| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.9977298524404086, | |
| "eval_steps": 500, | |
| "global_step": 660, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.003026863412788498, | |
| "grad_norm": 2.428103541813777, | |
| "learning_rate": 3.0303030303030305e-07, | |
| "loss": 0.6586, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.006053726825576996, | |
| "grad_norm": 2.5601497538714595, | |
| "learning_rate": 6.060606060606061e-07, | |
| "loss": 0.6737, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.009080590238365494, | |
| "grad_norm": 2.5096338543908274, | |
| "learning_rate": 9.090909090909091e-07, | |
| "loss": 0.6566, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.012107453651153992, | |
| "grad_norm": 2.4300470112462405, | |
| "learning_rate": 1.2121212121212122e-06, | |
| "loss": 0.6654, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.01513431706394249, | |
| "grad_norm": 2.4815072083554974, | |
| "learning_rate": 1.5151515151515152e-06, | |
| "loss": 0.675, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.018161180476730987, | |
| "grad_norm": 2.235626657391899, | |
| "learning_rate": 1.8181818181818183e-06, | |
| "loss": 0.6246, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.021188043889519486, | |
| "grad_norm": 2.0821007799777336, | |
| "learning_rate": 2.1212121212121216e-06, | |
| "loss": 0.6654, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.024214907302307985, | |
| "grad_norm": 1.8856706562493897, | |
| "learning_rate": 2.4242424242424244e-06, | |
| "loss": 0.6257, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.02724177071509648, | |
| "grad_norm": 1.7552483826707568, | |
| "learning_rate": 2.7272727272727272e-06, | |
| "loss": 0.6143, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.03026863412788498, | |
| "grad_norm": 1.7097151847992857, | |
| "learning_rate": 3.0303030303030305e-06, | |
| "loss": 0.6137, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.03329549754067348, | |
| "grad_norm": 1.7303379955902458, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 0.6011, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.036322360953461974, | |
| "grad_norm": 1.3895462219949413, | |
| "learning_rate": 3.6363636363636366e-06, | |
| "loss": 0.5715, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.03934922436625047, | |
| "grad_norm": 1.7691425641001932, | |
| "learning_rate": 3.93939393939394e-06, | |
| "loss": 0.5547, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.04237608777903897, | |
| "grad_norm": 1.6767314165523832, | |
| "learning_rate": 4.242424242424243e-06, | |
| "loss": 0.5648, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.04540295119182747, | |
| "grad_norm": 1.5820678541209858, | |
| "learning_rate": 4.5454545454545455e-06, | |
| "loss": 0.5598, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.04842981460461597, | |
| "grad_norm": 1.3724389051441108, | |
| "learning_rate": 4.848484848484849e-06, | |
| "loss": 0.5425, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.051456678017404466, | |
| "grad_norm": 1.2916651875538034, | |
| "learning_rate": 5.151515151515152e-06, | |
| "loss": 0.5237, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.05448354143019296, | |
| "grad_norm": 1.2604756576646161, | |
| "learning_rate": 5.4545454545454545e-06, | |
| "loss": 0.5296, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.057510404842981463, | |
| "grad_norm": 1.0397582907217255, | |
| "learning_rate": 5.7575757575757586e-06, | |
| "loss": 0.5105, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.06053726825576996, | |
| "grad_norm": 1.0096253325018365, | |
| "learning_rate": 6.060606060606061e-06, | |
| "loss": 0.5036, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.06356413166855845, | |
| "grad_norm": 1.0227826413788967, | |
| "learning_rate": 6.363636363636364e-06, | |
| "loss": 0.4835, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.06659099508134696, | |
| "grad_norm": 0.9294379959145279, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 0.4715, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.06961785849413545, | |
| "grad_norm": 0.9152055479199039, | |
| "learning_rate": 6.969696969696971e-06, | |
| "loss": 0.4934, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.07264472190692395, | |
| "grad_norm": 0.849182736987751, | |
| "learning_rate": 7.272727272727273e-06, | |
| "loss": 0.4808, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.07567158531971245, | |
| "grad_norm": 0.8677907650013278, | |
| "learning_rate": 7.5757575757575764e-06, | |
| "loss": 0.4832, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.07869844873250094, | |
| "grad_norm": 0.8863910923878754, | |
| "learning_rate": 7.87878787878788e-06, | |
| "loss": 0.4749, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.08172531214528944, | |
| "grad_norm": 0.8845029088973714, | |
| "learning_rate": 8.181818181818183e-06, | |
| "loss": 0.4831, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.08475217555807794, | |
| "grad_norm": 0.8629990435950339, | |
| "learning_rate": 8.484848484848486e-06, | |
| "loss": 0.4682, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.08777903897086645, | |
| "grad_norm": 0.8107329239073842, | |
| "learning_rate": 8.787878787878788e-06, | |
| "loss": 0.444, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.09080590238365494, | |
| "grad_norm": 0.8126339091203556, | |
| "learning_rate": 9.090909090909091e-06, | |
| "loss": 0.4426, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.09383276579644344, | |
| "grad_norm": 0.816122400966661, | |
| "learning_rate": 9.393939393939396e-06, | |
| "loss": 0.4582, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.09685962920923194, | |
| "grad_norm": 0.8351290090954328, | |
| "learning_rate": 9.696969696969698e-06, | |
| "loss": 0.4513, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.09988649262202043, | |
| "grad_norm": 0.8217783192034996, | |
| "learning_rate": 1e-05, | |
| "loss": 0.4425, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.10291335603480893, | |
| "grad_norm": 0.8216020465586724, | |
| "learning_rate": 1.0303030303030304e-05, | |
| "loss": 0.456, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.10594021944759743, | |
| "grad_norm": 0.8336333255012085, | |
| "learning_rate": 1.0606060606060606e-05, | |
| "loss": 0.451, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.10896708286038592, | |
| "grad_norm": 0.8219815536573118, | |
| "learning_rate": 1.0909090909090909e-05, | |
| "loss": 0.4453, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.11199394627317442, | |
| "grad_norm": 0.7857049631308176, | |
| "learning_rate": 1.1212121212121212e-05, | |
| "loss": 0.4516, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.11502080968596293, | |
| "grad_norm": 0.7567995931553804, | |
| "learning_rate": 1.1515151515151517e-05, | |
| "loss": 0.4238, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.11804767309875142, | |
| "grad_norm": 0.7627363188750758, | |
| "learning_rate": 1.181818181818182e-05, | |
| "loss": 0.4075, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.12107453651153992, | |
| "grad_norm": 0.7863643769755297, | |
| "learning_rate": 1.2121212121212122e-05, | |
| "loss": 0.436, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.12410139992432842, | |
| "grad_norm": 0.8213966288402991, | |
| "learning_rate": 1.2424242424242425e-05, | |
| "loss": 0.4404, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.1271282633371169, | |
| "grad_norm": 0.7878188402036066, | |
| "learning_rate": 1.2727272727272728e-05, | |
| "loss": 0.426, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.1301551267499054, | |
| "grad_norm": 0.7964997167425242, | |
| "learning_rate": 1.3030303030303032e-05, | |
| "loss": 0.4135, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.13318199016269391, | |
| "grad_norm": 0.840619345023061, | |
| "learning_rate": 1.3333333333333333e-05, | |
| "loss": 0.4419, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.1362088535754824, | |
| "grad_norm": 0.8831744636996385, | |
| "learning_rate": 1.3636363636363637e-05, | |
| "loss": 0.4553, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.1392357169882709, | |
| "grad_norm": 0.7478942004714882, | |
| "learning_rate": 1.3939393939393942e-05, | |
| "loss": 0.4283, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.1422625804010594, | |
| "grad_norm": 0.852898617889127, | |
| "learning_rate": 1.4242424242424245e-05, | |
| "loss": 0.4202, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.1452894438138479, | |
| "grad_norm": 0.7315552124327294, | |
| "learning_rate": 1.4545454545454546e-05, | |
| "loss": 0.4214, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.14831630722663638, | |
| "grad_norm": 0.7815526821692528, | |
| "learning_rate": 1.484848484848485e-05, | |
| "loss": 0.4217, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.1513431706394249, | |
| "grad_norm": 0.7745351426425916, | |
| "learning_rate": 1.5151515151515153e-05, | |
| "loss": 0.4285, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.1543700340522134, | |
| "grad_norm": 0.7297305661967034, | |
| "learning_rate": 1.5454545454545454e-05, | |
| "loss": 0.4111, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.15739689746500188, | |
| "grad_norm": 0.7664527677957759, | |
| "learning_rate": 1.575757575757576e-05, | |
| "loss": 0.414, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.1604237608777904, | |
| "grad_norm": 0.7682824777118996, | |
| "learning_rate": 1.606060606060606e-05, | |
| "loss": 0.4081, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.16345062429057888, | |
| "grad_norm": 0.7937451763349684, | |
| "learning_rate": 1.6363636363636366e-05, | |
| "loss": 0.4263, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.16647748770336737, | |
| "grad_norm": 0.7428048901912144, | |
| "learning_rate": 1.6666666666666667e-05, | |
| "loss": 0.4156, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.1695043511161559, | |
| "grad_norm": 0.8484288433584792, | |
| "learning_rate": 1.6969696969696972e-05, | |
| "loss": 0.4112, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.17253121452894438, | |
| "grad_norm": 0.7538626267707941, | |
| "learning_rate": 1.7272727272727274e-05, | |
| "loss": 0.4199, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.1755580779417329, | |
| "grad_norm": 0.7872885651357554, | |
| "learning_rate": 1.7575757575757576e-05, | |
| "loss": 0.3954, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.17858494135452138, | |
| "grad_norm": 0.769159727928397, | |
| "learning_rate": 1.787878787878788e-05, | |
| "loss": 0.4061, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.18161180476730987, | |
| "grad_norm": 0.8271079646364261, | |
| "learning_rate": 1.8181818181818182e-05, | |
| "loss": 0.4136, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.1846386681800984, | |
| "grad_norm": 0.8437369879266646, | |
| "learning_rate": 1.8484848484848487e-05, | |
| "loss": 0.396, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.18766553159288688, | |
| "grad_norm": 0.891590475023802, | |
| "learning_rate": 1.8787878787878792e-05, | |
| "loss": 0.4248, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.19069239500567536, | |
| "grad_norm": 0.8697842519316581, | |
| "learning_rate": 1.9090909090909094e-05, | |
| "loss": 0.4082, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.19371925841846388, | |
| "grad_norm": 0.7396648075503486, | |
| "learning_rate": 1.9393939393939395e-05, | |
| "loss": 0.3999, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.19674612183125237, | |
| "grad_norm": 0.7775140157628547, | |
| "learning_rate": 1.96969696969697e-05, | |
| "loss": 0.4215, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.19977298524404086, | |
| "grad_norm": 0.7608337308971462, | |
| "learning_rate": 2e-05, | |
| "loss": 0.4117, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.20279984865682937, | |
| "grad_norm": 0.7336156585259265, | |
| "learning_rate": 1.9999860139251737e-05, | |
| "loss": 0.3898, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.20582671206961786, | |
| "grad_norm": 0.8086972113395839, | |
| "learning_rate": 1.9999440560919153e-05, | |
| "loss": 0.4157, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.20885357548240635, | |
| "grad_norm": 0.7413649165310676, | |
| "learning_rate": 1.9998741276738753e-05, | |
| "loss": 0.4064, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.21188043889519487, | |
| "grad_norm": 0.7648055695668046, | |
| "learning_rate": 1.999776230627102e-05, | |
| "loss": 0.4124, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.21490730230798336, | |
| "grad_norm": 0.7835974392039421, | |
| "learning_rate": 1.9996503676899863e-05, | |
| "loss": 0.382, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.21793416572077184, | |
| "grad_norm": 0.7472259204783522, | |
| "learning_rate": 1.9994965423831853e-05, | |
| "loss": 0.3978, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.22096102913356036, | |
| "grad_norm": 0.7979559621474958, | |
| "learning_rate": 1.9993147590095232e-05, | |
| "loss": 0.4099, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.22398789254634885, | |
| "grad_norm": 0.7257644961735806, | |
| "learning_rate": 1.999105022653872e-05, | |
| "loss": 0.3791, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.22701475595913734, | |
| "grad_norm": 0.7881673089769411, | |
| "learning_rate": 1.9988673391830082e-05, | |
| "loss": 0.4078, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.23004161937192585, | |
| "grad_norm": 0.7342495057184855, | |
| "learning_rate": 1.9986017152454497e-05, | |
| "loss": 0.3927, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.23306848278471434, | |
| "grad_norm": 0.8390545933448048, | |
| "learning_rate": 1.9983081582712684e-05, | |
| "loss": 0.405, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.23609534619750283, | |
| "grad_norm": 0.8256201748238297, | |
| "learning_rate": 1.9979866764718846e-05, | |
| "loss": 0.4184, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.23912220961029135, | |
| "grad_norm": 0.884374198318761, | |
| "learning_rate": 1.997637278839835e-05, | |
| "loss": 0.4047, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.24214907302307984, | |
| "grad_norm": 0.7923235172326851, | |
| "learning_rate": 1.9972599751485225e-05, | |
| "loss": 0.3923, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.24517593643586832, | |
| "grad_norm": 0.8804287394001027, | |
| "learning_rate": 1.9968547759519426e-05, | |
| "loss": 0.4277, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.24820279984865684, | |
| "grad_norm": 0.8682834106758025, | |
| "learning_rate": 1.9964216925843876e-05, | |
| "loss": 0.4028, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.2512296632614453, | |
| "grad_norm": 0.8056297461993274, | |
| "learning_rate": 1.9959607371601303e-05, | |
| "loss": 0.3987, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.2542565266742338, | |
| "grad_norm": 0.8902877724632622, | |
| "learning_rate": 1.9954719225730847e-05, | |
| "loss": 0.3943, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.25728339008702233, | |
| "grad_norm": 0.7564669222554182, | |
| "learning_rate": 1.994955262496446e-05, | |
| "loss": 0.387, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.2603102534998108, | |
| "grad_norm": 0.9111977264345676, | |
| "learning_rate": 1.9944107713823068e-05, | |
| "loss": 0.4087, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.2633371169125993, | |
| "grad_norm": 0.7346269506943043, | |
| "learning_rate": 1.9938384644612542e-05, | |
| "loss": 0.4057, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.26636398032538783, | |
| "grad_norm": 0.7532036814239714, | |
| "learning_rate": 1.9932383577419432e-05, | |
| "loss": 0.3992, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.2693908437381763, | |
| "grad_norm": 0.7593926698417056, | |
| "learning_rate": 1.9926104680106484e-05, | |
| "loss": 0.3896, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.2724177071509648, | |
| "grad_norm": 0.7922557315510264, | |
| "learning_rate": 1.9919548128307954e-05, | |
| "loss": 0.3993, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.2754445705637533, | |
| "grad_norm": 0.7476337670627412, | |
| "learning_rate": 1.9912714105424694e-05, | |
| "loss": 0.3907, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.2784714339765418, | |
| "grad_norm": 0.743087554975147, | |
| "learning_rate": 1.990560280261901e-05, | |
| "loss": 0.3885, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.2814982973893303, | |
| "grad_norm": 0.7865952947121178, | |
| "learning_rate": 1.989821441880933e-05, | |
| "loss": 0.4053, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.2845251608021188, | |
| "grad_norm": 0.7318279291862737, | |
| "learning_rate": 1.9890549160664633e-05, | |
| "loss": 0.3804, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.2875520242149073, | |
| "grad_norm": 0.7096446891497972, | |
| "learning_rate": 1.9882607242598663e-05, | |
| "loss": 0.3874, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.2905788876276958, | |
| "grad_norm": 0.7446422123643741, | |
| "learning_rate": 1.9874388886763944e-05, | |
| "loss": 0.3833, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.2936057510404843, | |
| "grad_norm": 0.7594149238162948, | |
| "learning_rate": 1.9865894323045558e-05, | |
| "loss": 0.3624, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.29663261445327277, | |
| "grad_norm": 0.766612658533511, | |
| "learning_rate": 1.9857123789054707e-05, | |
| "loss": 0.4021, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.2996594778660613, | |
| "grad_norm": 0.8080120468881522, | |
| "learning_rate": 1.9848077530122083e-05, | |
| "loss": 0.4027, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.3026863412788498, | |
| "grad_norm": 0.7634058991194524, | |
| "learning_rate": 1.9838755799290993e-05, | |
| "loss": 0.4145, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.30571320469163826, | |
| "grad_norm": 0.6866040002467878, | |
| "learning_rate": 1.9829158857310288e-05, | |
| "loss": 0.4046, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.3087400681044268, | |
| "grad_norm": 0.7600568718649394, | |
| "learning_rate": 1.9819286972627066e-05, | |
| "loss": 0.3849, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.3117669315172153, | |
| "grad_norm": 0.7517330790665169, | |
| "learning_rate": 1.9809140421379168e-05, | |
| "loss": 0.4112, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.31479379493000376, | |
| "grad_norm": 0.7886624739575433, | |
| "learning_rate": 1.979871948738743e-05, | |
| "loss": 0.3965, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.3178206583427923, | |
| "grad_norm": 0.790063452148931, | |
| "learning_rate": 1.978802446214779e-05, | |
| "loss": 0.391, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.3208475217555808, | |
| "grad_norm": 0.7341444872973905, | |
| "learning_rate": 1.9777055644823087e-05, | |
| "loss": 0.3939, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.32387438516836925, | |
| "grad_norm": 0.7601884556847633, | |
| "learning_rate": 1.9765813342234726e-05, | |
| "loss": 0.3817, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.32690124858115777, | |
| "grad_norm": 0.7353332510573537, | |
| "learning_rate": 1.9754297868854075e-05, | |
| "loss": 0.3921, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.3299281119939463, | |
| "grad_norm": 0.8561603012030907, | |
| "learning_rate": 1.9742509546793673e-05, | |
| "loss": 0.3926, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.33295497540673474, | |
| "grad_norm": 0.7173620169366063, | |
| "learning_rate": 1.973044870579824e-05, | |
| "loss": 0.3737, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.33598183881952326, | |
| "grad_norm": 0.798814194298797, | |
| "learning_rate": 1.9718115683235418e-05, | |
| "loss": 0.4027, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.3390087022323118, | |
| "grad_norm": 0.7255738483858067, | |
| "learning_rate": 1.970551082408636e-05, | |
| "loss": 0.3905, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.3420355656451003, | |
| "grad_norm": 0.7362272279126251, | |
| "learning_rate": 1.969263448093608e-05, | |
| "loss": 0.3883, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.34506242905788875, | |
| "grad_norm": 0.7372784791546305, | |
| "learning_rate": 1.9679487013963566e-05, | |
| "loss": 0.3952, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.34808929247067727, | |
| "grad_norm": 0.7073711939849762, | |
| "learning_rate": 1.9666068790931733e-05, | |
| "loss": 0.3883, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.3511161558834658, | |
| "grad_norm": 0.7790275300522554, | |
| "learning_rate": 1.9652380187177128e-05, | |
| "loss": 0.3845, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.35414301929625425, | |
| "grad_norm": 0.7114624279551119, | |
| "learning_rate": 1.9638421585599422e-05, | |
| "loss": 0.3878, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.35716988270904276, | |
| "grad_norm": 0.7356217714339921, | |
| "learning_rate": 1.9624193376650708e-05, | |
| "loss": 0.3661, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.3601967461218313, | |
| "grad_norm": 0.7947276642308108, | |
| "learning_rate": 1.960969595832457e-05, | |
| "loss": 0.3838, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.36322360953461974, | |
| "grad_norm": 0.7943547882672132, | |
| "learning_rate": 1.9594929736144978e-05, | |
| "loss": 0.3754, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.36625047294740826, | |
| "grad_norm": 0.8187677995957213, | |
| "learning_rate": 1.957989512315489e-05, | |
| "loss": 0.395, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.3692773363601968, | |
| "grad_norm": 0.797006155697655, | |
| "learning_rate": 1.956459253990476e-05, | |
| "loss": 0.3732, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.37230419977298523, | |
| "grad_norm": 0.80588549575878, | |
| "learning_rate": 1.9549022414440738e-05, | |
| "loss": 0.3906, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.37533106318577375, | |
| "grad_norm": 0.8622319578845932, | |
| "learning_rate": 1.9533185182292705e-05, | |
| "loss": 0.3828, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.37835792659856227, | |
| "grad_norm": 0.7054568213021077, | |
| "learning_rate": 1.9517081286462082e-05, | |
| "loss": 0.387, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.3813847900113507, | |
| "grad_norm": 0.7422730729587885, | |
| "learning_rate": 1.9500711177409456e-05, | |
| "loss": 0.3621, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.38441165342413924, | |
| "grad_norm": 0.7932468099938381, | |
| "learning_rate": 1.9484075313041968e-05, | |
| "loss": 0.4096, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.38743851683692776, | |
| "grad_norm": 0.7670152209817788, | |
| "learning_rate": 1.9467174158700507e-05, | |
| "loss": 0.3897, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.3904653802497162, | |
| "grad_norm": 0.7634268493368022, | |
| "learning_rate": 1.9450008187146685e-05, | |
| "loss": 0.399, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.39349224366250474, | |
| "grad_norm": 0.7601258191659126, | |
| "learning_rate": 1.9432577878549635e-05, | |
| "loss": 0.3776, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.39651910707529325, | |
| "grad_norm": 0.7587607568972272, | |
| "learning_rate": 1.9414883720472557e-05, | |
| "loss": 0.3672, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.3995459704880817, | |
| "grad_norm": 0.7135580822390496, | |
| "learning_rate": 1.9396926207859085e-05, | |
| "loss": 0.3853, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.40257283390087023, | |
| "grad_norm": 0.7727128454159812, | |
| "learning_rate": 1.937870584301945e-05, | |
| "loss": 0.4053, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.40559969731365875, | |
| "grad_norm": 0.7162277140695203, | |
| "learning_rate": 1.9360223135616423e-05, | |
| "loss": 0.3927, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.4086265607264472, | |
| "grad_norm": 0.7959540139488714, | |
| "learning_rate": 1.9341478602651068e-05, | |
| "loss": 0.4057, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.4116534241392357, | |
| "grad_norm": 0.6730082626062208, | |
| "learning_rate": 1.932247276844826e-05, | |
| "loss": 0.3826, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.41468028755202424, | |
| "grad_norm": 0.7618307251586336, | |
| "learning_rate": 1.9303206164642037e-05, | |
| "loss": 0.3906, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.4177071509648127, | |
| "grad_norm": 0.7319463979984874, | |
| "learning_rate": 1.9283679330160726e-05, | |
| "loss": 0.3702, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.4207340143776012, | |
| "grad_norm": 0.7243867360012308, | |
| "learning_rate": 1.9263892811211865e-05, | |
| "loss": 0.3804, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.42376087779038973, | |
| "grad_norm": 0.7382814276063332, | |
| "learning_rate": 1.9243847161266924e-05, | |
| "loss": 0.381, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.4267877412031782, | |
| "grad_norm": 0.7570499606848203, | |
| "learning_rate": 1.9223542941045817e-05, | |
| "loss": 0.3805, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.4298146046159667, | |
| "grad_norm": 0.7913070717518623, | |
| "learning_rate": 1.920298071850123e-05, | |
| "loss": 0.3657, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.4328414680287552, | |
| "grad_norm": 0.7608237312622164, | |
| "learning_rate": 1.9182161068802742e-05, | |
| "loss": 0.3857, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.4358683314415437, | |
| "grad_norm": 0.7411608926161769, | |
| "learning_rate": 1.9161084574320696e-05, | |
| "loss": 0.3821, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.4388951948543322, | |
| "grad_norm": 0.8229525529244064, | |
| "learning_rate": 1.913975182460996e-05, | |
| "loss": 0.3764, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.4419220582671207, | |
| "grad_norm": 0.7703135609790647, | |
| "learning_rate": 1.9118163416393392e-05, | |
| "loss": 0.3869, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.4449489216799092, | |
| "grad_norm": 0.7933193973425723, | |
| "learning_rate": 1.9096319953545186e-05, | |
| "loss": 0.3689, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.4479757850926977, | |
| "grad_norm": 0.7177385181344123, | |
| "learning_rate": 1.9074222047073945e-05, | |
| "loss": 0.3693, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.4510026485054862, | |
| "grad_norm": 0.7248220497591469, | |
| "learning_rate": 1.9051870315105626e-05, | |
| "loss": 0.3786, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.4540295119182747, | |
| "grad_norm": 0.8296842795221794, | |
| "learning_rate": 1.9029265382866216e-05, | |
| "loss": 0.3683, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.4570563753310632, | |
| "grad_norm": 0.7429609372907842, | |
| "learning_rate": 1.9006407882664256e-05, | |
| "loss": 0.3684, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.4600832387438517, | |
| "grad_norm": 0.7446381382116806, | |
| "learning_rate": 1.8983298453873172e-05, | |
| "loss": 0.3617, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.46311010215664017, | |
| "grad_norm": 0.6893002794799369, | |
| "learning_rate": 1.895993774291336e-05, | |
| "loss": 0.3743, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.4661369655694287, | |
| "grad_norm": 0.8295593682248998, | |
| "learning_rate": 1.8936326403234125e-05, | |
| "loss": 0.3848, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.4691638289822172, | |
| "grad_norm": 0.7148067334453264, | |
| "learning_rate": 1.891246509529539e-05, | |
| "loss": 0.3722, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.47219069239500566, | |
| "grad_norm": 0.7285526140808942, | |
| "learning_rate": 1.8888354486549238e-05, | |
| "loss": 0.3813, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.4752175558077942, | |
| "grad_norm": 0.8237411740255886, | |
| "learning_rate": 1.886399525142122e-05, | |
| "loss": 0.399, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.4782444192205827, | |
| "grad_norm": 0.7501149361649864, | |
| "learning_rate": 1.8839388071291506e-05, | |
| "loss": 0.3751, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.48127128263337116, | |
| "grad_norm": 0.7857562754729291, | |
| "learning_rate": 1.881453363447582e-05, | |
| "loss": 0.3845, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.48429814604615967, | |
| "grad_norm": 0.7144190856085049, | |
| "learning_rate": 1.8789432636206197e-05, | |
| "loss": 0.3734, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.4873250094589482, | |
| "grad_norm": 0.6908493046837273, | |
| "learning_rate": 1.8764085778611507e-05, | |
| "loss": 0.3646, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.49035187287173665, | |
| "grad_norm": 0.742378982943462, | |
| "learning_rate": 1.873849377069785e-05, | |
| "loss": 0.3775, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.49337873628452517, | |
| "grad_norm": 0.7481859920559335, | |
| "learning_rate": 1.87126573283287e-05, | |
| "loss": 0.3748, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.4964055996973137, | |
| "grad_norm": 0.7246225504695799, | |
| "learning_rate": 1.8686577174204887e-05, | |
| "loss": 0.3749, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.49943246311010214, | |
| "grad_norm": 0.7230319014676211, | |
| "learning_rate": 1.866025403784439e-05, | |
| "loss": 0.3777, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.5024593265228906, | |
| "grad_norm": 0.6945638978960651, | |
| "learning_rate": 1.863368865556191e-05, | |
| "loss": 0.3693, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.5054861899356792, | |
| "grad_norm": 0.6859730116933777, | |
| "learning_rate": 1.8606881770448305e-05, | |
| "loss": 0.3721, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.5085130533484676, | |
| "grad_norm": 0.7944202733860734, | |
| "learning_rate": 1.8579834132349773e-05, | |
| "loss": 0.3957, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.5115399167612561, | |
| "grad_norm": 0.696166640074152, | |
| "learning_rate": 1.8552546497846893e-05, | |
| "loss": 0.3613, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.5145667801740447, | |
| "grad_norm": 0.7304623420920909, | |
| "learning_rate": 1.8525019630233463e-05, | |
| "loss": 0.3667, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.5175936435868331, | |
| "grad_norm": 0.7608880889476206, | |
| "learning_rate": 1.8497254299495147e-05, | |
| "loss": 0.3843, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.5206205069996216, | |
| "grad_norm": 0.6630072610595168, | |
| "learning_rate": 1.8469251282287925e-05, | |
| "loss": 0.3662, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.5236473704124102, | |
| "grad_norm": 0.7479426133136007, | |
| "learning_rate": 1.8441011361916387e-05, | |
| "loss": 0.3766, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.5266742338251986, | |
| "grad_norm": 0.6710891403388963, | |
| "learning_rate": 1.8412535328311813e-05, | |
| "loss": 0.3706, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.5297010972379871, | |
| "grad_norm": 0.6600703343986968, | |
| "learning_rate": 1.8383823978010077e-05, | |
| "loss": 0.3596, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.5327279606507757, | |
| "grad_norm": 0.6812146535273996, | |
| "learning_rate": 1.8354878114129368e-05, | |
| "loss": 0.3783, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.5357548240635641, | |
| "grad_norm": 0.7024745998255092, | |
| "learning_rate": 1.8325698546347714e-05, | |
| "loss": 0.381, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.5387816874763526, | |
| "grad_norm": 0.7120837077801814, | |
| "learning_rate": 1.8296286090880362e-05, | |
| "loss": 0.3649, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.5418085508891411, | |
| "grad_norm": 0.7235763994066224, | |
| "learning_rate": 1.8266641570456915e-05, | |
| "loss": 0.3607, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.5448354143019296, | |
| "grad_norm": 0.6983674580977625, | |
| "learning_rate": 1.8236765814298328e-05, | |
| "loss": 0.3677, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.5478622777147181, | |
| "grad_norm": 0.6725158126254324, | |
| "learning_rate": 1.820665965809373e-05, | |
| "loss": 0.3522, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.5508891411275066, | |
| "grad_norm": 0.6780152987818565, | |
| "learning_rate": 1.8176323943977034e-05, | |
| "loss": 0.3698, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.5539160045402951, | |
| "grad_norm": 0.7169395077357664, | |
| "learning_rate": 1.814575952050336e-05, | |
| "loss": 0.3801, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.5569428679530836, | |
| "grad_norm": 0.6549786214132481, | |
| "learning_rate": 1.8114967242625342e-05, | |
| "loss": 0.351, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.5599697313658721, | |
| "grad_norm": 0.696050078631106, | |
| "learning_rate": 1.808394797166919e-05, | |
| "loss": 0.3701, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.5629965947786606, | |
| "grad_norm": 0.6848359849120773, | |
| "learning_rate": 1.8052702575310588e-05, | |
| "loss": 0.3643, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.5660234581914491, | |
| "grad_norm": 0.7160811454365867, | |
| "learning_rate": 1.802123192755044e-05, | |
| "loss": 0.3612, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.5690503216042376, | |
| "grad_norm": 0.7312079121937604, | |
| "learning_rate": 1.7989536908690413e-05, | |
| "loss": 0.3773, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.5720771850170261, | |
| "grad_norm": 0.6523585805396673, | |
| "learning_rate": 1.7957618405308323e-05, | |
| "loss": 0.3663, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.5751040484298146, | |
| "grad_norm": 0.6723328256885663, | |
| "learning_rate": 1.792547731023332e-05, | |
| "loss": 0.3511, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.5781309118426031, | |
| "grad_norm": 0.6440739198880174, | |
| "learning_rate": 1.789311452252092e-05, | |
| "loss": 0.3603, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.5811577752553916, | |
| "grad_norm": 0.7151416935868069, | |
| "learning_rate": 1.7860530947427878e-05, | |
| "loss": 0.3654, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.58418463866818, | |
| "grad_norm": 0.6667460312559599, | |
| "learning_rate": 1.782772749638682e-05, | |
| "loss": 0.3601, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.5872115020809686, | |
| "grad_norm": 0.7054520730051989, | |
| "learning_rate": 1.779470508698079e-05, | |
| "loss": 0.3533, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.5902383654937571, | |
| "grad_norm": 0.7029595791407055, | |
| "learning_rate": 1.776146464291757e-05, | |
| "loss": 0.3611, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.5932652289065455, | |
| "grad_norm": 0.6835966769610827, | |
| "learning_rate": 1.772800709400383e-05, | |
| "loss": 0.3596, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.5962920923193341, | |
| "grad_norm": 0.7003974205824403, | |
| "learning_rate": 1.7694333376119144e-05, | |
| "loss": 0.3611, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.5993189557321226, | |
| "grad_norm": 0.7503867605561271, | |
| "learning_rate": 1.766044443118978e-05, | |
| "loss": 0.3672, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.602345819144911, | |
| "grad_norm": 0.697404607145072, | |
| "learning_rate": 1.762634120716238e-05, | |
| "loss": 0.37, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.6053726825576996, | |
| "grad_norm": 0.7550721739261597, | |
| "learning_rate": 1.7592024657977432e-05, | |
| "loss": 0.3567, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.6083995459704881, | |
| "grad_norm": 0.6878714741128076, | |
| "learning_rate": 1.7557495743542586e-05, | |
| "loss": 0.3588, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.6114264093832765, | |
| "grad_norm": 0.6778387556209182, | |
| "learning_rate": 1.75227554297058e-05, | |
| "loss": 0.3638, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.6144532727960651, | |
| "grad_norm": 0.6903024899839635, | |
| "learning_rate": 1.7487804688228327e-05, | |
| "loss": 0.38, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.6174801362088536, | |
| "grad_norm": 0.6645184589627329, | |
| "learning_rate": 1.745264449675755e-05, | |
| "loss": 0.3827, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.620506999621642, | |
| "grad_norm": 0.6693005390708127, | |
| "learning_rate": 1.7417275838799596e-05, | |
| "loss": 0.3733, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.6235338630344306, | |
| "grad_norm": 0.6906667930303119, | |
| "learning_rate": 1.7381699703691866e-05, | |
| "loss": 0.3806, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.626560726447219, | |
| "grad_norm": 0.6791274289737517, | |
| "learning_rate": 1.734591708657533e-05, | |
| "loss": 0.3469, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.6295875898600075, | |
| "grad_norm": 0.66553108298308, | |
| "learning_rate": 1.730992898836672e-05, | |
| "loss": 0.3787, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.6326144532727961, | |
| "grad_norm": 0.6368465235452057, | |
| "learning_rate": 1.7273736415730488e-05, | |
| "loss": 0.3527, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.6356413166855845, | |
| "grad_norm": 0.6837971152662418, | |
| "learning_rate": 1.72373403810507e-05, | |
| "loss": 0.3666, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.638668180098373, | |
| "grad_norm": 0.6867817526761967, | |
| "learning_rate": 1.720074190240269e-05, | |
| "loss": 0.3644, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.6416950435111616, | |
| "grad_norm": 0.6631251385376119, | |
| "learning_rate": 1.7163942003524574e-05, | |
| "loss": 0.354, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.64472190692395, | |
| "grad_norm": 0.7209026100177808, | |
| "learning_rate": 1.7126941713788633e-05, | |
| "loss": 0.3848, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.6477487703367385, | |
| "grad_norm": 0.6844399226660911, | |
| "learning_rate": 1.70897420681725e-05, | |
| "loss": 0.3662, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.6507756337495271, | |
| "grad_norm": 0.6077405301647013, | |
| "learning_rate": 1.7052344107230244e-05, | |
| "loss": 0.3478, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.6538024971623155, | |
| "grad_norm": 0.7176698303155052, | |
| "learning_rate": 1.7014748877063212e-05, | |
| "loss": 0.39, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.656829360575104, | |
| "grad_norm": 0.6881320372999566, | |
| "learning_rate": 1.697695742929082e-05, | |
| "loss": 0.3576, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.6598562239878926, | |
| "grad_norm": 0.6791092078569761, | |
| "learning_rate": 1.693897082102109e-05, | |
| "loss": 0.3719, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.662883087400681, | |
| "grad_norm": 0.6985859681359807, | |
| "learning_rate": 1.6900790114821122e-05, | |
| "loss": 0.3608, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.6659099508134695, | |
| "grad_norm": 0.6261027969828732, | |
| "learning_rate": 1.686241637868734e-05, | |
| "loss": 0.3362, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.6689368142262581, | |
| "grad_norm": 0.6536247217675197, | |
| "learning_rate": 1.682385068601563e-05, | |
| "loss": 0.3642, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.6719636776390465, | |
| "grad_norm": 0.6823497670604539, | |
| "learning_rate": 1.6785094115571323e-05, | |
| "loss": 0.3618, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.6749905410518351, | |
| "grad_norm": 0.6881816165910678, | |
| "learning_rate": 1.674614775145901e-05, | |
| "loss": 0.3671, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.6780174044646236, | |
| "grad_norm": 0.6415990375954306, | |
| "learning_rate": 1.670701268309221e-05, | |
| "loss": 0.3587, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.681044267877412, | |
| "grad_norm": 0.6818833860252491, | |
| "learning_rate": 1.666769000516292e-05, | |
| "loss": 0.3584, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.6840711312902006, | |
| "grad_norm": 0.6546529534461664, | |
| "learning_rate": 1.6628180817610963e-05, | |
| "loss": 0.3368, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.687097994702989, | |
| "grad_norm": 0.6488024941145758, | |
| "learning_rate": 1.658848622559325e-05, | |
| "loss": 0.3629, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.6901248581157775, | |
| "grad_norm": 0.6859480910881849, | |
| "learning_rate": 1.6548607339452853e-05, | |
| "loss": 0.3535, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.6931517215285661, | |
| "grad_norm": 0.7193973562314164, | |
| "learning_rate": 1.6508545274687936e-05, | |
| "loss": 0.3677, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.6961785849413545, | |
| "grad_norm": 0.6393640438298169, | |
| "learning_rate": 1.6468301151920576e-05, | |
| "loss": 0.349, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.699205448354143, | |
| "grad_norm": 0.7004035969173447, | |
| "learning_rate": 1.6427876096865394e-05, | |
| "loss": 0.3484, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.7022323117669316, | |
| "grad_norm": 0.7200642490693192, | |
| "learning_rate": 1.6387271240298082e-05, | |
| "loss": 0.3802, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.70525917517972, | |
| "grad_norm": 0.6586432016539763, | |
| "learning_rate": 1.6346487718023762e-05, | |
| "loss": 0.3585, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.7082860385925085, | |
| "grad_norm": 0.7485738265822466, | |
| "learning_rate": 1.6305526670845225e-05, | |
| "loss": 0.3662, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.7113129020052971, | |
| "grad_norm": 0.6747980737467446, | |
| "learning_rate": 1.6264389244531015e-05, | |
| "loss": 0.3613, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.7143397654180855, | |
| "grad_norm": 0.6822748829483241, | |
| "learning_rate": 1.6223076589783368e-05, | |
| "loss": 0.3698, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.717366628830874, | |
| "grad_norm": 0.75892127715738, | |
| "learning_rate": 1.6181589862206053e-05, | |
| "loss": 0.3793, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.7203934922436626, | |
| "grad_norm": 0.7083470227411819, | |
| "learning_rate": 1.613993022227202e-05, | |
| "loss": 0.3677, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.723420355656451, | |
| "grad_norm": 0.6967819406708023, | |
| "learning_rate": 1.6098098835290955e-05, | |
| "loss": 0.3673, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.7264472190692395, | |
| "grad_norm": 0.7256203028140483, | |
| "learning_rate": 1.6056096871376667e-05, | |
| "loss": 0.366, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.729474082482028, | |
| "grad_norm": 0.6603676106272461, | |
| "learning_rate": 1.6013925505414386e-05, | |
| "loss": 0.3378, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.7325009458948165, | |
| "grad_norm": 0.6271029834695545, | |
| "learning_rate": 1.5971585917027864e-05, | |
| "loss": 0.3414, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.735527809307605, | |
| "grad_norm": 0.7396551223633928, | |
| "learning_rate": 1.5929079290546408e-05, | |
| "loss": 0.3584, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.7385546727203935, | |
| "grad_norm": 0.618853025745765, | |
| "learning_rate": 1.5886406814971728e-05, | |
| "loss": 0.3495, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.741581536133182, | |
| "grad_norm": 0.7020144035350626, | |
| "learning_rate": 1.584356968394471e-05, | |
| "loss": 0.3473, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.7446083995459705, | |
| "grad_norm": 0.7243184782357341, | |
| "learning_rate": 1.5800569095711983e-05, | |
| "loss": 0.3646, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.747635262958759, | |
| "grad_norm": 0.6682648778369016, | |
| "learning_rate": 1.575740625309244e-05, | |
| "loss": 0.3651, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.7506621263715475, | |
| "grad_norm": 0.7134395811231984, | |
| "learning_rate": 1.5714082363443576e-05, | |
| "loss": 0.3615, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.753688989784336, | |
| "grad_norm": 0.6474807669082351, | |
| "learning_rate": 1.5670598638627707e-05, | |
| "loss": 0.346, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.7567158531971245, | |
| "grad_norm": 0.7004443713354839, | |
| "learning_rate": 1.5626956294978103e-05, | |
| "loss": 0.3751, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.759742716609913, | |
| "grad_norm": 0.6435964096310008, | |
| "learning_rate": 1.5583156553264923e-05, | |
| "loss": 0.3596, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.7627695800227015, | |
| "grad_norm": 0.6203657338792459, | |
| "learning_rate": 1.5539200638661106e-05, | |
| "loss": 0.3426, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.76579644343549, | |
| "grad_norm": 0.6580717318002014, | |
| "learning_rate": 1.5495089780708062e-05, | |
| "loss": 0.3706, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.7688233068482785, | |
| "grad_norm": 0.6967505139246781, | |
| "learning_rate": 1.5450825213281317e-05, | |
| "loss": 0.3528, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.771850170261067, | |
| "grad_norm": 0.6493194614766129, | |
| "learning_rate": 1.5406408174555978e-05, | |
| "loss": 0.3574, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.7748770336738555, | |
| "grad_norm": 0.6833731690290171, | |
| "learning_rate": 1.5361839906972095e-05, | |
| "loss": 0.3564, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.777903897086644, | |
| "grad_norm": 0.6429893423458913, | |
| "learning_rate": 1.531712165719992e-05, | |
| "loss": 0.3401, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.7809307604994324, | |
| "grad_norm": 0.6277377103470478, | |
| "learning_rate": 1.5272254676105026e-05, | |
| "loss": 0.3672, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.783957623912221, | |
| "grad_norm": 0.6743978957574767, | |
| "learning_rate": 1.5227240218713326e-05, | |
| "loss": 0.3615, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.7869844873250095, | |
| "grad_norm": 0.6741370010357317, | |
| "learning_rate": 1.5182079544175957e-05, | |
| "loss": 0.3537, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.7900113507377979, | |
| "grad_norm": 0.660553397506574, | |
| "learning_rate": 1.5136773915734067e-05, | |
| "loss": 0.3615, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.7930382141505865, | |
| "grad_norm": 0.695922678115741, | |
| "learning_rate": 1.5091324600683472e-05, | |
| "loss": 0.365, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.796065077563375, | |
| "grad_norm": 0.6265037792202043, | |
| "learning_rate": 1.5045732870339213e-05, | |
| "loss": 0.3522, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.7990919409761634, | |
| "grad_norm": 0.6517305705193905, | |
| "learning_rate": 1.5000000000000002e-05, | |
| "loss": 0.3531, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.802118804388952, | |
| "grad_norm": 0.6608936912204517, | |
| "learning_rate": 1.4954127268912525e-05, | |
| "loss": 0.3579, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.8051456678017405, | |
| "grad_norm": 0.6431369496730017, | |
| "learning_rate": 1.4908115960235683e-05, | |
| "loss": 0.3451, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.8081725312145289, | |
| "grad_norm": 0.688061151175042, | |
| "learning_rate": 1.4861967361004687e-05, | |
| "loss": 0.3458, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.8111993946273175, | |
| "grad_norm": 0.6626041785608937, | |
| "learning_rate": 1.4815682762095065e-05, | |
| "loss": 0.327, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.814226258040106, | |
| "grad_norm": 0.6968470566559201, | |
| "learning_rate": 1.476926345818654e-05, | |
| "loss": 0.343, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.8172531214528944, | |
| "grad_norm": 0.6383874179230478, | |
| "learning_rate": 1.472271074772683e-05, | |
| "loss": 0.3321, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.820279984865683, | |
| "grad_norm": 0.6588971365041115, | |
| "learning_rate": 1.4676025932895315e-05, | |
| "loss": 0.3644, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.8233068482784714, | |
| "grad_norm": 0.6826226132858566, | |
| "learning_rate": 1.4629210319566626e-05, | |
| "loss": 0.3439, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.8263337116912599, | |
| "grad_norm": 0.6263597212673647, | |
| "learning_rate": 1.4582265217274105e-05, | |
| "loss": 0.3484, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.8293605751040485, | |
| "grad_norm": 0.7136506894283627, | |
| "learning_rate": 1.4535191939173179e-05, | |
| "loss": 0.336, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.8323874385168369, | |
| "grad_norm": 0.7380891134024083, | |
| "learning_rate": 1.4487991802004625e-05, | |
| "loss": 0.3522, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.8354143019296254, | |
| "grad_norm": 0.6374772311756791, | |
| "learning_rate": 1.4440666126057743e-05, | |
| "loss": 0.3596, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.838441165342414, | |
| "grad_norm": 0.6661373466007278, | |
| "learning_rate": 1.4393216235133427e-05, | |
| "loss": 0.3472, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.8414680287552024, | |
| "grad_norm": 0.6994791731518725, | |
| "learning_rate": 1.4345643456507126e-05, | |
| "loss": 0.3595, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.8444948921679909, | |
| "grad_norm": 0.7804720693806786, | |
| "learning_rate": 1.4297949120891718e-05, | |
| "loss": 0.3713, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.8475217555807795, | |
| "grad_norm": 0.6733773978840016, | |
| "learning_rate": 1.4250134562400301e-05, | |
| "loss": 0.3433, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.8505486189935679, | |
| "grad_norm": 0.6566607068744286, | |
| "learning_rate": 1.4202201118508863e-05, | |
| "loss": 0.3404, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.8535754824063564, | |
| "grad_norm": 0.6454644913419311, | |
| "learning_rate": 1.4154150130018867e-05, | |
| "loss": 0.3553, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.856602345819145, | |
| "grad_norm": 0.6837835469571709, | |
| "learning_rate": 1.4105982941019751e-05, | |
| "loss": 0.3489, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.8596292092319334, | |
| "grad_norm": 0.7361009633555228, | |
| "learning_rate": 1.405770089885134e-05, | |
| "loss": 0.3578, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.8626560726447219, | |
| "grad_norm": 0.6134547169281902, | |
| "learning_rate": 1.4009305354066138e-05, | |
| "loss": 0.3391, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.8656829360575105, | |
| "grad_norm": 0.6763263511087825, | |
| "learning_rate": 1.396079766039157e-05, | |
| "loss": 0.3473, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.8687097994702989, | |
| "grad_norm": 0.6380256023139563, | |
| "learning_rate": 1.39121791746921e-05, | |
| "loss": 0.3487, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.8717366628830874, | |
| "grad_norm": 0.6476162225373854, | |
| "learning_rate": 1.3863451256931286e-05, | |
| "loss": 0.3283, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.874763526295876, | |
| "grad_norm": 0.6470116385807246, | |
| "learning_rate": 1.381461527013374e-05, | |
| "loss": 0.3399, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.8777903897086644, | |
| "grad_norm": 0.6952105969877852, | |
| "learning_rate": 1.3765672580346986e-05, | |
| "loss": 0.3505, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.8808172531214529, | |
| "grad_norm": 0.6487297669612282, | |
| "learning_rate": 1.3716624556603275e-05, | |
| "loss": 0.3397, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.8838441165342414, | |
| "grad_norm": 0.6868383209749898, | |
| "learning_rate": 1.3667472570881264e-05, | |
| "loss": 0.3553, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.8868709799470299, | |
| "grad_norm": 0.7142860693749593, | |
| "learning_rate": 1.361821799806765e-05, | |
| "loss": 0.3599, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.8898978433598184, | |
| "grad_norm": 0.777297341112277, | |
| "learning_rate": 1.356886221591872e-05, | |
| "loss": 0.3381, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.8929247067726069, | |
| "grad_norm": 0.7328571397190125, | |
| "learning_rate": 1.3519406605021797e-05, | |
| "loss": 0.3469, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.8959515701853954, | |
| "grad_norm": 0.6780736416646974, | |
| "learning_rate": 1.3469852548756626e-05, | |
| "loss": 0.3326, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.8989784335981839, | |
| "grad_norm": 0.5850608998491945, | |
| "learning_rate": 1.342020143325669e-05, | |
| "loss": 0.322, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.9020052970109724, | |
| "grad_norm": 0.6837448264967723, | |
| "learning_rate": 1.3370454647370418e-05, | |
| "loss": 0.3389, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.9050321604237609, | |
| "grad_norm": 0.7181925933874312, | |
| "learning_rate": 1.3320613582622354e-05, | |
| "loss": 0.3439, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.9080590238365494, | |
| "grad_norm": 0.6206074665982018, | |
| "learning_rate": 1.3270679633174219e-05, | |
| "loss": 0.3367, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.9110858872493379, | |
| "grad_norm": 0.6381056806047553, | |
| "learning_rate": 1.3220654195785917e-05, | |
| "loss": 0.3356, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.9141127506621264, | |
| "grad_norm": 0.6289038988657802, | |
| "learning_rate": 1.3170538669776469e-05, | |
| "loss": 0.3476, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.9171396140749148, | |
| "grad_norm": 0.6256041561811607, | |
| "learning_rate": 1.3120334456984871e-05, | |
| "loss": 0.3355, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.9201664774877034, | |
| "grad_norm": 0.6385860491003814, | |
| "learning_rate": 1.3070042961730878e-05, | |
| "loss": 0.3415, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.9231933409004919, | |
| "grad_norm": 0.6663872288001169, | |
| "learning_rate": 1.3019665590775717e-05, | |
| "loss": 0.3608, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.9262202043132803, | |
| "grad_norm": 0.6385870327689683, | |
| "learning_rate": 1.296920375328275e-05, | |
| "loss": 0.3414, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.9292470677260689, | |
| "grad_norm": 0.5995360564974304, | |
| "learning_rate": 1.2918658860778046e-05, | |
| "loss": 0.3316, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.9322739311388574, | |
| "grad_norm": 0.6028346758503628, | |
| "learning_rate": 1.2868032327110904e-05, | |
| "loss": 0.3339, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.9353007945516458, | |
| "grad_norm": 0.6069703216952733, | |
| "learning_rate": 1.2817325568414299e-05, | |
| "loss": 0.3351, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.9383276579644344, | |
| "grad_norm": 0.6397356877087157, | |
| "learning_rate": 1.2766540003065272e-05, | |
| "loss": 0.3628, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.9413545213772229, | |
| "grad_norm": 0.6441618010761438, | |
| "learning_rate": 1.2715677051645259e-05, | |
| "loss": 0.334, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.9443813847900113, | |
| "grad_norm": 0.6238054887881863, | |
| "learning_rate": 1.266473813690035e-05, | |
| "loss": 0.3412, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.9474082482027999, | |
| "grad_norm": 0.6303279157936459, | |
| "learning_rate": 1.2613724683701491e-05, | |
| "loss": 0.3327, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.9504351116155884, | |
| "grad_norm": 0.641775143243854, | |
| "learning_rate": 1.2562638119004627e-05, | |
| "loss": 0.3357, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.9534619750283768, | |
| "grad_norm": 0.6386762671096875, | |
| "learning_rate": 1.2511479871810792e-05, | |
| "loss": 0.3294, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.9564888384411654, | |
| "grad_norm": 0.6071900471679189, | |
| "learning_rate": 1.2460251373126136e-05, | |
| "loss": 0.3343, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.9595157018539539, | |
| "grad_norm": 0.6586140893838607, | |
| "learning_rate": 1.2408954055921884e-05, | |
| "loss": 0.3415, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.9625425652667423, | |
| "grad_norm": 0.6358998792309285, | |
| "learning_rate": 1.2357589355094275e-05, | |
| "loss": 0.3551, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.9655694286795309, | |
| "grad_norm": 0.6142886599971407, | |
| "learning_rate": 1.2306158707424402e-05, | |
| "loss": 0.3413, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.9685962920923193, | |
| "grad_norm": 0.6204726176920337, | |
| "learning_rate": 1.2254663551538047e-05, | |
| "loss": 0.3408, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.9716231555051078, | |
| "grad_norm": 0.6014227476763383, | |
| "learning_rate": 1.2203105327865407e-05, | |
| "loss": 0.3542, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.9746500189178964, | |
| "grad_norm": 0.6263706138359493, | |
| "learning_rate": 1.215148547860084e-05, | |
| "loss": 0.3222, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.9776768823306848, | |
| "grad_norm": 0.5948530467829928, | |
| "learning_rate": 1.2099805447662485e-05, | |
| "loss": 0.3362, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.9807037457434733, | |
| "grad_norm": 0.6177094580536964, | |
| "learning_rate": 1.2048066680651908e-05, | |
| "loss": 0.3586, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.9837306091562619, | |
| "grad_norm": 0.6566059127761604, | |
| "learning_rate": 1.1996270624813642e-05, | |
| "loss": 0.3418, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.9867574725690503, | |
| "grad_norm": 0.6071697379225318, | |
| "learning_rate": 1.194441872899471e-05, | |
| "loss": 0.3316, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.9897843359818388, | |
| "grad_norm": 0.6368969322093863, | |
| "learning_rate": 1.1892512443604103e-05, | |
| "loss": 0.3516, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.9928111993946274, | |
| "grad_norm": 0.6502519173549048, | |
| "learning_rate": 1.1840553220572204e-05, | |
| "loss": 0.3418, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.9958380628074158, | |
| "grad_norm": 0.6667689898744066, | |
| "learning_rate": 1.1788542513310178e-05, | |
| "loss": 0.3517, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.9988649262202043, | |
| "grad_norm": 0.6091893052023183, | |
| "learning_rate": 1.1736481776669307e-05, | |
| "loss": 0.3371, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 1.0018917896329929, | |
| "grad_norm": 0.7010640209006963, | |
| "learning_rate": 1.1684372466900306e-05, | |
| "loss": 0.2906, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 1.0049186530457812, | |
| "grad_norm": 0.7832057568290791, | |
| "learning_rate": 1.1632216041612595e-05, | |
| "loss": 0.2526, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 1.0079455164585698, | |
| "grad_norm": 0.6064599705914879, | |
| "learning_rate": 1.15800139597335e-05, | |
| "loss": 0.2261, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 1.0109723798713584, | |
| "grad_norm": 0.685790240307814, | |
| "learning_rate": 1.1527767681467472e-05, | |
| "loss": 0.2344, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 1.0139992432841467, | |
| "grad_norm": 0.8262392848334692, | |
| "learning_rate": 1.1475478668255223e-05, | |
| "loss": 0.2575, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 1.0170261066969353, | |
| "grad_norm": 0.7725837752682978, | |
| "learning_rate": 1.1423148382732854e-05, | |
| "loss": 0.235, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 1.0200529701097238, | |
| "grad_norm": 0.751693155717695, | |
| "learning_rate": 1.1370778288690947e-05, | |
| "loss": 0.247, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 1.0230798335225122, | |
| "grad_norm": 0.6594789513656627, | |
| "learning_rate": 1.1318369851033604e-05, | |
| "loss": 0.2324, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 1.0261066969353008, | |
| "grad_norm": 0.6530307801784562, | |
| "learning_rate": 1.1265924535737494e-05, | |
| "loss": 0.2301, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 1.0291335603480893, | |
| "grad_norm": 0.697619436082779, | |
| "learning_rate": 1.121344380981082e-05, | |
| "loss": 0.2415, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 1.0321604237608777, | |
| "grad_norm": 0.6302177170682544, | |
| "learning_rate": 1.1160929141252303e-05, | |
| "loss": 0.2387, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 1.0351872871736663, | |
| "grad_norm": 0.6764150309741347, | |
| "learning_rate": 1.1108381999010111e-05, | |
| "loss": 0.2367, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 1.0382141505864548, | |
| "grad_norm": 0.6428365717145489, | |
| "learning_rate": 1.1055803852940772e-05, | |
| "loss": 0.2352, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 1.0412410139992432, | |
| "grad_norm": 0.6768541113061189, | |
| "learning_rate": 1.1003196173768051e-05, | |
| "loss": 0.2364, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 1.0442678774120318, | |
| "grad_norm": 0.6938527219299584, | |
| "learning_rate": 1.0950560433041825e-05, | |
| "loss": 0.2416, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 1.0472947408248203, | |
| "grad_norm": 0.6822745645226103, | |
| "learning_rate": 1.0897898103096917e-05, | |
| "loss": 0.2317, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 1.0503216042376087, | |
| "grad_norm": 0.6999578365079624, | |
| "learning_rate": 1.0845210657011893e-05, | |
| "loss": 0.2452, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 1.0533484676503972, | |
| "grad_norm": 0.6977456518478019, | |
| "learning_rate": 1.0792499568567885e-05, | |
| "loss": 0.2392, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 1.0563753310631858, | |
| "grad_norm": 0.6751457131573062, | |
| "learning_rate": 1.0739766312207344e-05, | |
| "loss": 0.2332, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 1.0594021944759742, | |
| "grad_norm": 0.6385538314815781, | |
| "learning_rate": 1.068701236299281e-05, | |
| "loss": 0.2246, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.0624290578887627, | |
| "grad_norm": 0.647236871724086, | |
| "learning_rate": 1.0634239196565646e-05, | |
| "loss": 0.2382, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 1.0654559213015513, | |
| "grad_norm": 0.6427149764142498, | |
| "learning_rate": 1.0581448289104759e-05, | |
| "loss": 0.2261, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 1.0684827847143397, | |
| "grad_norm": 0.6410504041089251, | |
| "learning_rate": 1.0528641117285315e-05, | |
| "loss": 0.2276, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 1.0715096481271282, | |
| "grad_norm": 0.6814764878411548, | |
| "learning_rate": 1.0475819158237426e-05, | |
| "loss": 0.2256, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 1.0745365115399168, | |
| "grad_norm": 0.6747046800918909, | |
| "learning_rate": 1.0422983889504831e-05, | |
| "loss": 0.2355, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 1.0775633749527052, | |
| "grad_norm": 0.6439000684596956, | |
| "learning_rate": 1.0370136789003582e-05, | |
| "loss": 0.2207, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 1.0805902383654937, | |
| "grad_norm": 0.6294475791812397, | |
| "learning_rate": 1.031727933498068e-05, | |
| "loss": 0.2278, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 1.0836171017782823, | |
| "grad_norm": 0.660826924839849, | |
| "learning_rate": 1.0264413005972736e-05, | |
| "loss": 0.2196, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 1.0866439651910706, | |
| "grad_norm": 0.691224743438426, | |
| "learning_rate": 1.0211539280764617e-05, | |
| "loss": 0.2255, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 1.0896708286038592, | |
| "grad_norm": 0.6326164599746542, | |
| "learning_rate": 1.015865963834808e-05, | |
| "loss": 0.229, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.0926976920166478, | |
| "grad_norm": 0.6330645634744737, | |
| "learning_rate": 1.0105775557880398e-05, | |
| "loss": 0.2277, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 1.0957245554294364, | |
| "grad_norm": 0.660209528641675, | |
| "learning_rate": 1.0052888518642978e-05, | |
| "loss": 0.2323, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 1.0987514188422247, | |
| "grad_norm": 0.6560952206890839, | |
| "learning_rate": 1e-05, | |
| "loss": 0.2361, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 1.1017782822550133, | |
| "grad_norm": 0.630874571845756, | |
| "learning_rate": 9.947111481357023e-06, | |
| "loss": 0.2275, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 1.1048051456678016, | |
| "grad_norm": 0.6304553371162837, | |
| "learning_rate": 9.894224442119606e-06, | |
| "loss": 0.2246, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 1.1078320090805902, | |
| "grad_norm": 0.6614228523785296, | |
| "learning_rate": 9.841340361651921e-06, | |
| "loss": 0.2348, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 1.1108588724933788, | |
| "grad_norm": 0.6154784438617857, | |
| "learning_rate": 9.788460719235386e-06, | |
| "loss": 0.2255, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 1.1138857359061674, | |
| "grad_norm": 0.6137609979925743, | |
| "learning_rate": 9.735586994027267e-06, | |
| "loss": 0.2281, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 1.1169125993189557, | |
| "grad_norm": 0.6586354008218958, | |
| "learning_rate": 9.682720665019325e-06, | |
| "loss": 0.2246, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 1.1199394627317443, | |
| "grad_norm": 0.7391307576741606, | |
| "learning_rate": 9.62986321099642e-06, | |
| "loss": 0.2369, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 1.1229663261445326, | |
| "grad_norm": 0.64058215315683, | |
| "learning_rate": 9.57701611049517e-06, | |
| "loss": 0.2302, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 1.1259931895573212, | |
| "grad_norm": 0.6242995560769223, | |
| "learning_rate": 9.524180841762577e-06, | |
| "loss": 0.218, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 1.1290200529701098, | |
| "grad_norm": 0.6400819777701364, | |
| "learning_rate": 9.471358882714687e-06, | |
| "loss": 0.2335, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 1.1320469163828983, | |
| "grad_norm": 0.621781657977426, | |
| "learning_rate": 9.418551710895243e-06, | |
| "loss": 0.2309, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 1.1350737797956867, | |
| "grad_norm": 0.6330029543239697, | |
| "learning_rate": 9.365760803434356e-06, | |
| "loss": 0.2362, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 1.1381006432084753, | |
| "grad_norm": 0.6807397176564342, | |
| "learning_rate": 9.312987637007191e-06, | |
| "loss": 0.2328, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 1.1411275066212636, | |
| "grad_norm": 0.6463882683649799, | |
| "learning_rate": 9.260233687792657e-06, | |
| "loss": 0.2319, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 1.1441543700340522, | |
| "grad_norm": 0.6571184999461146, | |
| "learning_rate": 9.207500431432115e-06, | |
| "loss": 0.227, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 1.1471812334468408, | |
| "grad_norm": 0.6376308272468328, | |
| "learning_rate": 9.154789342988108e-06, | |
| "loss": 0.2324, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 1.1502080968596293, | |
| "grad_norm": 0.6680407306443344, | |
| "learning_rate": 9.102101896903084e-06, | |
| "loss": 0.2318, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.1532349602724177, | |
| "grad_norm": 0.6850885046496867, | |
| "learning_rate": 9.049439566958176e-06, | |
| "loss": 0.2383, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 1.1562618236852062, | |
| "grad_norm": 0.6175653949533372, | |
| "learning_rate": 8.99680382623195e-06, | |
| "loss": 0.2301, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 1.1592886870979946, | |
| "grad_norm": 0.6585323886906671, | |
| "learning_rate": 8.944196147059233e-06, | |
| "loss": 0.2292, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 1.1623155505107832, | |
| "grad_norm": 0.6208096285743232, | |
| "learning_rate": 8.89161800098989e-06, | |
| "loss": 0.2362, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 1.1653424139235717, | |
| "grad_norm": 0.647395775486307, | |
| "learning_rate": 8.839070858747697e-06, | |
| "loss": 0.2319, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 1.1683692773363603, | |
| "grad_norm": 0.6385810526792797, | |
| "learning_rate": 8.786556190189183e-06, | |
| "loss": 0.2407, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 1.1713961407491487, | |
| "grad_norm": 0.6687421755435028, | |
| "learning_rate": 8.734075464262507e-06, | |
| "loss": 0.2286, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 1.1744230041619372, | |
| "grad_norm": 0.6849689638067709, | |
| "learning_rate": 8.681630148966397e-06, | |
| "loss": 0.2381, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 1.1774498675747256, | |
| "grad_norm": 0.6179899285666828, | |
| "learning_rate": 8.629221711309056e-06, | |
| "loss": 0.2203, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 1.1804767309875142, | |
| "grad_norm": 0.6691948969188457, | |
| "learning_rate": 8.576851617267151e-06, | |
| "loss": 0.2412, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.1835035944003027, | |
| "grad_norm": 0.6654996556704351, | |
| "learning_rate": 8.52452133174478e-06, | |
| "loss": 0.2256, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 1.1865304578130913, | |
| "grad_norm": 0.6864749298390848, | |
| "learning_rate": 8.472232318532531e-06, | |
| "loss": 0.2392, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 1.1895573212258796, | |
| "grad_norm": 0.6165096033766236, | |
| "learning_rate": 8.419986040266502e-06, | |
| "loss": 0.2319, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 1.1925841846386682, | |
| "grad_norm": 0.6327129015431557, | |
| "learning_rate": 8.367783958387407e-06, | |
| "loss": 0.2277, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 1.1956110480514566, | |
| "grad_norm": 0.6870493301853312, | |
| "learning_rate": 8.315627533099697e-06, | |
| "loss": 0.23, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 1.1986379114642451, | |
| "grad_norm": 0.6298332524432041, | |
| "learning_rate": 8.263518223330698e-06, | |
| "loss": 0.2252, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 1.2016647748770337, | |
| "grad_norm": 0.639916284107376, | |
| "learning_rate": 8.211457486689829e-06, | |
| "loss": 0.2263, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 1.2046916382898223, | |
| "grad_norm": 0.62080506299087, | |
| "learning_rate": 8.159446779427798e-06, | |
| "loss": 0.2262, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 1.2077185017026106, | |
| "grad_norm": 0.6570782414631622, | |
| "learning_rate": 8.107487556395902e-06, | |
| "loss": 0.2241, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 1.2107453651153992, | |
| "grad_norm": 0.6441684223207326, | |
| "learning_rate": 8.055581271005292e-06, | |
| "loss": 0.2255, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.2137722285281876, | |
| "grad_norm": 0.6902716233875088, | |
| "learning_rate": 8.00372937518636e-06, | |
| "loss": 0.2445, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 1.2167990919409761, | |
| "grad_norm": 0.6265721849818264, | |
| "learning_rate": 7.951933319348095e-06, | |
| "loss": 0.2299, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 1.2198259553537647, | |
| "grad_norm": 0.6259028733015133, | |
| "learning_rate": 7.900194552337516e-06, | |
| "loss": 0.234, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 1.2228528187665533, | |
| "grad_norm": 0.679549461172087, | |
| "learning_rate": 7.848514521399167e-06, | |
| "loss": 0.2375, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 1.2258796821793416, | |
| "grad_norm": 0.6302741635948756, | |
| "learning_rate": 7.796894672134594e-06, | |
| "loss": 0.2176, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 1.2289065455921302, | |
| "grad_norm": 0.6287132683983069, | |
| "learning_rate": 7.745336448461958e-06, | |
| "loss": 0.2276, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 1.2319334090049185, | |
| "grad_norm": 0.6159772972498484, | |
| "learning_rate": 7.6938412925756e-06, | |
| "loss": 0.2288, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 1.2349602724177071, | |
| "grad_norm": 0.6081347913798519, | |
| "learning_rate": 7.642410644905726e-06, | |
| "loss": 0.2139, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 1.2379871358304957, | |
| "grad_norm": 0.6280369405799916, | |
| "learning_rate": 7.591045944078119e-06, | |
| "loss": 0.2215, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 1.2410139992432843, | |
| "grad_norm": 0.6495834293574575, | |
| "learning_rate": 7.539748626873866e-06, | |
| "loss": 0.2384, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.2440408626560726, | |
| "grad_norm": 0.6721472717421363, | |
| "learning_rate": 7.488520128189209e-06, | |
| "loss": 0.2417, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 1.2470677260688612, | |
| "grad_norm": 0.6731474696864634, | |
| "learning_rate": 7.4373618809953755e-06, | |
| "loss": 0.2199, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 1.2500945894816495, | |
| "grad_norm": 0.6357896346550611, | |
| "learning_rate": 7.386275316298513e-06, | |
| "loss": 0.2355, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 1.253121452894438, | |
| "grad_norm": 0.6618468905688569, | |
| "learning_rate": 7.335261863099652e-06, | |
| "loss": 0.2282, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 1.2561483163072267, | |
| "grad_norm": 0.6392305024750385, | |
| "learning_rate": 7.2843229483547405e-06, | |
| "loss": 0.2291, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 1.2591751797200152, | |
| "grad_norm": 0.658424865941322, | |
| "learning_rate": 7.233459996934731e-06, | |
| "loss": 0.2332, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 1.2622020431328036, | |
| "grad_norm": 0.630789937882681, | |
| "learning_rate": 7.182674431585703e-06, | |
| "loss": 0.225, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 1.2652289065455922, | |
| "grad_norm": 0.6674925516757622, | |
| "learning_rate": 7.131967672889101e-06, | |
| "loss": 0.22, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 1.2682557699583805, | |
| "grad_norm": 0.6596080731694776, | |
| "learning_rate": 7.081341139221955e-06, | |
| "loss": 0.2263, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 1.271282633371169, | |
| "grad_norm": 0.6741616793073207, | |
| "learning_rate": 7.0307962467172555e-06, | |
| "loss": 0.2286, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.2743094967839577, | |
| "grad_norm": 0.700457786726074, | |
| "learning_rate": 6.9803344092242855e-06, | |
| "loss": 0.232, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 1.2773363601967462, | |
| "grad_norm": 0.6536247428318708, | |
| "learning_rate": 6.929957038269123e-06, | |
| "loss": 0.2316, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 1.2803632236095346, | |
| "grad_norm": 0.6426840686443641, | |
| "learning_rate": 6.87966554301513e-06, | |
| "loss": 0.2234, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 1.2833900870223232, | |
| "grad_norm": 0.6423335138464212, | |
| "learning_rate": 6.8294613302235325e-06, | |
| "loss": 0.2195, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 1.2864169504351115, | |
| "grad_norm": 0.5979336942995529, | |
| "learning_rate": 6.779345804214088e-06, | |
| "loss": 0.2174, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.2894438138479, | |
| "grad_norm": 0.6575063651706292, | |
| "learning_rate": 6.729320366825785e-06, | |
| "loss": 0.2388, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 1.2924706772606886, | |
| "grad_norm": 0.6376411613100648, | |
| "learning_rate": 6.679386417377649e-06, | |
| "loss": 0.2203, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 1.2954975406734772, | |
| "grad_norm": 0.6407464308464492, | |
| "learning_rate": 6.629545352629583e-06, | |
| "loss": 0.2229, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 1.2985244040862656, | |
| "grad_norm": 0.6475968437027577, | |
| "learning_rate": 6.579798566743314e-06, | |
| "loss": 0.2203, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 1.3015512674990541, | |
| "grad_norm": 0.6172438731377891, | |
| "learning_rate": 6.530147451243377e-06, | |
| "loss": 0.2336, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.3045781309118425, | |
| "grad_norm": 0.6475474927303543, | |
| "learning_rate": 6.480593394978208e-06, | |
| "loss": 0.2362, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 1.307604994324631, | |
| "grad_norm": 0.6991787262564514, | |
| "learning_rate": 6.431137784081283e-06, | |
| "loss": 0.2365, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 1.3106318577374196, | |
| "grad_norm": 0.6607064860750409, | |
| "learning_rate": 6.381782001932352e-06, | |
| "loss": 0.2372, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 1.3136587211502082, | |
| "grad_norm": 0.6404132596649188, | |
| "learning_rate": 6.33252742911874e-06, | |
| "loss": 0.2266, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 1.3166855845629966, | |
| "grad_norm": 0.6195186003483328, | |
| "learning_rate": 6.283375443396726e-06, | |
| "loss": 0.2253, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 1.3197124479757851, | |
| "grad_norm": 0.6350275090900896, | |
| "learning_rate": 6.234327419653013e-06, | |
| "loss": 0.228, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 1.3227393113885735, | |
| "grad_norm": 0.6808840243041602, | |
| "learning_rate": 6.185384729866264e-06, | |
| "loss": 0.232, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 1.325766174801362, | |
| "grad_norm": 0.673027689005355, | |
| "learning_rate": 6.136548743068713e-06, | |
| "loss": 0.234, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 1.3287930382141506, | |
| "grad_norm": 0.6131590217594213, | |
| "learning_rate": 6.087820825307904e-06, | |
| "loss": 0.2224, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 1.3318199016269392, | |
| "grad_norm": 0.6469981084227346, | |
| "learning_rate": 6.039202339608432e-06, | |
| "loss": 0.2292, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 1.3348467650397275, | |
| "grad_norm": 0.6289948539984372, | |
| "learning_rate": 5.990694645933866e-06, | |
| "loss": 0.2226, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 1.3378736284525161, | |
| "grad_norm": 0.6683052274328736, | |
| "learning_rate": 5.9422991011486635e-06, | |
| "loss": 0.2346, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 1.3409004918653045, | |
| "grad_norm": 0.6528287889096593, | |
| "learning_rate": 5.894017058980249e-06, | |
| "loss": 0.2271, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 1.343927355278093, | |
| "grad_norm": 0.6097556954365678, | |
| "learning_rate": 5.845849869981137e-06, | |
| "loss": 0.218, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 1.3469542186908816, | |
| "grad_norm": 0.6403214424730346, | |
| "learning_rate": 5.797798881491138e-06, | |
| "loss": 0.2317, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 1.3499810821036702, | |
| "grad_norm": 0.6794700580428522, | |
| "learning_rate": 5.749865437599703e-06, | |
| "loss": 0.2392, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 1.3530079455164585, | |
| "grad_norm": 0.6749520850962962, | |
| "learning_rate": 5.702050879108284e-06, | |
| "loss": 0.2257, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 1.356034808929247, | |
| "grad_norm": 0.657157830576466, | |
| "learning_rate": 5.654356543492883e-06, | |
| "loss": 0.2248, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 1.3590616723420355, | |
| "grad_norm": 0.6016326049732952, | |
| "learning_rate": 5.606783764866576e-06, | |
| "loss": 0.2243, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 1.362088535754824, | |
| "grad_norm": 0.6218829982408461, | |
| "learning_rate": 5.559333873942259e-06, | |
| "loss": 0.2237, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.3651153991676126, | |
| "grad_norm": 0.641354259930125, | |
| "learning_rate": 5.512008197995379e-06, | |
| "loss": 0.2329, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 1.3681422625804012, | |
| "grad_norm": 0.6686866786635635, | |
| "learning_rate": 5.464808060826825e-06, | |
| "loss": 0.2315, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 1.3711691259931895, | |
| "grad_norm": 0.6109713947108018, | |
| "learning_rate": 5.417734782725896e-06, | |
| "loss": 0.2256, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 1.374195989405978, | |
| "grad_norm": 0.6306885594586495, | |
| "learning_rate": 5.370789680433376e-06, | |
| "loss": 0.2175, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 1.3772228528187664, | |
| "grad_norm": 0.6120054302916149, | |
| "learning_rate": 5.323974067104687e-06, | |
| "loss": 0.2233, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 1.380249716231555, | |
| "grad_norm": 0.6162403359489199, | |
| "learning_rate": 5.277289252273175e-06, | |
| "loss": 0.229, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 1.3832765796443436, | |
| "grad_norm": 0.6104933337207802, | |
| "learning_rate": 5.230736541813463e-06, | |
| "loss": 0.2157, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 1.3863034430571322, | |
| "grad_norm": 0.6131580786636313, | |
| "learning_rate": 5.184317237904939e-06, | |
| "loss": 0.2191, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 1.3893303064699205, | |
| "grad_norm": 0.6083381180500818, | |
| "learning_rate": 5.138032638995315e-06, | |
| "loss": 0.2162, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 1.392357169882709, | |
| "grad_norm": 0.6097581787025927, | |
| "learning_rate": 5.091884039764321e-06, | |
| "loss": 0.2121, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.3953840332954974, | |
| "grad_norm": 0.60049713518881, | |
| "learning_rate": 5.045872731087479e-06, | |
| "loss": 0.223, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 1.398410896708286, | |
| "grad_norm": 0.6303034310548725, | |
| "learning_rate": 5.000000000000003e-06, | |
| "loss": 0.2189, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 1.4014377601210746, | |
| "grad_norm": 0.6320145045640445, | |
| "learning_rate": 4.954267129660789e-06, | |
| "loss": 0.2176, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 1.4044646235338631, | |
| "grad_norm": 0.6510299734510754, | |
| "learning_rate": 4.908675399316534e-06, | |
| "loss": 0.2258, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 1.4074914869466515, | |
| "grad_norm": 0.615969555323373, | |
| "learning_rate": 4.863226084265939e-06, | |
| "loss": 0.2188, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 1.41051835035944, | |
| "grad_norm": 0.6388485765487084, | |
| "learning_rate": 4.817920455824045e-06, | |
| "loss": 0.2248, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 1.4135452137722284, | |
| "grad_norm": 0.6443990634564626, | |
| "learning_rate": 4.772759781286679e-06, | |
| "loss": 0.2224, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 1.416572077185017, | |
| "grad_norm": 0.6287367050037098, | |
| "learning_rate": 4.727745323894976e-06, | |
| "loss": 0.2171, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 1.4195989405978056, | |
| "grad_norm": 0.6331353021853174, | |
| "learning_rate": 4.682878342800087e-06, | |
| "loss": 0.2226, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 1.4226258040105941, | |
| "grad_norm": 0.6035815288545289, | |
| "learning_rate": 4.638160093027908e-06, | |
| "loss": 0.2202, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.4256526674233825, | |
| "grad_norm": 0.6328448063451526, | |
| "learning_rate": 4.593591825444028e-06, | |
| "loss": 0.2305, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 1.428679530836171, | |
| "grad_norm": 0.6454089373695652, | |
| "learning_rate": 4.549174786718684e-06, | |
| "loss": 0.2277, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 1.4317063942489594, | |
| "grad_norm": 0.6496607947935132, | |
| "learning_rate": 4.504910219291941e-06, | |
| "loss": 0.2197, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 1.434733257661748, | |
| "grad_norm": 0.5941961430748568, | |
| "learning_rate": 4.460799361338898e-06, | |
| "loss": 0.2163, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 1.4377601210745365, | |
| "grad_norm": 0.6365420395691762, | |
| "learning_rate": 4.416843446735077e-06, | |
| "loss": 0.2164, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.4407869844873251, | |
| "grad_norm": 0.6401612557876254, | |
| "learning_rate": 4.373043705021899e-06, | |
| "loss": 0.2228, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 1.4438138479001135, | |
| "grad_norm": 0.5999963428486789, | |
| "learning_rate": 4.3294013613722944e-06, | |
| "loss": 0.214, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 1.446840711312902, | |
| "grad_norm": 0.6300109668762238, | |
| "learning_rate": 4.2859176365564294e-06, | |
| "loss": 0.2213, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 1.4498675747256904, | |
| "grad_norm": 0.6373611137574444, | |
| "learning_rate": 4.2425937469075626e-06, | |
| "loss": 0.225, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 1.452894438138479, | |
| "grad_norm": 0.6260967812483024, | |
| "learning_rate": 4.19943090428802e-06, | |
| "loss": 0.2302, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.4559213015512675, | |
| "grad_norm": 0.6567401839086223, | |
| "learning_rate": 4.1564303160552935e-06, | |
| "loss": 0.2269, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 1.458948164964056, | |
| "grad_norm": 0.6334539811175335, | |
| "learning_rate": 4.113593185028273e-06, | |
| "loss": 0.2224, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 1.4619750283768445, | |
| "grad_norm": 0.5985712899157856, | |
| "learning_rate": 4.070920709453597e-06, | |
| "loss": 0.226, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 1.465001891789633, | |
| "grad_norm": 0.6159475178471709, | |
| "learning_rate": 4.028414082972141e-06, | |
| "loss": 0.2214, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 1.4680287552024214, | |
| "grad_norm": 0.6362124329836059, | |
| "learning_rate": 3.986074494585619e-06, | |
| "loss": 0.2191, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 1.47105561861521, | |
| "grad_norm": 0.6377608887772233, | |
| "learning_rate": 3.943903128623336e-06, | |
| "loss": 0.2177, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 1.4740824820279985, | |
| "grad_norm": 0.6128951805748513, | |
| "learning_rate": 3.9019011647090465e-06, | |
| "loss": 0.2198, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 1.477109345440787, | |
| "grad_norm": 0.6387334168296341, | |
| "learning_rate": 3.860069777727983e-06, | |
| "loss": 0.223, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 1.4801362088535754, | |
| "grad_norm": 0.6093054196710758, | |
| "learning_rate": 3.818410137793947e-06, | |
| "loss": 0.2215, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 1.483163072266364, | |
| "grad_norm": 0.6461723467882955, | |
| "learning_rate": 3.7769234102166365e-06, | |
| "loss": 0.2185, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.4861899356791524, | |
| "grad_norm": 0.6490106838212442, | |
| "learning_rate": 3.735610755468988e-06, | |
| "loss": 0.2229, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 1.489216799091941, | |
| "grad_norm": 0.6641734821313223, | |
| "learning_rate": 3.6944733291547784e-06, | |
| "loss": 0.2181, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 1.4922436625047295, | |
| "grad_norm": 0.635229168200597, | |
| "learning_rate": 3.653512281976238e-06, | |
| "loss": 0.229, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 1.495270525917518, | |
| "grad_norm": 0.6308230641960129, | |
| "learning_rate": 3.612728759701919e-06, | |
| "loss": 0.2232, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 1.4982973893303064, | |
| "grad_norm": 0.6211515229781885, | |
| "learning_rate": 3.5721239031346067e-06, | |
| "loss": 0.2159, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 1.501324252743095, | |
| "grad_norm": 0.6122468863270718, | |
| "learning_rate": 3.5316988480794255e-06, | |
| "loss": 0.2124, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 1.5043511161558833, | |
| "grad_norm": 0.6503014028703544, | |
| "learning_rate": 3.4914547253120655e-06, | |
| "loss": 0.2224, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 1.507377979568672, | |
| "grad_norm": 0.6147253605414733, | |
| "learning_rate": 3.4513926605471504e-06, | |
| "loss": 0.2109, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 1.5104048429814605, | |
| "grad_norm": 0.6322924236930667, | |
| "learning_rate": 3.4115137744067516e-06, | |
| "loss": 0.2148, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 1.513431706394249, | |
| "grad_norm": 0.6453343769041457, | |
| "learning_rate": 3.37181918238904e-06, | |
| "loss": 0.2272, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.5164585698070374, | |
| "grad_norm": 0.612411690829783, | |
| "learning_rate": 3.3323099948370853e-06, | |
| "loss": 0.2144, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 1.519485433219826, | |
| "grad_norm": 0.6430401517011031, | |
| "learning_rate": 3.292987316907792e-06, | |
| "loss": 0.2224, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 1.5225122966326143, | |
| "grad_norm": 0.6552131995874647, | |
| "learning_rate": 3.253852248540994e-06, | |
| "loss": 0.2315, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 1.525539160045403, | |
| "grad_norm": 0.6452107134806657, | |
| "learning_rate": 3.2149058844286796e-06, | |
| "loss": 0.2248, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 1.5285660234581915, | |
| "grad_norm": 0.6314171039726749, | |
| "learning_rate": 3.1761493139843734e-06, | |
| "loss": 0.2269, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 1.53159288687098, | |
| "grad_norm": 0.5935662111697132, | |
| "learning_rate": 3.1375836213126653e-06, | |
| "loss": 0.2251, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 1.5346197502837684, | |
| "grad_norm": 0.6301182217954552, | |
| "learning_rate": 3.099209885178882e-06, | |
| "loss": 0.2178, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 1.537646613696557, | |
| "grad_norm": 0.5910142341354517, | |
| "learning_rate": 3.0610291789789094e-06, | |
| "loss": 0.2118, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 1.5406734771093453, | |
| "grad_norm": 0.612713842099563, | |
| "learning_rate": 3.023042570709185e-06, | |
| "loss": 0.215, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 1.543700340522134, | |
| "grad_norm": 0.6137167106736944, | |
| "learning_rate": 2.9852511229367862e-06, | |
| "loss": 0.2139, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 1.5467272039349225, | |
| "grad_norm": 0.624726342205351, | |
| "learning_rate": 2.9476558927697605e-06, | |
| "loss": 0.2169, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 1.549754067347711, | |
| "grad_norm": 0.6360207261292684, | |
| "learning_rate": 2.9102579318274994e-06, | |
| "loss": 0.2134, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 1.5527809307604994, | |
| "grad_norm": 0.6287973506310153, | |
| "learning_rate": 2.8730582862113743e-06, | |
| "loss": 0.2135, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 1.555807794173288, | |
| "grad_norm": 0.6531353391577391, | |
| "learning_rate": 2.8360579964754277e-06, | |
| "loss": 0.2217, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 1.5588346575860763, | |
| "grad_norm": 0.6087596342151999, | |
| "learning_rate": 2.7992580975973136e-06, | |
| "loss": 0.2125, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 1.5618615209988649, | |
| "grad_norm": 0.6567850673389571, | |
| "learning_rate": 2.7626596189492983e-06, | |
| "loss": 0.2202, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 1.5648883844116535, | |
| "grad_norm": 0.6668643986435379, | |
| "learning_rate": 2.726263584269513e-06, | |
| "loss": 0.2215, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 1.567915247824442, | |
| "grad_norm": 0.6268959463137306, | |
| "learning_rate": 2.690071011633284e-06, | |
| "loss": 0.2178, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 1.5709421112372304, | |
| "grad_norm": 0.649521725593062, | |
| "learning_rate": 2.6540829134246683e-06, | |
| "loss": 0.2251, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 1.573968974650019, | |
| "grad_norm": 0.5923197102004909, | |
| "learning_rate": 2.618300296308135e-06, | |
| "loss": 0.2197, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 1.5769958380628073, | |
| "grad_norm": 0.635013504579372, | |
| "learning_rate": 2.582724161200405e-06, | |
| "loss": 0.2172, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 1.5800227014755959, | |
| "grad_norm": 0.6331460598256372, | |
| "learning_rate": 2.5473555032424534e-06, | |
| "loss": 0.2255, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 1.5830495648883844, | |
| "grad_norm": 0.6069842674037031, | |
| "learning_rate": 2.5121953117716744e-06, | |
| "loss": 0.2176, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 1.586076428301173, | |
| "grad_norm": 0.6551570714363433, | |
| "learning_rate": 2.477244570294206e-06, | |
| "loss": 0.2305, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 1.5891032917139614, | |
| "grad_norm": 0.6376452057400069, | |
| "learning_rate": 2.4425042564574186e-06, | |
| "loss": 0.2185, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.59213015512675, | |
| "grad_norm": 0.6263929612159485, | |
| "learning_rate": 2.4079753420225694e-06, | |
| "loss": 0.2262, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 1.5951570185395383, | |
| "grad_norm": 0.5981822625806432, | |
| "learning_rate": 2.3736587928376197e-06, | |
| "loss": 0.2165, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 1.5981838819523269, | |
| "grad_norm": 0.5732188009172626, | |
| "learning_rate": 2.339555568810221e-06, | |
| "loss": 0.2009, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 1.6012107453651154, | |
| "grad_norm": 0.6478707140204104, | |
| "learning_rate": 2.305666623880858e-06, | |
| "loss": 0.2255, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 1.604237608777904, | |
| "grad_norm": 0.6393834730019626, | |
| "learning_rate": 2.27199290599617e-06, | |
| "loss": 0.218, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.6072644721906924, | |
| "grad_norm": 0.6315797246696278, | |
| "learning_rate": 2.2385353570824308e-06, | |
| "loss": 0.2217, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 1.610291335603481, | |
| "grad_norm": 0.618483760062658, | |
| "learning_rate": 2.2052949130192136e-06, | |
| "loss": 0.222, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 1.6133181990162693, | |
| "grad_norm": 0.6448599225329523, | |
| "learning_rate": 2.172272503613183e-06, | |
| "loss": 0.2084, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 1.6163450624290578, | |
| "grad_norm": 0.615785139498038, | |
| "learning_rate": 2.1394690525721275e-06, | |
| "loss": 0.2181, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 1.6193719258418464, | |
| "grad_norm": 0.6452246087306891, | |
| "learning_rate": 2.1068854774790783e-06, | |
| "loss": 0.2357, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 1.622398789254635, | |
| "grad_norm": 0.6171094436204184, | |
| "learning_rate": 2.0745226897666858e-06, | |
| "loss": 0.2213, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 1.6254256526674233, | |
| "grad_norm": 0.5985447205213116, | |
| "learning_rate": 2.0423815946916783e-06, | |
| "loss": 0.2132, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 1.628452516080212, | |
| "grad_norm": 0.6347418633648162, | |
| "learning_rate": 2.010463091309587e-06, | |
| "loss": 0.2201, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 1.6314793794930003, | |
| "grad_norm": 0.6570037881749712, | |
| "learning_rate": 1.9787680724495617e-06, | |
| "loss": 0.2231, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 1.6345062429057888, | |
| "grad_norm": 0.5977427484713659, | |
| "learning_rate": 1.947297424689414e-06, | |
| "loss": 0.2158, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.6375331063185774, | |
| "grad_norm": 0.5902767096486045, | |
| "learning_rate": 1.9160520283308115e-06, | |
| "loss": 0.2104, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 1.640559969731366, | |
| "grad_norm": 0.6318353905792551, | |
| "learning_rate": 1.8850327573746584e-06, | |
| "loss": 0.2193, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 1.6435868331441543, | |
| "grad_norm": 0.643273275403014, | |
| "learning_rate": 1.854240479496643e-06, | |
| "loss": 0.2174, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 1.646613696556943, | |
| "grad_norm": 0.6293390379396012, | |
| "learning_rate": 1.8236760560229715e-06, | |
| "loss": 0.2213, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 1.6496405599697312, | |
| "grad_norm": 0.5922798884229773, | |
| "learning_rate": 1.7933403419062689e-06, | |
| "loss": 0.213, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 1.6526674233825198, | |
| "grad_norm": 0.6173586211053849, | |
| "learning_rate": 1.7632341857016733e-06, | |
| "loss": 0.2226, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 1.6556942867953084, | |
| "grad_norm": 0.617281724748909, | |
| "learning_rate": 1.7333584295430894e-06, | |
| "loss": 0.2181, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 1.658721150208097, | |
| "grad_norm": 0.6500866918049804, | |
| "learning_rate": 1.7037139091196396e-06, | |
| "loss": 0.2211, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 1.6617480136208853, | |
| "grad_norm": 0.6469866814140316, | |
| "learning_rate": 1.6743014536522872e-06, | |
| "loss": 0.225, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 1.6647748770336739, | |
| "grad_norm": 0.631651152091777, | |
| "learning_rate": 1.6451218858706374e-06, | |
| "loss": 0.2178, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.6678017404464622, | |
| "grad_norm": 0.598065762245239, | |
| "learning_rate": 1.616176021989926e-06, | |
| "loss": 0.2133, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 1.6708286038592508, | |
| "grad_norm": 0.6135156398450355, | |
| "learning_rate": 1.587464671688187e-06, | |
| "loss": 0.2169, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 1.6738554672720394, | |
| "grad_norm": 0.6322884028381904, | |
| "learning_rate": 1.558988638083616e-06, | |
| "loss": 0.2177, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 1.676882330684828, | |
| "grad_norm": 0.5999158208642941, | |
| "learning_rate": 1.5307487177120773e-06, | |
| "loss": 0.2159, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 1.6799091940976163, | |
| "grad_norm": 0.6565461009709168, | |
| "learning_rate": 1.5027457005048573e-06, | |
| "loss": 0.2265, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 1.6829360575104049, | |
| "grad_norm": 0.6193282148388332, | |
| "learning_rate": 1.4749803697665366e-06, | |
| "loss": 0.2186, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 1.6859629209231932, | |
| "grad_norm": 0.6475210199120864, | |
| "learning_rate": 1.4474535021531099e-06, | |
| "loss": 0.2226, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 1.6889897843359818, | |
| "grad_norm": 0.6255016640140482, | |
| "learning_rate": 1.4201658676502294e-06, | |
| "loss": 0.211, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 1.6920166477487704, | |
| "grad_norm": 0.635194404589593, | |
| "learning_rate": 1.3931182295516965e-06, | |
| "loss": 0.2074, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 1.695043511161559, | |
| "grad_norm": 0.59642643715763, | |
| "learning_rate": 1.3663113444380905e-06, | |
| "loss": 0.2161, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.6980703745743473, | |
| "grad_norm": 0.6174084608670078, | |
| "learning_rate": 1.339745962155613e-06, | |
| "loss": 0.2124, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 1.7010972379871359, | |
| "grad_norm": 0.6202145450074755, | |
| "learning_rate": 1.3134228257951142e-06, | |
| "loss": 0.2173, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 1.7041241013999242, | |
| "grad_norm": 0.6204001802516719, | |
| "learning_rate": 1.2873426716713012e-06, | |
| "loss": 0.2153, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 1.7071509648127128, | |
| "grad_norm": 0.6250794543994412, | |
| "learning_rate": 1.2615062293021508e-06, | |
| "loss": 0.2207, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 1.7101778282255014, | |
| "grad_norm": 0.5841897306778809, | |
| "learning_rate": 1.2359142213884933e-06, | |
| "loss": 0.2117, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 1.71320469163829, | |
| "grad_norm": 0.5934693567272207, | |
| "learning_rate": 1.2105673637938054e-06, | |
| "loss": 0.216, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 1.7162315550510783, | |
| "grad_norm": 0.6694891581684043, | |
| "learning_rate": 1.1854663655241804e-06, | |
| "loss": 0.2235, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 1.7192584184638668, | |
| "grad_norm": 0.6197925649756133, | |
| "learning_rate": 1.1606119287084982e-06, | |
| "loss": 0.2049, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 1.7222852818766552, | |
| "grad_norm": 0.6059201062000129, | |
| "learning_rate": 1.136004748578785e-06, | |
| "loss": 0.2107, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 1.7253121452894438, | |
| "grad_norm": 0.6548697659257185, | |
| "learning_rate": 1.1116455134507665e-06, | |
| "loss": 0.227, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.7283390087022323, | |
| "grad_norm": 0.59833137240142, | |
| "learning_rate": 1.0875349047046113e-06, | |
| "loss": 0.2043, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 1.731365872115021, | |
| "grad_norm": 0.641718543536004, | |
| "learning_rate": 1.0636735967658785e-06, | |
| "loss": 0.2211, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 1.7343927355278093, | |
| "grad_norm": 0.6186835952072806, | |
| "learning_rate": 1.0400622570866426e-06, | |
| "loss": 0.2134, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 1.7374195989405978, | |
| "grad_norm": 0.656071481454566, | |
| "learning_rate": 1.0167015461268303e-06, | |
| "loss": 0.2288, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 1.7404464623533862, | |
| "grad_norm": 0.6072315694780389, | |
| "learning_rate": 9.935921173357444e-07, | |
| "loss": 0.2176, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.7434733257661748, | |
| "grad_norm": 0.6226071859060439, | |
| "learning_rate": 9.707346171337895e-07, | |
| "loss": 0.217, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 1.7465001891789633, | |
| "grad_norm": 0.614680502409671, | |
| "learning_rate": 9.481296848943744e-07, | |
| "loss": 0.2159, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 1.749527052591752, | |
| "grad_norm": 0.6021228280219003, | |
| "learning_rate": 9.257779529260558e-07, | |
| "loss": 0.209, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 1.7525539160045402, | |
| "grad_norm": 0.6528067120310421, | |
| "learning_rate": 9.036800464548157e-07, | |
| "loss": 0.2321, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 1.7555807794173288, | |
| "grad_norm": 0.6229321313531281, | |
| "learning_rate": 8.818365836066101e-07, | |
| "loss": 0.2137, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.7586076428301172, | |
| "grad_norm": 0.6096118951700134, | |
| "learning_rate": 8.602481753900427e-07, | |
| "loss": 0.2187, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 1.7616345062429057, | |
| "grad_norm": 0.6268586420836513, | |
| "learning_rate": 8.389154256793042e-07, | |
| "loss": 0.2208, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 1.7646613696556943, | |
| "grad_norm": 0.5923316241208719, | |
| "learning_rate": 8.178389311972612e-07, | |
| "loss": 0.2108, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 1.7676882330684829, | |
| "grad_norm": 0.6401617670760633, | |
| "learning_rate": 7.970192814987676e-07, | |
| "loss": 0.2147, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 1.7707150964812712, | |
| "grad_norm": 0.6065266493426773, | |
| "learning_rate": 7.764570589541876e-07, | |
| "loss": 0.2179, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 1.7737419598940598, | |
| "grad_norm": 0.6056811053030827, | |
| "learning_rate": 7.561528387330797e-07, | |
| "loss": 0.2062, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 1.7767688233068482, | |
| "grad_norm": 0.615245451704518, | |
| "learning_rate": 7.361071887881376e-07, | |
| "loss": 0.222, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 1.7797956867196367, | |
| "grad_norm": 0.626003365126797, | |
| "learning_rate": 7.163206698392744e-07, | |
| "loss": 0.2184, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 1.7828225501324253, | |
| "grad_norm": 0.6178433914123657, | |
| "learning_rate": 6.96793835357964e-07, | |
| "loss": 0.2113, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 1.7858494135452139, | |
| "grad_norm": 0.6238590829523674, | |
| "learning_rate": 6.775272315517423e-07, | |
| "loss": 0.215, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.7888762769580022, | |
| "grad_norm": 0.6013692757858885, | |
| "learning_rate": 6.585213973489335e-07, | |
| "loss": 0.212, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 1.7919031403707908, | |
| "grad_norm": 0.6540240619364013, | |
| "learning_rate": 6.397768643835755e-07, | |
| "loss": 0.2309, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 1.7949300037835791, | |
| "grad_norm": 0.5860656802317855, | |
| "learning_rate": 6.212941569805508e-07, | |
| "loss": 0.2115, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 1.7979568671963677, | |
| "grad_norm": 0.6288184982100428, | |
| "learning_rate": 6.030737921409169e-07, | |
| "loss": 0.2074, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 1.8009837306091563, | |
| "grad_norm": 0.6176175191460814, | |
| "learning_rate": 5.851162795274445e-07, | |
| "loss": 0.2195, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 1.8040105940219449, | |
| "grad_norm": 0.6273418867343108, | |
| "learning_rate": 5.674221214503639e-07, | |
| "loss": 0.225, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 1.8070374574347332, | |
| "grad_norm": 0.588581351357251, | |
| "learning_rate": 5.499918128533155e-07, | |
| "loss": 0.207, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 1.8100643208475218, | |
| "grad_norm": 0.6211106468216192, | |
| "learning_rate": 5.328258412994958e-07, | |
| "loss": 0.2165, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 1.8130911842603101, | |
| "grad_norm": 0.6200054754133695, | |
| "learning_rate": 5.159246869580348e-07, | |
| "loss": 0.2257, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 1.8161180476730987, | |
| "grad_norm": 0.6040323317085101, | |
| "learning_rate": 4.992888225905467e-07, | |
| "loss": 0.2151, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.8191449110858873, | |
| "grad_norm": 0.6197797447831195, | |
| "learning_rate": 4.829187135379221e-07, | |
| "loss": 0.2222, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 1.8221717744986758, | |
| "grad_norm": 0.6101323752003237, | |
| "learning_rate": 4.6681481770729844e-07, | |
| "loss": 0.2053, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 1.8251986379114642, | |
| "grad_norm": 0.5916551656407122, | |
| "learning_rate": 4.509775855592613e-07, | |
| "loss": 0.2183, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 1.8282255013242528, | |
| "grad_norm": 0.6019259174373504, | |
| "learning_rate": 4.354074600952407e-07, | |
| "loss": 0.2045, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 1.8312523647370411, | |
| "grad_norm": 0.5970683787684781, | |
| "learning_rate": 4.2010487684511105e-07, | |
| "loss": 0.2174, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 1.8342792281498297, | |
| "grad_norm": 0.6107661297518818, | |
| "learning_rate": 4.0507026385502747e-07, | |
| "loss": 0.2079, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 1.8373060915626183, | |
| "grad_norm": 0.6196094186069546, | |
| "learning_rate": 3.9030404167542777e-07, | |
| "loss": 0.2129, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 1.8403329549754068, | |
| "grad_norm": 0.6245656824323225, | |
| "learning_rate": 3.7580662334929517e-07, | |
| "loss": 0.2142, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 1.8433598183881952, | |
| "grad_norm": 0.6261563188642632, | |
| "learning_rate": 3.615784144005796e-07, | |
| "loss": 0.2082, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 1.8463866818009838, | |
| "grad_norm": 0.6035889077740318, | |
| "learning_rate": 3.476198128228736e-07, | |
| "loss": 0.2072, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.849413545213772, | |
| "grad_norm": 0.6018688224473584, | |
| "learning_rate": 3.339312090682689e-07, | |
| "loss": 0.213, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 1.8524404086265607, | |
| "grad_norm": 0.6368077489016595, | |
| "learning_rate": 3.2051298603643754e-07, | |
| "loss": 0.2202, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 1.8554672720393492, | |
| "grad_norm": 0.6384719725357474, | |
| "learning_rate": 3.0736551906392354e-07, | |
| "loss": 0.218, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 1.8584941354521378, | |
| "grad_norm": 0.6108745923266727, | |
| "learning_rate": 2.9448917591363923e-07, | |
| "loss": 0.2021, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 1.8615209988649262, | |
| "grad_norm": 0.6401280789971274, | |
| "learning_rate": 2.818843167645835e-07, | |
| "loss": 0.2174, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 1.8645478622777147, | |
| "grad_norm": 0.6107898414846253, | |
| "learning_rate": 2.6955129420176193e-07, | |
| "loss": 0.2187, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 1.867574725690503, | |
| "grad_norm": 0.6353025665807718, | |
| "learning_rate": 2.5749045320632824e-07, | |
| "loss": 0.2152, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 1.8706015891032917, | |
| "grad_norm": 0.5991316639593026, | |
| "learning_rate": 2.4570213114592957e-07, | |
| "loss": 0.2094, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 1.8736284525160802, | |
| "grad_norm": 0.595107099698654, | |
| "learning_rate": 2.3418665776527738e-07, | |
| "loss": 0.2092, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 1.8766553159288688, | |
| "grad_norm": 0.6128185700572673, | |
| "learning_rate": 2.2294435517691504e-07, | |
| "loss": 0.2114, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.8796821793416572, | |
| "grad_norm": 0.5907469227255048, | |
| "learning_rate": 2.119755378522137e-07, | |
| "loss": 0.2142, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 1.8827090427544457, | |
| "grad_norm": 0.6125166555743226, | |
| "learning_rate": 2.0128051261257165e-07, | |
| "loss": 0.223, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 1.885735906167234, | |
| "grad_norm": 0.6222328648678676, | |
| "learning_rate": 1.908595786208367e-07, | |
| "loss": 0.2125, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 1.8887627695800226, | |
| "grad_norm": 0.6208949225757768, | |
| "learning_rate": 1.8071302737293294e-07, | |
| "loss": 0.2209, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 1.8917896329928112, | |
| "grad_norm": 0.5949954187177511, | |
| "learning_rate": 1.7084114268971275e-07, | |
| "loss": 0.2055, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.8948164964055998, | |
| "grad_norm": 0.6279787613827442, | |
| "learning_rate": 1.612442007090076e-07, | |
| "loss": 0.2198, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 1.8978433598183881, | |
| "grad_norm": 0.632568419768614, | |
| "learning_rate": 1.519224698779198e-07, | |
| "loss": 0.2219, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 1.9008702232311767, | |
| "grad_norm": 0.5978140886026606, | |
| "learning_rate": 1.4287621094529524e-07, | |
| "loss": 0.2176, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 1.903897086643965, | |
| "grad_norm": 0.6294482998886379, | |
| "learning_rate": 1.3410567695444576e-07, | |
| "loss": 0.2119, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 1.9069239500567536, | |
| "grad_norm": 0.6224470367064877, | |
| "learning_rate": 1.2561111323605714e-07, | |
| "loss": 0.2162, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.9099508134695422, | |
| "grad_norm": 0.628133186577987, | |
| "learning_rate": 1.1739275740134004e-07, | |
| "loss": 0.2232, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 1.9129776768823308, | |
| "grad_norm": 0.6283243441163742, | |
| "learning_rate": 1.0945083933537104e-07, | |
| "loss": 0.2175, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 1.9160045402951191, | |
| "grad_norm": 0.6361574810059025, | |
| "learning_rate": 1.0178558119067316e-07, | |
| "loss": 0.2195, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 1.9190314037079077, | |
| "grad_norm": 0.6237295830068026, | |
| "learning_rate": 9.439719738099318e-08, | |
| "loss": 0.2141, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 1.922058267120696, | |
| "grad_norm": 0.6200145946661065, | |
| "learning_rate": 8.728589457530857e-08, | |
| "loss": 0.2135, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 1.9250851305334846, | |
| "grad_norm": 0.6356561072850188, | |
| "learning_rate": 8.04518716920466e-08, | |
| "loss": 0.2126, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 1.9281119939462732, | |
| "grad_norm": 0.5945643492480908, | |
| "learning_rate": 7.389531989351773e-08, | |
| "loss": 0.2104, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 1.9311388573590618, | |
| "grad_norm": 0.6402439652276496, | |
| "learning_rate": 6.761642258056977e-08, | |
| "loss": 0.2187, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 1.9341657207718501, | |
| "grad_norm": 0.603567925495169, | |
| "learning_rate": 6.161535538745877e-08, | |
| "loss": 0.2131, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 1.9371925841846387, | |
| "grad_norm": 0.6002926930038147, | |
| "learning_rate": 5.5892286176932875e-08, | |
| "loss": 0.2113, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.940219447597427, | |
| "grad_norm": 0.6273153003750616, | |
| "learning_rate": 5.044737503554165e-08, | |
| "loss": 0.2286, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 1.9432463110102156, | |
| "grad_norm": 0.5826269280634375, | |
| "learning_rate": 4.528077426915412e-08, | |
| "loss": 0.202, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 1.9462731744230042, | |
| "grad_norm": 0.628486606139057, | |
| "learning_rate": 4.0392628398699954e-08, | |
| "loss": 0.2153, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 1.9493000378357928, | |
| "grad_norm": 0.6393233348032256, | |
| "learning_rate": 3.578307415612714e-08, | |
| "loss": 0.2216, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 1.952326901248581, | |
| "grad_norm": 0.6064717476805179, | |
| "learning_rate": 3.1452240480577265e-08, | |
| "loss": 0.2132, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 1.9553537646613697, | |
| "grad_norm": 0.5896315470014438, | |
| "learning_rate": 2.7400248514776184e-08, | |
| "loss": 0.205, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 1.958380628074158, | |
| "grad_norm": 0.6491973980872668, | |
| "learning_rate": 2.3627211601651157e-08, | |
| "loss": 0.2303, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 1.9614074914869466, | |
| "grad_norm": 0.6192374126779221, | |
| "learning_rate": 2.013323528115674e-08, | |
| "loss": 0.2251, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 1.9644343548997352, | |
| "grad_norm": 0.603425710878342, | |
| "learning_rate": 1.6918417287318245e-08, | |
| "loss": 0.2107, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 1.9674612183125237, | |
| "grad_norm": 0.6227981039567323, | |
| "learning_rate": 1.3982847545507271e-08, | |
| "loss": 0.219, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.970488081725312, | |
| "grad_norm": 0.62503343816089, | |
| "learning_rate": 1.1326608169920373e-08, | |
| "loss": 0.219, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 1.9735149451381007, | |
| "grad_norm": 0.6273063499433149, | |
| "learning_rate": 8.949773461282008e-09, | |
| "loss": 0.2194, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 1.976541808550889, | |
| "grad_norm": 0.6031346065641601, | |
| "learning_rate": 6.8524099047695415e-09, | |
| "loss": 0.2095, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 1.9795686719636776, | |
| "grad_norm": 0.6263699767303774, | |
| "learning_rate": 5.034576168149175e-09, | |
| "loss": 0.2123, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 1.9825955353764662, | |
| "grad_norm": 0.6398196516787586, | |
| "learning_rate": 3.4963231001383657e-09, | |
| "loss": 0.2187, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 1.9856223987892547, | |
| "grad_norm": 0.6141286388130531, | |
| "learning_rate": 2.237693728981416e-09, | |
| "loss": 0.2176, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 1.988649262202043, | |
| "grad_norm": 0.6132661711891038, | |
| "learning_rate": 1.2587232612493172e-09, | |
| "loss": 0.2158, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 1.9916761256148316, | |
| "grad_norm": 0.6106569698784038, | |
| "learning_rate": 5.594390808494332e-10, | |
| "loss": 0.2041, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 1.99470298902762, | |
| "grad_norm": 0.5959818832015624, | |
| "learning_rate": 1.3986074826388697e-10, | |
| "loss": 0.2103, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 1.9977298524404086, | |
| "grad_norm": 0.6209940539433801, | |
| "learning_rate": 0.0, | |
| "loss": 0.2286, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.9977298524404086, | |
| "step": 660, | |
| "total_flos": 2.1062994134617293e+17, | |
| "train_loss": 0.3068675411470009, | |
| "train_runtime": 2805.0436, | |
| "train_samples_per_second": 30.141, | |
| "train_steps_per_second": 0.235 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 660, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 2.1062994134617293e+17, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |