| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 144, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 2.065, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 2.0734, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 6e-06, | |
| "loss": 2.0409, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 1.9565, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1e-05, | |
| "loss": 1.9267, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.99872299773906e-06, | |
| "loss": 1.9743, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.994892643250147e-06, | |
| "loss": 1.9656, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.9885108930818e-06, | |
| "loss": 1.9553, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.979581007037776e-06, | |
| "loss": 1.8928, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.968107546511942e-06, | |
| "loss": 1.9141, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.95409637215831e-06, | |
| "loss": 1.9039, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.937554640897414e-06, | |
| "loss": 1.8886, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.918490802260538e-06, | |
| "loss": 1.871, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.896914594073703e-06, | |
| "loss": 1.8569, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.87283703748356e-06, | |
| "loss": 1.8244, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.846270431327793e-06, | |
| "loss": 1.797, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.817228345852853e-06, | |
| "loss": 1.833, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.785725615782262e-06, | |
| "loss": 1.8323, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.751778332739033e-06, | |
| "loss": 1.7928, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.715403837026046e-06, | |
| "loss": 1.839, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.676620708768608e-06, | |
| "loss": 1.8033, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.635448758423703e-06, | |
| "loss": 1.7964, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.591909016660806e-06, | |
| "loss": 1.7938, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.546023723619387e-06, | |
| "loss": 1.8156, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.497816317548625e-06, | |
| "loss": 1.8056, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.447311422835141e-06, | |
| "loss": 1.8169, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.39453483742483e-06, | |
| "loss": 1.7668, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.33951351964525e-06, | |
| "loss": 1.7645, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.28227557443528e-06, | |
| "loss": 1.7752, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.222850238989104e-06, | |
| "loss": 1.7713, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.161267867821802e-06, | |
| "loss": 1.7592, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.097559917264268e-06, | |
| "loss": 1.7947, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.031758929395259e-06, | |
| "loss": 1.7783, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.963898515418885e-06, | |
| "loss": 1.7728, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.89401333849598e-06, | |
| "loss": 1.747, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 8.82213909603812e-06, | |
| "loss": 1.7694, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 8.748312501473351e-06, | |
| "loss": 1.7631, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 8.672571265492944e-06, | |
| "loss": 1.7887, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 8.594954076788736e-06, | |
| "loss": 1.7259, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 8.515500582290914e-06, | |
| "loss": 1.7392, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 8.434251366916323e-06, | |
| "loss": 1.7701, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 8.351247932837655e-06, | |
| "loss": 1.7556, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 8.266532678284103e-06, | |
| "loss": 1.7706, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 8.18014887588431e-06, | |
| "loss": 1.744, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 8.092140650562665e-06, | |
| "loss": 1.7396, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 8.002552957000254e-06, | |
| "loss": 1.7493, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.911431556671967e-06, | |
| "loss": 1.7445, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.818822994471504e-06, | |
| "loss": 1.75, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 7.72477457493619e-06, | |
| "loss": 1.7622, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 7.629334338083774e-06, | |
| "loss": 1.7649, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 7.532551034873558e-06, | |
| "loss": 1.7704, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 7.43447410230435e-06, | |
| "loss": 1.7634, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 7.335153638162005e-06, | |
| "loss": 1.7069, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 7.234640375429427e-06, | |
| "loss": 1.7534, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 7.132985656372126e-06, | |
| "loss": 1.7701, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 7.030241406312528e-06, | |
| "loss": 1.7765, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.926460107106483e-06, | |
| "loss": 1.7523, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.8216947703354815e-06, | |
| "loss": 1.7728, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.715998910228296e-06, | |
| "loss": 1.7448, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 6.609426516325859e-06, | |
| "loss": 1.7383, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 6.502032025903356e-06, | |
| "loss": 1.7549, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 6.393870296163616e-06, | |
| "loss": 1.7524, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 6.284996576216014e-06, | |
| "loss": 1.7553, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 6.175466478855161e-06, | |
| "loss": 1.7088, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 6.065335952153846e-06, | |
| "loss": 1.725, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.954661250884704e-06, | |
| "loss": 1.7589, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.843498907785236e-06, | |
| "loss": 1.7447, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.731905704680834e-06, | |
| "loss": 1.7236, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.6199386434805615e-06, | |
| "loss": 1.7343, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.507654917060541e-06, | |
| "loss": 1.7354, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.395111880049775e-06, | |
| "loss": 1.683, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5.28236701953335e-06, | |
| "loss": 1.7345, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5.169477925687981e-06, | |
| "loss": 1.6867, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5.0565022623649e-06, | |
| "loss": 1.734, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.943497737635103e-06, | |
| "loss": 1.7232, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.830522074312019e-06, | |
| "loss": 1.7135, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.717632980466652e-06, | |
| "loss": 1.6894, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.6048881199502265e-06, | |
| "loss": 1.704, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.49234508293946e-06, | |
| "loss": 1.6857, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.38006135651944e-06, | |
| "loss": 1.7471, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.268094295319167e-06, | |
| "loss": 1.7243, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.1565010922147644e-06, | |
| "loss": 1.7297, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.045338749115299e-06, | |
| "loss": 1.7297, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 3.934664047846157e-06, | |
| "loss": 1.7569, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 3.8245335211448404e-06, | |
| "loss": 1.7305, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 3.715003423783986e-06, | |
| "loss": 1.7345, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 3.6061297038363853e-06, | |
| "loss": 1.7571, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.497967974096647e-06, | |
| "loss": 1.7386, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.3905734836741415e-06, | |
| "loss": 1.6932, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.2840010897717045e-06, | |
| "loss": 1.7224, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.178305229664519e-06, | |
| "loss": 1.7469, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.073539892893519e-06, | |
| "loss": 1.7062, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 2.969758593687475e-06, | |
| "loss": 1.7002, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 2.8670143436278757e-06, | |
| "loss": 1.7253, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 2.765359624570574e-06, | |
| "loss": 1.7215, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 2.664846361837997e-06, | |
| "loss": 1.6811, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 2.565525897695651e-06, | |
| "loss": 1.7047, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 2.4674489651264433e-06, | |
| "loss": 1.7149, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 2.3706656619162278e-06, | |
| "loss": 1.6934, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 2.275225425063813e-06, | |
| "loss": 1.6721, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 2.1811770055284968e-06, | |
| "loss": 1.7282, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 2.0885684433280336e-06, | |
| "loss": 1.7068, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.9974470429997482e-06, | |
| "loss": 1.727, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.907859349437336e-06, | |
| "loss": 1.6693, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.8198511241156902e-06, | |
| "loss": 1.7075, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.7334673217158976e-06, | |
| "loss": 1.6837, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.6487520671623469e-06, | |
| "loss": 1.743, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.5657486330836786e-06, | |
| "loss": 1.7136, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.4844994177090871e-06, | |
| "loss": 1.7284, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.4050459232112652e-06, | |
| "loss": 1.718, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.3274287345070564e-06, | |
| "loss": 1.6898, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.2516874985266508e-06, | |
| "loss": 1.704, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.1778609039618804e-06, | |
| "loss": 1.7421, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.1059866615040205e-06, | |
| "loss": 1.6848, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.036101484581117e-06, | |
| "loss": 1.7273, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 9.68241070604743e-07, | |
| "loss": 1.7132, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 9.024400827357344e-07, | |
| "loss": 1.7239, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 8.387321321781977e-07, | |
| "loss": 1.6926, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.771497610108981e-07, | |
| "loss": 1.6812, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.177244255647209e-07, | |
| "loss": 1.7045, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 6.604864803547511e-07, | |
| "loss": 1.7108, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 6.054651625751717e-07, | |
| "loss": 1.6969, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 5.526885771648599e-07, | |
| "loss": 1.7463, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 5.021836824513759e-07, | |
| "loss": 1.7348, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 4.5397627638061604e-07, | |
| "loss": 1.7276, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 4.080909833391944e-07, | |
| "loss": 1.6837, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.6455124157629805e-07, | |
| "loss": 1.7146, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.233792912313943e-07, | |
| "loss": 1.7009, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 2.8459616297395464e-07, | |
| "loss": 1.72, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 2.482216672609677e-07, | |
| "loss": 1.6926, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 2.142743842177386e-07, | |
| "loss": 1.7535, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.8277165414714858e-07, | |
| "loss": 1.7184, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.5372956867220678e-07, | |
| "loss": 1.7114, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.2716296251644e-07, | |
| "loss": 1.7173, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.0308540592629756e-07, | |
| "loss": 1.7009, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 8.150919773946165e-08, | |
| "loss": 1.7014, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 6.244535910258697e-08, | |
| "loss": 1.6968, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.590362784169022e-08, | |
| "loss": 1.6818, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.1892453488058803e-08, | |
| "loss": 1.7045, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.0418992962224495e-08, | |
| "loss": 1.6931, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.1489106918200487e-08, | |
| "loss": 1.6903, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 5.107356749853298e-09, | |
| "loss": 1.7394, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.2770022609409628e-09, | |
| "loss": 1.7425, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.0, | |
| "loss": 1.7309, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 144, | |
| "total_flos": 0.0, | |
| "train_loss": 1.7571394882268376, | |
| "train_runtime": 4539.073, | |
| "train_samples_per_second": 3.561, | |
| "train_steps_per_second": 0.032 | |
| } | |
| ], | |
| "logging_steps": 1.0, | |
| "max_steps": 144, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 100, | |
| "total_flos": 0.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |