0226, "step": 46910 }, { "epoch": 20.997986126650257, "grad_norm": 0.07947182655334473, "learning_rate": 1.243982485229559e-05, "loss": 0.0257, "step": 46920 }, { "epoch": 21.002461400760797, "grad_norm": 0.0947101041674614, "learning_rate": 1.2421640442443055e-05, "loss": 0.0235, "step": 46930 }, { "epoch": 21.006936674871337, "grad_norm": 0.08604966849088669, "learning_rate": 1.2403467447903943e-05, "loss": 0.0227, "step": 46940 }, { "epoch": 21.011411948981873, "grad_norm": 0.1013873741030693, "learning_rate": 1.2385305874198776e-05, "loss": 0.0248, "step": 46950 }, { "epoch": 21.015887223092413, "grad_norm": 0.09641207009553909, "learning_rate": 1.2367155726844492e-05, "loss": 0.0258, "step": 46960 }, { "epoch": 21.020362497202953, "grad_norm": 0.09853753447532654, "learning_rate": 1.2349017011354675e-05, "loss": 0.026, "step": 46970 }, { "epoch": 21.024837771313493, "grad_norm": 0.08945076167583466, "learning_rate": 1.233088973323937e-05, "loss": 0.022, "step": 46980 }, { "epoch": 21.029313045424033, "grad_norm": 0.07534501701593399, "learning_rate": 1.2312773898005175e-05, "loss": 0.0234, "step": 46990 }, { "epoch": 21.033788319534573, "grad_norm": 0.08849978446960449, "learning_rate": 1.2294669511155193e-05, "loss": 0.025, "step": 47000 }, { "epoch": 21.038263593645112, "grad_norm": 0.0868368074297905, "learning_rate": 1.2276576578189064e-05, "loss": 0.0227, "step": 47010 }, { "epoch": 21.04273886775565, "grad_norm": 0.10549423098564148, "learning_rate": 1.2258495104602924e-05, "loss": 0.0228, "step": 47020 }, { "epoch": 21.04721414186619, "grad_norm": 0.10304857790470123, "learning_rate": 1.2240425095889495e-05, "loss": 0.0292, "step": 47030 }, { "epoch": 21.05168941597673, "grad_norm": 0.08447135984897614, "learning_rate": 1.2222366557537911e-05, "loss": 0.0274, "step": 47040 }, { "epoch": 21.056164690087268, "grad_norm": 0.08368151634931564, "learning_rate": 1.2204319495033916e-05, "loss": 0.0258, "step": 47050 }, { "epoch": 21.060639964197808, "grad_norm": 0.08094821870326996, "learning_rate": 1.2186283913859726e-05, "loss": 0.0214, "step": 47060 }, { "epoch": 21.065115238308348, "grad_norm": 0.08380810916423798, "learning_rate": 1.2168259819494066e-05, "loss": 0.0249, "step": 47070 }, { "epoch": 21.069590512418884, "grad_norm": 0.09549648314714432, "learning_rate": 1.2150247217412186e-05, "loss": 0.0242, "step": 47080 }, { "epoch": 21.074065786529424, "grad_norm": 0.08347193896770477, "learning_rate": 1.2132246113085822e-05, "loss": 0.0249, "step": 47090 }, { "epoch": 21.078541060639964, "grad_norm": 0.0982908383011818, "learning_rate": 1.2114256511983274e-05, "loss": 0.0251, "step": 47100 }, { "epoch": 21.083016334750504, "grad_norm": 0.08727549016475677, "learning_rate": 1.2096278419569257e-05, "loss": 0.0238, "step": 47110 }, { "epoch": 21.087491608861043, "grad_norm": 0.09163574874401093, "learning_rate": 1.2078311841305084e-05, "loss": 0.0245, "step": 47120 }, { "epoch": 21.091966882971583, "grad_norm": 0.1053156927227974, "learning_rate": 1.2060356782648503e-05, "loss": 0.0249, "step": 47130 }, { "epoch": 21.096442157082123, "grad_norm": 0.10119204968214035, "learning_rate": 1.2042413249053796e-05, "loss": 0.0251, "step": 47140 }, { "epoch": 21.10091743119266, "grad_norm": 0.10073500871658325, "learning_rate": 1.202448124597173e-05, "loss": 0.0262, "step": 47150 }, { "epoch": 21.1053927053032, "grad_norm": 0.11531460285186768, "learning_rate": 1.2006560778849578e-05, "loss": 0.0274, "step": 47160 }, { "epoch": 21.10986797941374, "grad_norm": 0.09656641632318497, "learning_rate": 1.1988651853131088e-05, "loss": 0.0265, "step": 47170 }, { "epoch": 21.11434325352428, "grad_norm": 0.11120959371328354, "learning_rate": 1.1970754474256563e-05, "loss": 0.0227, "step": 47180 }, { "epoch": 21.11881852763482, "grad_norm": 0.10340343415737152, "learning_rate": 1.1952868647662696e-05, "loss": 0.0253, "step": 47190 }, { "epoch": 21.12329380174536, "grad_norm": 0.07914996147155762, "learning_rate": 1.1934994378782772e-05, "loss": 0.024, "step": 47200 }, { "epoch": 21.127769075855895, "grad_norm": 0.10459490120410919, "learning_rate": 1.1917131673046506e-05, "loss": 0.0247, "step": 47210 }, { "epoch": 21.132244349966435, "grad_norm": 0.09689809381961823, "learning_rate": 1.1899280535880119e-05, "loss": 0.0245, "step": 47220 }, { "epoch": 21.136719624076974, "grad_norm": 0.07684337347745895, "learning_rate": 1.1881440972706315e-05, "loss": 0.026, "step": 47230 }, { "epoch": 21.141194898187514, "grad_norm": 0.10918211191892624, "learning_rate": 1.1863612988944267e-05, "loss": 0.026, "step": 47240 }, { "epoch": 21.145670172298054, "grad_norm": 0.11015146970748901, "learning_rate": 1.1845796590009683e-05, "loss": 0.0289, "step": 47250 }, { "epoch": 21.150145446408594, "grad_norm": 0.09092772006988525, "learning_rate": 1.1827991781314667e-05, "loss": 0.0231, "step": 47260 }, { "epoch": 21.15462072051913, "grad_norm": 0.08652061969041824, "learning_rate": 1.1810198568267905e-05, "loss": 0.0241, "step": 47270 }, { "epoch": 21.15909599462967, "grad_norm": 0.09434568136930466, "learning_rate": 1.1792416956274444e-05, "loss": 0.024, "step": 47280 }, { "epoch": 21.16357126874021, "grad_norm": 0.10248246043920517, "learning_rate": 1.1774646950735913e-05, "loss": 0.0242, "step": 47290 }, { "epoch": 21.16804654285075, "grad_norm": 0.07740329205989838, "learning_rate": 1.1756888557050355e-05, "loss": 0.0243, "step": 47300 }, { "epoch": 21.17252181696129, "grad_norm": 0.10629229247570038, "learning_rate": 1.1739141780612306e-05, "loss": 0.0255, "step": 47310 }, { "epoch": 21.17699709107183, "grad_norm": 0.08893154561519623, "learning_rate": 1.1721406626812764e-05, "loss": 0.025, "step": 47320 }, { "epoch": 21.18147236518237, "grad_norm": 0.08280885219573975, "learning_rate": 1.1703683101039197e-05, "loss": 0.023, "step": 47330 }, { "epoch": 21.185947639292905, "grad_norm": 0.09359458088874817, "learning_rate": 1.1685971208675539e-05, "loss": 0.0236, "step": 47340 }, { "epoch": 21.190422913403445, "grad_norm": 0.08273698389530182, "learning_rate": 1.1668270955102218e-05, "loss": 0.0266, "step": 47350 }, { "epoch": 21.194898187513985, "grad_norm": 0.0764322429895401, "learning_rate": 1.1650582345696088e-05, "loss": 0.025, "step": 47360 }, { "epoch": 21.199373461624525, "grad_norm": 0.08327654749155045, "learning_rate": 1.1632905385830484e-05, "loss": 0.0236, "step": 47370 }, { "epoch": 21.203848735735065, "grad_norm": 0.07372863590717316, "learning_rate": 1.16152400808752e-05, "loss": 0.0232, "step": 47380 }, { "epoch": 21.208324009845605, "grad_norm": 0.09215111285448074, "learning_rate": 1.1597586436196473e-05, "loss": 0.0245, "step": 47390 }, { "epoch": 21.21279928395614, "grad_norm": 0.08859506994485855, "learning_rate": 1.157994445715706e-05, "loss": 0.0237, "step": 47400 }, { "epoch": 21.21727455806668, "grad_norm": 0.08390745520591736, "learning_rate": 1.1562314149116065e-05, "loss": 0.0223, "step": 47410 }, { "epoch": 21.22174983217722, "grad_norm": 0.07174240797758102, "learning_rate": 1.1544695517429178e-05, "loss": 0.0206, "step": 47420 }, { "epoch": 21.22622510628776, "grad_norm": 0.09077949821949005, "learning_rate": 1.1527088567448407e-05, "loss": 0.0211, "step": 47430 }, { "epoch": 21.2307003803983, "grad_norm": 0.11408227682113647, "learning_rate": 1.1509493304522329e-05, "loss": 0.0242, "step": 47440 }, { "epoch": 21.23517565450884, "grad_norm": 0.07845432311296463, "learning_rate": 1.1491909733995898e-05, "loss": 0.0254, "step": 47450 }, { "epoch": 21.239650928619376, "grad_norm": 0.09295940399169922, "learning_rate": 1.1474337861210543e-05, "loss": 0.0252, "step": 47460 }, { "epoch": 21.244126202729916, "grad_norm": 0.0829380676150322, "learning_rate": 1.145677769150414e-05, "loss": 0.0252, "step": 47470 }, { "epoch": 21.248601476840456, "grad_norm": 0.08022494614124298, "learning_rate": 1.143922923021099e-05, "loss": 0.023, "step": 47480 }, { "epoch": 21.253076750950996, "grad_norm": 0.10370538383722305, "learning_rate": 1.1421692482661856e-05, "loss": 0.026, "step": 47490 }, { "epoch": 21.257552025061536, "grad_norm": 0.1088540256023407, "learning_rate": 1.1404167454183957e-05, "loss": 0.0241, "step": 47500 }, { "epoch": 21.262027299172075, "grad_norm": 0.1158289685845375, "learning_rate": 1.1386654150100918e-05, "loss": 0.0222, "step": 47510 }, { "epoch": 21.266502573282615, "grad_norm": 0.08533205837011337, "learning_rate": 1.1369152575732822e-05, "loss": 0.027, "step": 47520 }, { "epoch": 21.27097784739315, "grad_norm": 0.07742300629615784, "learning_rate": 1.135166273639619e-05, "loss": 0.0218, "step": 47530 }, { "epoch": 21.27545312150369, "grad_norm": 0.10542160272598267, "learning_rate": 1.133418463740395e-05, "loss": 0.0249, "step": 47540 }, { "epoch": 21.27992839561423, "grad_norm": 0.10211008787155151, "learning_rate": 1.1316718284065537e-05, "loss": 0.0258, "step": 47550 }, { "epoch": 21.28440366972477, "grad_norm": 0.08675959706306458, "learning_rate": 1.1299263681686706e-05, "loss": 0.0235, "step": 47560 }, { "epoch": 21.28887894383531, "grad_norm": 0.09651634842157364, "learning_rate": 1.1281820835569761e-05, "loss": 0.0219, "step": 47570 }, { "epoch": 21.29335421794585, "grad_norm": 0.08597514778375626, "learning_rate": 1.1264389751013326e-05, "loss": 0.0265, "step": 47580 }, { "epoch": 21.297829492056387, "grad_norm": 0.10241416841745377, "learning_rate": 1.124697043331256e-05, "loss": 0.0237, "step": 47590 }, { "epoch": 21.302304766166927, "grad_norm": 0.07940967381000519, "learning_rate": 1.1229562887758926e-05, "loss": 0.0246, "step": 47600 }, { "epoch": 21.306780040277467, "grad_norm": 0.08493480831384659, "learning_rate": 1.1212167119640438e-05, "loss": 0.0258, "step": 47610 }, { "epoch": 21.311255314388006, "grad_norm": 0.08682496100664139, "learning_rate": 1.1194783134241437e-05, "loss": 0.0243, "step": 47620 }, { "epoch": 21.315730588498546, "grad_norm": 0.09927640110254288, "learning_rate": 1.1177410936842719e-05, "loss": 0.0253, "step": 47630 }, { "epoch": 21.320205862609086, "grad_norm": 0.09082260727882385, "learning_rate": 1.1160050532721528e-05, "loss": 0.0242, "step": 47640 }, { "epoch": 21.324681136719626, "grad_norm": 0.07632376253604889, "learning_rate": 1.1142701927151456e-05, "loss": 0.0249, "step": 47650 }, { "epoch": 21.329156410830162, "grad_norm": 0.0680294930934906, "learning_rate": 1.1125365125402582e-05, "loss": 0.0226, "step": 47660 }, { "epoch": 21.333631684940702, "grad_norm": 0.07456374168395996, "learning_rate": 1.1108040132741354e-05, "loss": 0.0221, "step": 47670 }, { "epoch": 21.33810695905124, "grad_norm": 0.08764128386974335, "learning_rate": 1.1090726954430658e-05, "loss": 0.0237, "step": 47680 }, { "epoch": 21.34258223316178, "grad_norm": 0.08605851978063583, "learning_rate": 1.107342559572977e-05, "loss": 0.0266, "step": 47690 }, { "epoch": 21.34705750727232, "grad_norm": 0.08110889792442322, "learning_rate": 1.1056136061894384e-05, "loss": 0.0228, "step": 47700 }, { "epoch": 21.35153278138286, "grad_norm": 0.09233580529689789, "learning_rate": 1.1038858358176602e-05, "loss": 0.0253, "step": 47710 }, { "epoch": 21.356008055493398, "grad_norm": 0.08460871875286102, "learning_rate": 1.1021592489824967e-05, "loss": 0.0226, "step": 47720 }, { "epoch": 21.360483329603937, "grad_norm": 0.08266440778970718, "learning_rate": 1.100433846208434e-05, "loss": 0.0249, "step": 47730 }, { "epoch": 21.364958603714477, "grad_norm": 0.08168362826108932, "learning_rate": 1.0987096280196086e-05, "loss": 0.0237, "step": 47740 }, { "epoch": 21.369433877825017, "grad_norm": 0.0661042109131813, "learning_rate": 1.0969865949397901e-05, "loss": 0.0239, "step": 47750 }, { "epoch": 21.373909151935557, "grad_norm": 0.09523556381464005, "learning_rate": 1.095264747492391e-05, "loss": 0.0241, "step": 47760 }, { "epoch": 21.378384426046097, "grad_norm": 0.06792537122964859, "learning_rate": 1.093544086200463e-05, "loss": 0.0241, "step": 47770 }, { "epoch": 21.382859700156633, "grad_norm": 0.06914783269166946, "learning_rate": 1.0918246115866964e-05, "loss": 0.0223, "step": 47780 }, { "epoch": 21.387334974267173, "grad_norm": 0.08259846270084381, "learning_rate": 1.090106324173426e-05, "loss": 0.0213, "step": 47790 }, { "epoch": 21.391810248377713, "grad_norm": 0.08658700436353683, "learning_rate": 1.0883892244826172e-05, "loss": 0.0214, "step": 47800 }, { "epoch": 21.396285522488252, "grad_norm": 0.09235323965549469, "learning_rate": 1.086673313035883e-05, "loss": 0.0249, "step": 47810 }, { "epoch": 21.400760796598792, "grad_norm": 0.0959194153547287, "learning_rate": 1.0849585903544706e-05, "loss": 0.0244, "step": 47820 }, { "epoch": 21.405236070709332, "grad_norm": 0.07817860692739487, "learning_rate": 1.0832450569592684e-05, "loss": 0.0272, "step": 47830 }, { "epoch": 21.409711344819872, "grad_norm": 0.08589743822813034, "learning_rate": 1.0815327133708015e-05, "loss": 0.0262, "step": 47840 }, { "epoch": 21.414186618930408, "grad_norm": 0.11413529515266418, "learning_rate": 1.0798215601092354e-05, "loss": 0.0256, "step": 47850 }, { "epoch": 21.418661893040948, "grad_norm": 0.10147131234407425, "learning_rate": 1.0781115976943717e-05, "loss": 0.0236, "step": 47860 }, { "epoch": 21.423137167151488, "grad_norm": 0.08611639589071274, "learning_rate": 1.0764028266456555e-05, "loss": 0.0232, "step": 47870 }, { "epoch": 21.427612441262028, "grad_norm": 0.09582439064979553, "learning_rate": 1.0746952474821614e-05, "loss": 0.0231, "step": 47880 }, { "epoch": 21.432087715372568, "grad_norm": 0.0803178995847702, "learning_rate": 1.0729888607226113e-05, "loss": 0.0224, "step": 47890 }, { "epoch": 21.436562989483107, "grad_norm": 0.08358190953731537, "learning_rate": 1.0712836668853582e-05, "loss": 0.0247, "step": 47900 }, { "epoch": 21.441038263593644, "grad_norm": 0.10104536265134811, "learning_rate": 1.069579666488395e-05, "loss": 0.0221, "step": 47910 }, { "epoch": 21.445513537704183, "grad_norm": 0.09562094509601593, "learning_rate": 1.0678768600493528e-05, "loss": 0.026, "step": 47920 }, { "epoch": 21.449988811814723, "grad_norm": 0.0784793570637703, "learning_rate": 1.0661752480854975e-05, "loss": 0.0268, "step": 47930 }, { "epoch": 21.454464085925263, "grad_norm": 0.06222847104072571, "learning_rate": 1.0644748311137376e-05, "loss": 0.023, "step": 47940 }, { "epoch": 21.458939360035803, "grad_norm": 0.0772080048918724, "learning_rate": 1.0627756096506102e-05, "loss": 0.0235, "step": 47950 }, { "epoch": 21.463414634146343, "grad_norm": 0.09534073621034622, "learning_rate": 1.0610775842122972e-05, "loss": 0.0242, "step": 47960 }, { "epoch": 21.46788990825688, "grad_norm": 0.09838719666004181, "learning_rate": 1.059380755314613e-05, "loss": 0.024, "step": 47970 }, { "epoch": 21.47236518236742, "grad_norm": 0.07060188800096512, "learning_rate": 1.0576851234730095e-05, "loss": 0.0264, "step": 47980 }, { "epoch": 21.47684045647796, "grad_norm": 0.08927878737449646, "learning_rate": 1.0559906892025745e-05, "loss": 0.0233, "step": 47990 }, { "epoch": 21.4813157305885, "grad_norm": 0.08746800571680069, "learning_rate": 1.0542974530180327e-05, "loss": 0.0226, "step": 48000 }, { "epoch": 21.48579100469904, "grad_norm": 0.091275155544281, "learning_rate": 1.0526054154337445e-05, "loss": 0.0242, "step": 48010 }, { "epoch": 21.490266278809578, "grad_norm": 0.0870751291513443, "learning_rate": 1.0509145769637057e-05, "loss": 0.0195, "step": 48020 }, { "epoch": 21.494741552920118, "grad_norm": 0.10205098986625671, "learning_rate": 1.049224938121548e-05, "loss": 0.023, "step": 48030 }, { "epoch": 21.499216827030654, "grad_norm": 0.08245690912008286, "learning_rate": 1.0475364994205411e-05, "loss": 0.0219, "step": 48040 }, { "epoch": 21.503692101141194, "grad_norm": 0.07998043298721313, "learning_rate": 1.045849261373587e-05, "loss": 0.0242, "step": 48050 }, { "epoch": 21.508167375251734, "grad_norm": 0.07799527794122696, "learning_rate": 1.0441632244932237e-05, "loss": 0.0223, "step": 48060 }, { "epoch": 21.512642649362274, "grad_norm": 0.09791392832994461, "learning_rate": 1.0424783892916257e-05, "loss": 0.0214, "step": 48070 }, { "epoch": 21.517117923472814, "grad_norm": 0.09022922068834305, "learning_rate": 1.0407947562805986e-05, "loss": 0.0218, "step": 48080 }, { "epoch": 21.521593197583353, "grad_norm": 0.07924144715070724, "learning_rate": 1.0391123259715906e-05, "loss": 0.023, "step": 48090 }, { "epoch": 21.52606847169389, "grad_norm": 0.0870908722281456, "learning_rate": 1.0374310988756747e-05, "loss": 0.026, "step": 48100 }, { "epoch": 21.53054374580443, "grad_norm": 0.09906847029924393, "learning_rate": 1.0357510755035677e-05, "loss": 0.0246, "step": 48110 }, { "epoch": 21.53501901991497, "grad_norm": 0.09108643978834152, "learning_rate": 1.0340722563656107e-05, "loss": 0.022, "step": 48120 }, { "epoch": 21.53949429402551, "grad_norm": 0.08519630134105682, "learning_rate": 1.03239464197179e-05, "loss": 0.0227, "step": 48130 }, { "epoch": 21.54396956813605, "grad_norm": 0.07861601561307907, "learning_rate": 1.0307182328317188e-05, "loss": 0.0244, "step": 48140 }, { "epoch": 21.54844484224659, "grad_norm": 0.08417921513319016, "learning_rate": 1.0290430294546449e-05, "loss": 0.0244, "step": 48150 }, { "epoch": 21.55292011635713, "grad_norm": 0.09345526993274689, "learning_rate": 1.0273690323494523e-05, "loss": 0.0246, "step": 48160 }, { "epoch": 21.557395390467665, "grad_norm": 0.09487895667552948, "learning_rate": 1.0256962420246557e-05, "loss": 0.0218, "step": 48170 }, { "epoch": 21.561870664578205, "grad_norm": 0.08964231610298157, "learning_rate": 1.0240246589884044e-05, "loss": 0.0205, "step": 48180 }, { "epoch": 21.566345938688745, "grad_norm": 0.09474246948957443, "learning_rate": 1.0223542837484839e-05, "loss": 0.026, "step": 48190 }, { "epoch": 21.570821212799284, "grad_norm": 0.09279607981443405, "learning_rate": 1.0206851168123077e-05, "loss": 0.0227, "step": 48200 }, { "epoch": 21.575296486909824, "grad_norm": 0.09180163592100143, "learning_rate": 1.0190171586869258e-05, "loss": 0.0267, "step": 48210 }, { "epoch": 21.579771761020364, "grad_norm": 0.09252879023551941, "learning_rate": 1.0173504098790187e-05, "loss": 0.0241, "step": 48220 }, { "epoch": 21.5842470351309, "grad_norm": 0.08702791482210159, "learning_rate": 1.0156848708949006e-05, "loss": 0.0261, "step": 48230 }, { "epoch": 21.58872230924144, "grad_norm": 0.0739879310131073, "learning_rate": 1.0140205422405214e-05, "loss": 0.0262, "step": 48240 }, { "epoch": 21.59319758335198, "grad_norm": 0.08559790998697281, "learning_rate": 1.012357424421455e-05, "loss": 0.0214, "step": 48250 }, { "epoch": 21.59767285746252, "grad_norm": 0.10232649743556976, "learning_rate": 1.0106955179429183e-05, "loss": 0.0228, "step": 48260 }, { "epoch": 21.60214813157306, "grad_norm": 0.09137039631605148, "learning_rate": 1.009034823309749e-05, "loss": 0.0234, "step": 48270 }, { "epoch": 21.6066234056836, "grad_norm": 0.08212830126285553, "learning_rate": 1.0073753410264263e-05, "loss": 0.0227, "step": 48280 }, { "epoch": 21.611098679794136, "grad_norm": 0.10262913256883621, "learning_rate": 1.0057170715970559e-05, "loss": 0.0229, "step": 48290 }, { "epoch": 21.615573953904676, "grad_norm": 0.09065995365381241, "learning_rate": 1.0040600155253765e-05, "loss": 0.0233, "step": 48300 }, { "epoch": 21.620049228015215, "grad_norm": 0.09514353424310684, "learning_rate": 1.0024041733147577e-05, "loss": 0.0264, "step": 48310 }, { "epoch": 21.624524502125755, "grad_norm": 0.09987451881170273, "learning_rate": 1.0007495454682004e-05, "loss": 0.0276, "step": 48320 }, { "epoch": 21.628999776236295, "grad_norm": 0.08098875731229782, "learning_rate": 9.990961324883358e-06, "loss": 0.0222, "step": 48330 }, { "epoch": 21.633475050346835, "grad_norm": 0.0919468104839325, "learning_rate": 9.974439348774295e-06, "loss": 0.0233, "step": 48340 }, { "epoch": 21.63795032445737, "grad_norm": 0.0988708883523941, "learning_rate": 9.95792953137375e-06, "loss": 0.0219, "step": 48350 }, { "epoch": 21.64242559856791, "grad_norm": 0.11409702897071838, "learning_rate": 9.941431877696955e-06, "loss": 0.0251, "step": 48360 }, { "epoch": 21.64690087267845, "grad_norm": 0.09445685148239136, "learning_rate": 9.924946392755479e-06, "loss": 0.0233, "step": 48370 }, { "epoch": 21.65137614678899, "grad_norm": 0.09220165759325027, "learning_rate": 9.908473081557151e-06, "loss": 0.0242, "step": 48380 }, { "epoch": 21.65585142089953, "grad_norm": 0.0812411829829216, "learning_rate": 9.892011949106172e-06, "loss": 0.0252, "step": 48390 }, { "epoch": 21.66032669501007, "grad_norm": 0.07404196262359619, "learning_rate": 9.875563000402948e-06, "loss": 0.0246, "step": 48400 }, { "epoch": 21.66480196912061, "grad_norm": 0.08655797690153122, "learning_rate": 9.859126240444283e-06, "loss": 0.0275, "step": 48410 }, { "epoch": 21.669277243231146, "grad_norm": 0.08323128521442413, "learning_rate": 9.842701674223187e-06, "loss": 0.0295, "step": 48420 }, { "epoch": 21.673752517341686, "grad_norm": 0.06149947643280029, "learning_rate": 9.826289306729052e-06, "loss": 0.0227, "step": 48430 }, { "epoch": 21.678227791452226, "grad_norm": 0.0817159116268158, "learning_rate": 9.809889142947476e-06, "loss": 0.0248, "step": 48440 }, { "epoch": 21.682703065562766, "grad_norm": 0.0842001661658287, "learning_rate": 9.793501187860432e-06, "loss": 0.0286, "step": 48450 }, { "epoch": 21.687178339673306, "grad_norm": 0.07235726714134216, "learning_rate": 9.777125446446133e-06, "loss": 0.0245, "step": 48460 }, { "epoch": 21.691653613783846, "grad_norm": 0.07320089638233185, "learning_rate": 9.760761923679107e-06, "loss": 0.0222, "step": 48470 }, { "epoch": 21.696128887894382, "grad_norm": 0.08523433655500412, "learning_rate": 9.744410624530148e-06, "loss": 0.0293, "step": 48480 }, { "epoch": 21.70060416200492, "grad_norm": 0.07691114395856857, "learning_rate": 9.728071553966339e-06, "loss": 0.0245, "step": 48490 }, { "epoch": 21.70507943611546, "grad_norm": 0.09943945705890656, "learning_rate": 9.711744716951093e-06, "loss": 0.0261, "step": 48500 }, { "epoch": 21.709554710226, "grad_norm": 0.09611576050519943, "learning_rate": 9.695430118444048e-06, "loss": 0.0244, "step": 48510 }, { "epoch": 21.71402998433654, "grad_norm": 0.0835796594619751, "learning_rate": 9.679127763401152e-06, "loss": 0.0268, "step": 48520 }, { "epoch": 21.71850525844708, "grad_norm": 0.09617006033658981, "learning_rate": 9.662837656774632e-06, "loss": 0.0229, "step": 48530 }, { "epoch": 21.72298053255762, "grad_norm": 0.08651832491159439, "learning_rate": 9.646559803512994e-06, "loss": 0.0234, "step": 48540 }, { "epoch": 21.727455806668157, "grad_norm": 0.09351154416799545, "learning_rate": 9.630294208560998e-06, "loss": 0.0245, "step": 48550 }, { "epoch": 21.731931080778697, "grad_norm": 0.07889202982187271, "learning_rate": 9.614040876859748e-06, "loss": 0.0225, "step": 48560 }, { "epoch": 21.736406354889237, "grad_norm": 0.09850514680147171, "learning_rate": 9.597799813346525e-06, "loss": 0.0225, "step": 48570 }, { "epoch": 21.740881628999777, "grad_norm": 0.09318459033966064, "learning_rate": 9.581571022954988e-06, "loss": 0.0261, "step": 48580 }, { "epoch": 21.745356903110316, "grad_norm": 0.08182897418737411, "learning_rate": 9.56535451061496e-06, "loss": 0.0274, "step": 48590 }, { "epoch": 21.749832177220856, "grad_norm": 0.09541220963001251, "learning_rate": 9.549150281252633e-06, "loss": 0.0225, "step": 48600 }, { "epoch": 21.754307451331393, "grad_norm": 0.09791054576635361, "learning_rate": 9.532958339790404e-06, "loss": 0.0235, "step": 48610 }, { "epoch": 21.758782725441932, "grad_norm": 0.0650411918759346, "learning_rate": 9.51677869114696e-06, "loss": 0.0232, "step": 48620 }, { "epoch": 21.763257999552472, "grad_norm": 0.1029086783528328, "learning_rate": 9.500611340237258e-06, "loss": 0.0247, "step": 48630 }, { "epoch": 21.767733273663012, "grad_norm": 0.08944419026374817, "learning_rate": 9.484456291972487e-06, "loss": 0.0271, "step": 48640 }, { "epoch": 21.772208547773552, "grad_norm": 0.11000005155801773, "learning_rate": 9.468313551260161e-06, "loss": 0.0246, "step": 48650 }, { "epoch": 21.77668382188409, "grad_norm": 0.0891297236084938, "learning_rate": 9.452183123004e-06, "loss": 0.024, "step": 48660 }, { "epoch": 21.781159095994628, "grad_norm": 0.09401003271341324, "learning_rate": 9.436065012104001e-06, "loss": 0.0245, "step": 48670 }, { "epoch": 21.785634370105168, "grad_norm": 0.09874605387449265, "learning_rate": 9.41995922345642e-06, "loss": 0.0276, "step": 48680 }, { "epoch": 21.790109644215708, "grad_norm": 0.10050936788320541, "learning_rate": 9.403865761953779e-06, "loss": 0.0261, "step": 48690 }, { "epoch": 21.794584918326247, "grad_norm": 0.12043560296297073, "learning_rate": 9.387784632484826e-06, "loss": 0.0227, "step": 48700 }, { "epoch": 21.799060192436787, "grad_norm": 0.08037620037794113, "learning_rate": 9.371715839934625e-06, "loss": 0.0231, "step": 48710 }, { "epoch": 21.803535466547327, "grad_norm": 0.08988785743713379, "learning_rate": 9.355659389184396e-06, "loss": 0.022, "step": 48720 }, { "epoch": 21.808010740657867, "grad_norm": 0.08129216730594635, "learning_rate": 9.33961528511172e-06, "loss": 0.0234, "step": 48730 }, { "epoch": 21.812486014768403, "grad_norm": 0.08067493885755539, "learning_rate": 9.32358353259032e-06, "loss": 0.0237, "step": 48740 }, { "epoch": 21.816961288878943, "grad_norm": 0.09318855404853821, "learning_rate": 9.307564136490254e-06, "loss": 0.0234, "step": 48750 }, { "epoch": 21.821436562989483, "grad_norm": 0.09091762453317642, "learning_rate": 9.291557101677784e-06, "loss": 0.0241, "step": 48760 }, { "epoch": 21.825911837100023, "grad_norm": 0.07676625996828079, "learning_rate": 9.275562433015422e-06, "loss": 0.0235, "step": 48770 }, { "epoch": 21.830387111210563, "grad_norm": 0.1131792664527893, "learning_rate": 9.259580135361929e-06, "loss": 0.0247, "step": 48780 }, { "epoch": 21.834862385321102, "grad_norm": 0.08487926423549652, "learning_rate": 9.243610213572285e-06, "loss": 0.0235, "step": 48790 }, { "epoch": 21.83933765943164, "grad_norm": 0.08006255328655243, "learning_rate": 9.227652672497761e-06, "loss": 0.0244, "step": 48800 }, { "epoch": 21.84381293354218, "grad_norm": 0.07135213166475296, "learning_rate": 9.211707516985829e-06, "loss": 0.0231, "step": 48810 }, { "epoch": 21.84828820765272, "grad_norm": 0.0737660750746727, "learning_rate": 9.195774751880198e-06, "loss": 0.0199, "step": 48820 }, { "epoch": 21.852763481763258, "grad_norm": 0.07661952823400497, "learning_rate": 9.17985438202082e-06, "loss": 0.0255, "step": 48830 }, { "epoch": 21.857238755873798, "grad_norm": 0.09426913410425186, "learning_rate": 9.163946412243896e-06, "loss": 0.024, "step": 48840 }, { "epoch": 21.861714029984338, "grad_norm": 0.07639282941818237, "learning_rate": 9.148050847381828e-06, "loss": 0.0214, "step": 48850 }, { "epoch": 21.866189304094874, "grad_norm": 0.0831066220998764, "learning_rate": 9.132167692263289e-06, "loss": 0.0255, "step": 48860 }, { "epoch": 21.870664578205414, "grad_norm": 0.08834046870470047, "learning_rate": 9.116296951713133e-06, "loss": 0.0232, "step": 48870 }, { "epoch": 21.875139852315954, "grad_norm": 0.09502347558736801, "learning_rate": 9.100438630552521e-06, "loss": 0.0228, "step": 48880 }, { "epoch": 21.879615126426494, "grad_norm": 0.08701560646295547, "learning_rate": 9.084592733598735e-06, "loss": 0.0226, "step": 48890 }, { "epoch": 21.884090400537033, "grad_norm": 0.09642309695482254, "learning_rate": 9.068759265665384e-06, "loss": 0.0269, "step": 48900 }, { "epoch": 21.888565674647573, "grad_norm": 0.08391109108924866, "learning_rate": 9.052938231562236e-06, "loss": 0.0242, "step": 48910 }, { "epoch": 21.893040948758113, "grad_norm": 0.0774226263165474, "learning_rate": 9.037129636095309e-06, "loss": 0.0249, "step": 48920 }, { "epoch": 21.89751622286865, "grad_norm": 0.06956174969673157, "learning_rate": 9.02133348406684e-06, "loss": 0.0196, "step": 48930 }, { "epoch": 21.90199149697919, "grad_norm": 0.0725146159529686, "learning_rate": 9.005549780275263e-06, "loss": 0.0239, "step": 48940 }, { "epoch": 21.90646677108973, "grad_norm": 0.0886022299528122, "learning_rate": 8.989778529515298e-06, "loss": 0.0265, "step": 48950 }, { "epoch": 21.91094204520027, "grad_norm": 0.07827109098434448, "learning_rate": 8.974019736577777e-06, "loss": 0.0232, "step": 48960 }, { "epoch": 21.91541731931081, "grad_norm": 0.06689652800559998, "learning_rate": 8.958273406249839e-06, "loss": 0.0225, "step": 48970 }, { "epoch": 21.91989259342135, "grad_norm": 0.09782480448484421, "learning_rate": 8.9425395433148e-06, "loss": 0.0227, "step": 48980 }, { "epoch": 21.924367867531885, "grad_norm": 0.10284829884767532, "learning_rate": 8.92681815255219e-06, "loss": 0.0261, "step": 48990 }, { "epoch": 21.928843141642425, "grad_norm": 0.10407264530658722, "learning_rate": 8.911109238737747e-06, "loss": 0.0218, "step": 49000 }, { "epoch": 21.933318415752964, "grad_norm": 0.08798869699239731, "learning_rate": 8.895412806643427e-06, "loss": 0.026, "step": 49010 }, { "epoch": 21.937793689863504, "grad_norm": 0.09403622895479202, "learning_rate": 8.879728861037384e-06, "loss": 0.025, "step": 49020 }, { "epoch": 21.942268963974044, "grad_norm": 0.08667270839214325, "learning_rate": 8.864057406684023e-06, "loss": 0.0255, "step": 49030 }, { "epoch": 21.946744238084584, "grad_norm": 0.09008600562810898, "learning_rate": 8.848398448343859e-06, "loss": 0.0217, "step": 49040 }, { "epoch": 21.951219512195124, "grad_norm": 0.10093718022108078, "learning_rate": 8.832751990773714e-06, "loss": 0.023, "step": 49050 }, { "epoch": 21.95569478630566, "grad_norm": 0.0832148939371109, "learning_rate": 8.817118038726558e-06, "loss": 0.0237, "step": 49060 }, { "epoch": 21.9601700604162, "grad_norm": 0.07564684003591537, "learning_rate": 8.801496596951575e-06, "loss": 0.0218, "step": 49070 }, { "epoch": 21.96464533452674, "grad_norm": 0.06754831224679947, "learning_rate": 8.785887670194138e-06, "loss": 0.0213, "step": 49080 }, { "epoch": 21.96912060863728, "grad_norm": 0.07300498336553574, "learning_rate": 8.770291263195819e-06, "loss": 0.027, "step": 49090 }, { "epoch": 21.97359588274782, "grad_norm": 0.07877938449382782, "learning_rate": 8.754707380694427e-06, "loss": 0.0244, "step": 49100 }, { "epoch": 21.97807115685836, "grad_norm": 0.07811074703931808, "learning_rate": 8.739136027423894e-06, "loss": 0.0229, "step": 49110 }, { "epoch": 21.982546430968895, "grad_norm": 0.07777675241231918, "learning_rate": 8.723577208114419e-06, "loss": 0.0221, "step": 49120 }, { "epoch": 21.987021705079435, "grad_norm": 0.07911471277475357, "learning_rate": 8.708030927492345e-06, "loss": 0.0245, "step": 49130 }, { "epoch": 21.991496979189975, "grad_norm": 0.07316448539495468, "learning_rate": 8.692497190280224e-06, "loss": 0.0222, "step": 49140 }, { "epoch": 21.995972253300515, "grad_norm": 0.08764529973268509, "learning_rate": 8.676976001196801e-06, "loss": 0.0266, "step": 49150 }, { "epoch": 22.000447527411055, "grad_norm": 0.08983173221349716, "learning_rate": 8.661467364956993e-06, "loss": 0.0222, "step": 49160 }, { "epoch": 22.004922801521595, "grad_norm": 0.08454525470733643, "learning_rate": 8.645971286271904e-06, "loss": 0.0246, "step": 49170 }, { "epoch": 22.00939807563213, "grad_norm": 0.11046009510755539, "learning_rate": 8.630487769848877e-06, "loss": 0.0278, "step": 49180 }, { "epoch": 22.01387334974267, "grad_norm": 0.09507972002029419, "learning_rate": 8.615016820391342e-06, "loss": 0.0197, "step": 49190 }, { "epoch": 22.01834862385321, "grad_norm": 0.08592724800109863, "learning_rate": 8.599558442598998e-06, "loss": 0.0252, "step": 49200 }, { "epoch": 22.02282389796375, "grad_norm": 0.07833247631788254, "learning_rate": 8.584112641167686e-06, "loss": 0.0274, "step": 49210 }, { "epoch": 22.02729917207429, "grad_norm": 0.07960409671068192, "learning_rate": 8.568679420789433e-06, "loss": 0.0239, "step": 49220 }, { "epoch": 22.03177444618483, "grad_norm": 0.0811333954334259, "learning_rate": 8.55325878615244e-06, "loss": 0.0205, "step": 49230 }, { "epoch": 22.03624972029537, "grad_norm": 0.08003690838813782, "learning_rate": 8.537850741941073e-06, "loss": 0.0233, "step": 49240 }, { "epoch": 22.040724994405906, "grad_norm": 0.07691966742277145, "learning_rate": 8.522455292835934e-06, "loss": 0.0243, "step": 49250 }, { "epoch": 22.045200268516446, "grad_norm": 0.12514854967594147, "learning_rate": 8.507072443513702e-06, "loss": 0.0247, "step": 49260 }, { "epoch": 22.049675542626986, "grad_norm": 0.09107660502195358, "learning_rate": 8.491702198647333e-06, "loss": 0.0258, "step": 49270 }, { "epoch": 22.054150816737526, "grad_norm": 0.07401914149522781, "learning_rate": 8.476344562905841e-06, "loss": 0.0224, "step": 49280 }, { "epoch": 22.058626090848065, "grad_norm": 0.0777893140912056, "learning_rate": 8.460999540954517e-06, "loss": 0.026, "step": 49290 }, { "epoch": 22.063101364958605, "grad_norm": 0.08081375807523727, "learning_rate": 8.445667137454761e-06, "loss": 0.0228, "step": 49300 }, { "epoch": 22.06757663906914, "grad_norm": 0.09568096697330475, "learning_rate": 8.43034735706415e-06, "loss": 0.0231, "step": 49310 }, { "epoch": 22.07205191317968, "grad_norm": 0.0745314359664917, "learning_rate": 8.415040204436426e-06, "loss": 0.0228, "step": 49320 }, { "epoch": 22.07652718729022, "grad_norm": 0.10883776098489761, "learning_rate": 8.399745684221499e-06, "loss": 0.0252, "step": 49330 }, { "epoch": 22.08100246140076, "grad_norm": 0.09222814440727234, "learning_rate": 8.384463801065434e-06, "loss": 0.025, "step": 49340 }, { "epoch": 22.0854777355113, "grad_norm": 0.07827534526586533, "learning_rate": 8.369194559610482e-06, "loss": 0.0256, "step": 49350 }, { "epoch": 22.08995300962184, "grad_norm": 0.08667495101690292, "learning_rate": 8.353937964495029e-06, "loss": 0.0224, "step": 49360 }, { "epoch": 22.094428283732377, "grad_norm": 0.08458145707845688, "learning_rate": 8.338694020353616e-06, "loss": 0.0257, "step": 49370 }, { "epoch": 22.098903557842917, "grad_norm": 0.07043365389108658, "learning_rate": 8.323462731816961e-06, "loss": 0.025, "step": 49380 }, { "epoch": 22.103378831953457, "grad_norm": 0.0979226753115654, "learning_rate": 8.308244103511909e-06, "loss": 0.0208, "step": 49390 }, { "epoch": 22.107854106063996, "grad_norm": 0.08202345669269562, "learning_rate": 8.293038140061515e-06, "loss": 0.025, "step": 49400 }, { "epoch": 22.112329380174536, "grad_norm": 0.08659054338932037, "learning_rate": 8.277844846084898e-06, "loss": 0.0235, "step": 49410 }, { "epoch": 22.116804654285076, "grad_norm": 0.09969870746135712, "learning_rate": 8.262664226197436e-06, "loss": 0.0251, "step": 49420 }, { "epoch": 22.121279928395616, "grad_norm": 0.06412165611982346, "learning_rate": 8.247496285010548e-06, "loss": 0.0228, "step": 49430 }, { "epoch": 22.125755202506152, "grad_norm": 0.07460356503725052, "learning_rate": 8.232341027131885e-06, "loss": 0.025, "step": 49440 }, { "epoch": 22.130230476616692, "grad_norm": 0.08297042548656464, "learning_rate": 8.21719845716521e-06, "loss": 0.0258, "step": 49450 }, { "epoch": 22.134705750727232, "grad_norm": 0.08794965595006943, "learning_rate": 8.202068579710431e-06, "loss": 0.0234, "step": 49460 }, { "epoch": 22.13918102483777, "grad_norm": 0.09068941324949265, "learning_rate": 8.186951399363613e-06, "loss": 0.0251, "step": 49470 }, { "epoch": 22.14365629894831, "grad_norm": 0.10086594521999359, "learning_rate": 8.17184692071694e-06, "loss": 0.0234, "step": 49480 }, { "epoch": 22.14813157305885, "grad_norm": 0.10773791372776031, "learning_rate": 8.156755148358764e-06, "loss": 0.0222, "step": 49490 }, { "epoch": 22.152606847169388, "grad_norm": 0.12008818238973618, "learning_rate": 8.141676086873572e-06, "loss": 0.0215, "step": 49500 }, { "epoch": 22.157082121279927, "grad_norm": 0.09116268903017044, "learning_rate": 8.126609740841979e-06, "loss": 0.0252, "step": 49510 }, { "epoch": 22.161557395390467, "grad_norm": 0.09682969003915787, "learning_rate": 8.111556114840746e-06, "loss": 0.0241, "step": 49520 }, { "epoch": 22.166032669501007, "grad_norm": 0.0919514074921608, "learning_rate": 8.096515213442762e-06, "loss": 0.0218, "step": 49530 }, { "epoch": 22.170507943611547, "grad_norm": 0.14774897694587708, "learning_rate": 8.08148704121705e-06, "loss": 0.0223, "step": 49540 }, { "epoch": 22.174983217722087, "grad_norm": 0.08057010173797607, "learning_rate": 8.066471602728803e-06, "loss": 0.0204, "step": 49550 }, { "epoch": 22.179458491832627, "grad_norm": 0.08747602254152298, "learning_rate": 8.051468902539272e-06, "loss": 0.025, "step": 49560 }, { "epoch": 22.183933765943163, "grad_norm": 0.0816688984632492, "learning_rate": 8.036478945205922e-06, "loss": 0.024, "step": 49570 }, { "epoch": 22.188409040053703, "grad_norm": 0.06770048290491104, "learning_rate": 8.021501735282266e-06, "loss": 0.0237, "step": 49580 }, { "epoch": 22.192884314164242, "grad_norm": 0.08938726037740707, "learning_rate": 8.00653727731801e-06, "loss": 0.0211, "step": 49590 }, { "epoch": 22.197359588274782, "grad_norm": 0.09414972364902496, "learning_rate": 7.991585575858961e-06, "loss": 0.0244, "step": 49600 }, { "epoch": 22.201834862385322, "grad_norm": 0.09291639178991318, "learning_rate": 7.976646635447044e-06, "loss": 0.0224, "step": 49610 }, { "epoch": 22.206310136495862, "grad_norm": 0.08882784098386765, "learning_rate": 7.96172046062032e-06, "loss": 0.0268, "step": 49620 }, { "epoch": 22.210785410606398, "grad_norm": 0.09479376673698425, "learning_rate": 7.946807055912959e-06, "loss": 0.0252, "step": 49630 }, { "epoch": 22.215260684716938, "grad_norm": 0.0586528554558754, "learning_rate": 7.931906425855268e-06, "loss": 0.0215, "step": 49640 }, { "epoch": 22.219735958827478, "grad_norm": 0.10342291742563248, "learning_rate": 7.917018574973645e-06, "loss": 0.0258, "step": 49650 }, { "epoch": 22.224211232938018, "grad_norm": 0.09359610080718994, "learning_rate": 7.902143507790661e-06, "loss": 0.0252, "step": 49660 }, { "epoch": 22.228686507048558, "grad_norm": 0.07112628221511841, "learning_rate": 7.887281228824944e-06, "loss": 0.0227, "step": 49670 }, { "epoch": 22.233161781159097, "grad_norm": 0.09844347089529037, "learning_rate": 7.872431742591268e-06, "loss": 0.0237, "step": 49680 }, { "epoch": 22.237637055269634, "grad_norm": 0.0980263203382492, "learning_rate": 7.857595053600513e-06, "loss": 0.024, "step": 49690 }, { "epoch": 22.242112329380173, "grad_norm": 0.07568585127592087, "learning_rate": 7.842771166359681e-06, "loss": 0.0244, "step": 49700 }, { "epoch": 22.246587603490713, "grad_norm": 0.09344524145126343, "learning_rate": 7.827960085371855e-06, "loss": 0.0245, "step": 49710 }, { "epoch": 22.251062877601253, "grad_norm": 0.0917460024356842, "learning_rate": 7.813161815136294e-06, "loss": 0.0234, "step": 49720 }, { "epoch": 22.255538151711793, "grad_norm": 0.079647496342659, "learning_rate": 7.79837636014827e-06, "loss": 0.0271, "step": 49730 }, { "epoch": 22.260013425822333, "grad_norm": 0.07224371284246445, "learning_rate": 7.783603724899257e-06, "loss": 0.0214, "step": 49740 }, { "epoch": 22.264488699932873, "grad_norm": 0.10346174985170364, "learning_rate": 7.768843913876756e-06, "loss": 0.0269, "step": 49750 }, { "epoch": 22.26896397404341, "grad_norm": 0.10401077568531036, "learning_rate": 7.754096931564431e-06, "loss": 0.022, "step": 49760 }, { "epoch": 22.27343924815395, "grad_norm": 0.06143175810575485, "learning_rate": 7.739362782442021e-06, "loss": 0.0264, "step": 49770 }, { "epoch": 22.27791452226449, "grad_norm": 0.07999242842197418, "learning_rate": 7.724641470985378e-06, "loss": 0.0257, "step": 49780 }, { "epoch": 22.28238979637503, "grad_norm": 0.07898827642202377, "learning_rate": 7.709933001666431e-06, "loss": 0.0242, "step": 49790 }, { "epoch": 22.286865070485568, "grad_norm": 0.0929323062300682, "learning_rate": 7.695237378953223e-06, "loss": 0.0214, "step": 49800 }, { "epoch": 22.291340344596108, "grad_norm": 0.08778774738311768, "learning_rate": 7.680554607309926e-06, "loss": 0.0283, "step": 49810 }, { "epoch": 22.295815618706644, "grad_norm": 0.08867551386356354, "learning_rate": 7.66588469119675e-06, "loss": 0.0224, "step": 49820 }, { "epoch": 22.300290892817184, "grad_norm": 0.09392254799604416, "learning_rate": 7.651227635070041e-06, "loss": 0.0226, "step": 49830 }, { "epoch": 22.304766166927724, "grad_norm": 0.09163708239793777, "learning_rate": 7.636583443382223e-06, "loss": 0.025, "step": 49840 }, { "epoch": 22.309241441038264, "grad_norm": 0.0793265849351883, "learning_rate": 7.62195212058181e-06, "loss": 0.0219, "step": 49850 }, { "epoch": 22.313716715148804, "grad_norm": 0.08413085341453552, "learning_rate": 7.607333671113409e-06, "loss": 0.0232, "step": 49860 }, { "epoch": 22.318191989259343, "grad_norm": 0.07164056599140167, "learning_rate": 7.592728099417745e-06, "loss": 0.0235, "step": 49870 }, { "epoch": 22.32266726336988, "grad_norm": 0.07618215680122375, "learning_rate": 7.578135409931558e-06, "loss": 0.0211, "step": 49880 }, { "epoch": 22.32714253748042, "grad_norm": 0.08784940093755722, "learning_rate": 7.56355560708778e-06, "loss": 0.0232, "step": 49890 }, { "epoch": 22.33161781159096, "grad_norm": 0.08804085105657578, "learning_rate": 7.5489886953153125e-06, "loss": 0.0233, "step": 49900 }, { "epoch": 22.3360930857015, "grad_norm": 0.08339692652225494, "learning_rate": 7.5344346790392375e-06, "loss": 0.0233, "step": 49910 }, { "epoch": 22.34056835981204, "grad_norm": 0.09312570095062256, "learning_rate": 7.519893562680663e-06, "loss": 0.0242, "step": 49920 }, { "epoch": 22.34504363392258, "grad_norm": 0.07170523703098297, "learning_rate": 7.505365350656812e-06, "loss": 0.0223, "step": 49930 }, { "epoch": 22.34951890803312, "grad_norm": 0.07363861799240112, "learning_rate": 7.490850047380954e-06, "loss": 0.0219, "step": 49940 }, { "epoch": 22.353994182143655, "grad_norm": 0.07788985222578049, "learning_rate": 7.476347657262456e-06, "loss": 0.024, "step": 49950 }, { "epoch": 22.358469456254195, "grad_norm": 0.0886843204498291, "learning_rate": 7.461858184706777e-06, "loss": 0.0223, "step": 49960 }, { "epoch": 22.362944730364735, "grad_norm": 0.087040476500988, "learning_rate": 7.4473816341154245e-06, "loss": 0.0217, "step": 49970 }, { "epoch": 22.367420004475274, "grad_norm": 0.08607640117406845, "learning_rate": 7.432918009885997e-06, "loss": 0.0238, "step": 49980 }, { "epoch": 22.371895278585814, "grad_norm": 0.07851847261190414, "learning_rate": 7.418467316412158e-06, "loss": 0.0231, "step": 49990 }, { "epoch": 22.376370552696354, "grad_norm": 0.08979766070842743, "learning_rate": 7.404029558083653e-06, "loss": 0.0256, "step": 50000 }, { "epoch": 22.38084582680689, "grad_norm": 0.0607261098921299, "learning_rate": 7.389604739286271e-06, "loss": 0.02, "step": 50010 }, { "epoch": 22.38532110091743, "grad_norm": 0.08410079032182693, "learning_rate": 7.375192864401931e-06, "loss": 0.0226, "step": 50020 }, { "epoch": 22.38979637502797, "grad_norm": 0.08384741097688675, "learning_rate": 7.36079393780853e-06, "loss": 0.0228, "step": 50030 }, { "epoch": 22.39427164913851, "grad_norm": 0.09945249557495117, "learning_rate": 7.3464079638801365e-06, "loss": 0.0241, "step": 50040 }, { "epoch": 22.39874692324905, "grad_norm": 0.11672493815422058, "learning_rate": 7.332034946986771e-06, "loss": 0.0251, "step": 50050 }, { "epoch": 22.40322219735959, "grad_norm": 0.07971016317605972, "learning_rate": 7.317674891494625e-06, "loss": 0.0209, "step": 50060 }, { "epoch": 22.407697471470126, "grad_norm": 0.10983368009328842, "learning_rate": 7.30332780176588e-06, "loss": 0.0226, "step": 50070 }, { "epoch": 22.412172745580666, "grad_norm": 0.09338223189115524, "learning_rate": 7.2889936821588125e-06, "loss": 0.0242, "step": 50080 }, { "epoch": 22.416648019691205, "grad_norm": 0.09292897582054138, "learning_rate": 7.2746725370277435e-06, "loss": 0.0227, "step": 50090 }, { "epoch": 22.421123293801745, "grad_norm": 0.0841149166226387, "learning_rate": 7.260364370723044e-06, "loss": 0.0266, "step": 50100 }, { "epoch": 22.425598567912285, "grad_norm": 0.07816166430711746, "learning_rate": 7.246069187591204e-06, "loss": 0.0286, "step": 50110 }, { "epoch": 22.430073842022825, "grad_norm": 0.0654289722442627, "learning_rate": 7.2317869919746705e-06, "loss": 0.0233, "step": 50120 }, { "epoch": 22.434549116133365, "grad_norm": 0.06337681412696838, "learning_rate": 7.217517788212025e-06, "loss": 0.0228, "step": 50130 }, { "epoch": 22.4390243902439, "grad_norm": 0.11201604455709457, "learning_rate": 7.203261580637877e-06, "loss": 0.025, "step": 50140 }, { "epoch": 22.44349966435444, "grad_norm": 0.07675135135650635, "learning_rate": 7.189018373582873e-06, "loss": 0.0226, "step": 50150 }, { "epoch": 22.44797493846498, "grad_norm": 0.06358262896537781, "learning_rate": 7.174788171373731e-06, "loss": 0.0229, "step": 50160 }, { "epoch": 22.45245021257552, "grad_norm": 0.05660836026072502, "learning_rate": 7.160570978333203e-06, "loss": 0.0215, "step": 50170 }, { "epoch": 22.45692548668606, "grad_norm": 0.06939582526683807, "learning_rate": 7.146366798780096e-06, "loss": 0.022, "step": 50180 }, { "epoch": 22.4614007607966, "grad_norm": 0.08394255489110947, "learning_rate": 7.132175637029293e-06, "loss": 0.0199, "step": 50190 }, { "epoch": 22.465876034907136, "grad_norm": 0.0994134172797203, "learning_rate": 7.1179974973916486e-06, "loss": 0.0231, "step": 50200 }, { "epoch": 22.470351309017676, "grad_norm": 0.10102816671133041, "learning_rate": 7.103832384174142e-06, "loss": 0.0244, "step": 50210 }, { "epoch": 22.474826583128216, "grad_norm": 0.07130665332078934, "learning_rate": 7.089680301679752e-06, "loss": 0.0225, "step": 50220 }, { "epoch": 22.479301857238756, "grad_norm": 0.097496896982193, "learning_rate": 7.075541254207502e-06, "loss": 0.0254, "step": 50230 }, { "epoch": 22.483777131349296, "grad_norm": 0.08947519958019257, "learning_rate": 7.061415246052466e-06, "loss": 0.0255, "step": 50240 }, { "epoch": 22.488252405459836, "grad_norm": 0.07277868688106537, "learning_rate": 7.047302281505736e-06, "loss": 0.0222, "step": 50250 }, { "epoch": 22.492727679570372, "grad_norm": 0.09191254526376724, "learning_rate": 7.0332023648544965e-06, "loss": 0.0223, "step": 50260 }, { "epoch": 22.49720295368091, "grad_norm": 0.0675560012459755, "learning_rate": 7.01911550038189e-06, "loss": 0.0233, "step": 50270 }, { "epoch": 22.50167822779145, "grad_norm": 0.09441602230072021, "learning_rate": 7.005041692367154e-06, "loss": 0.0248, "step": 50280 }, { "epoch": 22.50615350190199, "grad_norm": 0.08409996330738068, "learning_rate": 6.9909809450855345e-06, "loss": 0.0225, "step": 50290 }, { "epoch": 22.51062877601253, "grad_norm": 0.07708868384361267, "learning_rate": 6.976933262808322e-06, "loss": 0.025, "step": 50300 }, { "epoch": 22.51510405012307, "grad_norm": 0.09187404066324234, "learning_rate": 6.962898649802823e-06, "loss": 0.0228, "step": 50310 }, { "epoch": 22.51957932423361, "grad_norm": 0.08168166130781174, "learning_rate": 6.948877110332386e-06, "loss": 0.0221, "step": 50320 }, { "epoch": 22.524054598344147, "grad_norm": 0.07750482857227325, "learning_rate": 6.934868648656373e-06, "loss": 0.0213, "step": 50330 }, { "epoch": 22.528529872454687, "grad_norm": 0.08196471631526947, "learning_rate": 6.92087326903022e-06, "loss": 0.0213, "step": 50340 }, { "epoch": 22.533005146565227, "grad_norm": 0.0747394859790802, "learning_rate": 6.906890975705305e-06, "loss": 0.0228, "step": 50350 }, { "epoch": 22.537480420675767, "grad_norm": 0.06756541132926941, "learning_rate": 6.892921772929112e-06, "loss": 0.0247, "step": 50360 }, { "epoch": 22.541955694786306, "grad_norm": 0.08683958649635315, "learning_rate": 6.878965664945108e-06, "loss": 0.0221, "step": 50370 }, { "epoch": 22.546430968896846, "grad_norm": 0.08484320342540741, "learning_rate": 6.865022655992798e-06, "loss": 0.022, "step": 50380 }, { "epoch": 22.550906243007383, "grad_norm": 0.07522361725568771, "learning_rate": 6.851092750307686e-06, "loss": 0.023, "step": 50390 }, { "epoch": 22.555381517117922, "grad_norm": 0.07274962961673737, "learning_rate": 6.837175952121306e-06, "loss": 0.0249, "step": 50400 }, { "epoch": 22.559856791228462, "grad_norm": 0.08217878639698029, "learning_rate": 6.8232722656612405e-06, "loss": 0.023, "step": 50410 }, { "epoch": 22.564332065339002, "grad_norm": 0.06927512586116791, "learning_rate": 6.809381695151029e-06, "loss": 0.0212, "step": 50420 }, { "epoch": 22.568807339449542, "grad_norm": 0.05670664831995964, "learning_rate": 6.795504244810285e-06, "loss": 0.0224, "step": 50430 }, { "epoch": 22.57328261356008, "grad_norm": 0.0910080298781395, "learning_rate": 6.781639918854604e-06, "loss": 0.0267, "step": 50440 }, { "epoch": 22.57775788767062, "grad_norm": 0.10051415115594864, "learning_rate": 6.7677887214956e-06, "loss": 0.0231, "step": 50450 }, { "epoch": 22.582233161781158, "grad_norm": 0.08855152875185013, "learning_rate": 6.753950656940905e-06, "loss": 0.0226, "step": 50460 }, { "epoch": 22.586708435891698, "grad_norm": 0.07941560447216034, "learning_rate": 6.740125729394148e-06, "loss": 0.0233, "step": 50470 }, { "epoch": 22.591183710002237, "grad_norm": 0.07697858661413193, "learning_rate": 6.726313943054991e-06, "loss": 0.0232, "step": 50480 }, { "epoch": 22.595658984112777, "grad_norm": 0.09071075171232224, "learning_rate": 6.712515302119077e-06, "loss": 0.0227, "step": 50490 }, { "epoch": 22.600134258223317, "grad_norm": 0.05286739766597748, "learning_rate": 6.698729810778065e-06, "loss": 0.0247, "step": 50500 }, { "epoch": 22.604609532333857, "grad_norm": 0.07504855841398239, "learning_rate": 6.684957473219644e-06, "loss": 0.0241, "step": 50510 }, { "epoch": 22.609084806444393, "grad_norm": 0.09101551026105881, "learning_rate": 6.671198293627479e-06, "loss": 0.0216, "step": 50520 }, { "epoch": 22.613560080554933, "grad_norm": 0.07185620069503784, "learning_rate": 6.6574522761812366e-06, "loss": 0.021, "step": 50530 }, { "epoch": 22.618035354665473, "grad_norm": 0.08230318874120712, "learning_rate": 6.6437194250566e-06, "loss": 0.027, "step": 50540 }, { "epoch": 22.622510628776013, "grad_norm": 0.08249399811029434, "learning_rate": 6.629999744425236e-06, "loss": 0.0244, "step": 50550 }, { "epoch": 22.626985902886553, "grad_norm": 0.06431064009666443, "learning_rate": 6.6162932384548515e-06, "loss": 0.0237, "step": 50560 }, { "epoch": 22.631461176997092, "grad_norm": 0.08386485278606415, "learning_rate": 6.602599911309082e-06, "loss": 0.0213, "step": 50570 }, { "epoch": 22.63593645110763, "grad_norm": 0.0672784075140953, "learning_rate": 6.588919767147639e-06, "loss": 0.0224, "step": 50580 }, { "epoch": 22.64041172521817, "grad_norm": 0.07327983528375626, "learning_rate": 6.575252810126143e-06, "loss": 0.0245, "step": 50590 }, { "epoch": 22.64488699932871, "grad_norm": 0.08883638679981232, "learning_rate": 6.561599044396288e-06, "loss": 0.0235, "step": 50600 }, { "epoch": 22.649362273439248, "grad_norm": 0.09044286608695984, "learning_rate": 6.5479584741057255e-06, "loss": 0.0248, "step": 50610 }, { "epoch": 22.653837547549788, "grad_norm": 0.08399950712919235, "learning_rate": 6.5343311033980895e-06, "loss": 0.0231, "step": 50620 }, { "epoch": 22.658312821660328, "grad_norm": 0.09142995625734329, "learning_rate": 6.520716936413018e-06, "loss": 0.0239, "step": 50630 }, { "epoch": 22.662788095770868, "grad_norm": 0.09724117070436478, "learning_rate": 6.5071159772861436e-06, "loss": 0.0234, "step": 50640 }, { "epoch": 22.667263369881404, "grad_norm": 0.08262860774993896, "learning_rate": 6.493528230149054e-06, "loss": 0.0259, "step": 50650 }, { "epoch": 22.671738643991944, "grad_norm": 0.09185628592967987, "learning_rate": 6.479953699129382e-06, "loss": 0.0246, "step": 50660 }, { "epoch": 22.676213918102484, "grad_norm": 0.08545278757810593, "learning_rate": 6.466392388350695e-06, "loss": 0.0225, "step": 50670 }, { "epoch": 22.680689192213023, "grad_norm": 0.08079634606838226, "learning_rate": 6.452844301932559e-06, "loss": 0.0204, "step": 50680 }, { "epoch": 22.685164466323563, "grad_norm": 0.11417687684297562, "learning_rate": 6.439309443990532e-06, "loss": 0.0231, "step": 50690 }, { "epoch": 22.689639740434103, "grad_norm": 0.09191784262657166, "learning_rate": 6.425787818636131e-06, "loss": 0.0203, "step": 50700 }, { "epoch": 22.69411501454464, "grad_norm": 0.08832793682813644, "learning_rate": 6.412279429976903e-06, "loss": 0.0215, "step": 50710 }, { "epoch": 22.69859028865518, "grad_norm": 0.07404299080371857, "learning_rate": 6.398784282116293e-06, "loss": 0.0226, "step": 50720 }, { "epoch": 22.70306556276572, "grad_norm": 0.11435121297836304, "learning_rate": 6.385302379153818e-06, "loss": 0.0247, "step": 50730 }, { "epoch": 22.70754083687626, "grad_norm": 0.07030846178531647, "learning_rate": 6.3718337251848785e-06, "loss": 0.0242, "step": 50740 }, { "epoch": 22.7120161109868, "grad_norm": 0.0825062096118927, "learning_rate": 6.3583783243009285e-06, "loss": 0.0217, "step": 50750 }, { "epoch": 22.71649138509734, "grad_norm": 0.08046552538871765, "learning_rate": 6.344936180589351e-06, "loss": 0.024, "step": 50760 }, { "epoch": 22.720966659207875, "grad_norm": 0.07824260741472244, "learning_rate": 6.3315072981335215e-06, "loss": 0.0224, "step": 50770 }, { "epoch": 22.725441933318415, "grad_norm": 0.07224129885435104, "learning_rate": 6.318091681012772e-06, "loss": 0.0241, "step": 50780 }, { "epoch": 22.729917207428954, "grad_norm": 0.15485042333602905, "learning_rate": 6.304689333302416e-06, "loss": 0.0247, "step": 50790 }, { "epoch": 22.734392481539494, "grad_norm": 0.090182363986969, "learning_rate": 6.291300259073724e-06, "loss": 0.0236, "step": 50800 }, { "epoch": 22.738867755650034, "grad_norm": 0.07598832249641418, "learning_rate": 6.277924462393958e-06, "loss": 0.0223, "step": 50810 }, { "epoch": 22.743343029760574, "grad_norm": 0.09158561378717422, "learning_rate": 6.264561947326331e-06, "loss": 0.0208, "step": 50820 }, { "epoch": 22.747818303871114, "grad_norm": 0.08485899865627289, "learning_rate": 6.251212717930017e-06, "loss": 0.0266, "step": 50830 }, { "epoch": 22.75229357798165, "grad_norm": 0.09215296059846878, "learning_rate": 6.237876778260155e-06, "loss": 0.0216, "step": 50840 }, { "epoch": 22.75676885209219, "grad_norm": 0.06673050671815872, "learning_rate": 6.22455413236786e-06, "loss": 0.0246, "step": 50850 }, { "epoch": 22.76124412620273, "grad_norm": 0.0791642963886261, "learning_rate": 6.211244784300197e-06, "loss": 0.025, "step": 50860 }, { "epoch": 22.76571940031327, "grad_norm": 0.08515533804893494, "learning_rate": 6.1979487381001786e-06, "loss": 0.0231, "step": 50870 }, { "epoch": 22.77019467442381, "grad_norm": 0.07632050663232803, "learning_rate": 6.184665997806832e-06, "loss": 0.0256, "step": 50880 }, { "epoch": 22.77466994853435, "grad_norm": 0.10391424596309662, "learning_rate": 6.171396567455051e-06, "loss": 0.0228, "step": 50890 }, { "epoch": 22.779145222644885, "grad_norm": 0.08698688447475433, "learning_rate": 6.158140451075795e-06, "loss": 0.027, "step": 50900 }, { "epoch": 22.783620496755425, "grad_norm": 0.09098260849714279, "learning_rate": 6.144897652695864e-06, "loss": 0.0237, "step": 50910 }, { "epoch": 22.788095770865965, "grad_norm": 0.09082435071468353, "learning_rate": 6.131668176338118e-06, "loss": 0.0237, "step": 50920 }, { "epoch": 22.792571044976505, "grad_norm": 0.060685865581035614, "learning_rate": 6.118452026021299e-06, "loss": 0.02, "step": 50930 }, { "epoch": 22.797046319087045, "grad_norm": 0.06650228798389435, "learning_rate": 6.1052492057601275e-06, "loss": 0.0252, "step": 50940 }, { "epoch": 22.801521593197585, "grad_norm": 0.09199979901313782, "learning_rate": 6.092059719565274e-06, "loss": 0.0266, "step": 50950 }, { "epoch": 22.805996867308124, "grad_norm": 0.06827617436647415, "learning_rate": 6.07888357144335e-06, "loss": 0.0221, "step": 50960 }, { "epoch": 22.81047214141866, "grad_norm": 0.07980797439813614, "learning_rate": 6.0657207653969315e-06, "loss": 0.0233, "step": 50970 }, { "epoch": 22.8149474155292, "grad_norm": 0.08974282443523407, "learning_rate": 6.052571305424531e-06, "loss": 0.0214, "step": 50980 }, { "epoch": 22.81942268963974, "grad_norm": 0.08716230094432831, "learning_rate": 6.039435195520604e-06, "loss": 0.0255, "step": 50990 }, { "epoch": 22.82389796375028, "grad_norm": 0.09405400604009628, "learning_rate": 6.026312439675552e-06, "loss": 0.0236, "step": 51000 }, { "epoch": 22.82837323786082, "grad_norm": 0.08845628052949905, "learning_rate": 6.01320304187572e-06, "loss": 0.0244, "step": 51010 }, { "epoch": 22.83284851197136, "grad_norm": 0.0859757661819458, "learning_rate": 6.0001070061033945e-06, "loss": 0.0277, "step": 51020 }, { "epoch": 22.837323786081896, "grad_norm": 0.09407148510217667, "learning_rate": 5.9870243363368275e-06, "loss": 0.0216, "step": 51030 }, { "epoch": 22.841799060192436, "grad_norm": 0.06780661642551422, "learning_rate": 5.9739550365501494e-06, "loss": 0.0242, "step": 51040 }, { "epoch": 22.846274334302976, "grad_norm": 0.08222386240959167, "learning_rate": 5.960899110713519e-06, "loss": 0.0256, "step": 51050 }, { "epoch": 22.850749608413516, "grad_norm": 0.09215040504932404, "learning_rate": 5.947856562792925e-06, "loss": 0.0246, "step": 51060 }, { "epoch": 22.855224882524055, "grad_norm": 0.09325899928808212, "learning_rate": 5.934827396750392e-06, "loss": 0.0226, "step": 51070 }, { "epoch": 22.859700156634595, "grad_norm": 0.07758355140686035, "learning_rate": 5.921811616543821e-06, "loss": 0.0226, "step": 51080 }, { "epoch": 22.86417543074513, "grad_norm": 0.08533790707588196, "learning_rate": 5.908809226127054e-06, "loss": 0.0225, "step": 51090 }, { "epoch": 22.86865070485567, "grad_norm": 0.0790102481842041, "learning_rate": 5.895820229449906e-06, "loss": 0.0206, "step": 51100 }, { "epoch": 22.87312597896621, "grad_norm": 0.0823378935456276, "learning_rate": 5.882844630458045e-06, "loss": 0.022, "step": 51110 }, { "epoch": 22.87760125307675, "grad_norm": 0.08467613905668259, "learning_rate": 5.869882433093155e-06, "loss": 0.021, "step": 51120 }, { "epoch": 22.88207652718729, "grad_norm": 0.08854441344738007, "learning_rate": 5.856933641292789e-06, "loss": 0.0239, "step": 51130 }, { "epoch": 22.88655180129783, "grad_norm": 0.07479114085435867, "learning_rate": 5.843998258990452e-06, "loss": 0.0222, "step": 51140 }, { "epoch": 22.89102707540837, "grad_norm": 0.08658897131681442, "learning_rate": 5.831076290115573e-06, "loss": 0.0225, "step": 51150 }, { "epoch": 22.895502349518907, "grad_norm": 0.07119441032409668, "learning_rate": 5.818167738593505e-06, "loss": 0.0222, "step": 51160 }, { "epoch": 22.899977623629447, "grad_norm": 0.06270641833543777, "learning_rate": 5.805272608345513e-06, "loss": 0.0233, "step": 51170 }, { "epoch": 22.904452897739986, "grad_norm": 0.08473464101552963, "learning_rate": 5.79239090328883e-06, "loss": 0.0219, "step": 51180 }, { "epoch": 22.908928171850526, "grad_norm": 0.08221861720085144, "learning_rate": 5.779522627336537e-06, "loss": 0.0226, "step": 51190 }, { "epoch": 22.913403445961066, "grad_norm": 0.07944405823945999, "learning_rate": 5.766667784397706e-06, "loss": 0.023, "step": 51200 }, { "epoch": 22.917878720071606, "grad_norm": 0.07375848293304443, "learning_rate": 5.753826378377286e-06, "loss": 0.021, "step": 51210 }, { "epoch": 22.922353994182142, "grad_norm": 0.07806754112243652, "learning_rate": 5.740998413176163e-06, "loss": 0.0229, "step": 51220 }, { "epoch": 22.926829268292682, "grad_norm": 0.08026893436908722, "learning_rate": 5.72818389269113e-06, "loss": 0.0245, "step": 51230 }, { "epoch": 22.931304542403222, "grad_norm": 0.07907789945602417, "learning_rate": 5.715382820814885e-06, "loss": 0.0267, "step": 51240 }, { "epoch": 22.93577981651376, "grad_norm": 0.08006113767623901, "learning_rate": 5.702595201436101e-06, "loss": 0.0294, "step": 51250 }, { "epoch": 22.9402550906243, "grad_norm": 0.10551148653030396, "learning_rate": 5.689821038439263e-06, "loss": 0.0275, "step": 51260 }, { "epoch": 22.94473036473484, "grad_norm": 0.08993925154209137, "learning_rate": 5.67706033570487e-06, "loss": 0.0246, "step": 51270 }, { "epoch": 22.949205638845378, "grad_norm": 0.09229132533073425, "learning_rate": 5.6643130971092525e-06, "loss": 0.0233, "step": 51280 }, { "epoch": 22.953680912955917, "grad_norm": 0.08382605016231537, "learning_rate": 5.651579326524709e-06, "loss": 0.0231, "step": 51290 }, { "epoch": 22.958156187066457, "grad_norm": 0.0830502137541771, "learning_rate": 5.6388590278194096e-06, "loss": 0.0272, "step": 51300 }, { "epoch": 22.962631461176997, "grad_norm": 0.08344598859548569, "learning_rate": 5.626152204857454e-06, "loss": 0.0291, "step": 51310 }, { "epoch": 22.967106735287537, "grad_norm": 0.07680463790893555, "learning_rate": 5.613458861498832e-06, "loss": 0.0229, "step": 51320 }, { "epoch": 22.971582009398077, "grad_norm": 0.08639951795339584, "learning_rate": 5.600779001599455e-06, "loss": 0.026, "step": 51330 }, { "epoch": 22.976057283508617, "grad_norm": 0.07581108808517456, "learning_rate": 5.58811262901111e-06, "loss": 0.0217, "step": 51340 }, { "epoch": 22.980532557619153, "grad_norm": 0.0743539035320282, "learning_rate": 5.57545974758153e-06, "loss": 0.0227, "step": 51350 }, { "epoch": 22.985007831729693, "grad_norm": 0.08059895038604736, "learning_rate": 5.562820361154314e-06, "loss": 0.0199, "step": 51360 }, { "epoch": 22.989483105840232, "grad_norm": 0.09815160185098648, "learning_rate": 5.550194473568981e-06, "loss": 0.0256, "step": 51370 }, { "epoch": 22.993958379950772, "grad_norm": 0.06792279332876205, "learning_rate": 5.537582088660937e-06, "loss": 0.0253, "step": 51380 }, { "epoch": 22.998433654061312, "grad_norm": 0.0687752515077591, "learning_rate": 5.524983210261481e-06, "loss": 0.0252, "step": 51390 }, { "epoch": 23.002908928171852, "grad_norm": 0.06884366273880005, "learning_rate": 5.512397842197847e-06, "loss": 0.0254, "step": 51400 }, { "epoch": 23.00738420228239, "grad_norm": 0.0718134194612503, "learning_rate": 5.499825988293106e-06, "loss": 0.0218, "step": 51410 }, { "epoch": 23.011859476392928, "grad_norm": 0.07381029427051544, "learning_rate": 5.48726765236629e-06, "loss": 0.0199, "step": 51420 }, { "epoch": 23.016334750503468, "grad_norm": 0.07511071860790253, "learning_rate": 5.474722838232254e-06, "loss": 0.0234, "step": 51430 }, { "epoch": 23.020810024614008, "grad_norm": 0.07367487996816635, "learning_rate": 5.462191549701806e-06, "loss": 0.0234, "step": 51440 }, { "epoch": 23.025285298724548, "grad_norm": 0.08542537689208984, "learning_rate": 5.449673790581611e-06, "loss": 0.0199, "step": 51450 }, { "epoch": 23.029760572835087, "grad_norm": 0.08553268760442734, "learning_rate": 5.437169564674233e-06, "loss": 0.0252, "step": 51460 }, { "epoch": 23.034235846945624, "grad_norm": 0.07653208076953888, "learning_rate": 5.424678875778133e-06, "loss": 0.022, "step": 51470 }, { "epoch": 23.038711121056163, "grad_norm": 0.08136743307113647, "learning_rate": 5.412201727687644e-06, "loss": 0.0228, "step": 51480 }, { "epoch": 23.043186395166703, "grad_norm": 0.09052679687738419, "learning_rate": 5.399738124192988e-06, "loss": 0.0217, "step": 51490 }, { "epoch": 23.047661669277243, "grad_norm": 0.054103583097457886, "learning_rate": 5.387288069080299e-06, "loss": 0.0206, "step": 51500 }, { "epoch": 23.052136943387783, "grad_norm": 0.08594919741153717, "learning_rate": 5.374851566131561e-06, "loss": 0.0221, "step": 51510 }, { "epoch": 23.056612217498323, "grad_norm": 0.07732944190502167, "learning_rate": 5.362428619124666e-06, "loss": 0.0243, "step": 51520 }, { "epoch": 23.061087491608863, "grad_norm": 0.0790652185678482, "learning_rate": 5.350019231833364e-06, "loss": 0.0258, "step": 51530 }, { "epoch": 23.0655627657194, "grad_norm": 0.07864946126937866, "learning_rate": 5.337623408027293e-06, "loss": 0.0245, "step": 51540 }, { "epoch": 23.07003803982994, "grad_norm": 0.1135566234588623, "learning_rate": 5.32524115147201e-06, "loss": 0.0247, "step": 51550 }, { "epoch": 23.07451331394048, "grad_norm": 0.08329805731773376, "learning_rate": 5.312872465928881e-06, "loss": 0.0217, "step": 51560 }, { "epoch": 23.07898858805102, "grad_norm": 0.07227176427841187, "learning_rate": 5.300517355155215e-06, "loss": 0.0219, "step": 51570 }, { "epoch": 23.083463862161558, "grad_norm": 0.08753903210163116, "learning_rate": 5.28817582290414e-06, "loss": 0.0251, "step": 51580 }, { "epoch": 23.087939136272098, "grad_norm": 0.07081082463264465, "learning_rate": 5.2758478729247164e-06, "loss": 0.0199, "step": 51590 }, { "epoch": 23.092414410382634, "grad_norm": 0.07955548912286758, "learning_rate": 5.263533508961827e-06, "loss": 0.0276, "step": 51600 }, { "epoch": 23.096889684493174, "grad_norm": 0.09326305985450745, "learning_rate": 5.251232734756268e-06, "loss": 0.0233, "step": 51610 }, { "epoch": 23.101364958603714, "grad_norm": 0.07042413204908371, "learning_rate": 5.238945554044672e-06, "loss": 0.0236, "step": 51620 }, { "epoch": 23.105840232714254, "grad_norm": 0.07065989822149277, "learning_rate": 5.226671970559577e-06, "loss": 0.0251, "step": 51630 }, { "epoch": 23.110315506824794, "grad_norm": 0.07922409474849701, "learning_rate": 5.214411988029355e-06, "loss": 0.0245, "step": 51640 }, { "epoch": 23.114790780935333, "grad_norm": 0.057473134249448776, "learning_rate": 5.202165610178284e-06, "loss": 0.0221, "step": 51650 }, { "epoch": 23.119266055045873, "grad_norm": 0.0772484764456749, "learning_rate": 5.1899328407264855e-06, "loss": 0.0215, "step": 51660 }, { "epoch": 23.12374132915641, "grad_norm": 0.09488261491060257, "learning_rate": 5.177713683389945e-06, "loss": 0.0268, "step": 51670 }, { "epoch": 23.12821660326695, "grad_norm": 0.07365269213914871, "learning_rate": 5.165508141880526e-06, "loss": 0.024, "step": 51680 }, { "epoch": 23.13269187737749, "grad_norm": 0.08985736221075058, "learning_rate": 5.153316219905946e-06, "loss": 0.0238, "step": 51690 }, { "epoch": 23.13716715148803, "grad_norm": 0.08007828146219254, "learning_rate": 5.141137921169792e-06, "loss": 0.0223, "step": 51700 }, { "epoch": 23.14164242559857, "grad_norm": 0.0736691802740097, "learning_rate": 5.128973249371494e-06, "loss": 0.0253, "step": 51710 }, { "epoch": 23.14611769970911, "grad_norm": 0.08379337936639786, "learning_rate": 5.116822208206396e-06, "loss": 0.0235, "step": 51720 }, { "epoch": 23.150592973819645, "grad_norm": 0.07617625594139099, "learning_rate": 5.1046848013656165e-06, "loss": 0.0249, "step": 51730 }, { "epoch": 23.155068247930185, "grad_norm": 0.07281877845525742, "learning_rate": 5.092561032536225e-06, "loss": 0.0241, "step": 51740 }, { "epoch": 23.159543522040725, "grad_norm": 0.07741906493902206, "learning_rate": 5.080450905401057e-06, "loss": 0.0229, "step": 51750 }, { "epoch": 23.164018796151264, "grad_norm": 0.13230668008327484, "learning_rate": 5.068354423638882e-06, "loss": 0.0211, "step": 51760 }, { "epoch": 23.168494070261804, "grad_norm": 0.08912190794944763, "learning_rate": 5.056271590924283e-06, "loss": 0.0234, "step": 51770 }, { "epoch": 23.172969344372344, "grad_norm": 0.08027001470327377, "learning_rate": 5.044202410927706e-06, "loss": 0.0257, "step": 51780 }, { "epoch": 23.17744461848288, "grad_norm": 0.09494546055793762, "learning_rate": 5.032146887315448e-06, "loss": 0.0223, "step": 51790 }, { "epoch": 23.18191989259342, "grad_norm": 0.07495315372943878, "learning_rate": 5.020105023749644e-06, "loss": 0.0221, "step": 51800 }, { "epoch": 23.18639516670396, "grad_norm": 0.07635717839002609, "learning_rate": 5.008076823888319e-06, "loss": 0.0226, "step": 51810 }, { "epoch": 23.1908704408145, "grad_norm": 0.06563503295183182, "learning_rate": 4.996062291385317e-06, "loss": 0.0238, "step": 51820 }, { "epoch": 23.19534571492504, "grad_norm": 0.09091918170452118, "learning_rate": 4.984061429890324e-06, "loss": 0.0224, "step": 51830 }, { "epoch": 23.19982098903558, "grad_norm": 0.08123914897441864, "learning_rate": 4.972074243048897e-06, "loss": 0.0266, "step": 51840 }, { "epoch": 23.20429626314612, "grad_norm": 0.04360344260931015, "learning_rate": 4.960100734502421e-06, "loss": 0.0234, "step": 51850 }, { "epoch": 23.208771537256656, "grad_norm": 0.08101079612970352, "learning_rate": 4.948140907888121e-06, "loss": 0.0246, "step": 51860 }, { "epoch": 23.213246811367195, "grad_norm": 0.07686589658260345, "learning_rate": 4.936194766839103e-06, "loss": 0.0242, "step": 51870 }, { "epoch": 23.217722085477735, "grad_norm": 0.06921537965536118, "learning_rate": 4.924262314984262e-06, "loss": 0.0218, "step": 51880 }, { "epoch": 23.222197359588275, "grad_norm": 0.10044244676828384, "learning_rate": 4.91234355594839e-06, "loss": 0.0208, "step": 51890 }, { "epoch": 23.226672633698815, "grad_norm": 0.07175976783037186, "learning_rate": 4.900438493352055e-06, "loss": 0.0267, "step": 51900 }, { "epoch": 23.231147907809355, "grad_norm": 0.07653314620256424, "learning_rate": 4.888547130811732e-06, "loss": 0.0227, "step": 51910 }, { "epoch": 23.23562318191989, "grad_norm": 0.09256741404533386, "learning_rate": 4.8766694719396875e-06, "loss": 0.0221, "step": 51920 }, { "epoch": 23.24009845603043, "grad_norm": 0.08346337080001831, "learning_rate": 4.864805520344051e-06, "loss": 0.0263, "step": 51930 }, { "epoch": 23.24457373014097, "grad_norm": 0.07306932657957077, "learning_rate": 4.852955279628768e-06, "loss": 0.0214, "step": 51940 }, { "epoch": 23.24904900425151, "grad_norm": 0.060868650674819946, "learning_rate": 4.8411187533936195e-06, "loss": 0.0209, "step": 51950 }, { "epoch": 23.25352427836205, "grad_norm": 0.0673288032412529, "learning_rate": 4.829295945234258e-06, "loss": 0.0228, "step": 51960 }, { "epoch": 23.25799955247259, "grad_norm": 0.08280345797538757, "learning_rate": 4.817486858742127e-06, "loss": 0.0232, "step": 51970 }, { "epoch": 23.262474826583126, "grad_norm": 0.09179087728261948, "learning_rate": 4.805691497504505e-06, "loss": 0.0247, "step": 51980 }, { "epoch": 23.266950100693666, "grad_norm": 0.06317055970430374, "learning_rate": 4.7939098651045235e-06, "loss": 0.0231, "step": 51990 }, { "epoch": 23.271425374804206, "grad_norm": 0.09458164870738983, "learning_rate": 4.782141965121128e-06, "loss": 0.0199, "step": 52000 }, { "epoch": 23.275900648914746, "grad_norm": 0.0738387480378151, "learning_rate": 4.770387801129084e-06, "loss": 0.0203, "step": 52010 }, { "epoch": 23.280375923025286, "grad_norm": 0.09234633296728134, "learning_rate": 4.758647376699032e-06, "loss": 0.0258, "step": 52020 }, { "epoch": 23.284851197135826, "grad_norm": 0.07854770869016647, "learning_rate": 4.7469206953973495e-06, "loss": 0.0205, "step": 52030 }, { "epoch": 23.289326471246365, "grad_norm": 0.05745663493871689, "learning_rate": 4.7352077607863475e-06, "loss": 0.0207, "step": 52040 }, { "epoch": 23.2938017453569, "grad_norm": 0.08521430194377899, "learning_rate": 4.723508576424062e-06, "loss": 0.0221, "step": 52050 }, { "epoch": 23.29827701946744, "grad_norm": 0.07691773772239685, "learning_rate": 4.711823145864419e-06, "loss": 0.021, "step": 52060 }, { "epoch": 23.30275229357798, "grad_norm": 0.06865062564611435, "learning_rate": 4.700151472657144e-06, "loss": 0.023, "step": 52070 }, { "epoch": 23.30722756768852, "grad_norm": 0.0690888985991478, "learning_rate": 4.688493560347773e-06, "loss": 0.021, "step": 52080 }, { "epoch": 23.31170284179906, "grad_norm": 0.07212747633457184, "learning_rate": 4.67684941247768e-06, "loss": 0.0256, "step": 52090 }, { "epoch": 23.3161781159096, "grad_norm": 0.07521888613700867, "learning_rate": 4.66521903258404e-06, "loss": 0.0231, "step": 52100 }, { "epoch": 23.320653390020137, "grad_norm": 0.07317917793989182, "learning_rate": 4.653602424199876e-06, "loss": 0.0243, "step": 52110 }, { "epoch": 23.325128664130677, "grad_norm": 0.06658344715833664, "learning_rate": 4.64199959085398e-06, "loss": 0.0223, "step": 52120 }, { "epoch": 23.329603938241217, "grad_norm": 0.0723426565527916, "learning_rate": 4.630410536071006e-06, "loss": 0.0234, "step": 52130 }, { "epoch": 23.334079212351757, "grad_norm": 0.053838100284338, "learning_rate": 4.618835263371396e-06, "loss": 0.0225, "step": 52140 }, { "epoch": 23.338554486462296, "grad_norm": 0.08203171193599701, "learning_rate": 4.6072737762714144e-06, "loss": 0.0226, "step": 52150 }, { "epoch": 23.343029760572836, "grad_norm": 0.08772063255310059, "learning_rate": 4.595726078283136e-06, "loss": 0.0234, "step": 52160 }, { "epoch": 23.347505034683373, "grad_norm": 0.13812965154647827, "learning_rate": 4.5841921729144424e-06, "loss": 0.0235, "step": 52170 }, { "epoch": 23.351980308793912, "grad_norm": 0.06732098758220673, "learning_rate": 4.57267206366902e-06, "loss": 0.0204, "step": 52180 }, { "epoch": 23.356455582904452, "grad_norm": 0.08119688928127289, "learning_rate": 4.5611657540464036e-06, "loss": 0.0224, "step": 52190 }, { "epoch": 23.360930857014992, "grad_norm": 0.07257939875125885, "learning_rate": 4.549673247541875e-06, "loss": 0.0229, "step": 52200 }, { "epoch": 23.365406131125532, "grad_norm": 0.08159693330526352, "learning_rate": 4.538194547646574e-06, "loss": 0.0257, "step": 52210 }, { "epoch": 23.36988140523607, "grad_norm": 0.0976032018661499, "learning_rate": 4.526729657847423e-06, "loss": 0.0224, "step": 52220 }, { "epoch": 23.37435667934661, "grad_norm": 0.07807765901088715, "learning_rate": 4.515278581627141e-06, "loss": 0.0241, "step": 52230 }, { "epoch": 23.378831953457148, "grad_norm": 0.07890791445970535, "learning_rate": 4.5038413224642805e-06, "loss": 0.0225, "step": 52240 }, { "epoch": 23.383307227567688, "grad_norm": 0.0954548567533493, "learning_rate": 4.492417883833155e-06, "loss": 0.0221, "step": 52250 }, { "epoch": 23.387782501678227, "grad_norm": 0.08639536052942276, "learning_rate": 4.48100826920394e-06, "loss": 0.023, "step": 52260 }, { "epoch": 23.392257775788767, "grad_norm": 0.090311698615551, "learning_rate": 4.469612482042534e-06, "loss": 0.0206, "step": 52270 }, { "epoch": 23.396733049899307, "grad_norm": 0.06812968850135803, "learning_rate": 4.458230525810708e-06, "loss": 0.0239, "step": 52280 }, { "epoch": 23.401208324009847, "grad_norm": 0.08922536671161652, "learning_rate": 4.446862403965984e-06, "loss": 0.0228, "step": 52290 }, { "epoch": 23.405683598120383, "grad_norm": 0.07718385756015778, "learning_rate": 4.435508119961701e-06, "loss": 0.022, "step": 52300 }, { "epoch": 23.410158872230923, "grad_norm": 0.07350536435842514, "learning_rate": 4.4241676772469875e-06, "loss": 0.0189, "step": 52310 }, { "epoch": 23.414634146341463, "grad_norm": 0.06075897812843323, "learning_rate": 4.412841079266777e-06, "loss": 0.0203, "step": 52320 }, { "epoch": 23.419109420452003, "grad_norm": 0.06970494985580444, "learning_rate": 4.401528329461779e-06, "loss": 0.0246, "step": 52330 }, { "epoch": 23.423584694562543, "grad_norm": 0.07008381932973862, "learning_rate": 4.390229431268534e-06, "loss": 0.0234, "step": 52340 }, { "epoch": 23.428059968673082, "grad_norm": 0.06264961510896683, "learning_rate": 4.378944388119311e-06, "loss": 0.0241, "step": 52350 }, { "epoch": 23.432535242783622, "grad_norm": 0.07906997203826904, "learning_rate": 4.367673203442241e-06, "loss": 0.0258, "step": 52360 }, { "epoch": 23.43701051689416, "grad_norm": 0.08315768837928772, "learning_rate": 4.356415880661197e-06, "loss": 0.0266, "step": 52370 }, { "epoch": 23.4414857910047, "grad_norm": 0.09067816287279129, "learning_rate": 4.3451724231958644e-06, "loss": 0.0245, "step": 52380 }, { "epoch": 23.445961065115238, "grad_norm": 0.06255968660116196, "learning_rate": 4.333942834461702e-06, "loss": 0.0192, "step": 52390 }, { "epoch": 23.450436339225778, "grad_norm": 0.06484623998403549, "learning_rate": 4.322727117869951e-06, "loss": 0.0228, "step": 52400 }, { "epoch": 23.454911613336318, "grad_norm": 0.08035222440958023, "learning_rate": 4.311525276827682e-06, "loss": 0.0215, "step": 52410 }, { "epoch": 23.459386887446858, "grad_norm": 0.07450442016124725, "learning_rate": 4.300337314737685e-06, "loss": 0.0235, "step": 52420 }, { "epoch": 23.463862161557394, "grad_norm": 0.08308423310518265, "learning_rate": 4.289163234998589e-06, "loss": 0.025, "step": 52430 }, { "epoch": 23.468337435667934, "grad_norm": 0.06247877702116966, "learning_rate": 4.27800304100478e-06, "loss": 0.0255, "step": 52440 }, { "epoch": 23.472812709778474, "grad_norm": 0.0723431408405304, "learning_rate": 4.266856736146424e-06, "loss": 0.0211, "step": 52450 }, { "epoch": 23.477287983889013, "grad_norm": 0.08544034510850906, "learning_rate": 4.255724323809479e-06, "loss": 0.0238, "step": 52460 }, { "epoch": 23.481763257999553, "grad_norm": 0.07400292903184891, "learning_rate": 4.244605807375679e-06, "loss": 0.0231, "step": 52470 }, { "epoch": 23.486238532110093, "grad_norm": 0.09516981989145279, "learning_rate": 4.233501190222533e-06, "loss": 0.0232, "step": 52480 }, { "epoch": 23.49071380622063, "grad_norm": 0.0829157680273056, "learning_rate": 4.222410475723326e-06, "loss": 0.0234, "step": 52490 }, { "epoch": 23.49518908033117, "grad_norm": 0.08240066468715668, "learning_rate": 4.2113336672471245e-06, "loss": 0.0267, "step": 52500 }, { "epoch": 23.49966435444171, "grad_norm": 0.06642580777406693, "learning_rate": 4.2002707681587874e-06, "loss": 0.0221, "step": 52510 }, { "epoch": 23.50413962855225, "grad_norm": 0.0620366670191288, "learning_rate": 4.189221781818914e-06, "loss": 0.0208, "step": 52520 }, { "epoch": 23.50861490266279, "grad_norm": 0.06853140145540237, "learning_rate": 4.178186711583904e-06, "loss": 0.0208, "step": 52530 }, { "epoch": 23.51309017677333, "grad_norm": 0.09080667048692703, "learning_rate": 4.167165560805914e-06, "loss": 0.0219, "step": 52540 }, { "epoch": 23.51756545088387, "grad_norm": 0.06423128396272659, "learning_rate": 4.1561583328328716e-06, "loss": 0.0231, "step": 52550 }, { "epoch": 23.522040724994405, "grad_norm": 0.06752847880125046, "learning_rate": 4.145165031008508e-06, "loss": 0.0249, "step": 52560 }, { "epoch": 23.526515999104944, "grad_norm": 0.09116256237030029, "learning_rate": 4.1341856586722625e-06, "loss": 0.0225, "step": 52570 }, { "epoch": 23.530991273215484, "grad_norm": 0.06540344655513763, "learning_rate": 4.123220219159418e-06, "loss": 0.0209, "step": 52580 }, { "epoch": 23.535466547326024, "grad_norm": 0.0947365090250969, "learning_rate": 4.112268715800943e-06, "loss": 0.0258, "step": 52590 }, { "epoch": 23.539941821436564, "grad_norm": 0.0778503566980362, "learning_rate": 4.101331151923649e-06, "loss": 0.0269, "step": 52600 }, { "epoch": 23.544417095547104, "grad_norm": 0.09024624526500702, "learning_rate": 4.090407530850065e-06, "loss": 0.0237, "step": 52610 }, { "epoch": 23.54889236965764, "grad_norm": 0.07321001589298248, "learning_rate": 4.079497855898501e-06, "loss": 0.0233, "step": 52620 }, { "epoch": 23.55336764376818, "grad_norm": 0.0788361132144928, "learning_rate": 4.068602130383031e-06, "loss": 0.0216, "step": 52630 }, { "epoch": 23.55784291787872, "grad_norm": 0.0768953189253807, "learning_rate": 4.057720357613482e-06, "loss": 0.0243, "step": 52640 }, { "epoch": 23.56231819198926, "grad_norm": 0.08764490485191345, "learning_rate": 4.046852540895446e-06, "loss": 0.0204, "step": 52650 }, { "epoch": 23.5667934660998, "grad_norm": 0.06134377419948578, "learning_rate": 4.03599868353029e-06, "loss": 0.0241, "step": 52660 }, { "epoch": 23.57126874021034, "grad_norm": 0.06246524676680565, "learning_rate": 4.025158788815131e-06, "loss": 0.0235, "step": 52670 }, { "epoch": 23.575744014320875, "grad_norm": 0.08147556334733963, "learning_rate": 4.01433286004283e-06, "loss": 0.0236, "step": 52680 }, { "epoch": 23.580219288431415, "grad_norm": 0.07084918022155762, "learning_rate": 4.003520900502028e-06, "loss": 0.02, "step": 52690 }, { "epoch": 23.584694562541955, "grad_norm": 0.0896819457411766, "learning_rate": 3.9927229134771035e-06, "loss": 0.0273, "step": 52700 }, { "epoch": 23.589169836652495, "grad_norm": 0.07086925208568573, "learning_rate": 3.981938902248222e-06, "loss": 0.0248, "step": 52710 }, { "epoch": 23.593645110763035, "grad_norm": 0.07034340500831604, "learning_rate": 3.971168870091247e-06, "loss": 0.0241, "step": 52720 }, { "epoch": 23.598120384873575, "grad_norm": 0.06036965548992157, "learning_rate": 3.960412820277865e-06, "loss": 0.0219, "step": 52730 }, { "epoch": 23.602595658984114, "grad_norm": 0.08050708472728729, "learning_rate": 3.949670756075447e-06, "loss": 0.0204, "step": 52740 }, { "epoch": 23.60707093309465, "grad_norm": 0.054876185953617096, "learning_rate": 3.9389426807471766e-06, "loss": 0.0248, "step": 52750 }, { "epoch": 23.61154620720519, "grad_norm": 0.08320722728967667, "learning_rate": 3.928228597551947e-06, "loss": 0.0251, "step": 52760 }, { "epoch": 23.61602148131573, "grad_norm": 0.09047843515872955, "learning_rate": 3.917528509744412e-06, "loss": 0.0237, "step": 52770 }, { "epoch": 23.62049675542627, "grad_norm": 0.06490593403577805, "learning_rate": 3.90684242057498e-06, "loss": 0.0199, "step": 52780 }, { "epoch": 23.62497202953681, "grad_norm": 0.05966830626130104, "learning_rate": 3.8961703332898e-06, "loss": 0.0232, "step": 52790 }, { "epoch": 23.62944730364735, "grad_norm": 0.06197010725736618, "learning_rate": 3.885512251130763e-06, "loss": 0.024, "step": 52800 }, { "epoch": 23.633922577757886, "grad_norm": 0.08344081789255142, "learning_rate": 3.8748681773355335e-06, "loss": 0.023, "step": 52810 }, { "epoch": 23.638397851868426, "grad_norm": 0.07255501300096512, "learning_rate": 3.864238115137481e-06, "loss": 0.0212, "step": 52820 }, { "epoch": 23.642873125978966, "grad_norm": 0.0789855495095253, "learning_rate": 3.8536220677657495e-06, "loss": 0.0227, "step": 52830 }, { "epoch": 23.647348400089506, "grad_norm": 0.0818982645869255, "learning_rate": 3.843020038445211e-06, "loss": 0.0233, "step": 52840 }, { "epoch": 23.651823674200045, "grad_norm": 0.09021732211112976, "learning_rate": 3.832432030396471e-06, "loss": 0.0268, "step": 52850 }, { "epoch": 23.656298948310585, "grad_norm": 0.08103825151920319, "learning_rate": 3.821858046835913e-06, "loss": 0.0198, "step": 52860 }, { "epoch": 23.660774222421125, "grad_norm": 0.06625567376613617, "learning_rate": 3.8112980909756014e-06, "loss": 0.0238, "step": 52870 }, { "epoch": 23.66524949653166, "grad_norm": 0.07608969509601593, "learning_rate": 3.8007521660234023e-06, "loss": 0.0219, "step": 52880 }, { "epoch": 23.6697247706422, "grad_norm": 0.08815110474824905, "learning_rate": 3.790220275182854e-06, "loss": 0.0231, "step": 52890 }, { "epoch": 23.67420004475274, "grad_norm": 0.07977914065122604, "learning_rate": 3.7797024216533138e-06, "loss": 0.0214, "step": 52900 }, { "epoch": 23.67867531886328, "grad_norm": 0.07330012321472168, "learning_rate": 3.769198608629776e-06, "loss": 0.0221, "step": 52910 }, { "epoch": 23.68315059297382, "grad_norm": 0.06778915971517563, "learning_rate": 3.75870883930306e-06, "loss": 0.0215, "step": 52920 }, { "epoch": 23.68762586708436, "grad_norm": 0.06948694586753845, "learning_rate": 3.7482331168596675e-06, "loss": 0.024, "step": 52930 }, { "epoch": 23.692101141194897, "grad_norm": 0.10543704777956009, "learning_rate": 3.7377714444818468e-06, "loss": 0.0239, "step": 52940 }, { "epoch": 23.696576415305437, "grad_norm": 0.0662974864244461, "learning_rate": 3.7273238253475785e-06, "loss": 0.0235, "step": 52950 }, { "epoch": 23.701051689415976, "grad_norm": 0.07462503015995026, "learning_rate": 3.7168902626305622e-06, "loss": 0.0228, "step": 52960 }, { "epoch": 23.705526963526516, "grad_norm": 0.08010649681091309, "learning_rate": 3.7064707595002635e-06, "loss": 0.0211, "step": 52970 }, { "epoch": 23.710002237637056, "grad_norm": 0.06453663110733032, "learning_rate": 3.696065319121833e-06, "loss": 0.022, "step": 52980 }, { "epoch": 23.714477511747596, "grad_norm": 0.10316333919763565, "learning_rate": 3.685673944656176e-06, "loss": 0.0227, "step": 52990 }, { "epoch": 23.718952785858132, "grad_norm": 0.07141587138175964, "learning_rate": 3.675296639259912e-06, "loss": 0.0214, "step": 53000 }, { "epoch": 23.723428059968672, "grad_norm": 0.09997265040874481, "learning_rate": 3.664933406085402e-06, "loss": 0.0277, "step": 53010 }, { "epoch": 23.727903334079212, "grad_norm": 0.07096680253744125, "learning_rate": 3.654584248280707e-06, "loss": 0.0236, "step": 53020 }, { "epoch": 23.73237860818975, "grad_norm": 0.0798213928937912, "learning_rate": 3.64424916898965e-06, "loss": 0.0218, "step": 53030 }, { "epoch": 23.73685388230029, "grad_norm": 0.09661639481782913, "learning_rate": 3.6339281713517303e-06, "loss": 0.0253, "step": 53040 }, { "epoch": 23.74132915641083, "grad_norm": 0.05425408482551575, "learning_rate": 3.6236212585022187e-06, "loss": 0.0269, "step": 53050 }, { "epoch": 23.74580443052137, "grad_norm": 0.07936439663171768, "learning_rate": 3.6133284335720605e-06, "loss": 0.0231, "step": 53060 }, { "epoch": 23.750279704631907, "grad_norm": 0.09110789000988007, "learning_rate": 3.60304969968796e-06, "loss": 0.0228, "step": 53070 }, { "epoch": 23.754754978742447, "grad_norm": 0.07704000174999237, "learning_rate": 3.59278505997232e-06, "loss": 0.023, "step": 53080 }, { "epoch": 23.759230252852987, "grad_norm": 0.08012982457876205, "learning_rate": 3.582534517543268e-06, "loss": 0.0239, "step": 53090 }, { "epoch": 23.763705526963527, "grad_norm": 0.08873404562473297, "learning_rate": 3.5722980755146517e-06, "loss": 0.022, "step": 53100 }, { "epoch": 23.768180801074067, "grad_norm": 0.06876230984926224, "learning_rate": 3.5620757369960123e-06, "loss": 0.0229, "step": 53110 }, { "epoch": 23.772656075184607, "grad_norm": 0.08080927282571793, "learning_rate": 3.5518675050926544e-06, "loss": 0.0203, "step": 53120 }, { "epoch": 23.777131349295143, "grad_norm": 0.09455467015504837, "learning_rate": 3.541673382905558e-06, "loss": 0.0207, "step": 53130 }, { "epoch": 23.781606623405683, "grad_norm": 0.04879579320549965, "learning_rate": 3.531493373531419e-06, "loss": 0.0198, "step": 53140 }, { "epoch": 23.786081897516222, "grad_norm": 0.08498048782348633, "learning_rate": 3.5213274800626692e-06, "loss": 0.0241, "step": 53150 }, { "epoch": 23.790557171626762, "grad_norm": 0.07402295619249344, "learning_rate": 3.511175705587433e-06, "loss": 0.0216, "step": 53160 }, { "epoch": 23.795032445737302, "grad_norm": 0.06166928634047508, "learning_rate": 3.5010380531895426e-06, "loss": 0.0232, "step": 53170 }, { "epoch": 23.799507719847842, "grad_norm": 0.08594506978988647, "learning_rate": 3.4909145259485744e-06, "loss": 0.0189, "step": 53180 }, { "epoch": 23.80398299395838, "grad_norm": 0.0734560638666153, "learning_rate": 3.4808051269397512e-06, "loss": 0.0261, "step": 53190 }, { "epoch": 23.808458268068918, "grad_norm": 0.08464151620864868, "learning_rate": 3.470709859234084e-06, "loss": 0.0189, "step": 53200 }, { "epoch": 23.812933542179458, "grad_norm": 0.0758490040898323, "learning_rate": 3.460628725898207e-06, "loss": 0.0215, "step": 53210 }, { "epoch": 23.817408816289998, "grad_norm": 0.06583757698535919, "learning_rate": 3.4505617299945336e-06, "loss": 0.0245, "step": 53220 }, { "epoch": 23.821884090400538, "grad_norm": 0.058528922498226166, "learning_rate": 3.440508874581139e-06, "loss": 0.0223, "step": 53230 }, { "epoch": 23.826359364511077, "grad_norm": 0.08182013034820557, "learning_rate": 3.430470162711813e-06, "loss": 0.0248, "step": 53240 }, { "epoch": 23.830834638621617, "grad_norm": 0.08812297880649567, "learning_rate": 3.420445597436056e-06, "loss": 0.0236, "step": 53250 }, { "epoch": 23.835309912732153, "grad_norm": 0.07709847390651703, "learning_rate": 3.41043518179906e-06, "loss": 0.0265, "step": 53260 }, { "epoch": 23.839785186842693, "grad_norm": 0.08325950801372528, "learning_rate": 3.4004389188417305e-06, "loss": 0.0256, "step": 53270 }, { "epoch": 23.844260460953233, "grad_norm": 0.0747787281870842, "learning_rate": 3.390456811600673e-06, "loss": 0.023, "step": 53280 }, { "epoch": 23.848735735063773, "grad_norm": 0.07154136151075363, "learning_rate": 3.380488863108183e-06, "loss": 0.0204, "step": 53290 }, { "epoch": 23.853211009174313, "grad_norm": 0.07366862893104553, "learning_rate": 3.3705350763922562e-06, "loss": 0.0204, "step": 53300 }, { "epoch": 23.857686283284853, "grad_norm": 0.07470109313726425, "learning_rate": 3.360595454476595e-06, "loss": 0.023, "step": 53310 }, { "epoch": 23.86216155739539, "grad_norm": 0.04958013817667961, "learning_rate": 3.35067000038059e-06, "loss": 0.0227, "step": 53320 }, { "epoch": 23.86663683150593, "grad_norm": 0.07001104205846786, "learning_rate": 3.3407587171193354e-06, "loss": 0.0228, "step": 53330 }, { "epoch": 23.87111210561647, "grad_norm": 0.0733620822429657, "learning_rate": 3.3308616077036115e-06, "loss": 0.0247, "step": 53340 }, { "epoch": 23.87558737972701, "grad_norm": 0.08004922419786453, "learning_rate": 3.3209786751399187e-06, "loss": 0.0261, "step": 53350 }, { "epoch": 23.88006265383755, "grad_norm": 0.06528070569038391, "learning_rate": 3.3111099224304e-06, "loss": 0.0247, "step": 53360 }, { "epoch": 23.884537927948088, "grad_norm": 0.06327774375677109, "learning_rate": 3.301255352572946e-06, "loss": 0.0236, "step": 53370 }, { "epoch": 23.889013202058628, "grad_norm": 0.07230264693498611, "learning_rate": 3.2914149685611073e-06, "loss": 0.0239, "step": 53380 }, { "epoch": 23.893488476169164, "grad_norm": 0.07136935740709305, "learning_rate": 3.2815887733841365e-06, "loss": 0.0256, "step": 53390 }, { "epoch": 23.897963750279704, "grad_norm": 0.07671236991882324, "learning_rate": 3.271776770026963e-06, "loss": 0.021, "step": 53400 }, { "epoch": 23.902439024390244, "grad_norm": 0.0637376606464386, "learning_rate": 3.2619789614702135e-06, "loss": 0.0198, "step": 53410 }, { "epoch": 23.906914298500784, "grad_norm": 0.06495695561170578, "learning_rate": 3.2521953506902237e-06, "loss": 0.0216, "step": 53420 }, { "epoch": 23.911389572611323, "grad_norm": 0.05875464528799057, "learning_rate": 3.2424259406589664e-06, "loss": 0.0237, "step": 53430 }, { "epoch": 23.915864846721863, "grad_norm": 0.07758409529924393, "learning_rate": 3.2326707343441566e-06, "loss": 0.0235, "step": 53440 }, { "epoch": 23.9203401208324, "grad_norm": 0.0808217003941536, "learning_rate": 3.2229297347091514e-06, "loss": 0.0268, "step": 53450 }, { "epoch": 23.92481539494294, "grad_norm": 0.07034505158662796, "learning_rate": 3.213202944713023e-06, "loss": 0.0262, "step": 53460 }, { "epoch": 23.92929066905348, "grad_norm": 0.058344241231679916, "learning_rate": 3.203490367310502e-06, "loss": 0.0248, "step": 53470 }, { "epoch": 23.93376594316402, "grad_norm": 0.06909004598855972, "learning_rate": 3.193792005452018e-06, "loss": 0.023, "step": 53480 }, { "epoch": 23.93824121727456, "grad_norm": 0.0858546793460846, "learning_rate": 3.1841078620836683e-06, "loss": 0.0243, "step": 53490 }, { "epoch": 23.9427164913851, "grad_norm": 0.07899477332830429, "learning_rate": 3.1744379401472677e-06, "loss": 0.0205, "step": 53500 }, { "epoch": 23.947191765495635, "grad_norm": 0.07068207859992981, "learning_rate": 3.164782242580244e-06, "loss": 0.0255, "step": 53510 }, { "epoch": 23.951667039606175, "grad_norm": 0.055991530418395996, "learning_rate": 3.155140772315773e-06, "loss": 0.0225, "step": 53520 }, { "epoch": 23.956142313716715, "grad_norm": 0.054140474647283554, "learning_rate": 3.1455135322826678e-06, "loss": 0.0215, "step": 53530 }, { "epoch": 23.960617587827254, "grad_norm": 0.0900380089879036, "learning_rate": 3.1359005254054273e-06, "loss": 0.0216, "step": 53540 }, { "epoch": 23.965092861937794, "grad_norm": 0.06451699137687683, "learning_rate": 3.126301754604233e-06, "loss": 0.0242, "step": 53550 }, { "epoch": 23.969568136048334, "grad_norm": 0.06250254064798355, "learning_rate": 3.1167172227949347e-06, "loss": 0.0198, "step": 53560 }, { "epoch": 23.97404341015887, "grad_norm": 0.07814963161945343, "learning_rate": 3.107146932889071e-06, "loss": 0.022, "step": 53570 }, { "epoch": 23.97851868426941, "grad_norm": 0.07888756692409515, "learning_rate": 3.0975908877938277e-06, "loss": 0.023, "step": 53580 }, { "epoch": 23.98299395837995, "grad_norm": 0.06567969918251038, "learning_rate": 3.0880490904120874e-06, "loss": 0.0214, "step": 53590 }, { "epoch": 23.98746923249049, "grad_norm": 0.09527292847633362, "learning_rate": 3.078521543642399e-06, "loss": 0.0214, "step": 53600 }, { "epoch": 23.99194450660103, "grad_norm": 0.06205030530691147, "learning_rate": 3.0690082503789742e-06, "loss": 0.0239, "step": 53610 }, { "epoch": 23.99641978071157, "grad_norm": 0.07862520217895508, "learning_rate": 3.059509213511702e-06, "loss": 0.0209, "step": 53620 }, { "epoch": 24.00089505482211, "grad_norm": 0.08200357854366302, "learning_rate": 3.0500244359261355e-06, "loss": 0.0221, "step": 53630 }, { "epoch": 24.005370328932646, "grad_norm": 0.07215242087841034, "learning_rate": 3.040553920503503e-06, "loss": 0.0224, "step": 53640 }, { "epoch": 24.009845603043185, "grad_norm": 0.06514335423707962, "learning_rate": 3.031097670120703e-06, "loss": 0.0215, "step": 53650 }, { "epoch": 24.014320877153725, "grad_norm": 0.06802577525377274, "learning_rate": 3.021655687650282e-06, "loss": 0.0259, "step": 53660 }, { "epoch": 24.018796151264265, "grad_norm": 0.08363520354032516, "learning_rate": 3.0122279759604745e-06, "loss": 0.0222, "step": 53670 }, { "epoch": 24.023271425374805, "grad_norm": 0.07811006903648376, "learning_rate": 3.0028145379151716e-06, "loss": 0.0244, "step": 53680 }, { "epoch": 24.027746699485345, "grad_norm": 0.08562299609184265, "learning_rate": 2.9934153763739205e-06, "loss": 0.0231, "step": 53690 }, { "epoch": 24.03222197359588, "grad_norm": 0.0688583180308342, "learning_rate": 2.9840304941919415e-06, "loss": 0.0212, "step": 53700 }, { "epoch": 24.03669724770642, "grad_norm": 0.09055883437395096, "learning_rate": 2.9746598942201054e-06, "loss": 0.0254, "step": 53710 }, { "epoch": 24.04117252181696, "grad_norm": 0.0751492902636528, "learning_rate": 2.965303579304973e-06, "loss": 0.0212, "step": 53720 }, { "epoch": 24.0456477959275, "grad_norm": 0.07248644530773163, "learning_rate": 2.9559615522887273e-06, "loss": 0.0211, "step": 53730 }, { "epoch": 24.05012307003804, "grad_norm": 0.08126058429479599, "learning_rate": 2.946633816009242e-06, "loss": 0.024, "step": 53740 }, { "epoch": 24.05459834414858, "grad_norm": 0.06903546303510666, "learning_rate": 2.9373203733000232e-06, "loss": 0.0222, "step": 53750 }, { "epoch": 24.05907361825912, "grad_norm": 0.06579052656888962, "learning_rate": 2.928021226990263e-06, "loss": 0.0213, "step": 53760 }, { "epoch": 24.063548892369656, "grad_norm": 0.04762301221489906, "learning_rate": 2.9187363799047905e-06, "loss": 0.0203, "step": 53770 }, { "epoch": 24.068024166480196, "grad_norm": 0.08766445517539978, "learning_rate": 2.9094658348640945e-06, "loss": 0.0225, "step": 53780 }, { "epoch": 24.072499440590736, "grad_norm": 0.09292762726545334, "learning_rate": 2.9002095946843277e-06, "loss": 0.0232, "step": 53790 }, { "epoch": 24.076974714701276, "grad_norm": 0.0651782974600792, "learning_rate": 2.890967662177285e-06, "loss": 0.0219, "step": 53800 }, { "epoch": 24.081449988811816, "grad_norm": 0.061731427907943726, "learning_rate": 2.881740040150416e-06, "loss": 0.0259, "step": 53810 }, { "epoch": 24.085925262922355, "grad_norm": 0.07401297241449356, "learning_rate": 2.8725267314068495e-06, "loss": 0.0226, "step": 53820 }, { "epoch": 24.09040053703289, "grad_norm": 0.06570841372013092, "learning_rate": 2.8633277387453308e-06, "loss": 0.0222, "step": 53830 }, { "epoch": 24.09487581114343, "grad_norm": 0.06562380492687225, "learning_rate": 2.854143064960274e-06, "loss": 0.0206, "step": 53840 }, { "epoch": 24.09935108525397, "grad_norm": 0.08514630049467087, "learning_rate": 2.844972712841737e-06, "loss": 0.0198, "step": 53850 }, { "epoch": 24.10382635936451, "grad_norm": 0.09477756172418594, "learning_rate": 2.8358166851754297e-06, "loss": 0.0273, "step": 53860 }, { "epoch": 24.10830163347505, "grad_norm": 0.07060982286930084, "learning_rate": 2.8266749847427275e-06, "loss": 0.0227, "step": 53870 }, { "epoch": 24.11277690758559, "grad_norm": 0.06823202967643738, "learning_rate": 2.817547614320615e-06, "loss": 0.0255, "step": 53880 }, { "epoch": 24.117252181696127, "grad_norm": 0.09395600110292435, "learning_rate": 2.8084345766817676e-06, "loss": 0.022, "step": 53890 }, { "epoch": 24.121727455806667, "grad_norm": 0.07605749368667603, "learning_rate": 2.7993358745944608e-06, "loss": 0.0232, "step": 53900 }, { "epoch": 24.126202729917207, "grad_norm": 0.06726285070180893, "learning_rate": 2.790251510822661e-06, "loss": 0.0204, "step": 53910 }, { "epoch": 24.130678004027747, "grad_norm": 0.08940169960260391, "learning_rate": 2.7811814881259503e-06, "loss": 0.023, "step": 53920 }, { "epoch": 24.135153278138286, "grad_norm": 0.06747826933860779, "learning_rate": 2.7721258092595627e-06, "loss": 0.0233, "step": 53930 }, { "epoch": 24.139628552248826, "grad_norm": 0.07480014860630035, "learning_rate": 2.7630844769743757e-06, "loss": 0.025, "step": 53940 }, { "epoch": 24.144103826359366, "grad_norm": 0.07241522520780563, "learning_rate": 2.754057494016904e-06, "loss": 0.0231, "step": 53950 }, { "epoch": 24.148579100469902, "grad_norm": 0.08207716792821884, "learning_rate": 2.7450448631293036e-06, "loss": 0.0239, "step": 53960 }, { "epoch": 24.153054374580442, "grad_norm": 0.0708828791975975, "learning_rate": 2.73604658704939e-06, "loss": 0.0209, "step": 53970 }, { "epoch": 24.157529648690982, "grad_norm": 0.06481197476387024, "learning_rate": 2.7270626685105828e-06, "loss": 0.0203, "step": 53980 }, { "epoch": 24.162004922801522, "grad_norm": 0.08898180723190308, "learning_rate": 2.718093110241976e-06, "loss": 0.0243, "step": 53990 }, { "epoch": 24.16648019691206, "grad_norm": 0.06895019859075546, "learning_rate": 2.7091379149682685e-06, "loss": 0.0221, "step": 54000 }, { "epoch": 24.1709554710226, "grad_norm": 0.06938222795724869, "learning_rate": 2.700197085409817e-06, "loss": 0.023, "step": 54010 }, { "epoch": 24.175430745133138, "grad_norm": 0.08013592660427094, "learning_rate": 2.691270624282621e-06, "loss": 0.022, "step": 54020 }, { "epoch": 24.179906019243678, "grad_norm": 0.08827582746744156, "learning_rate": 2.682358534298285e-06, "loss": 0.0242, "step": 54030 }, { "epoch": 24.184381293354217, "grad_norm": 0.07764533162117004, "learning_rate": 2.6734608181640917e-06, "loss": 0.024, "step": 54040 }, { "epoch": 24.188856567464757, "grad_norm": 0.0778035894036293, "learning_rate": 2.664577478582897e-06, "loss": 0.0219, "step": 54050 }, { "epoch": 24.193331841575297, "grad_norm": 0.0779801607131958, "learning_rate": 2.6557085182532582e-06, "loss": 0.0197, "step": 54060 }, { "epoch": 24.197807115685837, "grad_norm": 0.08702041953802109, "learning_rate": 2.646853939869315e-06, "loss": 0.0211, "step": 54070 }, { "epoch": 24.202282389796373, "grad_norm": 0.09265492111444473, "learning_rate": 2.63801374612086e-06, "loss": 0.023, "step": 54080 }, { "epoch": 24.206757663906913, "grad_norm": 0.08463715016841888, "learning_rate": 2.6291879396933004e-06, "loss": 0.0231, "step": 54090 }, { "epoch": 24.211232938017453, "grad_norm": 0.09311825782060623, "learning_rate": 2.620376523267698e-06, "loss": 0.0235, "step": 54100 }, { "epoch": 24.215708212127993, "grad_norm": 0.07246855646371841, "learning_rate": 2.611579499520722e-06, "loss": 0.0238, "step": 54110 }, { "epoch": 24.220183486238533, "grad_norm": 0.06132378429174423, "learning_rate": 2.602796871124663e-06, "loss": 0.022, "step": 54120 }, { "epoch": 24.224658760349072, "grad_norm": 0.08110727369785309, "learning_rate": 2.594028640747476e-06, "loss": 0.0249, "step": 54130 }, { "epoch": 24.229134034459612, "grad_norm": 0.08023639768362045, "learning_rate": 2.5852748110527014e-06, "loss": 0.0203, "step": 54140 }, { "epoch": 24.23360930857015, "grad_norm": 0.07442878186702728, "learning_rate": 2.57653538469953e-06, "loss": 0.0233, "step": 54150 }, { "epoch": 24.23808458268069, "grad_norm": 0.06439995020627975, "learning_rate": 2.567810364342765e-06, "loss": 0.0242, "step": 54160 }, { "epoch": 24.242559856791228, "grad_norm": 0.05214157700538635, "learning_rate": 2.559099752632832e-06, "loss": 0.0227, "step": 54170 }, { "epoch": 24.247035130901768, "grad_norm": 0.0686146691441536, "learning_rate": 2.5504035522157854e-06, "loss": 0.0201, "step": 54180 }, { "epoch": 24.251510405012308, "grad_norm": 0.06670372933149338, "learning_rate": 2.541721765733318e-06, "loss": 0.0239, "step": 54190 }, { "epoch": 24.255985679122848, "grad_norm": 0.0859556719660759, "learning_rate": 2.533054395822704e-06, "loss": 0.0215, "step": 54200 }, { "epoch": 24.260460953233384, "grad_norm": 0.06678090244531631, "learning_rate": 2.5244014451168863e-06, "loss": 0.0231, "step": 54210 }, { "epoch": 24.264936227343924, "grad_norm": 0.067989282310009, "learning_rate": 2.515762916244374e-06, "loss": 0.0211, "step": 54220 }, { "epoch": 24.269411501454464, "grad_norm": 0.10129965841770172, "learning_rate": 2.507138811829346e-06, "loss": 0.0244, "step": 54230 }, { "epoch": 24.273886775565003, "grad_norm": 0.052622850984334946, "learning_rate": 2.4985291344915674e-06, "loss": 0.0222, "step": 54240 }, { "epoch": 24.278362049675543, "grad_norm": 0.06744063645601273, "learning_rate": 2.4899338868464404e-06, "loss": 0.0257, "step": 54250 }, { "epoch": 24.282837323786083, "grad_norm": 0.11061716824769974, "learning_rate": 2.481353071504966e-06, "loss": 0.0237, "step": 54260 }, { "epoch": 24.287312597896623, "grad_norm": 0.06950024515390396, "learning_rate": 2.4727866910737583e-06, "loss": 0.0223, "step": 54270 }, { "epoch": 24.29178787200716, "grad_norm": 0.08634299784898758, "learning_rate": 2.4642347481550865e-06, "loss": 0.0249, "step": 54280 }, { "epoch": 24.2962631461177, "grad_norm": 0.0600302517414093, "learning_rate": 2.455697245346783e-06, "loss": 0.0242, "step": 54290 }, { "epoch": 24.30073842022824, "grad_norm": 0.06893378496170044, "learning_rate": 2.4471741852423237e-06, "loss": 0.0248, "step": 54300 }, { "epoch": 24.30521369433878, "grad_norm": 0.08881773054599762, "learning_rate": 2.438665570430787e-06, "loss": 0.021, "step": 54310 }, { "epoch": 24.30968896844932, "grad_norm": 0.05471443384885788, "learning_rate": 2.430171403496867e-06, "loss": 0.0212, "step": 54320 }, { "epoch": 24.31416424255986, "grad_norm": 0.059036023914813995, "learning_rate": 2.421691687020855e-06, "loss": 0.0238, "step": 54330 }, { "epoch": 24.318639516670395, "grad_norm": 0.05757422745227814, "learning_rate": 2.413226423578696e-06, "loss": 0.0215, "step": 54340 }, { "epoch": 24.323114790780934, "grad_norm": 0.08602998405694962, "learning_rate": 2.404775615741872e-06, "loss": 0.0228, "step": 54350 }, { "epoch": 24.327590064891474, "grad_norm": 0.07187015563249588, "learning_rate": 2.3963392660775575e-06, "loss": 0.0193, "step": 54360 }, { "epoch": 24.332065339002014, "grad_norm": 0.0724884569644928, "learning_rate": 2.3879173771484576e-06, "loss": 0.0248, "step": 54370 }, { "epoch": 24.336540613112554, "grad_norm": 0.07563534379005432, "learning_rate": 2.379509951512937e-06, "loss": 0.022, "step": 54380 }, { "epoch": 24.341015887223094, "grad_norm": 0.09506145864725113, "learning_rate": 2.371116991724953e-06, "loss": 0.024, "step": 54390 }, { "epoch": 24.34549116133363, "grad_norm": 0.056106630712747574, "learning_rate": 2.362738500334055e-06, "loss": 0.0216, "step": 54400 }, { "epoch": 24.34996643544417, "grad_norm": 0.06572757661342621, "learning_rate": 2.354374479885413e-06, "loss": 0.0232, "step": 54410 }, { "epoch": 24.35444170955471, "grad_norm": 0.06769069284200668, "learning_rate": 2.3460249329197824e-06, "loss": 0.0227, "step": 54420 }, { "epoch": 24.35891698366525, "grad_norm": 0.08487554639577866, "learning_rate": 2.3376898619735577e-06, "loss": 0.0227, "step": 54430 }, { "epoch": 24.36339225777579, "grad_norm": 0.0768161192536354, "learning_rate": 2.3293692695787017e-06, "loss": 0.0223, "step": 54440 }, { "epoch": 24.36786753188633, "grad_norm": 0.07444394379854202, "learning_rate": 2.321063158262793e-06, "loss": 0.0244, "step": 54450 }, { "epoch": 24.37234280599687, "grad_norm": 0.051563557237386703, "learning_rate": 2.3127715305490073e-06, "loss": 0.0221, "step": 54460 }, { "epoch": 24.376818080107405, "grad_norm": 0.07184053957462311, "learning_rate": 2.3044943889561245e-06, "loss": 0.0238, "step": 54470 }, { "epoch": 24.381293354217945, "grad_norm": 0.06035039201378822, "learning_rate": 2.296231735998511e-06, "loss": 0.0205, "step": 54480 }, { "epoch": 24.385768628328485, "grad_norm": 0.05006234347820282, "learning_rate": 2.2879835741861586e-06, "loss": 0.0204, "step": 54490 }, { "epoch": 24.390243902439025, "grad_norm": 0.07459007948637009, "learning_rate": 2.2797499060246253e-06, "loss": 0.0251, "step": 54500 }, { "epoch": 24.394719176549565, "grad_norm": 0.06826278567314148, "learning_rate": 2.271530734015104e-06, "loss": 0.0201, "step": 54510 }, { "epoch": 24.399194450660104, "grad_norm": 0.0586746446788311, "learning_rate": 2.263326060654336e-06, "loss": 0.0248, "step": 54520 }, { "epoch": 24.40366972477064, "grad_norm": 0.07128124684095383, "learning_rate": 2.2551358884347007e-06, "loss": 0.0252, "step": 54530 }, { "epoch": 24.40814499888118, "grad_norm": 0.06602166593074799, "learning_rate": 2.2469602198441573e-06, "loss": 0.0227, "step": 54540 }, { "epoch": 24.41262027299172, "grad_norm": 0.08904732018709183, "learning_rate": 2.2387990573662477e-06, "loss": 0.0238, "step": 54550 }, { "epoch": 24.41709554710226, "grad_norm": 0.05620776116847992, "learning_rate": 2.230652403480127e-06, "loss": 0.0212, "step": 54560 }, { "epoch": 24.4215708212128, "grad_norm": 0.05790606886148453, "learning_rate": 2.222520260660521e-06, "loss": 0.0242, "step": 54570 }, { "epoch": 24.42604609532334, "grad_norm": 0.08354925364255905, "learning_rate": 2.214402631377782e-06, "loss": 0.0224, "step": 54580 }, { "epoch": 24.430521369433876, "grad_norm": 0.07554090768098831, "learning_rate": 2.206299518097804e-06, "loss": 0.0242, "step": 54590 }, { "epoch": 24.434996643544416, "grad_norm": 0.08461830765008926, "learning_rate": 2.1982109232821178e-06, "loss": 0.0231, "step": 54600 }, { "epoch": 24.439471917654956, "grad_norm": 0.07708386331796646, "learning_rate": 2.1901368493878248e-06, "loss": 0.0239, "step": 54610 }, { "epoch": 24.443947191765496, "grad_norm": 0.08364702761173248, "learning_rate": 2.1820772988676076e-06, "loss": 0.0227, "step": 54620 }, { "epoch": 24.448422465876035, "grad_norm": 0.07337280362844467, "learning_rate": 2.174032274169746e-06, "loss": 0.0214, "step": 54630 }, { "epoch": 24.452897739986575, "grad_norm": 0.07168141007423401, "learning_rate": 2.1660017777381135e-06, "loss": 0.0195, "step": 54640 }, { "epoch": 24.457373014097115, "grad_norm": 0.06169183924794197, "learning_rate": 2.157985812012148e-06, "loss": 0.0184, "step": 54650 }, { "epoch": 24.46184828820765, "grad_norm": 0.06445149332284927, "learning_rate": 2.149984379426906e-06, "loss": 0.0229, "step": 54660 }, { "epoch": 24.46632356231819, "grad_norm": 0.05976565554738045, "learning_rate": 2.141997482412994e-06, "loss": 0.0222, "step": 54670 }, { "epoch": 24.47079883642873, "grad_norm": 0.06244267150759697, "learning_rate": 2.134025123396638e-06, "loss": 0.0243, "step": 54680 }, { "epoch": 24.47527411053927, "grad_norm": 0.08231230080127716, "learning_rate": 2.1260673047996227e-06, "loss": 0.0221, "step": 54690 }, { "epoch": 24.47974938464981, "grad_norm": 0.0695924386382103, "learning_rate": 2.118124029039309e-06, "loss": 0.0245, "step": 54700 }, { "epoch": 24.48422465876035, "grad_norm": 0.041952453553676605, "learning_rate": 2.110195298528683e-06, "loss": 0.021, "step": 54710 }, { "epoch": 24.488699932870887, "grad_norm": 0.07579559087753296, "learning_rate": 2.102281115676258e-06, "loss": 0.0228, "step": 54720 }, { "epoch": 24.493175206981427, "grad_norm": 0.06540646404027939, "learning_rate": 2.0943814828861762e-06, "loss": 0.0209, "step": 54730 }, { "epoch": 24.497650481091966, "grad_norm": 0.09959695488214493, "learning_rate": 2.0864964025581135e-06, "loss": 0.0233, "step": 54740 }, { "epoch": 24.502125755202506, "grad_norm": 0.08303980529308319, "learning_rate": 2.0786258770873647e-06, "loss": 0.0222, "step": 54750 }, { "epoch": 24.506601029313046, "grad_norm": 0.07981248944997787, "learning_rate": 2.0707699088647836e-06, "loss": 0.0245, "step": 54760 }, { "epoch": 24.511076303423586, "grad_norm": 0.07550721615552902, "learning_rate": 2.062928500276812e-06, "loss": 0.0222, "step": 54770 }, { "epoch": 24.515551577534126, "grad_norm": 0.06629084795713425, "learning_rate": 2.0551016537054493e-06, "loss": 0.0212, "step": 54780 }, { "epoch": 24.520026851644662, "grad_norm": 0.05256744846701622, "learning_rate": 2.047289371528299e-06, "loss": 0.0198, "step": 54790 }, { "epoch": 24.524502125755202, "grad_norm": 0.07153604179620743, "learning_rate": 2.0394916561185083e-06, "loss": 0.0214, "step": 54800 }, { "epoch": 24.52897739986574, "grad_norm": 0.07168957591056824, "learning_rate": 2.0317085098448372e-06, "loss": 0.0233, "step": 54810 }, { "epoch": 24.53345267397628, "grad_norm": 0.050101932138204575, "learning_rate": 2.0239399350715895e-06, "loss": 0.0216, "step": 54820 }, { "epoch": 24.53792794808682, "grad_norm": 0.07386177033185959, "learning_rate": 2.0161859341586597e-06, "loss": 0.0251, "step": 54830 }, { "epoch": 24.54240322219736, "grad_norm": 0.05987437441945076, "learning_rate": 2.008446509461498e-06, "loss": 0.0225, "step": 54840 }, { "epoch": 24.546878496307897, "grad_norm": 0.07112149149179459, "learning_rate": 2.00072166333114e-06, "loss": 0.0238, "step": 54850 }, { "epoch": 24.551353770418437, "grad_norm": 0.060330141335725784, "learning_rate": 1.9930113981142028e-06, "loss": 0.0231, "step": 54860 }, { "epoch": 24.555829044528977, "grad_norm": 0.07626284658908844, "learning_rate": 1.985315716152847e-06, "loss": 0.0226, "step": 54870 }, { "epoch": 24.560304318639517, "grad_norm": 0.07087820768356323, "learning_rate": 1.9776346197848296e-06, "loss": 0.0212, "step": 54880 }, { "epoch": 24.564779592750057, "grad_norm": 0.09254059940576553, "learning_rate": 1.96996811134344e-06, "loss": 0.021, "step": 54890 }, { "epoch": 24.569254866860597, "grad_norm": 0.06089954450726509, "learning_rate": 1.962316193157593e-06, "loss": 0.02, "step": 54900 }, { "epoch": 24.573730140971133, "grad_norm": 0.06487387418746948, "learning_rate": 1.9546788675517236e-06, "loss": 0.0198, "step": 54910 }, { "epoch": 24.578205415081673, "grad_norm": 0.06607669591903687, "learning_rate": 1.9470561368458485e-06, "loss": 0.023, "step": 54920 }, { "epoch": 24.582680689192212, "grad_norm": 0.05367652326822281, "learning_rate": 1.939448003355554e-06, "loss": 0.0238, "step": 54930 }, { "epoch": 24.587155963302752, "grad_norm": 0.05975595861673355, "learning_rate": 1.9318544693919916e-06, "loss": 0.0238, "step": 54940 }, { "epoch": 24.591631237413292, "grad_norm": 0.08256828784942627, "learning_rate": 1.924275537261877e-06, "loss": 0.0213, "step": 54950 }, { "epoch": 24.596106511523832, "grad_norm": 0.06718707084655762, "learning_rate": 1.91671120926748e-06, "loss": 0.0267, "step": 54960 }, { "epoch": 24.600581785634372, "grad_norm": 0.09060722589492798, "learning_rate": 1.9091614877066622e-06, "loss": 0.024, "step": 54970 }, { "epoch": 24.605057059744908, "grad_norm": 0.0723431408405304, "learning_rate": 1.9016263748728114e-06, "loss": 0.0215, "step": 54980 }, { "epoch": 24.609532333855448, "grad_norm": 0.05194584280252457, "learning_rate": 1.8941058730549132e-06, "loss": 0.0205, "step": 54990 }, { "epoch": 24.614007607965988, "grad_norm": 0.07367053627967834, "learning_rate": 1.8865999845374793e-06, "loss": 0.0234, "step": 55000 }, { "epoch": 24.618482882076528, "grad_norm": 0.08319585770368576, "learning_rate": 1.8791087116006189e-06, "loss": 0.0227, "step": 55010 }, { "epoch": 24.622958156187067, "grad_norm": 0.07312959432601929, "learning_rate": 1.8716320565199618e-06, "loss": 0.024, "step": 55020 }, { "epoch": 24.627433430297607, "grad_norm": 0.07664021849632263, "learning_rate": 1.8641700215667413e-06, "loss": 0.0187, "step": 55030 }, { "epoch": 24.631908704408144, "grad_norm": 0.05208143591880798, "learning_rate": 1.856722609007705e-06, "loss": 0.0234, "step": 55040 }, { "epoch": 24.636383978518683, "grad_norm": 0.067894347012043, "learning_rate": 1.849289821105199e-06, "loss": 0.0226, "step": 55050 }, { "epoch": 24.640859252629223, "grad_norm": 0.08627787232398987, "learning_rate": 1.841871660117095e-06, "loss": 0.0263, "step": 55060 }, { "epoch": 24.645334526739763, "grad_norm": 0.09832891076803207, "learning_rate": 1.8344681282968401e-06, "loss": 0.023, "step": 55070 }, { "epoch": 24.649809800850303, "grad_norm": 0.07158241420984268, "learning_rate": 1.8270792278934302e-06, "loss": 0.0241, "step": 55080 }, { "epoch": 24.654285074960843, "grad_norm": 0.07948105782270432, "learning_rate": 1.8197049611514194e-06, "loss": 0.0215, "step": 55090 }, { "epoch": 24.65876034907138, "grad_norm": 0.06579402089118958, "learning_rate": 1.812345330310916e-06, "loss": 0.021, "step": 55100 }, { "epoch": 24.66323562318192, "grad_norm": 0.07106617093086243, "learning_rate": 1.8050003376075707e-06, "loss": 0.0207, "step": 55110 }, { "epoch": 24.66771089729246, "grad_norm": 0.06572968512773514, "learning_rate": 1.7976699852726153e-06, "loss": 0.0239, "step": 55120 }, { "epoch": 24.672186171403, "grad_norm": 0.07418856024742126, "learning_rate": 1.7903542755328073e-06, "loss": 0.0245, "step": 55130 }, { "epoch": 24.67666144551354, "grad_norm": 0.0806424468755722, "learning_rate": 1.7830532106104747e-06, "loss": 0.022, "step": 55140 }, { "epoch": 24.681136719624078, "grad_norm": 0.08224750310182571, "learning_rate": 1.7757667927234767e-06, "loss": 0.0208, "step": 55150 }, { "epoch": 24.685611993734618, "grad_norm": 0.06456077843904495, "learning_rate": 1.7684950240852372e-06, "loss": 0.0236, "step": 55160 }, { "epoch": 24.690087267845154, "grad_norm": 0.10692466795444489, "learning_rate": 1.7612379069047335e-06, "loss": 0.0226, "step": 55170 }, { "epoch": 24.694562541955694, "grad_norm": 0.057007018476724625, "learning_rate": 1.7539954433864858e-06, "loss": 0.0225, "step": 55180 }, { "epoch": 24.699037816066234, "grad_norm": 0.06507768481969833, "learning_rate": 1.7467676357305561e-06, "loss": 0.0228, "step": 55190 }, { "epoch": 24.703513090176774, "grad_norm": 0.09661465883255005, "learning_rate": 1.7395544861325718e-06, "loss": 0.0248, "step": 55200 }, { "epoch": 24.707988364287313, "grad_norm": 0.06663944572210312, "learning_rate": 1.73235599678368e-06, "loss": 0.0216, "step": 55210 }, { "epoch": 24.712463638397853, "grad_norm": 0.07049502432346344, "learning_rate": 1.7251721698706147e-06, "loss": 0.0255, "step": 55220 }, { "epoch": 24.71693891250839, "grad_norm": 0.0713144838809967, "learning_rate": 1.7180030075756136e-06, "loss": 0.0246, "step": 55230 }, { "epoch": 24.72141418661893, "grad_norm": 0.06739949434995651, "learning_rate": 1.7108485120764905e-06, "loss": 0.0221, "step": 55240 }, { "epoch": 24.72588946072947, "grad_norm": 0.06400047242641449, "learning_rate": 1.70370868554659e-06, "loss": 0.0232, "step": 55250 }, { "epoch": 24.73036473484001, "grad_norm": 0.06279453635215759, "learning_rate": 1.696583530154794e-06, "loss": 0.022, "step": 55260 }, { "epoch": 24.73484000895055, "grad_norm": 0.059762731194496155, "learning_rate": 1.6894730480655484e-06, "loss": 0.0259, "step": 55270 }, { "epoch": 24.73931528306109, "grad_norm": 0.08874127268791199, "learning_rate": 1.682377241438826e-06, "loss": 0.0235, "step": 55280 }, { "epoch": 24.74379055717163, "grad_norm": 0.08374295383691788, "learning_rate": 1.6752961124301415e-06, "loss": 0.022, "step": 55290 }, { "epoch": 24.748265831282165, "grad_norm": 0.06800059229135513, "learning_rate": 1.6682296631905626e-06, "loss": 0.021, "step": 55300 }, { "epoch": 24.752741105392705, "grad_norm": 0.07017699629068375, "learning_rate": 1.6611778958666835e-06, "loss": 0.02, "step": 55310 }, { "epoch": 24.757216379503244, "grad_norm": 0.06063047796487808, "learning_rate": 1.6541408126006463e-06, "loss": 0.0247, "step": 55320 }, { "epoch": 24.761691653613784, "grad_norm": 0.05950850620865822, "learning_rate": 1.6471184155301355e-06, "loss": 0.023, "step": 55330 }, { "epoch": 24.766166927724324, "grad_norm": 0.07234857976436615, "learning_rate": 1.6401107067883559e-06, "loss": 0.0227, "step": 55340 }, { "epoch": 24.770642201834864, "grad_norm": 0.0929342731833458, "learning_rate": 1.6331176885040878e-06, "loss": 0.0243, "step": 55350 }, { "epoch": 24.7751174759454, "grad_norm": 0.08487287163734436, "learning_rate": 1.626139362801604e-06, "loss": 0.0232, "step": 55360 }, { "epoch": 24.77959275005594, "grad_norm": 0.0649556890130043, "learning_rate": 1.6191757318007429e-06, "loss": 0.0221, "step": 55370 }, { "epoch": 24.78406802416648, "grad_norm": 0.06731665879487991, "learning_rate": 1.6122267976168781e-06, "loss": 0.0221, "step": 55380 }, { "epoch": 24.78854329827702, "grad_norm": 0.06942565739154816, "learning_rate": 1.6052925623609049e-06, "loss": 0.0237, "step": 55390 }, { "epoch": 24.79301857238756, "grad_norm": 0.06615299731492996, "learning_rate": 1.5983730281392662e-06, "loss": 0.0282, "step": 55400 }, { "epoch": 24.7974938464981, "grad_norm": 0.06998937577009201, "learning_rate": 1.5914681970539192e-06, "loss": 0.0249, "step": 55410 }, { "epoch": 24.801969120608636, "grad_norm": 0.0549243688583374, "learning_rate": 1.5845780712023973e-06, "loss": 0.024, "step": 55420 }, { "epoch": 24.806444394719176, "grad_norm": 0.05713542923331261, "learning_rate": 1.5777026526777094e-06, "loss": 0.0219, "step": 55430 }, { "epoch": 24.810919668829715, "grad_norm": 0.07164763659238815, "learning_rate": 1.5708419435684462e-06, "loss": 0.0232, "step": 55440 }, { "epoch": 24.815394942940255, "grad_norm": 0.06512518972158432, "learning_rate": 1.5639959459586962e-06, "loss": 0.025, "step": 55450 }, { "epoch": 24.819870217050795, "grad_norm": 0.052343979477882385, "learning_rate": 1.5571646619281066e-06, "loss": 0.0197, "step": 55460 }, { "epoch": 24.824345491161335, "grad_norm": 0.08559450507164001, "learning_rate": 1.550348093551829e-06, "loss": 0.0209, "step": 55470 }, { "epoch": 24.82882076527187, "grad_norm": 0.08014874160289764, "learning_rate": 1.5435462429005675e-06, "loss": 0.0229, "step": 55480 }, { "epoch": 24.83329603938241, "grad_norm": 0.051554810255765915, "learning_rate": 1.5367591120405256e-06, "loss": 0.0213, "step": 55490 }, { "epoch": 24.83777131349295, "grad_norm": 0.059295158833265305, "learning_rate": 1.5299867030334814e-06, "loss": 0.0236, "step": 55500 }, { "epoch": 24.84224658760349, "grad_norm": 0.07206369936466217, "learning_rate": 1.5232290179366948e-06, "loss": 0.0217, "step": 55510 }, { "epoch": 24.84672186171403, "grad_norm": 0.0768774077296257, "learning_rate": 1.516486058802974e-06, "loss": 0.0273, "step": 55520 }, { "epoch": 24.85119713582457, "grad_norm": 0.07755032181739807, "learning_rate": 1.5097578276806633e-06, "loss": 0.0229, "step": 55530 }, { "epoch": 24.85567240993511, "grad_norm": 0.07142321020364761, "learning_rate": 1.5030443266136118e-06, "loss": 0.024, "step": 55540 }, { "epoch": 24.860147684045646, "grad_norm": 0.07453295588493347, "learning_rate": 1.4963455576412045e-06, "loss": 0.0212, "step": 55550 }, { "epoch": 24.864622958156186, "grad_norm": 0.06895986199378967, "learning_rate": 1.4896615227983468e-06, "loss": 0.0212, "step": 55560 }, { "epoch": 24.869098232266726, "grad_norm": 0.06309611350297928, "learning_rate": 1.482992224115487e-06, "loss": 0.0253, "step": 55570 }, { "epoch": 24.873573506377266, "grad_norm": 0.06005359813570976, "learning_rate": 1.4763376636185599e-06, "loss": 0.021, "step": 55580 }, { "epoch": 24.878048780487806, "grad_norm": 0.09445089846849442, "learning_rate": 1.4696978433290653e-06, "loss": 0.0234, "step": 55590 }, { "epoch": 24.882524054598345, "grad_norm": 0.059752605855464935, "learning_rate": 1.463072765264001e-06, "loss": 0.022, "step": 55600 }, { "epoch": 24.88699932870888, "grad_norm": 0.07366074621677399, "learning_rate": 1.4564624314358844e-06, "loss": 0.0197, "step": 55610 }, { "epoch": 24.89147460281942, "grad_norm": 0.056271251291036606, "learning_rate": 1.4498668438527597e-06, "loss": 0.0206, "step": 55620 }, { "epoch": 24.89594987692996, "grad_norm": 0.07687298208475113, "learning_rate": 1.4432860045182017e-06, "loss": 0.0229, "step": 55630 }, { "epoch": 24.9004251510405, "grad_norm": 0.07928509265184402, "learning_rate": 1.4367199154312783e-06, "loss": 0.024, "step": 55640 }, { "epoch": 24.90490042515104, "grad_norm": 0.06705622375011444, "learning_rate": 1.4301685785866214e-06, "loss": 0.0211, "step": 55650 }, { "epoch": 24.90937569926158, "grad_norm": 0.06711945682764053, "learning_rate": 1.4236319959743227e-06, "loss": 0.022, "step": 55660 }, { "epoch": 24.91385097337212, "grad_norm": 0.08166247606277466, "learning_rate": 1.4171101695800493e-06, "loss": 0.0219, "step": 55670 }, { "epoch": 24.918326247482657, "grad_norm": 0.0793272852897644, "learning_rate": 1.4106031013849496e-06, "loss": 0.0211, "step": 55680 }, { "epoch": 24.922801521593197, "grad_norm": 0.06609385460615158, "learning_rate": 1.4041107933656928e-06, "loss": 0.021, "step": 55690 }, { "epoch": 24.927276795703737, "grad_norm": 0.0748477652668953, "learning_rate": 1.3976332474944843e-06, "loss": 0.024, "step": 55700 }, { "epoch": 24.931752069814276, "grad_norm": 0.07651635259389877, "learning_rate": 1.3911704657390113e-06, "loss": 0.0271, "step": 55710 }, { "epoch": 24.936227343924816, "grad_norm": 0.07949122786521912, "learning_rate": 1.3847224500625256e-06, "loss": 0.0243, "step": 55720 }, { "epoch": 24.940702618035356, "grad_norm": 0.07523266226053238, "learning_rate": 1.3782892024237327e-06, "loss": 0.0222, "step": 55730 }, { "epoch": 24.945177892145892, "grad_norm": 0.07533029466867447, "learning_rate": 1.3718707247769135e-06, "loss": 0.0244, "step": 55740 }, { "epoch": 24.949653166256432, "grad_norm": 0.07837705314159393, "learning_rate": 1.3654670190718034e-06, "loss": 0.0236, "step": 55750 }, { "epoch": 24.954128440366972, "grad_norm": 0.05873294547200203, "learning_rate": 1.3590780872536958e-06, "loss": 0.0224, "step": 55760 }, { "epoch": 24.958603714477512, "grad_norm": 0.06668862700462341, "learning_rate": 1.3527039312633827e-06, "loss": 0.0228, "step": 55770 }, { "epoch": 24.96307898858805, "grad_norm": 0.08060525357723236, "learning_rate": 1.3463445530371488e-06, "loss": 0.0267, "step": 55780 }, { "epoch": 24.96755426269859, "grad_norm": 0.06818985193967819, "learning_rate": 1.339999954506821e-06, "loss": 0.0215, "step": 55790 }, { "epoch": 24.972029536809128, "grad_norm": 0.07653996348381042, "learning_rate": 1.333670137599713e-06, "loss": 0.0246, "step": 55800 }, { "epoch": 24.976504810919668, "grad_norm": 0.049161527305841446, "learning_rate": 1.3273551042386534e-06, "loss": 0.0209, "step": 55810 }, { "epoch": 24.980980085030208, "grad_norm": 0.0467340424656868, "learning_rate": 1.3210548563419856e-06, "loss": 0.0235, "step": 55820 }, { "epoch": 24.985455359140747, "grad_norm": 0.056863363832235336, "learning_rate": 1.3147693958235618e-06, "loss": 0.0212, "step": 55830 }, { "epoch": 24.989930633251287, "grad_norm": 0.06590430438518524, "learning_rate": 1.3084987245927383e-06, "loss": 0.0241, "step": 55840 }, { "epoch": 24.994405907361827, "grad_norm": 0.07825790345668793, "learning_rate": 1.3022428445543799e-06, "loss": 0.0209, "step": 55850 }, { "epoch": 24.998881181472367, "grad_norm": 0.11514896154403687, "learning_rate": 1.2960017576088446e-06, "loss": 0.0243, "step": 55860 }, { "epoch": 25.003356455582903, "grad_norm": 0.07707114517688751, "learning_rate": 1.2897754656520379e-06, "loss": 0.0219, "step": 55870 }, { "epoch": 25.007831729693443, "grad_norm": 0.08042573928833008, "learning_rate": 1.2835639705753078e-06, "loss": 0.0214, "step": 55880 }, { "epoch": 25.012307003803983, "grad_norm": 0.07317277044057846, "learning_rate": 1.2773672742655784e-06, "loss": 0.0219, "step": 55890 }, { "epoch": 25.016782277914523, "grad_norm": 0.0737830251455307, "learning_rate": 1.2711853786052109e-06, "loss": 0.0207, "step": 55900 }, { "epoch": 25.021257552025062, "grad_norm": 0.07439964264631271, "learning_rate": 1.2650182854721193e-06, "loss": 0.0233, "step": 55910 }, { "epoch": 25.025732826135602, "grad_norm": 0.06926017999649048, "learning_rate": 1.2588659967397e-06, "loss": 0.0215, "step": 55920 }, { "epoch": 25.03020810024614, "grad_norm": 0.12256717681884766, "learning_rate": 1.2527285142768574e-06, "loss": 0.021, "step": 55930 }, { "epoch": 25.03468337435668, "grad_norm": 0.06868284195661545, "learning_rate": 1.2466058399479952e-06, "loss": 0.0226, "step": 55940 }, { "epoch": 25.039158648467218, "grad_norm": 0.055817101150751114, "learning_rate": 1.2404979756130142e-06, "loss": 0.0219, "step": 55950 }, { "epoch": 25.043633922577758, "grad_norm": 0.08006437122821808, "learning_rate": 1.2344049231273302e-06, "loss": 0.0238, "step": 55960 }, { "epoch": 25.048109196688298, "grad_norm": 0.06408238410949707, "learning_rate": 1.2283266843418517e-06, "loss": 0.025, "step": 55970 }, { "epoch": 25.052584470798838, "grad_norm": 0.06536125391721725, "learning_rate": 1.222263261102985e-06, "loss": 0.0245, "step": 55980 }, { "epoch": 25.057059744909374, "grad_norm": 0.06789366900920868, "learning_rate": 1.2162146552526399e-06, "loss": 0.0205, "step": 55990 }, { "epoch": 25.061535019019914, "grad_norm": 0.06389251351356506, "learning_rate": 1.210180868628219e-06, "loss": 0.0243, "step": 56000 }, { "epoch": 25.066010293130454, "grad_norm": 0.0703028216958046, "learning_rate": 1.2041619030626284e-06, "loss": 0.0224, "step": 56010 }, { "epoch": 25.070485567240993, "grad_norm": 0.06217474490404129, "learning_rate": 1.1981577603842776e-06, "loss": 0.0242, "step": 56020 }, { "epoch": 25.074960841351533, "grad_norm": 0.056424517184495926, "learning_rate": 1.192168442417052e-06, "loss": 0.0247, "step": 56030 }, { "epoch": 25.079436115462073, "grad_norm": 0.06277111172676086, "learning_rate": 1.1861939509803687e-06, "loss": 0.0248, "step": 56040 }, { "epoch": 25.083911389572613, "grad_norm": 0.06645642220973969, "learning_rate": 1.1802342878891037e-06, "loss": 0.0238, "step": 56050 }, { "epoch": 25.08838666368315, "grad_norm": 0.07822145521640778, "learning_rate": 1.1742894549536477e-06, "loss": 0.0235, "step": 56060 }, { "epoch": 25.09286193779369, "grad_norm": 0.07143453508615494, "learning_rate": 1.1683594539798893e-06, "loss": 0.0239, "step": 56070 }, { "epoch": 25.09733721190423, "grad_norm": 0.06185320392251015, "learning_rate": 1.16244428676921e-06, "loss": 0.0214, "step": 56080 }, { "epoch": 25.10181248601477, "grad_norm": 0.05661202594637871, "learning_rate": 1.1565439551184664e-06, "loss": 0.0218, "step": 56090 }, { "epoch": 25.10628776012531, "grad_norm": 0.0709112137556076, "learning_rate": 1.1506584608200367e-06, "loss": 0.021, "step": 56100 }, { "epoch": 25.11076303423585, "grad_norm": 0.06833823025226593, "learning_rate": 1.1447878056617734e-06, "loss": 0.0217, "step": 56110 }, { "epoch": 25.115238308346385, "grad_norm": 0.051342546939849854, "learning_rate": 1.138931991427028e-06, "loss": 0.0203, "step": 56120 }, { "epoch": 25.119713582456924, "grad_norm": 0.07749678939580917, "learning_rate": 1.1330910198946442e-06, "loss": 0.0238, "step": 56130 }, { "epoch": 25.124188856567464, "grad_norm": 0.0988159328699112, "learning_rate": 1.1272648928389473e-06, "loss": 0.022, "step": 56140 }, { "epoch": 25.128664130678004, "grad_norm": 0.07116661220788956, "learning_rate": 1.1214536120297714e-06, "loss": 0.0224, "step": 56150 }, { "epoch": 25.133139404788544, "grad_norm": 0.05679116025567055, "learning_rate": 1.1156571792324211e-06, "loss": 0.021, "step": 56160 }, { "epoch": 25.137614678899084, "grad_norm": 0.08271782845258713, "learning_rate": 1.10987559620771e-06, "loss": 0.021, "step": 56170 }, { "epoch": 25.142089953009624, "grad_norm": 0.06823473423719406, "learning_rate": 1.1041088647119114e-06, "loss": 0.0224, "step": 56180 }, { "epoch": 25.14656522712016, "grad_norm": 0.06622306257486343, "learning_rate": 1.0983569864968346e-06, "loss": 0.0209, "step": 56190 }, { "epoch": 25.1510405012307, "grad_norm": 0.06375796347856522, "learning_rate": 1.0926199633097157e-06, "loss": 0.0235, "step": 56200 }, { "epoch": 25.15551577534124, "grad_norm": 0.059044212102890015, "learning_rate": 1.0868977968933436e-06, "loss": 0.0202, "step": 56210 }, { "epoch": 25.15999104945178, "grad_norm": 0.06610739231109619, "learning_rate": 1.0811904889859336e-06, "loss": 0.0226, "step": 56220 }, { "epoch": 25.16446632356232, "grad_norm": 0.06062132120132446, "learning_rate": 1.0754980413212268e-06, "loss": 0.0234, "step": 56230 }, { "epoch": 25.16894159767286, "grad_norm": 0.08091790229082108, "learning_rate": 1.0698204556284452e-06, "loss": 0.0223, "step": 56240 }, { "epoch": 25.173416871783395, "grad_norm": 0.08410315960645676, "learning_rate": 1.064157733632276e-06, "loss": 0.0231, "step": 56250 }, { "epoch": 25.177892145893935, "grad_norm": 0.08399749547243118, "learning_rate": 1.0585098770529157e-06, "loss": 0.0191, "step": 56260 }, { "epoch": 25.182367420004475, "grad_norm": 0.09635680168867111, "learning_rate": 1.0528768876060246e-06, "loss": 0.0228, "step": 56270 }, { "epoch": 25.186842694115015, "grad_norm": 0.08886772394180298, "learning_rate": 1.0472587670027678e-06, "loss": 0.024, "step": 56280 }, { "epoch": 25.191317968225555, "grad_norm": 0.06938537210226059, "learning_rate": 1.0416555169497688e-06, "loss": 0.0231, "step": 56290 }, { "epoch": 25.195793242336094, "grad_norm": 0.07173626869916916, "learning_rate": 1.0360671391491606e-06, "loss": 0.02, "step": 56300 }, { "epoch": 25.20026851644663, "grad_norm": 0.05579651519656181, "learning_rate": 1.030493635298535e-06, "loss": 0.0242, "step": 56310 }, { "epoch": 25.20474379055717, "grad_norm": 0.04501989111304283, "learning_rate": 1.0249350070909768e-06, "loss": 0.0169, "step": 56320 }, { "epoch": 25.20921906466771, "grad_norm": 0.07293622940778732, "learning_rate": 1.0193912562150464e-06, "loss": 0.0209, "step": 56330 }, { "epoch": 25.21369433877825, "grad_norm": 0.0622115358710289, "learning_rate": 1.0138623843548078e-06, "loss": 0.0244, "step": 56340 }, { "epoch": 25.21816961288879, "grad_norm": 0.11200553178787231, "learning_rate": 1.008348393189762e-06, "loss": 0.0211, "step": 56350 }, { "epoch": 25.22264488699933, "grad_norm": 0.054070256650447845, "learning_rate": 1.00284928439493e-06, "loss": 0.0232, "step": 56360 }, { "epoch": 25.22712016110987, "grad_norm": 0.07153208553791046, "learning_rate": 9.97365059640787e-07, "loss": 0.0236, "step": 56370 }, { "epoch": 25.231595435220406, "grad_norm": 0.07880662381649017, "learning_rate": 9.918957205933e-07, "loss": 0.0232, "step": 56380 }, { "epoch": 25.236070709330946, "grad_norm": 0.06348978728055954, "learning_rate": 9.864412689139123e-07, "loss": 0.0214, "step": 56390 }, { "epoch": 25.240545983441486, "grad_norm": 0.08854231983423233, "learning_rate": 9.810017062595322e-07, "loss": 0.0213, "step": 56400 }, { "epoch": 25.245021257552025, "grad_norm": 0.06854210048913956, "learning_rate": 9.75577034282571e-07, "loss": 0.0204, "step": 56410 }, { "epoch": 25.249496531662565, "grad_norm": 0.08261562138795853, "learning_rate": 9.701672546308827e-07, "loss": 0.0236, "step": 56420 }, { "epoch": 25.253971805773105, "grad_norm": 0.05344239994883537, "learning_rate": 9.647723689478305e-07, "loss": 0.0235, "step": 56430 }, { "epoch": 25.25844707988364, "grad_norm": 0.0740654319524765, "learning_rate": 9.593923788722315e-07, "loss": 0.0228, "step": 56440 }, { "epoch": 25.26292235399418, "grad_norm": 0.05753658711910248, "learning_rate": 9.540272860383837e-07, "loss": 0.0195, "step": 56450 }, { "epoch": 25.26739762810472, "grad_norm": 0.06747250258922577, "learning_rate": 9.486770920760668e-07, "loss": 0.0232, "step": 56460 }, { "epoch": 25.27187290221526, "grad_norm": 0.07606132328510284, "learning_rate": 9.433417986105198e-07, "loss": 0.0239, "step": 56470 }, { "epoch": 25.2763481763258, "grad_norm": 0.07842040807008743, "learning_rate": 9.380214072624682e-07, "loss": 0.0241, "step": 56480 }, { "epoch": 25.28082345043634, "grad_norm": 0.06946040689945221, "learning_rate": 9.327159196481138e-07, "loss": 0.0231, "step": 56490 }, { "epoch": 25.285298724546877, "grad_norm": 0.0681198462843895, "learning_rate": 9.274253373791064e-07, "loss": 0.0231, "step": 56500 }, { "epoch": 25.289773998657417, "grad_norm": 0.0496402308344841, "learning_rate": 9.221496620626047e-07, "loss": 0.0222, "step": 56510 }, { "epoch": 25.294249272767956, "grad_norm": 0.056960176676511765, "learning_rate": 9.168888953011989e-07, "loss": 0.021, "step": 56520 }, { "epoch": 25.298724546878496, "grad_norm": 0.06181725114583969, "learning_rate": 9.116430386929886e-07, "loss": 0.0221, "step": 56530 }, { "epoch": 25.303199820989036, "grad_norm": 0.08477290719747543, "learning_rate": 9.064120938315213e-07, "loss": 0.0208, "step": 56540 }, { "epoch": 25.307675095099576, "grad_norm": 0.07493223994970322, "learning_rate": 9.011960623058202e-07, "loss": 0.0207, "step": 56550 }, { "epoch": 25.312150369210116, "grad_norm": 0.08671266585588455, "learning_rate": 8.959949457003736e-07, "loss": 0.0232, "step": 56560 }, { "epoch": 25.316625643320652, "grad_norm": 0.09413482248783112, "learning_rate": 8.908087455951397e-07, "loss": 0.0272, "step": 56570 }, { "epoch": 25.321100917431192, "grad_norm": 0.09046152234077454, "learning_rate": 8.856374635655695e-07, "loss": 0.0267, "step": 56580 }, { "epoch": 25.32557619154173, "grad_norm": 0.08605136722326279, "learning_rate": 8.804811011825398e-07, "loss": 0.0216, "step": 56590 }, { "epoch": 25.33005146565227, "grad_norm": 0.04153513163328171, "learning_rate": 8.753396600124253e-07, "loss": 0.0242, "step": 56600 }, { "epoch": 25.33452673976281, "grad_norm": 0.08193585276603699, "learning_rate": 8.702131416170656e-07, "loss": 0.0229, "step": 56610 }, { "epoch": 25.33900201387335, "grad_norm": 0.06712353974580765, "learning_rate": 8.651015475537538e-07, "loss": 0.0199, "step": 56620 }, { "epoch": 25.343477287983887, "grad_norm": 0.06901484727859497, "learning_rate": 8.60004879375259e-07, "loss": 0.0199, "step": 56630 }, { "epoch": 25.347952562094427, "grad_norm": 0.09550810605287552, "learning_rate": 8.549231386298151e-07, "loss": 0.0219, "step": 56640 }, { "epoch": 25.352427836204967, "grad_norm": 0.053468137979507446, "learning_rate": 8.498563268611204e-07, "loss": 0.0216, "step": 56650 }, { "epoch": 25.356903110315507, "grad_norm": 0.08376485854387283, "learning_rate": 8.448044456083493e-07, "loss": 0.0197, "step": 56660 }, { "epoch": 25.361378384426047, "grad_norm": 0.08258692175149918, "learning_rate": 8.397674964061075e-07, "loss": 0.0229, "step": 56670 }, { "epoch": 25.365853658536587, "grad_norm": 0.06031908467411995, "learning_rate": 8.347454807845045e-07, "loss": 0.0239, "step": 56680 }, { "epoch": 25.370328932647126, "grad_norm": 0.06836334615945816, "learning_rate": 8.297384002690866e-07, "loss": 0.0255, "step": 56690 }, { "epoch": 25.374804206757663, "grad_norm": 0.07623302191495895, "learning_rate": 8.247462563808817e-07, "loss": 0.0237, "step": 56700 }, { "epoch": 25.379279480868203, "grad_norm": 0.05918397381901741, "learning_rate": 8.197690506363709e-07, "loss": 0.0241, "step": 56710 }, { "epoch": 25.383754754978742, "grad_norm": 0.08051345497369766, "learning_rate": 8.148067845474838e-07, "loss": 0.0244, "step": 56720 }, { "epoch": 25.388230029089282, "grad_norm": 0.057610418647527695, "learning_rate": 8.098594596216424e-07, "loss": 0.02, "step": 56730 }, { "epoch": 25.392705303199822, "grad_norm": 0.06585419178009033, "learning_rate": 8.049270773617057e-07, "loss": 0.0194, "step": 56740 }, { "epoch": 25.397180577310362, "grad_norm": 0.06759890168905258, "learning_rate": 8.000096392660029e-07, "loss": 0.0251, "step": 56750 }, { "epoch": 25.401655851420898, "grad_norm": 0.08605509251356125, "learning_rate": 7.951071468283167e-07, "loss": 0.0246, "step": 56760 }, { "epoch": 25.406131125531438, "grad_norm": 0.06207672506570816, "learning_rate": 7.90219601537906e-07, "loss": 0.0214, "step": 56770 }, { "epoch": 25.410606399641978, "grad_norm": 0.07434020936489105, "learning_rate": 7.853470048794664e-07, "loss": 0.0243, "step": 56780 }, { "epoch": 25.415081673752518, "grad_norm": 0.08009161800146103, "learning_rate": 7.804893583331696e-07, "loss": 0.0209, "step": 56790 }, { "epoch": 25.419556947863057, "grad_norm": 0.05070391669869423, "learning_rate": 7.756466633746407e-07, "loss": 0.0228, "step": 56800 }, { "epoch": 25.424032221973597, "grad_norm": 0.06248873099684715, "learning_rate": 7.708189214749639e-07, "loss": 0.0237, "step": 56810 }, { "epoch": 25.428507496084134, "grad_norm": 0.08381694555282593, "learning_rate": 7.66006134100672e-07, "loss": 0.0239, "step": 56820 }, { "epoch": 25.432982770194673, "grad_norm": 0.06586989015340805, "learning_rate": 7.612083027137728e-07, "loss": 0.0227, "step": 56830 }, { "epoch": 25.437458044305213, "grad_norm": 0.09457169473171234, "learning_rate": 7.564254287717176e-07, "loss": 0.0251, "step": 56840 }, { "epoch": 25.441933318415753, "grad_norm": 0.06602965295314789, "learning_rate": 7.516575137274162e-07, "loss": 0.022, "step": 56850 }, { "epoch": 25.446408592526293, "grad_norm": 0.07529473304748535, "learning_rate": 7.469045590292323e-07, "loss": 0.0211, "step": 56860 }, { "epoch": 25.450883866636833, "grad_norm": 0.08305945992469788, "learning_rate": 7.421665661209887e-07, "loss": 0.0227, "step": 56870 }, { "epoch": 25.455359140747372, "grad_norm": 0.07642974704504013, "learning_rate": 7.374435364419674e-07, "loss": 0.0195, "step": 56880 }, { "epoch": 25.45983441485791, "grad_norm": 0.06564078480005264, "learning_rate": 7.32735471426893e-07, "loss": 0.0243, "step": 56890 }, { "epoch": 25.46430968896845, "grad_norm": 0.05645906925201416, "learning_rate": 7.280423725059604e-07, "loss": 0.0228, "step": 56900 }, { "epoch": 25.46878496307899, "grad_norm": 0.07024683058261871, "learning_rate": 7.233642411048014e-07, "loss": 0.0243, "step": 56910 }, { "epoch": 25.47326023718953, "grad_norm": 0.06360854208469391, "learning_rate": 7.187010786445181e-07, "loss": 0.0214, "step": 56920 }, { "epoch": 25.477735511300068, "grad_norm": 0.05010396987199783, "learning_rate": 7.140528865416441e-07, "loss": 0.0185, "step": 56930 }, { "epoch": 25.482210785410608, "grad_norm": 0.09008659422397614, "learning_rate": 7.094196662081831e-07, "loss": 0.0215, "step": 56940 }, { "epoch": 25.486686059521144, "grad_norm": 0.05953236296772957, "learning_rate": 7.048014190515872e-07, "loss": 0.0222, "step": 56950 }, { "epoch": 25.491161333631684, "grad_norm": 0.07052052021026611, "learning_rate": 7.001981464747565e-07, "loss": 0.0251, "step": 56960 }, { "epoch": 25.495636607742224, "grad_norm": 0.06229604780673981, "learning_rate": 6.956098498760389e-07, "loss": 0.0222, "step": 56970 }, { "epoch": 25.500111881852764, "grad_norm": 0.0718589499592781, "learning_rate": 6.910365306492416e-07, "loss": 0.024, "step": 56980 }, { "epoch": 25.504587155963304, "grad_norm": 0.07143209874629974, "learning_rate": 6.864781901836259e-07, "loss": 0.0247, "step": 56990 }, { "epoch": 25.509062430073843, "grad_norm": 0.06800755113363266, "learning_rate": 6.819348298638839e-07, "loss": 0.022, "step": 57000 }, { "epoch": 25.51353770418438, "grad_norm": 0.06281273066997528, "learning_rate": 6.774064510701727e-07, "loss": 0.0221, "step": 57010 }, { "epoch": 25.51801297829492, "grad_norm": 0.0752604678273201, "learning_rate": 6.728930551780865e-07, "loss": 0.0224, "step": 57020 }, { "epoch": 25.52248825240546, "grad_norm": 0.06732075661420822, "learning_rate": 6.683946435586952e-07, "loss": 0.0257, "step": 57030 }, { "epoch": 25.526963526516, "grad_norm": 0.07980488985776901, "learning_rate": 6.639112175784778e-07, "loss": 0.0241, "step": 57040 }, { "epoch": 25.53143880062654, "grad_norm": 0.07229304313659668, "learning_rate": 6.594427785993951e-07, "loss": 0.0233, "step": 57050 }, { "epoch": 25.53591407473708, "grad_norm": 0.04880115017294884, "learning_rate": 6.549893279788277e-07, "loss": 0.0252, "step": 57060 }, { "epoch": 25.54038934884762, "grad_norm": 0.047120098024606705, "learning_rate": 6.50550867069627e-07, "loss": 0.023, "step": 57070 }, { "epoch": 25.544864622958155, "grad_norm": 0.05165456235408783, "learning_rate": 6.461273972200755e-07, "loss": 0.0216, "step": 57080 }, { "epoch": 25.549339897068695, "grad_norm": 0.07018270343542099, "learning_rate": 6.417189197739093e-07, "loss": 0.0226, "step": 57090 }, { "epoch": 25.553815171179235, "grad_norm": 0.07274896651506424, "learning_rate": 6.373254360703018e-07, "loss": 0.0234, "step": 57100 }, { "epoch": 25.558290445289774, "grad_norm": 0.07148301601409912, "learning_rate": 6.329469474438798e-07, "loss": 0.0223, "step": 57110 }, { "epoch": 25.562765719400314, "grad_norm": 0.07854261249303818, "learning_rate": 6.285834552247128e-07, "loss": 0.0214, "step": 57120 }, { "epoch": 25.567240993510854, "grad_norm": 0.04887008294463158, "learning_rate": 6.24234960738318e-07, "loss": 0.0222, "step": 57130 }, { "epoch": 25.57171626762139, "grad_norm": 0.08692841976881027, "learning_rate": 6.1990146530565e-07, "loss": 0.0253, "step": 57140 }, { "epoch": 25.57619154173193, "grad_norm": 0.054064832627773285, "learning_rate": 6.15582970243117e-07, "loss": 0.0199, "step": 57150 }, { "epoch": 25.58066681584247, "grad_norm": 0.06283891201019287, "learning_rate": 6.11279476862553e-07, "loss": 0.0246, "step": 57160 }, { "epoch": 25.58514208995301, "grad_norm": 0.0598238930106163, "learning_rate": 6.069909864712453e-07, "loss": 0.0226, "step": 57170 }, { "epoch": 25.58961736406355, "grad_norm": 0.06692572683095932, "learning_rate": 6.027175003719354e-07, "loss": 0.0212, "step": 57180 }, { "epoch": 25.59409263817409, "grad_norm": 0.07482142746448517, "learning_rate": 5.984590198627849e-07, "loss": 0.0209, "step": 57190 }, { "epoch": 25.59856791228463, "grad_norm": 0.08595924079418182, "learning_rate": 5.9421554623742e-07, "loss": 0.0234, "step": 57200 }, { "epoch": 25.603043186395166, "grad_norm": 0.05554866045713425, "learning_rate": 5.899870807848762e-07, "loss": 0.0258, "step": 57210 }, { "epoch": 25.607518460505705, "grad_norm": 0.06492603570222855, "learning_rate": 5.857736247896706e-07, "loss": 0.0236, "step": 57220 }, { "epoch": 25.611993734616245, "grad_norm": 0.06323204934597015, "learning_rate": 5.815751795317237e-07, "loss": 0.0226, "step": 57230 }, { "epoch": 25.616469008726785, "grad_norm": 0.060541894286870956, "learning_rate": 5.773917462864264e-07, "loss": 0.0253, "step": 57240 }, { "epoch": 25.620944282837325, "grad_norm": 0.06966357678174973, "learning_rate": 5.732233263245845e-07, "loss": 0.0226, "step": 57250 }, { "epoch": 25.625419556947865, "grad_norm": 0.07388221472501755, "learning_rate": 5.690699209124573e-07, "loss": 0.0235, "step": 57260 }, { "epoch": 25.6298948310584, "grad_norm": 0.06355644762516022, "learning_rate": 5.64931531311741e-07, "loss": 0.0239, "step": 57270 }, { "epoch": 25.63437010516894, "grad_norm": 0.06467900425195694, "learning_rate": 5.608081587795688e-07, "loss": 0.0235, "step": 57280 }, { "epoch": 25.63884537927948, "grad_norm": 0.09136252105236053, "learning_rate": 5.566998045685112e-07, "loss": 0.0226, "step": 57290 }, { "epoch": 25.64332065339002, "grad_norm": 0.0777633935213089, "learning_rate": 5.526064699265753e-07, "loss": 0.0231, "step": 57300 }, { "epoch": 25.64779592750056, "grad_norm": 0.05192317068576813, "learning_rate": 5.485281560972166e-07, "loss": 0.0204, "step": 57310 }, { "epoch": 25.6522712016111, "grad_norm": 0.08918178081512451, "learning_rate": 5.444648643193051e-07, "loss": 0.0253, "step": 57320 }, { "epoch": 25.656746475721636, "grad_norm": 0.06555275619029999, "learning_rate": 5.404165958271811e-07, "loss": 0.0241, "step": 57330 }, { "epoch": 25.661221749832176, "grad_norm": 0.06413660943508148, "learning_rate": 5.363833518505834e-07, "loss": 0.023, "step": 57340 }, { "epoch": 25.665697023942716, "grad_norm": 0.053515587002038956, "learning_rate": 5.323651336147151e-07, "loss": 0.0248, "step": 57350 }, { "epoch": 25.670172298053256, "grad_norm": 0.07225658744573593, "learning_rate": 5.283619423401998e-07, "loss": 0.0203, "step": 57360 }, { "epoch": 25.674647572163796, "grad_norm": 0.05709230154752731, "learning_rate": 5.243737792431091e-07, "loss": 0.0223, "step": 57370 }, { "epoch": 25.679122846274336, "grad_norm": 0.09590768814086914, "learning_rate": 5.204006455349297e-07, "loss": 0.0262, "step": 57380 }, { "epoch": 25.683598120384872, "grad_norm": 0.07395821809768677, "learning_rate": 5.164425424226016e-07, "loss": 0.0256, "step": 57390 }, { "epoch": 25.68807339449541, "grad_norm": 0.04904337227344513, "learning_rate": 5.124994711084963e-07, "loss": 0.0221, "step": 57400 }, { "epoch": 25.69254866860595, "grad_norm": 0.09547413885593414, "learning_rate": 5.085714327904056e-07, "loss": 0.0218, "step": 57410 }, { "epoch": 25.69702394271649, "grad_norm": 0.07093130797147751, "learning_rate": 5.046584286615697e-07, "loss": 0.023, "step": 57420 }, { "epoch": 25.70149921682703, "grad_norm": 0.06443009525537491, "learning_rate": 5.007604599106486e-07, "loss": 0.0214, "step": 57430 }, { "epoch": 25.70597449093757, "grad_norm": 0.05292917788028717, "learning_rate": 4.968775277217563e-07, "loss": 0.0245, "step": 57440 }, { "epoch": 25.71044976504811, "grad_norm": 0.07611612230539322, "learning_rate": 4.930096332744105e-07, "loss": 0.0235, "step": 57450 }, { "epoch": 25.714925039158647, "grad_norm": 0.0682113990187645, "learning_rate": 4.891567777435879e-07, "loss": 0.0214, "step": 57460 }, { "epoch": 25.719400313269187, "grad_norm": 0.0853809341788292, "learning_rate": 4.853189622996745e-07, "loss": 0.0244, "step": 57470 }, { "epoch": 25.723875587379727, "grad_norm": 0.04127541184425354, "learning_rate": 4.814961881085045e-07, "loss": 0.022, "step": 57480 }, { "epoch": 25.728350861490267, "grad_norm": 0.04565267637372017, "learning_rate": 4.776884563313266e-07, "loss": 0.0202, "step": 57490 }, { "epoch": 25.732826135600806, "grad_norm": 0.04913035035133362, "learning_rate": 4.738957681248379e-07, "loss": 0.0235, "step": 57500 }, { "epoch": 25.737301409711346, "grad_norm": 0.0635208711028099, "learning_rate": 4.701181246411501e-07, "loss": 0.0207, "step": 57510 }, { "epoch": 25.741776683821882, "grad_norm": 0.06521882861852646, "learning_rate": 4.6635552702782305e-07, "loss": 0.0218, "step": 57520 }, { "epoch": 25.746251957932422, "grad_norm": 0.09167583286762238, "learning_rate": 4.6260797642782014e-07, "loss": 0.0248, "step": 57530 }, { "epoch": 25.750727232042962, "grad_norm": 0.06447575986385345, "learning_rate": 4.5887547397955864e-07, "loss": 0.0247, "step": 57540 }, { "epoch": 25.755202506153502, "grad_norm": 0.04552532359957695, "learning_rate": 4.5515802081687045e-07, "loss": 0.0244, "step": 57550 }, { "epoch": 25.75967778026404, "grad_norm": 0.07142572849988937, "learning_rate": 4.514556180690188e-07, "loss": 0.022, "step": 57560 }, { "epoch": 25.76415305437458, "grad_norm": 0.05820276215672493, "learning_rate": 4.4776826686069305e-07, "loss": 0.0219, "step": 57570 }, { "epoch": 25.76862832848512, "grad_norm": 0.0469278022646904, "learning_rate": 4.440959683120194e-07, "loss": 0.0203, "step": 57580 }, { "epoch": 25.773103602595658, "grad_norm": 0.09274917840957642, "learning_rate": 4.404387235385443e-07, "loss": 0.0222, "step": 57590 }, { "epoch": 25.777578876706198, "grad_norm": 0.05638033524155617, "learning_rate": 4.367965336512403e-07, "loss": 0.0265, "step": 57600 }, { "epoch": 25.782054150816737, "grad_norm": 0.056418225169181824, "learning_rate": 4.331693997565056e-07, "loss": 0.0202, "step": 57610 }, { "epoch": 25.786529424927277, "grad_norm": 0.05705961585044861, "learning_rate": 4.2955732295617554e-07, "loss": 0.0236, "step": 57620 }, { "epoch": 25.791004699037817, "grad_norm": 0.08392651379108429, "learning_rate": 4.259603043475002e-07, "loss": 0.0242, "step": 57630 }, { "epoch": 25.795479973148357, "grad_norm": 0.06348030269145966, "learning_rate": 4.2237834502314997e-07, "loss": 0.0233, "step": 57640 }, { "epoch": 25.799955247258893, "grad_norm": 0.05546289682388306, "learning_rate": 4.1881144607124334e-07, "loss": 0.0197, "step": 57650 }, { "epoch": 25.804430521369433, "grad_norm": 0.062221623957157135, "learning_rate": 4.1525960857530243e-07, "loss": 0.0211, "step": 57660 }, { "epoch": 25.808905795479973, "grad_norm": 0.06065545231103897, "learning_rate": 4.1172283361428644e-07, "loss": 0.0237, "step": 57670 }, { "epoch": 25.813381069590513, "grad_norm": 0.06535425782203674, "learning_rate": 4.082011222625637e-07, "loss": 0.0234, "step": 57680 }, { "epoch": 25.817856343701052, "grad_norm": 0.10467549413442612, "learning_rate": 4.0469447558995065e-07, "loss": 0.0248, "step": 57690 }, { "epoch": 25.822331617811592, "grad_norm": 0.07924742251634598, "learning_rate": 4.012028946616675e-07, "loss": 0.0236, "step": 57700 }, { "epoch": 25.82680689192213, "grad_norm": 0.07494198530912399, "learning_rate": 3.977263805383602e-07, "loss": 0.0243, "step": 57710 }, { "epoch": 25.83128216603267, "grad_norm": 0.06047989800572395, "learning_rate": 3.9426493427611177e-07, "loss": 0.0232, "step": 57720 }, { "epoch": 25.835757440143208, "grad_norm": 0.06641564518213272, "learning_rate": 3.9081855692640333e-07, "loss": 0.0245, "step": 57730 }, { "epoch": 25.840232714253748, "grad_norm": 0.054606836289167404, "learning_rate": 3.873872495361697e-07, "loss": 0.0222, "step": 57740 }, { "epoch": 25.844707988364288, "grad_norm": 0.04894019663333893, "learning_rate": 3.839710131477492e-07, "loss": 0.0212, "step": 57750 }, { "epoch": 25.849183262474828, "grad_norm": 0.08328894525766373, "learning_rate": 3.805698487988951e-07, "loss": 0.0262, "step": 57760 }, { "epoch": 25.853658536585368, "grad_norm": 0.06159064546227455, "learning_rate": 3.7718375752279214e-07, "loss": 0.0232, "step": 57770 }, { "epoch": 25.858133810695904, "grad_norm": 0.08781348913908005, "learning_rate": 3.738127403480507e-07, "loss": 0.0228, "step": 57780 }, { "epoch": 25.862609084806444, "grad_norm": 0.07347183674573898, "learning_rate": 3.7045679829870175e-07, "loss": 0.0235, "step": 57790 }, { "epoch": 25.867084358916983, "grad_norm": 0.08802875131368637, "learning_rate": 3.6711593239417973e-07, "loss": 0.0242, "step": 57800 }, { "epoch": 25.871559633027523, "grad_norm": 0.08460371196269989, "learning_rate": 3.637901436493507e-07, "loss": 0.0217, "step": 57810 }, { "epoch": 25.876034907138063, "grad_norm": 0.05279671028256416, "learning_rate": 3.604794330745176e-07, "loss": 0.0212, "step": 57820 }, { "epoch": 25.880510181248603, "grad_norm": 0.05377067252993584, "learning_rate": 3.571838016753759e-07, "loss": 0.0207, "step": 57830 }, { "epoch": 25.88498545535914, "grad_norm": 0.07551734894514084, "learning_rate": 3.5390325045304706e-07, "loss": 0.0251, "step": 57840 }, { "epoch": 25.88946072946968, "grad_norm": 0.06460294872522354, "learning_rate": 3.5063778040408947e-07, "loss": 0.0225, "step": 57850 }, { "epoch": 25.89393600358022, "grad_norm": 0.09511226415634155, "learning_rate": 3.4738739252045405e-07, "loss": 0.0231, "step": 57860 }, { "epoch": 25.89841127769076, "grad_norm": 0.059966959059238434, "learning_rate": 3.441520877895288e-07, "loss": 0.0247, "step": 57870 }, { "epoch": 25.9028865518013, "grad_norm": 0.05421395227313042, "learning_rate": 3.4093186719411085e-07, "loss": 0.0239, "step": 57880 }, { "epoch": 25.90736182591184, "grad_norm": 0.07263540476560593, "learning_rate": 3.377267317124233e-07, "loss": 0.0216, "step": 57890 }, { "epoch": 25.911837100022375, "grad_norm": 0.09991221129894257, "learning_rate": 3.3453668231809286e-07, "loss": 0.0286, "step": 57900 }, { "epoch": 25.916312374132914, "grad_norm": 0.07993126660585403, "learning_rate": 3.3136171998017775e-07, "loss": 0.022, "step": 57910 }, { "epoch": 25.920787648243454, "grad_norm": 0.06201104819774628, "learning_rate": 3.2820184566315084e-07, "loss": 0.0225, "step": 57920 }, { "epoch": 25.925262922353994, "grad_norm": 0.08561783283948898, "learning_rate": 3.250570603268943e-07, "loss": 0.0222, "step": 57930 }, { "epoch": 25.929738196464534, "grad_norm": 0.06355354934930801, "learning_rate": 3.219273649267163e-07, "loss": 0.024, "step": 57940 }, { "epoch": 25.934213470575074, "grad_norm": 0.07155746221542358, "learning_rate": 3.188127604133284e-07, "loss": 0.0267, "step": 57950 }, { "epoch": 25.938688744685614, "grad_norm": 0.07050029933452606, "learning_rate": 3.157132477328628e-07, "loss": 0.0245, "step": 57960 }, { "epoch": 25.94316401879615, "grad_norm": 0.06695582717657089, "learning_rate": 3.12628827826883e-07, "loss": 0.0188, "step": 57970 }, { "epoch": 25.94763929290669, "grad_norm": 0.11098793894052505, "learning_rate": 3.095595016323394e-07, "loss": 0.0221, "step": 57980 }, { "epoch": 25.95211456701723, "grad_norm": 0.05937023088335991, "learning_rate": 3.0650527008162513e-07, "loss": 0.0226, "step": 57990 }, { "epoch": 25.95658984112777, "grad_norm": 0.07331227511167526, "learning_rate": 3.034661341025258e-07, "loss": 0.024, "step": 58000 }, { "epoch": 25.96106511523831, "grad_norm": 0.04874490946531296, "learning_rate": 3.0044209461825844e-07, "loss": 0.0232, "step": 58010 }, { "epoch": 25.96554038934885, "grad_norm": 0.07524160295724869, "learning_rate": 2.9743315254743833e-07, "loss": 0.0219, "step": 58020 }, { "epoch": 25.970015663459385, "grad_norm": 0.05447525158524513, "learning_rate": 2.944393088041009e-07, "loss": 0.0219, "step": 58030 }, { "epoch": 25.974490937569925, "grad_norm": 0.06417597085237503, "learning_rate": 2.9146056429771305e-07, "loss": 0.024, "step": 58040 }, { "epoch": 25.978966211680465, "grad_norm": 0.08669592440128326, "learning_rate": 2.8849691993311777e-07, "loss": 0.0222, "step": 58050 }, { "epoch": 25.983441485791005, "grad_norm": 0.07234430313110352, "learning_rate": 2.85548376610606e-07, "loss": 0.0234, "step": 58060 }, { "epoch": 25.987916759901545, "grad_norm": 0.05328509211540222, "learning_rate": 2.826149352258667e-07, "loss": 0.0226, "step": 58070 }, { "epoch": 25.992392034012084, "grad_norm": 0.08461876958608627, "learning_rate": 2.796965966699927e-07, "loss": 0.0234, "step": 58080 }, { "epoch": 25.996867308122624, "grad_norm": 0.0602012537419796, "learning_rate": 2.767933618295082e-07, "loss": 0.0236, "step": 58090 }, { "epoch": 26.00134258223316, "grad_norm": 0.06657827645540237, "learning_rate": 2.7390523158633554e-07, "loss": 0.025, "step": 58100 }, { "epoch": 26.0058178563437, "grad_norm": 0.0737830102443695, "learning_rate": 2.7103220681780615e-07, "loss": 0.0246, "step": 58110 }, { "epoch": 26.01029313045424, "grad_norm": 0.07189750671386719, "learning_rate": 2.6817428839668315e-07, "loss": 0.0208, "step": 58120 }, { "epoch": 26.01476840456478, "grad_norm": 0.04863140359520912, "learning_rate": 2.653314771911108e-07, "loss": 0.018, "step": 58130 }, { "epoch": 26.01924367867532, "grad_norm": 0.05414518713951111, "learning_rate": 2.625037740646763e-07, "loss": 0.02, "step": 58140 }, { "epoch": 26.02371895278586, "grad_norm": 0.07391583174467087, "learning_rate": 2.5969117987634797e-07, "loss": 0.0192, "step": 58150 }, { "epoch": 26.028194226896396, "grad_norm": 0.05717112496495247, "learning_rate": 2.568936954805201e-07, "loss": 0.0213, "step": 58160 }, { "epoch": 26.032669501006936, "grad_norm": 0.06542834639549255, "learning_rate": 2.5411132172700194e-07, "loss": 0.0223, "step": 58170 }, { "epoch": 26.037144775117476, "grad_norm": 0.06287791579961777, "learning_rate": 2.51344059460995e-07, "loss": 0.0222, "step": 58180 }, { "epoch": 26.041620049228015, "grad_norm": 0.10678178071975708, "learning_rate": 2.485919095231326e-07, "loss": 0.0253, "step": 58190 }, { "epoch": 26.046095323338555, "grad_norm": 0.05249643325805664, "learning_rate": 2.458548727494292e-07, "loss": 0.0202, "step": 58200 }, { "epoch": 26.050570597449095, "grad_norm": 0.06843402236700058, "learning_rate": 2.4313294997134195e-07, "loss": 0.021, "step": 58210 }, { "epoch": 26.05504587155963, "grad_norm": 0.07518023252487183, "learning_rate": 2.404261420157039e-07, "loss": 0.0239, "step": 58220 }, { "epoch": 26.05952114567017, "grad_norm": 0.05914663150906563, "learning_rate": 2.3773444970477955e-07, "loss": 0.0244, "step": 58230 }, { "epoch": 26.06399641978071, "grad_norm": 0.06298981606960297, "learning_rate": 2.3505787385623702e-07, "loss": 0.0238, "step": 58240 }, { "epoch": 26.06847169389125, "grad_norm": 0.0840022936463356, "learning_rate": 2.323964152831426e-07, "loss": 0.0223, "step": 58250 }, { "epoch": 26.07294696800179, "grad_norm": 0.06034082919359207, "learning_rate": 2.2975007479397738e-07, "loss": 0.0188, "step": 58260 }, { "epoch": 26.07742224211233, "grad_norm": 0.06084949150681496, "learning_rate": 2.2711885319263714e-07, "loss": 0.0227, "step": 58270 }, { "epoch": 26.08189751622287, "grad_norm": 0.06335198879241943, "learning_rate": 2.2450275127841036e-07, "loss": 0.0221, "step": 58280 }, { "epoch": 26.086372790333407, "grad_norm": 0.058572590351104736, "learning_rate": 2.219017698460002e-07, "loss": 0.0248, "step": 58290 }, { "epoch": 26.090848064443946, "grad_norm": 0.06669831275939941, "learning_rate": 2.1931590968551908e-07, "loss": 0.0233, "step": 58300 }, { "epoch": 26.095323338554486, "grad_norm": 0.04715944826602936, "learning_rate": 2.1674517158248308e-07, "loss": 0.0198, "step": 58310 }, { "epoch": 26.099798612665026, "grad_norm": 0.04746556654572487, "learning_rate": 2.1418955631781202e-07, "loss": 0.02, "step": 58320 }, { "epoch": 26.104273886775566, "grad_norm": 0.07543343305587769, "learning_rate": 2.1164906466783485e-07, "loss": 0.0209, "step": 58330 }, { "epoch": 26.108749160886106, "grad_norm": 0.06113037094473839, "learning_rate": 2.0912369740428983e-07, "loss": 0.0232, "step": 58340 }, { "epoch": 26.113224434996642, "grad_norm": 0.08488951623439789, "learning_rate": 2.0661345529430775e-07, "loss": 0.026, "step": 58350 }, { "epoch": 26.117699709107182, "grad_norm": 0.0648958757519722, "learning_rate": 2.041183391004453e-07, "loss": 0.0207, "step": 58360 }, { "epoch": 26.12217498321772, "grad_norm": 0.06348193436861038, "learning_rate": 2.0163834958064065e-07, "loss": 0.0218, "step": 58370 }, { "epoch": 26.12665025732826, "grad_norm": 0.08929307758808136, "learning_rate": 1.9917348748826335e-07, "loss": 0.0236, "step": 58380 }, { "epoch": 26.1311255314388, "grad_norm": 0.047524288296699524, "learning_rate": 1.9672375357206452e-07, "loss": 0.0206, "step": 58390 }, { "epoch": 26.13560080554934, "grad_norm": 0.06464558839797974, "learning_rate": 1.942891485762044e-07, "loss": 0.0246, "step": 58400 }, { "epoch": 26.140076079659877, "grad_norm": 0.059254005551338196, "learning_rate": 1.918696732402636e-07, "loss": 0.0237, "step": 58410 }, { "epoch": 26.144551353770417, "grad_norm": 0.15276814997196198, "learning_rate": 1.8946532829920426e-07, "loss": 0.0203, "step": 58420 }, { "epoch": 26.149026627880957, "grad_norm": 0.06135411933064461, "learning_rate": 1.870761144834088e-07, "loss": 0.0217, "step": 58430 }, { "epoch": 26.153501901991497, "grad_norm": 0.06562868505716324, "learning_rate": 1.847020325186577e-07, "loss": 0.0216, "step": 58440 }, { "epoch": 26.157977176102037, "grad_norm": 0.05730767920613289, "learning_rate": 1.8234308312612968e-07, "loss": 0.0212, "step": 58450 }, { "epoch": 26.162452450212577, "grad_norm": 0.09145381301641464, "learning_rate": 1.799992670224182e-07, "loss": 0.0232, "step": 58460 }, { "epoch": 26.166927724323116, "grad_norm": 0.053647637367248535, "learning_rate": 1.776705849195037e-07, "loss": 0.0192, "step": 58470 }, { "epoch": 26.171402998433653, "grad_norm": 0.05717506632208824, "learning_rate": 1.753570375247815e-07, "loss": 0.0213, "step": 58480 }, { "epoch": 26.175878272544193, "grad_norm": 0.061378251761198044, "learning_rate": 1.73058625541056e-07, "loss": 0.0225, "step": 58490 }, { "epoch": 26.180353546654732, "grad_norm": 0.06538954377174377, "learning_rate": 1.7077534966650766e-07, "loss": 0.0222, "step": 58500 }, { "epoch": 26.184828820765272, "grad_norm": 0.056034669280052185, "learning_rate": 1.6850721059474827e-07, "loss": 0.0248, "step": 58510 }, { "epoch": 26.189304094875812, "grad_norm": 0.07992403209209442, "learning_rate": 1.662542090147712e-07, "loss": 0.0238, "step": 58520 }, { "epoch": 26.193779368986352, "grad_norm": 0.08174225687980652, "learning_rate": 1.6401634561098444e-07, "loss": 0.0254, "step": 58530 }, { "epoch": 26.198254643096888, "grad_norm": 0.07148127257823944, "learning_rate": 1.6179362106318874e-07, "loss": 0.0198, "step": 58540 }, { "epoch": 26.202729917207428, "grad_norm": 0.07896694540977478, "learning_rate": 1.5958603604658838e-07, "loss": 0.0191, "step": 58550 }, { "epoch": 26.207205191317968, "grad_norm": 0.07047120481729507, "learning_rate": 1.5739359123178587e-07, "loss": 0.0224, "step": 58560 }, { "epoch": 26.211680465428508, "grad_norm": 0.06681754440069199, "learning_rate": 1.5521628728479843e-07, "loss": 0.0257, "step": 58570 }, { "epoch": 26.216155739539047, "grad_norm": 0.07776746898889542, "learning_rate": 1.5305412486702474e-07, "loss": 0.023, "step": 58580 }, { "epoch": 26.220631013649587, "grad_norm": 0.07226645201444626, "learning_rate": 1.5090710463527836e-07, "loss": 0.0225, "step": 58590 }, { "epoch": 26.225106287760127, "grad_norm": 0.06434139609336853, "learning_rate": 1.4877522724175973e-07, "loss": 0.0222, "step": 58600 }, { "epoch": 26.229581561870663, "grad_norm": 0.06106953322887421, "learning_rate": 1.4665849333408976e-07, "loss": 0.0197, "step": 58610 }, { "epoch": 26.234056835981203, "grad_norm": 0.04300588741898537, "learning_rate": 1.4455690355525964e-07, "loss": 0.0196, "step": 58620 }, { "epoch": 26.238532110091743, "grad_norm": 0.07763839513063431, "learning_rate": 1.42470458543692e-07, "loss": 0.0221, "step": 58630 }, { "epoch": 26.243007384202283, "grad_norm": 0.06847019493579865, "learning_rate": 1.4039915893318544e-07, "loss": 0.0227, "step": 58640 }, { "epoch": 26.247482658312823, "grad_norm": 0.07677921652793884, "learning_rate": 1.383430053529422e-07, "loss": 0.0234, "step": 58650 }, { "epoch": 26.251957932423363, "grad_norm": 0.06887707859277725, "learning_rate": 1.3630199842758484e-07, "loss": 0.0252, "step": 58660 }, { "epoch": 26.2564332065339, "grad_norm": 0.07506993412971497, "learning_rate": 1.342761387770952e-07, "loss": 0.0224, "step": 58670 }, { "epoch": 26.26090848064444, "grad_norm": 0.0680224671959877, "learning_rate": 1.3226542701689215e-07, "loss": 0.0229, "step": 58680 }, { "epoch": 26.26538375475498, "grad_norm": 0.06771422177553177, "learning_rate": 1.3026986375776485e-07, "loss": 0.021, "step": 58690 }, { "epoch": 26.26985902886552, "grad_norm": 0.06614864617586136, "learning_rate": 1.2828944960592836e-07, "loss": 0.0228, "step": 58700 }, { "epoch": 26.274334302976058, "grad_norm": 0.05896759033203125, "learning_rate": 1.2632418516296262e-07, "loss": 0.0219, "step": 58710 }, { "epoch": 26.278809577086598, "grad_norm": 0.08669628202915192, "learning_rate": 1.243740710258734e-07, "loss": 0.0223, "step": 58720 }, { "epoch": 26.283284851197134, "grad_norm": 0.06502196937799454, "learning_rate": 1.2243910778705348e-07, "loss": 0.0231, "step": 58730 }, { "epoch": 26.287760125307674, "grad_norm": 0.061579860746860504, "learning_rate": 1.2051929603428825e-07, "loss": 0.0205, "step": 58740 }, { "epoch": 26.292235399418214, "grad_norm": 0.0753893107175827, "learning_rate": 1.1861463635077785e-07, "loss": 0.0188, "step": 58750 }, { "epoch": 26.296710673528754, "grad_norm": 0.07319600135087967, "learning_rate": 1.1672512931509283e-07, "loss": 0.0224, "step": 58760 }, { "epoch": 26.301185947639294, "grad_norm": 0.09119873493909836, "learning_rate": 1.1485077550122402e-07, "loss": 0.0243, "step": 58770 }, { "epoch": 26.305661221749833, "grad_norm": 0.06299026310443878, "learning_rate": 1.1299157547854377e-07, "loss": 0.0209, "step": 58780 }, { "epoch": 26.310136495860373, "grad_norm": 0.06686893850564957, "learning_rate": 1.1114752981183917e-07, "loss": 0.0255, "step": 58790 }, { "epoch": 26.31461176997091, "grad_norm": 0.07171625643968582, "learning_rate": 1.0931863906127327e-07, "loss": 0.0237, "step": 58800 }, { "epoch": 26.31908704408145, "grad_norm": 0.10314973443746567, "learning_rate": 1.0750490378241828e-07, "loss": 0.0209, "step": 58810 }, { "epoch": 26.32356231819199, "grad_norm": 0.06685611605644226, "learning_rate": 1.0570632452623353e-07, "loss": 0.0259, "step": 58820 }, { "epoch": 26.32803759230253, "grad_norm": 0.07183421403169632, "learning_rate": 1.0392290183909304e-07, "loss": 0.0242, "step": 58830 }, { "epoch": 26.33251286641307, "grad_norm": 0.05232860520482063, "learning_rate": 1.0215463626274125e-07, "loss": 0.0237, "step": 58840 }, { "epoch": 26.33698814052361, "grad_norm": 0.049633994698524475, "learning_rate": 1.0040152833433736e-07, "loss": 0.0217, "step": 58850 }, { "epoch": 26.341463414634145, "grad_norm": 0.07174717634916306, "learning_rate": 9.866357858642205e-08, "loss": 0.0242, "step": 58860 }, { "epoch": 26.345938688744685, "grad_norm": 0.06676443666219711, "learning_rate": 9.694078754694524e-08, "loss": 0.0222, "step": 58870 }, { "epoch": 26.350413962855225, "grad_norm": 0.06434806436300278, "learning_rate": 9.523315573924385e-08, "loss": 0.0202, "step": 58880 }, { "epoch": 26.354889236965764, "grad_norm": 0.06560681760311127, "learning_rate": 9.354068368204739e-08, "loss": 0.0237, "step": 58890 }, { "epoch": 26.359364511076304, "grad_norm": 0.07899449020624161, "learning_rate": 9.186337188949457e-08, "loss": 0.0233, "step": 58900 }, { "epoch": 26.363839785186844, "grad_norm": 0.07699385285377502, "learning_rate": 9.02012208710945e-08, "loss": 0.0217, "step": 58910 }, { "epoch": 26.36831505929738, "grad_norm": 0.06213834509253502, "learning_rate": 8.855423113177664e-08, "loss": 0.0196, "step": 58920 }, { "epoch": 26.37279033340792, "grad_norm": 0.08124905079603195, "learning_rate": 8.69224031718463e-08, "loss": 0.0228, "step": 58930 }, { "epoch": 26.37726560751846, "grad_norm": 0.06966391205787659, "learning_rate": 8.530573748701253e-08, "loss": 0.0228, "step": 58940 }, { "epoch": 26.381740881629, "grad_norm": 0.06352026760578156, "learning_rate": 8.37042345683714e-08, "loss": 0.0234, "step": 58950 }, { "epoch": 26.38621615573954, "grad_norm": 0.06371227651834488, "learning_rate": 8.211789490242261e-08, "loss": 0.0247, "step": 58960 }, { "epoch": 26.39069142985008, "grad_norm": 0.0687239021062851, "learning_rate": 8.054671897104738e-08, "loss": 0.0228, "step": 58970 }, { "epoch": 26.39516670396062, "grad_norm": 0.05256184935569763, "learning_rate": 7.899070725153613e-08, "loss": 0.0212, "step": 58980 }, { "epoch": 26.399641978071156, "grad_norm": 0.08363822102546692, "learning_rate": 7.744986021656076e-08, "loss": 0.0235, "step": 58990 }, { "epoch": 26.404117252181695, "grad_norm": 0.05136530101299286, "learning_rate": 7.59241783341913e-08, "loss": 0.0237, "step": 59000 }, { "epoch": 26.408592526292235, "grad_norm": 0.06615179032087326, "learning_rate": 7.44136620678848e-08, "loss": 0.0242, "step": 59010 }, { "epoch": 26.413067800402775, "grad_norm": 0.04483947902917862, "learning_rate": 7.291831187649645e-08, "loss": 0.0217, "step": 59020 }, { "epoch": 26.417543074513315, "grad_norm": 0.06423760950565338, "learning_rate": 7.143812821427953e-08, "loss": 0.0252, "step": 59030 }, { "epoch": 26.422018348623855, "grad_norm": 0.06592671573162079, "learning_rate": 6.997311153086883e-08, "loss": 0.0216, "step": 59040 }, { "epoch": 26.42649362273439, "grad_norm": 0.06237265467643738, "learning_rate": 6.852326227130834e-08, "loss": 0.023, "step": 59050 }, { "epoch": 26.43096889684493, "grad_norm": 0.10880421847105026, "learning_rate": 6.708858087601244e-08, "loss": 0.022, "step": 59060 }, { "epoch": 26.43544417095547, "grad_norm": 0.051121585071086884, "learning_rate": 6.566906778079917e-08, "loss": 0.0215, "step": 59070 }, { "epoch": 26.43991944506601, "grad_norm": 0.0690193921327591, "learning_rate": 6.426472341689027e-08, "loss": 0.0206, "step": 59080 }, { "epoch": 26.44439471917655, "grad_norm": 0.06857118010520935, "learning_rate": 6.287554821087783e-08, "loss": 0.0247, "step": 59090 }, { "epoch": 26.44886999328709, "grad_norm": 0.06529562175273895, "learning_rate": 6.150154258476315e-08, "loss": 0.0236, "step": 59100 }, { "epoch": 26.453345267397626, "grad_norm": 0.05311225727200508, "learning_rate": 6.014270695592905e-08, "loss": 0.0236, "step": 59110 }, { "epoch": 26.457820541508166, "grad_norm": 0.054983094334602356, "learning_rate": 5.8799041737150896e-08, "loss": 0.0204, "step": 59120 }, { "epoch": 26.462295815618706, "grad_norm": 0.06641404330730438, "learning_rate": 5.747054733660773e-08, "loss": 0.0244, "step": 59130 }, { "epoch": 26.466771089729246, "grad_norm": 0.06570218503475189, "learning_rate": 5.615722415785451e-08, "loss": 0.0225, "step": 59140 }, { "epoch": 26.471246363839786, "grad_norm": 0.050688620656728745, "learning_rate": 5.485907259984435e-08, "loss": 0.0222, "step": 59150 }, { "epoch": 26.475721637950326, "grad_norm": 0.05588148161768913, "learning_rate": 5.3576093056922906e-08, "loss": 0.0228, "step": 59160 }, { "epoch": 26.480196912060865, "grad_norm": 0.060879964381456375, "learning_rate": 5.2308285918828415e-08, "loss": 0.0209, "step": 59170 }, { "epoch": 26.4846721861714, "grad_norm": 0.057583075016736984, "learning_rate": 5.105565157068615e-08, "loss": 0.0246, "step": 59180 }, { "epoch": 26.48914746028194, "grad_norm": 0.052534546703100204, "learning_rate": 4.981819039300284e-08, "loss": 0.0223, "step": 59190 }, { "epoch": 26.49362273439248, "grad_norm": 0.050794366747140884, "learning_rate": 4.859590276170556e-08, "loss": 0.0218, "step": 59200 }, { "epoch": 26.49809800850302, "grad_norm": 0.05678899958729744, "learning_rate": 4.7388789048075086e-08, "loss": 0.0216, "step": 59210 }, { "epoch": 26.50257328261356, "grad_norm": 0.07461439818143845, "learning_rate": 4.619684961881254e-08, "loss": 0.0215, "step": 59220 }, { "epoch": 26.5070485567241, "grad_norm": 0.07537055015563965, "learning_rate": 4.502008483598941e-08, "loss": 0.0239, "step": 59230 }, { "epoch": 26.511523830834637, "grad_norm": 0.051813021302223206, "learning_rate": 4.385849505708084e-08, "loss": 0.0211, "step": 59240 }, { "epoch": 26.515999104945177, "grad_norm": 0.08162394911050797, "learning_rate": 4.2712080634949024e-08, "loss": 0.0239, "step": 59250 }, { "epoch": 26.520474379055717, "grad_norm": 0.07307862490415573, "learning_rate": 4.158084191783762e-08, "loss": 0.0204, "step": 59260 }, { "epoch": 26.524949653166257, "grad_norm": 0.06319037824869156, "learning_rate": 4.046477924939396e-08, "loss": 0.0224, "step": 59270 }, { "epoch": 26.529424927276796, "grad_norm": 0.05866989865899086, "learning_rate": 3.936389296864129e-08, "loss": 0.0231, "step": 59280 }, { "epoch": 26.533900201387336, "grad_norm": 0.05466695874929428, "learning_rate": 3.827818341000655e-08, "loss": 0.0248, "step": 59290 }, { "epoch": 26.538375475497872, "grad_norm": 0.08271347731351852, "learning_rate": 3.720765090329814e-08, "loss": 0.025, "step": 59300 }, { "epoch": 26.542850749608412, "grad_norm": 0.05553743988275528, "learning_rate": 3.615229577371149e-08, "loss": 0.0223, "step": 59310 }, { "epoch": 26.547326023718952, "grad_norm": 0.04895065352320671, "learning_rate": 3.511211834184014e-08, "loss": 0.0219, "step": 59320 }, { "epoch": 26.551801297829492, "grad_norm": 0.052615560591220856, "learning_rate": 3.4087118923659125e-08, "loss": 0.0241, "step": 59330 }, { "epoch": 26.55627657194003, "grad_norm": 0.06700743734836578, "learning_rate": 3.3077297830541584e-08, "loss": 0.0184, "step": 59340 }, { "epoch": 26.56075184605057, "grad_norm": 0.05515635758638382, "learning_rate": 3.20826553692366e-08, "loss": 0.0205, "step": 59350 }, { "epoch": 26.56522712016111, "grad_norm": 0.06251371651887894, "learning_rate": 3.110319184189692e-08, "loss": 0.0236, "step": 59360 }, { "epoch": 26.569702394271648, "grad_norm": 0.07180698215961456, "learning_rate": 3.01389075460512e-08, "loss": 0.0248, "step": 59370 }, { "epoch": 26.574177668382188, "grad_norm": 0.09491918236017227, "learning_rate": 2.9189802774631792e-08, "loss": 0.0217, "step": 59380 }, { "epoch": 26.578652942492727, "grad_norm": 0.05813685432076454, "learning_rate": 2.8255877815946963e-08, "loss": 0.0222, "step": 59390 }, { "epoch": 26.583128216603267, "grad_norm": 0.06120353192090988, "learning_rate": 2.7337132953697554e-08, "loss": 0.0202, "step": 59400 }, { "epoch": 26.587603490713807, "grad_norm": 0.05727482959628105, "learning_rate": 2.6433568466976978e-08, "loss": 0.0246, "step": 59410 }, { "epoch": 26.592078764824347, "grad_norm": 0.07771178334951401, "learning_rate": 2.5545184630265672e-08, "loss": 0.0246, "step": 59420 }, { "epoch": 26.596554038934883, "grad_norm": 0.04921739175915718, "learning_rate": 2.467198171342e-08, "loss": 0.0196, "step": 59430 }, { "epoch": 26.601029313045423, "grad_norm": 0.07105813920497894, "learning_rate": 2.3813959981711097e-08, "loss": 0.0221, "step": 59440 }, { "epoch": 26.605504587155963, "grad_norm": 0.0673651322722435, "learning_rate": 2.2971119695774925e-08, "loss": 0.0195, "step": 59450 }, { "epoch": 26.609979861266503, "grad_norm": 0.07278452813625336, "learning_rate": 2.214346111164556e-08, "loss": 0.024, "step": 59460 }, { "epoch": 26.614455135377042, "grad_norm": 0.06796965003013611, "learning_rate": 2.133098448074411e-08, "loss": 0.0225, "step": 59470 }, { "epoch": 26.618930409487582, "grad_norm": 0.05786443129181862, "learning_rate": 2.0533690049878707e-08, "loss": 0.0204, "step": 59480 }, { "epoch": 26.623405683598122, "grad_norm": 0.0721493512392044, "learning_rate": 1.9751578061244504e-08, "loss": 0.023, "step": 59490 }, { "epoch": 26.62788095770866, "grad_norm": 0.06310580670833588, "learning_rate": 1.8984648752429225e-08, "loss": 0.0204, "step": 59500 }, { "epoch": 26.632356231819198, "grad_norm": 0.054285719990730286, "learning_rate": 1.8232902356407622e-08, "loss": 0.0206, "step": 59510 }, { "epoch": 26.636831505929738, "grad_norm": 0.06012050434947014, "learning_rate": 1.749633910153592e-08, "loss": 0.022, "step": 59520 }, { "epoch": 26.641306780040278, "grad_norm": 0.0827544629573822, "learning_rate": 1.6774959211568465e-08, "loss": 0.0211, "step": 59530 }, { "epoch": 26.645782054150818, "grad_norm": 0.0696718841791153, "learning_rate": 1.6068762905635527e-08, "loss": 0.0215, "step": 59540 }, { "epoch": 26.650257328261358, "grad_norm": 0.07523687928915024, "learning_rate": 1.5377750398265502e-08, "loss": 0.0255, "step": 59550 }, { "epoch": 26.654732602371894, "grad_norm": 0.09217715263366699, "learning_rate": 1.4701921899362703e-08, "loss": 0.0226, "step": 59560 }, { "epoch": 26.659207876482434, "grad_norm": 0.07120350003242493, "learning_rate": 1.4041277614235127e-08, "loss": 0.0255, "step": 59570 }, { "epoch": 26.663683150592973, "grad_norm": 0.040611185133457184, "learning_rate": 1.3395817743561134e-08, "loss": 0.0211, "step": 59580 }, { "epoch": 26.668158424703513, "grad_norm": 0.07435055077075958, "learning_rate": 1.2765542483417214e-08, "loss": 0.0233, "step": 59590 }, { "epoch": 26.672633698814053, "grad_norm": 0.061821430921554565, "learning_rate": 1.215045202527243e-08, "loss": 0.0245, "step": 59600 }, { "epoch": 26.677108972924593, "grad_norm": 0.06730903685092926, "learning_rate": 1.1550546555960662e-08, "loss": 0.0203, "step": 59610 }, { "epoch": 26.68158424703513, "grad_norm": 0.09366670995950699, "learning_rate": 1.096582625772502e-08, "loss": 0.0211, "step": 59620 }, { "epoch": 26.68605952114567, "grad_norm": 0.06460500508546829, "learning_rate": 1.0396291308190087e-08, "loss": 0.0225, "step": 59630 }, { "epoch": 26.69053479525621, "grad_norm": 0.06131056323647499, "learning_rate": 9.841941880361916e-09, "loss": 0.0212, "step": 59640 }, { "epoch": 26.69501006936675, "grad_norm": 0.07516928762197495, "learning_rate": 9.302778142639135e-09, "loss": 0.0255, "step": 59650 }, { "epoch": 26.69948534347729, "grad_norm": 0.05964084342122078, "learning_rate": 8.778800258801844e-09, "loss": 0.0223, "step": 59660 }, { "epoch": 26.70396061758783, "grad_norm": 0.056050751358270645, "learning_rate": 8.270008388022721e-09, "loss": 0.0234, "step": 59670 }, { "epoch": 26.708435891698368, "grad_norm": 0.07037303596735, "learning_rate": 7.77640268486146e-09, "loss": 0.0215, "step": 59680 }, { "epoch": 26.712911165808904, "grad_norm": 0.07427025586366653, "learning_rate": 7.2979832992592365e-09, "loss": 0.0267, "step": 59690 }, { "epoch": 26.717386439919444, "grad_norm": 0.07832662016153336, "learning_rate": 6.834750376549792e-09, "loss": 0.0218, "step": 59700 }, { "epoch": 26.721861714029984, "grad_norm": 0.06600495427846909, "learning_rate": 6.386704057453896e-09, "loss": 0.0237, "step": 59710 }, { "epoch": 26.726336988140524, "grad_norm": 0.061857253313064575, "learning_rate": 5.953844478068238e-09, "loss": 0.0207, "step": 59720 }, { "epoch": 26.730812262251064, "grad_norm": 0.06441668421030045, "learning_rate": 5.536171769887632e-09, "loss": 0.0248, "step": 59730 }, { "epoch": 26.735287536361604, "grad_norm": 0.07410154491662979, "learning_rate": 5.133686059793918e-09, "loss": 0.0245, "step": 59740 }, { "epoch": 26.73976281047214, "grad_norm": 0.06976128369569778, "learning_rate": 4.746387470044855e-09, "loss": 0.0205, "step": 59750 }, { "epoch": 26.74423808458268, "grad_norm": 0.06970130652189255, "learning_rate": 4.3742761183018784e-09, "loss": 0.0246, "step": 59760 }, { "epoch": 26.74871335869322, "grad_norm": 0.053498461842536926, "learning_rate": 4.017352117591244e-09, "loss": 0.0222, "step": 59770 }, { "epoch": 26.75318863280376, "grad_norm": 0.06728889793157578, "learning_rate": 3.6756155763373323e-09, "loss": 0.0224, "step": 59780 }, { "epoch": 26.7576639069143, "grad_norm": 0.05588546767830849, "learning_rate": 3.349066598362649e-09, "loss": 0.0233, "step": 59790 }, { "epoch": 26.76213918102484, "grad_norm": 0.06189928576350212, "learning_rate": 3.0377052828489683e-09, "loss": 0.0212, "step": 59800 }, { "epoch": 26.766614455135375, "grad_norm": 0.07702004909515381, "learning_rate": 2.741531724392843e-09, "loss": 0.0255, "step": 59810 }, { "epoch": 26.771089729245915, "grad_norm": 0.05724051594734192, "learning_rate": 2.4605460129556445e-09, "loss": 0.0203, "step": 59820 }, { "epoch": 26.775565003356455, "grad_norm": 0.055842164903879166, "learning_rate": 2.1947482338968705e-09, "loss": 0.0219, "step": 59830 }, { "epoch": 26.780040277466995, "grad_norm": 0.08473437279462814, "learning_rate": 1.9441384679574903e-09, "loss": 0.0244, "step": 59840 }, { "epoch": 26.784515551577535, "grad_norm": 0.05445104092359543, "learning_rate": 1.7087167912710478e-09, "loss": 0.0199, "step": 59850 }, { "epoch": 26.788990825688074, "grad_norm": 0.06051673740148544, "learning_rate": 1.4884832753414569e-09, "loss": 0.0227, "step": 59860 }, { "epoch": 26.793466099798614, "grad_norm": 0.08454134315252304, "learning_rate": 1.2834379870818592e-09, "loss": 0.0265, "step": 59870 }, { "epoch": 26.79794137390915, "grad_norm": 0.06696601212024689, "learning_rate": 1.0935809887702154e-09, "loss": 0.0255, "step": 59880 }, { "epoch": 26.80241664801969, "grad_norm": 0.06742631644010544, "learning_rate": 9.189123380826114e-10, "loss": 0.0211, "step": 59890 }, { "epoch": 26.80689192213023, "grad_norm": 0.06380529701709747, "learning_rate": 7.594320880821571e-10, "loss": 0.0203, "step": 59900 }, { "epoch": 26.81136719624077, "grad_norm": 0.07061690092086792, "learning_rate": 6.151402872134337e-10, "loss": 0.0245, "step": 59910 }, { "epoch": 26.81584247035131, "grad_norm": 0.07539287209510803, "learning_rate": 4.860369793080466e-10, "loss": 0.0203, "step": 59920 }, { "epoch": 26.82031774446185, "grad_norm": 0.07547426968812943, "learning_rate": 3.721222035846239e-10, "loss": 0.0214, "step": 59930 }, { "epoch": 26.824793018572386, "grad_norm": 0.06281965225934982, "learning_rate": 2.7339599464326627e-10, "loss": 0.0214, "step": 59940 }, { "epoch": 26.829268292682926, "grad_norm": 0.06106727197766304, "learning_rate": 1.8985838248219978e-10, "loss": 0.022, "step": 59950 }, { "epoch": 26.833743566793466, "grad_norm": 0.06055526062846184, "learning_rate": 1.2150939247002058e-10, "loss": 0.0237, "step": 59960 }, { "epoch": 26.838218840904005, "grad_norm": 0.05976271256804466, "learning_rate": 6.834904537900144e-11, "loss": 0.0245, "step": 59970 }, { "epoch": 26.842694115014545, "grad_norm": 0.0725957453250885, "learning_rate": 3.037735734623404e-11, "loss": 0.0228, "step": 59980 }, { "epoch": 26.847169389125085, "grad_norm": 0.0691622793674469, "learning_rate": 7.594339912486703e-12, "loss": 0.0209, "step": 59990 }, { "epoch": 26.851644663235625, "grad_norm": 0.06690680980682373, "learning_rate": 0.0, "loss": 0.0213, "step": 60000 }, { "epoch": 26.851644663235625, "step": 60000, "total_flos": 2.48759038144512e+19, "train_loss": 0.05932763454367717, "train_runtime": 112442.0393, "train_samples_per_second": 68.302, "train_steps_per_second": 0.534 } ], "logging_steps": 10, "max_steps": 60000, "num_input_tokens_seen": 0, "num_train_epochs": 27, "save_steps": 1000, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 2.48759038144512e+19, "train_batch_size": 64, "trial_name": null, "trial_params": null }