| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.7777777777777777, | |
| "eval_steps": 500, | |
| "global_step": 400, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.692307692307694e-07, | |
| "loss": 2.065, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.5384615384615387e-06, | |
| "loss": 2.0734, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 2.307692307692308e-06, | |
| "loss": 2.0542, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 3.0769230769230774e-06, | |
| "loss": 2.0148, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 3.846153846153847e-06, | |
| "loss": 1.9962, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.615384615384616e-06, | |
| "loss": 1.9589, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 5.384615384615385e-06, | |
| "loss": 1.9399, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 6.153846153846155e-06, | |
| "loss": 1.932, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 6.923076923076923e-06, | |
| "loss": 1.9223, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 7.692307692307694e-06, | |
| "loss": 1.9535, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 8.461538461538462e-06, | |
| "loss": 1.9152, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.230769230769232e-06, | |
| "loss": 1.9174, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1e-05, | |
| "loss": 1.8982, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.999859456853116e-06, | |
| "loss": 1.8829, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.99943783531341e-06, | |
| "loss": 1.8493, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.998735159083295e-06, | |
| "loss": 1.8207, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.997751467665295e-06, | |
| "loss": 1.8492, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.996486816359851e-06, | |
| "loss": 1.8488, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.994941276262188e-06, | |
| "loss": 1.807, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.99311493425834e-06, | |
| "loss": 1.8532, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.991007893020242e-06, | |
| "loss": 1.8169, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.98862027099998e-06, | |
| "loss": 1.8074, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.985952202423116e-06, | |
| "loss": 1.802, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.983003837281152e-06, | |
| "loss": 1.8236, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.979775341323097e-06, | |
| "loss": 1.8138, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.976266896046143e-06, | |
| "loss": 1.8273, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.972478698685463e-06, | |
| "loss": 1.7713, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.968410962203131e-06, | |
| "loss": 1.7721, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.964063915276141e-06, | |
| "loss": 1.7798, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.959437802283552e-06, | |
| "loss": 1.7767, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.954532883292761e-06, | |
| "loss": 1.7657, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.949349434044862e-06, | |
| "loss": 1.7999, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.943887745939164e-06, | |
| "loss": 1.785, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.938148126016805e-06, | |
| "loss": 1.7797, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.932130896943477e-06, | |
| "loss": 1.752, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.925836396991309e-06, | |
| "loss": 1.7729, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.919264980019829e-06, | |
| "loss": 1.7665, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.912417015456088e-06, | |
| "loss": 1.7936, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.905292888273883e-06, | |
| "loss": 1.7327, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.897892998972113e-06, | |
| "loss": 1.7431, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.89021776355227e-06, | |
| "loss": 1.775, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.882267613495049e-06, | |
| "loss": 1.7589, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 9.874042995736095e-06, | |
| "loss": 1.7761, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 9.865544372640872e-06, | |
| "loss": 1.749, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 9.85677222197867e-06, | |
| "loss": 1.7459, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 9.847727036895759e-06, | |
| "loss": 1.7583, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 9.838409325887643e-06, | |
| "loss": 1.7499, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 9.828819612770497e-06, | |
| "loss": 1.7571, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 9.818958436651704e-06, | |
| "loss": 1.7692, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 9.808826351899551e-06, | |
| "loss": 1.7719, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 9.798423928112071e-06, | |
| "loss": 1.7769, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 9.787751750085014e-06, | |
| "loss": 1.7687, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 9.77681041777897e-06, | |
| "loss": 1.7137, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 9.765600546285654e-06, | |
| "loss": 1.7616, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 9.754122765793306e-06, | |
| "loss": 1.7783, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 9.742377721551286e-06, | |
| "loss": 1.782, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 9.730366073833785e-06, | |
| "loss": 1.7578, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 9.718088497902709e-06, | |
| "loss": 1.779, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 9.705545683969722e-06, | |
| "loss": 1.7519, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 9.692738337157441e-06, | |
| "loss": 1.7469, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 9.679667177459794e-06, | |
| "loss": 1.7611, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 9.66633293970155e-06, | |
| "loss": 1.758, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 9.652736373497001e-06, | |
| "loss": 1.7635, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 9.63887824320783e-06, | |
| "loss": 1.7144, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 9.624759327900131e-06, | |
| "loss": 1.7299, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 9.610380421300623e-06, | |
| "loss": 1.7645, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 9.595742331752014e-06, | |
| "loss": 1.7515, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 9.580845882167574e-06, | |
| "loss": 1.7299, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 9.565691909984864e-06, | |
| "loss": 1.7419, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 9.550281267118659e-06, | |
| "loss": 1.7442, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 9.534614819913056e-06, | |
| "loss": 1.6911, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 9.518693449092772e-06, | |
| "loss": 1.7422, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 9.502518049713633e-06, | |
| "loss": 1.6952, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 9.486089531112247e-06, | |
| "loss": 1.7407, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.469408816854898e-06, | |
| "loss": 1.7308, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.452476844685611e-06, | |
| "loss": 1.7235, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.435294566473453e-06, | |
| "loss": 1.6946, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.417862948158997e-06, | |
| "loss": 1.7177, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 9.40018296970005e-06, | |
| "loss": 1.6954, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 9.382255625016527e-06, | |
| "loss": 1.76, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 9.364081921934607e-06, | |
| "loss": 1.7361, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 9.345662882130056e-06, | |
| "loss": 1.739, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 9.326999541070804e-06, | |
| "loss": 1.7371, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 9.308092947958725e-06, | |
| "loss": 1.7674, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 9.288944165670651e-06, | |
| "loss": 1.7409, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 9.269554270698636e-06, | |
| "loss": 1.7435, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 9.24992435308942e-06, | |
| "loss": 1.7686, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 9.23005551638316e-06, | |
| "loss": 1.7498, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 9.209948877551393e-06, | |
| "loss": 1.7058, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 9.189605566934235e-06, | |
| "loss": 1.7349, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 9.169026728176845e-06, | |
| "loss": 1.7551, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 9.148213518165121e-06, | |
| "loss": 1.7166, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 9.127167106960682e-06, | |
| "loss": 1.7136, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 9.105888677735069e-06, | |
| "loss": 1.7347, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 9.084379426703245e-06, | |
| "loss": 1.7336, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 9.062640563056339e-06, | |
| "loss": 1.693, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 9.040673308893677e-06, | |
| "loss": 1.7118, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 9.018478899154068e-06, | |
| "loss": 1.7265, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 8.996058581546386e-06, | |
| "loss": 1.7073, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 8.973413616479429e-06, | |
| "loss": 1.6815, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 8.95054527699106e-06, | |
| "loss": 1.7396, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 8.927454848676633e-06, | |
| "loss": 1.7189, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 8.904143629616735e-06, | |
| "loss": 1.7371, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 8.880612930304196e-06, | |
| "loss": 1.6812, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 8.856864073570429e-06, | |
| "loss": 1.7201, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 8.83289839451106e-06, | |
| "loss": 1.6916, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 8.80871724041087e-06, | |
| "loss": 1.7577, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 8.784321970668054e-06, | |
| "loss": 1.7217, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 8.759713956717804e-06, | |
| "loss": 1.7391, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 8.734894581955208e-06, | |
| "loss": 1.7288, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 8.70986524165748e-06, | |
| "loss": 1.6997, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 8.684627342905519e-06, | |
| "loss": 1.7126, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 8.65918230450481e-06, | |
| "loss": 1.7519, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 8.63353155690566e-06, | |
| "loss": 1.696, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 8.607676542122782e-06, | |
| "loss": 1.7394, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 8.581618713654239e-06, | |
| "loss": 1.724, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 8.55535953639971e-06, | |
| "loss": 1.7395, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 8.528900486578158e-06, | |
| "loss": 1.7046, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 8.502243051644838e-06, | |
| "loss": 1.6919, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 8.475388730207662e-06, | |
| "loss": 1.7136, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 8.44833903194297e-06, | |
| "loss": 1.7178, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 8.421095477510648e-06, | |
| "loss": 1.7065, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 8.393659598468644e-06, | |
| "loss": 1.7563, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 8.366032937186869e-06, | |
| "loss": 1.7431, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 8.33821704676049e-06, | |
| "loss": 1.7337, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 8.310213490922616e-06, | |
| "loss": 1.6932, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 8.282023843956392e-06, | |
| "loss": 1.7132, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 8.253649690606495e-06, | |
| "loss": 1.7068, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 8.225092625990047e-06, | |
| "loss": 1.7296, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 8.196354255506937e-06, | |
| "loss": 1.6946, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 8.167436194749576e-06, | |
| "loss": 1.7605, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 8.138340069412069e-06, | |
| "loss": 1.724, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 8.109067515198822e-06, | |
| "loss": 1.7175, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 8.079620177732587e-06, | |
| "loss": 1.725, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 8.049999712461956e-06, | |
| "loss": 1.7061, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 8.020207784568293e-06, | |
| "loss": 1.7019, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 7.990246068872111e-06, | |
| "loss": 1.7027, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 7.960116249738939e-06, | |
| "loss": 1.6869, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 7.92982002098461e-06, | |
| "loss": 1.7039, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 7.899359085780062e-06, | |
| "loss": 1.6941, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 7.868735156555567e-06, | |
| "loss": 1.694, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 7.83794995490448e-06, | |
| "loss": 1.7478, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 7.807005211486445e-06, | |
| "loss": 1.745, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 7.775902665930114e-06, | |
| "loss": 1.7377, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 7.744644066735335e-06, | |
| "loss": 1.6321, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 7.713231171174868e-06, | |
| "loss": 1.639, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 7.681665745195593e-06, | |
| "loss": 1.6131, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 7.649949563319228e-06, | |
| "loss": 1.5813, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 7.618084408542576e-06, | |
| "loss": 1.5663, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 7.586072072237291e-06, | |
| "loss": 1.5375, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 7.5539143540491635e-06, | |
| "loss": 1.5059, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 7.521613061796957e-06, | |
| "loss": 1.53, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 7.48917001137078e-06, | |
| "loss": 1.5345, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 7.456587026629991e-06, | |
| "loss": 1.5892, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 7.423865939300674e-06, | |
| "loss": 1.5414, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 7.391008588872661e-06, | |
| "loss": 1.5132, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 7.358016822496126e-06, | |
| "loss": 1.4788, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 7.324892494877734e-06, | |
| "loss": 1.4733, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 7.29163746817638e-06, | |
| "loss": 1.4576, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 7.258253611898509e-06, | |
| "loss": 1.3702, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 7.224742802793005e-06, | |
| "loss": 1.4155, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 7.191106924745695e-06, | |
| "loss": 1.4572, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 7.157347868673441e-06, | |
| "loss": 1.4039, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 7.1234675324178295e-06, | |
| "loss": 1.4494, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 7.089467820638491e-06, | |
| "loss": 1.4039, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 7.055350644706023e-06, | |
| "loss": 1.4112, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 7.021117922594532e-06, | |
| "loss": 1.3878, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 6.986771578773812e-06, | |
| "loss": 1.4093, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 6.952313544101165e-06, | |
| "loss": 1.4049, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 6.917745755712839e-06, | |
| "loss": 1.4388, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 6.8830701569151394e-06, | |
| "loss": 1.3516, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 6.8482886970751785e-06, | |
| "loss": 1.3582, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 6.81340333151128e-06, | |
| "loss": 1.3546, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 6.7784160213830696e-06, | |
| "loss": 1.3703, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 6.743328733581211e-06, | |
| "loss": 1.3635, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 6.708143440616845e-06, | |
| "loss": 1.4029, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 6.672862120510688e-06, | |
| "loss": 1.3707, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 6.637486756681843e-06, | |
| "loss": 1.3997, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 6.602019337836291e-06, | |
| "loss": 1.3705, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 6.566461857855096e-06, | |
| "loss": 1.3657, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 6.5308163156823064e-06, | |
| "loss": 1.3924, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 6.495084715212597e-06, | |
| "loss": 1.4033, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 6.459269065178592e-06, | |
| "loss": 1.3618, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 6.423371379037957e-06, | |
| "loss": 1.3641, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 6.387393674860205e-06, | |
| "loss": 1.4002, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 6.351337975213239e-06, | |
| "loss": 1.3755, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 6.315206307049656e-06, | |
| "loss": 1.4215, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 6.2790007015927946e-06, | |
| "loss": 1.3548, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 6.242723194222546e-06, | |
| "loss": 1.3778, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 6.2063758243609275e-06, | |
| "loss": 1.3512, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 6.169960635357437e-06, | |
| "loss": 1.3625, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 6.133479674374176e-06, | |
| "loss": 1.3779, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 6.0969349922707675e-06, | |
| "loss": 1.4155, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 6.060328643489064e-06, | |
| "loss": 1.3929, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 6.023662685937643e-06, | |
| "loss": 1.4223, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 5.9869391808761315e-06, | |
| "loss": 1.381, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 5.9501601927993135e-06, | |
| "loss": 1.3046, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 5.9133277893210785e-06, | |
| "loss": 1.4541, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 5.8764440410581846e-06, | |
| "loss": 1.4183, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 5.839511021513853e-06, | |
| "loss": 1.3938, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 5.802530806961195e-06, | |
| "loss": 1.371, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 5.765505476326505e-06, | |
| "loss": 1.4197, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 5.728437111072376e-06, | |
| "loss": 1.3727, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 5.691327795080685e-06, | |
| "loss": 1.3876, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 5.654179614535457e-06, | |
| "loss": 1.3934, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 5.616994657805566e-06, | |
| "loss": 1.3715, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 5.579775015327347e-06, | |
| "loss": 1.4043, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 5.5425227794870715e-06, | |
| "loss": 1.3275, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 5.505240044503324e-06, | |
| "loss": 1.3601, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 5.46792890630926e-06, | |
| "loss": 1.4168, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 5.430591462434792e-06, | |
| "loss": 1.3915, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 5.393229811888663e-06, | |
| "loss": 1.3551, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 5.355846055040449e-06, | |
| "loss": 1.3995, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 5.318442293502482e-06, | |
| "loss": 1.3818, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 5.281020630011703e-06, | |
| "loss": 1.3313, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 5.2435831683114515e-06, | |
| "loss": 1.3647, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 5.206132013033199e-06, | |
| "loss": 1.3176, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 5.1686692695782325e-06, | |
| "loss": 1.3863, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 5.131197043999294e-06, | |
| "loss": 1.4079, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 5.093717442882185e-06, | |
| "loss": 1.3518, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 5.0562325732273405e-06, | |
| "loss": 1.3355, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 5.018744542331376e-06, | |
| "loss": 1.3873, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 4.981255457668625e-06, | |
| "loss": 1.3487, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 4.94376742677266e-06, | |
| "loss": 1.4021, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 4.906282557117817e-06, | |
| "loss": 1.3814, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 4.8688029560007064e-06, | |
| "loss": 1.3848, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 4.831330730421769e-06, | |
| "loss": 1.4255, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 4.793867986966802e-06, | |
| "loss": 1.3916, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 4.756416831688549e-06, | |
| "loss": 1.3955, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 4.718979369988299e-06, | |
| "loss": 1.4237, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 4.681557706497519e-06, | |
| "loss": 1.4423, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 4.644153944959553e-06, | |
| "loss": 1.405, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 4.606770188111339e-06, | |
| "loss": 1.3454, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 4.5694085375652105e-06, | |
| "loss": 1.4008, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 4.532071093690741e-06, | |
| "loss": 1.4151, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 4.494759955496678e-06, | |
| "loss": 1.3832, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 4.457477220512929e-06, | |
| "loss": 1.3788, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 4.420224984672654e-06, | |
| "loss": 1.3766, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 4.383005342194436e-06, | |
| "loss": 1.4192, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 4.345820385464543e-06, | |
| "loss": 1.3277, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 4.308672204919316e-06, | |
| "loss": 1.3533, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 4.271562888927626e-06, | |
| "loss": 1.3872, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 4.234494523673497e-06, | |
| "loss": 1.3871, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 4.1974691930388055e-06, | |
| "loss": 1.3475, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 4.16048897848615e-06, | |
| "loss": 1.3901, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 4.123555958941817e-06, | |
| "loss": 1.3747, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 4.0866722106789214e-06, | |
| "loss": 1.3803, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 4.049839807200688e-06, | |
| "loss": 1.3585, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 4.013060819123869e-06, | |
| "loss": 1.3699, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 3.976337314062358e-06, | |
| "loss": 1.3383, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 3.9396713565109375e-06, | |
| "loss": 1.4172, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 3.903065007729234e-06, | |
| "loss": 1.3812, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 3.866520325625825e-06, | |
| "loss": 1.3922, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 3.830039364642566e-06, | |
| "loss": 1.4036, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 3.7936241756390746e-06, | |
| "loss": 1.3689, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 3.7572768057774543e-06, | |
| "loss": 1.3692, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 3.7209992984072062e-06, | |
| "loss": 1.4185, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 3.6847936929503446e-06, | |
| "loss": 1.3918, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 3.6486620247867625e-06, | |
| "loss": 1.4451, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 3.6126063251397968e-06, | |
| "loss": 1.4484, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 3.576628620962045e-06, | |
| "loss": 1.4314, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 3.5407309348214094e-06, | |
| "loss": 1.3911, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 3.5049152847874053e-06, | |
| "loss": 1.3604, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 3.469183684317694e-06, | |
| "loss": 1.3757, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 3.4335381421449056e-06, | |
| "loss": 1.3903, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 3.39798066216371e-06, | |
| "loss": 1.3951, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 3.3625132433181573e-06, | |
| "loss": 1.4243, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 3.327137879489313e-06, | |
| "loss": 1.447, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 3.2918565593831565e-06, | |
| "loss": 1.3961, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 3.2566712664187907e-06, | |
| "loss": 1.3795, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 3.221583978616932e-06, | |
| "loss": 1.3906, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 3.1865966684887222e-06, | |
| "loss": 1.3857, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 3.1517113029248236e-06, | |
| "loss": 1.4236, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 3.1169298430848605e-06, | |
| "loss": 1.3731, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 3.082254244287163e-06, | |
| "loss": 1.4524, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 3.0476864558988364e-06, | |
| "loss": 1.4195, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 3.0132284212261886e-06, | |
| "loss": 1.4213, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 2.97888207740547e-06, | |
| "loss": 1.4078, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 2.944649355293979e-06, | |
| "loss": 1.3755, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 2.9105321793615106e-06, | |
| "loss": 1.3934, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 2.876532467582174e-06, | |
| "loss": 1.3902, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 2.842652131326562e-06, | |
| "loss": 1.3704, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 2.8088930752543063e-06, | |
| "loss": 1.3678, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 2.775257197206996e-06, | |
| "loss": 1.3996, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 2.741746388101493e-06, | |
| "loss": 1.4083, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 2.7083625318236213e-06, | |
| "loss": 1.4332, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 2.6751075051222684e-06, | |
| "loss": 1.4242, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 2.6419831775038763e-06, | |
| "loss": 1.467, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 2.6089914111273398e-06, | |
| "loss": 1.4311, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 2.576134060699328e-06, | |
| "loss": 1.3565, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 2.54341297337001e-06, | |
| "loss": 1.3257, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 2.510829988629222e-06, | |
| "loss": 1.2748, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 2.478386938203043e-06, | |
| "loss": 1.2655, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 2.4460856459508374e-06, | |
| "loss": 1.2386, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 2.4139279277627113e-06, | |
| "loss": 1.1905, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 2.381915591457424e-06, | |
| "loss": 1.1825, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 2.3500504366807743e-06, | |
| "loss": 1.2096, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 2.3183342548044067e-06, | |
| "loss": 1.2918, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 2.286768828825133e-06, | |
| "loss": 1.2533, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 2.2553559332646675e-06, | |
| "loss": 1.223, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 2.2240973340698886e-06, | |
| "loss": 1.1928, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 2.1929947885135567e-06, | |
| "loss": 1.1685, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 2.1620500450955224e-06, | |
| "loss": 1.182, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 2.1312648434444342e-06, | |
| "loss": 1.0815, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 2.100640914219939e-06, | |
| "loss": 1.1353, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 2.0701799790153897e-06, | |
| "loss": 1.1641, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 2.039883750261063e-06, | |
| "loss": 1.0813, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 2.00975393112789e-06, | |
| "loss": 1.1248, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 1.979792215431709e-06, | |
| "loss": 1.0653, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 1.9500002875380458e-06, | |
| "loss": 1.101, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 1.920379822267414e-06, | |
| "loss": 1.0595, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 1.8909324848011802e-06, | |
| "loss": 1.0685, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 1.8616599305879334e-06, | |
| "loss": 1.0513, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 1.8325638052504235e-06, | |
| "loss": 1.11, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 1.8036457444930643e-06, | |
| "loss": 1.0086, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 1.774907374009953e-06, | |
| "loss": 1.0186, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 1.7463503093935063e-06, | |
| "loss": 1.0377, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 1.7179761560436097e-06, | |
| "loss": 1.0262, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 1.6897865090773858e-06, | |
| "loss": 1.0674, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 1.661782953239512e-06, | |
| "loss": 1.1015, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 1.6339670628131327e-06, | |
| "loss": 1.0721, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 1.6063404015313583e-06, | |
| "loss": 1.0939, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 1.578904522489354e-06, | |
| "loss": 1.0636, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 1.5516609680570316e-06, | |
| "loss": 1.065, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 1.5246112697923389e-06, | |
| "loss": 1.0918, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 1.4977569483551634e-06, | |
| "loss": 1.0895, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 1.471099513421842e-06, | |
| "loss": 1.0478, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 1.4446404636002931e-06, | |
| "loss": 1.0685, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 1.4183812863457624e-06, | |
| "loss": 1.1131, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 1.3923234578772177e-06, | |
| "loss": 1.084, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 1.3664684430943431e-06, | |
| "loss": 1.1083, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 1.3408176954951912e-06, | |
| "loss": 1.0363, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 1.315372657094483e-06, | |
| "loss": 1.0793, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 1.29013475834252e-06, | |
| "loss": 1.0365, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 1.265105418044793e-06, | |
| "loss": 1.0729, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 1.2402860432821972e-06, | |
| "loss": 1.0885, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 1.2156780293319476e-06, | |
| "loss": 1.1265, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 1.1912827595891313e-06, | |
| "loss": 1.1007, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 1.1671016054889407e-06, | |
| "loss": 1.1211, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 1.1431359264295717e-06, | |
| "loss": 1.0842, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 1.1193870696958058e-06, | |
| "loss": 1.0116, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 1.0958563703832675e-06, | |
| "loss": 1.159, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 1.0725451513233676e-06, | |
| "loss": 1.1288, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 1.0494547230089413e-06, | |
| "loss": 1.0911, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 1.0265863835205709e-06, | |
| "loss": 1.0699, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 1.0039414184536161e-06, | |
| "loss": 1.1496, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 9.815211008459336e-07, | |
| "loss": 1.1034, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 9.593266911063253e-07, | |
| "loss": 1.0986, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 9.373594369436611e-07, | |
| "loss": 1.0866, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 9.15620573296756e-07, | |
| "loss": 1.0892, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 8.941113222649328e-07, | |
| "loss": 1.1079, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 8.728328930393188e-07, | |
| "loss": 1.0632, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 8.517864818348804e-07, | |
| "loss": 1.0748, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 8.309732718231578e-07, | |
| "loss": 1.1431, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 8.103944330657665e-07, | |
| "loss": 1.104, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 7.900511224486085e-07, | |
| "loss": 1.074, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 7.699444836168413e-07, | |
| "loss": 1.1328, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 7.500756469105819e-07, | |
| "loss": 1.0849, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 7.304457293013656e-07, | |
| "loss": 1.0606, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 7.1105583432935e-07, | |
| "loss": 1.1129, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 6.919070520412768e-07, | |
| "loss": 1.0376, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 6.730004589291961e-07, | |
| "loss": 1.1137, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 6.543371178699442e-07, | |
| "loss": 1.15, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 6.359180780653957e-07, | |
| "loss": 1.0695, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 6.177443749834743e-07, | |
| "loss": 1.0922, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 5.998170302999529e-07, | |
| "loss": 1.122, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 5.821370518410019e-07, | |
| "loss": 1.0837, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 5.647054335265489e-07, | |
| "loss": 1.1392, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 5.475231553143906e-07, | |
| "loss": 1.1152, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 5.305911831451044e-07, | |
| "loss": 1.1638, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 5.139104688877549e-07, | |
| "loss": 1.1806, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 4.974819502863687e-07, | |
| "loss": 1.1411, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 4.813065509072279e-07, | |
| "loss": 1.1415, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 4.6538518008694465e-07, | |
| "loss": 1.1661, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 4.4971873288134237e-07, | |
| "loss": 1.2095, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 4.343080900151375e-07, | |
| "loss": 1.1528, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 4.1915411783242766e-07, | |
| "loss": 1.0978, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 4.0425766824798817e-07, | |
| "loss": 1.1445, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 3.8961957869937893e-07, | |
| "loss": 1.169, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 3.752406720998691e-07, | |
| "loss": 1.1319, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 3.61121756792171e-07, | |
| "loss": 1.1205, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 3.472636265030005e-07, | |
| "loss": 1.1078, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 3.3366706029845097e-07, | |
| "loss": 1.1676, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 3.2033282254020747e-07, | |
| "loss": 1.0505, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 3.072616628425601e-07, | |
| "loss": 1.0925, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 2.9445431603027876e-07, | |
| "loss": 1.1381, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 2.8191150209729233e-07, | |
| "loss": 1.1355, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 2.696339261662156e-07, | |
| "loss": 1.1112, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 2.576222784487148e-07, | |
| "loss": 1.1411, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 2.45877234206694e-07, | |
| "loss": 1.1366, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 2.3439945371434792e-07, | |
| "loss": 1.1416, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 2.2318958222103004e-07, | |
| "loss": 1.113, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 2.1224824991498695e-07, | |
| "loss": 1.1216, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 2.0157607188792894e-07, | |
| "loss": 1.1175, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 1.911736481004489e-07, | |
| "loss": 1.1901, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 1.8104156334829703e-07, | |
| "loss": 1.1377, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 1.7118038722950313e-07, | |
| "loss": 1.1619, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 1.615906741123574e-07, | |
| "loss": 1.187, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 1.5227296310424244e-07, | |
| "loss": 1.1571, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 1.432277780213298e-07, | |
| "loss": 1.1464, | |
| "step": 400 | |
| } | |
| ], | |
| "logging_steps": 1.0, | |
| "max_steps": 432, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 100, | |
| "total_flos": 0.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |