| { | |
| "best_metric": 0.06827498227357864, | |
| "best_model_checkpoint": "llava-v1.5-7b-chess/checkpoint-390", | |
| "epoch": 2.4074074074074074, | |
| "eval_steps": 130, | |
| "global_step": 520, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9.984567901234569e-05, | |
| "loss": 2.5107, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.969135802469136e-05, | |
| "loss": 2.3561, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.953703703703704e-05, | |
| "loss": 2.163, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.938271604938272e-05, | |
| "loss": 1.9937, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.92283950617284e-05, | |
| "loss": 1.4659, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.907407407407407e-05, | |
| "loss": 1.3839, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.891975308641976e-05, | |
| "loss": 1.4814, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.876543209876543e-05, | |
| "loss": 1.0705, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.861111111111112e-05, | |
| "loss": 0.952, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.84567901234568e-05, | |
| "loss": 0.636, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.830246913580248e-05, | |
| "loss": 0.5561, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.814814814814815e-05, | |
| "loss": 0.3848, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.799382716049384e-05, | |
| "loss": 0.2741, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.78395061728395e-05, | |
| "loss": 0.3006, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.768518518518519e-05, | |
| "loss": 0.1832, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.753086419753087e-05, | |
| "loss": 0.3091, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.737654320987655e-05, | |
| "loss": 0.1454, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.722222222222223e-05, | |
| "loss": 0.1904, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.70679012345679e-05, | |
| "loss": 0.1467, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.691358024691359e-05, | |
| "loss": 0.1522, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.675925925925926e-05, | |
| "loss": 0.1504, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.660493827160494e-05, | |
| "loss": 0.1238, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.645061728395061e-05, | |
| "loss": 0.0872, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.62962962962963e-05, | |
| "loss": 0.2049, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.614197530864198e-05, | |
| "loss": 0.1473, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.598765432098767e-05, | |
| "loss": 0.0626, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.583333333333334e-05, | |
| "loss": 0.2654, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.567901234567902e-05, | |
| "loss": 0.1939, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.552469135802469e-05, | |
| "loss": 0.1079, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.537037037037038e-05, | |
| "loss": 0.1935, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.521604938271606e-05, | |
| "loss": 0.2057, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.506172839506173e-05, | |
| "loss": 0.0422, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.490740740740742e-05, | |
| "loss": 0.074, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.475308641975309e-05, | |
| "loss": 0.351, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.459876543209877e-05, | |
| "loss": 0.0666, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.444444444444444e-05, | |
| "loss": 0.0813, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.429012345679013e-05, | |
| "loss": 0.0498, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.41358024691358e-05, | |
| "loss": 0.0692, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.398148148148148e-05, | |
| "loss": 0.0168, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.382716049382717e-05, | |
| "loss": 0.2658, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.367283950617285e-05, | |
| "loss": 0.0337, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.351851851851852e-05, | |
| "loss": 0.0826, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.33641975308642e-05, | |
| "loss": 0.0924, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.320987654320989e-05, | |
| "loss": 0.0671, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.305555555555556e-05, | |
| "loss": 0.2447, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.290123456790125e-05, | |
| "loss": 0.3208, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.274691358024692e-05, | |
| "loss": 0.0711, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.25925925925926e-05, | |
| "loss": 0.0612, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.243827160493827e-05, | |
| "loss": 0.0505, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.228395061728396e-05, | |
| "loss": 0.12, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.212962962962963e-05, | |
| "loss": 0.0821, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.197530864197531e-05, | |
| "loss": 0.0987, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.182098765432098e-05, | |
| "loss": 0.083, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.166666666666667e-05, | |
| "loss": 0.0302, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.151234567901235e-05, | |
| "loss": 0.0583, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.135802469135802e-05, | |
| "loss": 0.0338, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.120370370370371e-05, | |
| "loss": 0.1319, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.104938271604939e-05, | |
| "loss": 0.0691, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.089506172839508e-05, | |
| "loss": 0.0636, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.074074074074075e-05, | |
| "loss": 0.1005, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.058641975308643e-05, | |
| "loss": 0.1563, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.04320987654321e-05, | |
| "loss": 0.0541, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.027777777777779e-05, | |
| "loss": 0.0599, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 9.012345679012346e-05, | |
| "loss": 0.0733, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 8.996913580246914e-05, | |
| "loss": 0.0252, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 8.981481481481481e-05, | |
| "loss": 0.0559, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 8.96604938271605e-05, | |
| "loss": 0.1204, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 8.950617283950617e-05, | |
| "loss": 0.0218, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 8.935185185185185e-05, | |
| "loss": 0.062, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 8.919753086419754e-05, | |
| "loss": 0.0455, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 8.904320987654321e-05, | |
| "loss": 0.0314, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 8.888888888888889e-05, | |
| "loss": 0.155, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 8.873456790123458e-05, | |
| "loss": 0.1251, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 8.858024691358026e-05, | |
| "loss": 0.0518, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 8.842592592592593e-05, | |
| "loss": 0.0892, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 8.827160493827162e-05, | |
| "loss": 0.0873, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 8.811728395061729e-05, | |
| "loss": 0.0324, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 8.796296296296297e-05, | |
| "loss": 0.0898, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 8.780864197530864e-05, | |
| "loss": 0.1661, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 8.765432098765433e-05, | |
| "loss": 0.0439, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 8.75e-05, | |
| "loss": 0.1008, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 8.734567901234568e-05, | |
| "loss": 0.058, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 8.719135802469135e-05, | |
| "loss": 0.0703, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 8.703703703703704e-05, | |
| "loss": 0.065, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 8.688271604938272e-05, | |
| "loss": 0.07, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 8.67283950617284e-05, | |
| "loss": 0.0911, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 8.657407407407408e-05, | |
| "loss": 0.1903, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 8.641975308641975e-05, | |
| "loss": 0.1781, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 8.626543209876543e-05, | |
| "loss": 0.0393, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 8.611111111111112e-05, | |
| "loss": 0.0683, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 8.59567901234568e-05, | |
| "loss": 0.1688, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 8.580246913580247e-05, | |
| "loss": 0.0357, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 8.564814814814816e-05, | |
| "loss": 0.1529, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 8.549382716049383e-05, | |
| "loss": 0.1086, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 8.533950617283951e-05, | |
| "loss": 0.0686, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 8.518518518518518e-05, | |
| "loss": 0.0557, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 8.503086419753087e-05, | |
| "loss": 0.1763, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 8.487654320987654e-05, | |
| "loss": 0.0451, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 8.472222222222222e-05, | |
| "loss": 0.0453, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 8.456790123456791e-05, | |
| "loss": 0.1907, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 8.441358024691358e-05, | |
| "loss": 0.05, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 8.425925925925926e-05, | |
| "loss": 0.1193, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 8.410493827160494e-05, | |
| "loss": 0.0753, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 8.395061728395062e-05, | |
| "loss": 0.0587, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 8.379629629629629e-05, | |
| "loss": 0.0572, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 8.364197530864199e-05, | |
| "loss": 0.118, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 8.348765432098766e-05, | |
| "loss": 0.1059, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 8.333333333333334e-05, | |
| "loss": 0.1691, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 8.317901234567902e-05, | |
| "loss": 0.1887, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 8.30246913580247e-05, | |
| "loss": 0.0661, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 8.287037037037037e-05, | |
| "loss": 0.2594, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 8.271604938271605e-05, | |
| "loss": 0.0911, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 8.256172839506174e-05, | |
| "loss": 0.0227, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 8.240740740740741e-05, | |
| "loss": 0.0945, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 8.22530864197531e-05, | |
| "loss": 0.0717, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 8.209876543209877e-05, | |
| "loss": 0.0953, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 8.194444444444445e-05, | |
| "loss": 0.154, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.179012345679012e-05, | |
| "loss": 0.0985, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.16358024691358e-05, | |
| "loss": 0.0707, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.148148148148148e-05, | |
| "loss": 0.0374, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.132716049382716e-05, | |
| "loss": 0.0593, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.117283950617285e-05, | |
| "loss": 0.0411, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.101851851851853e-05, | |
| "loss": 0.0654, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.08641975308642e-05, | |
| "loss": 0.0299, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 8.070987654320989e-05, | |
| "loss": 0.0134, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 8.055555555555556e-05, | |
| "loss": 0.0687, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 8.040123456790124e-05, | |
| "loss": 0.0241, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 8.024691358024692e-05, | |
| "loss": 0.1784, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 8.00925925925926e-05, | |
| "loss": 0.0642, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 7.993827160493828e-05, | |
| "loss": 0.0599, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "eval_loss": 0.08161517977714539, | |
| "eval_runtime": 45.6818, | |
| "eval_samples_per_second": 4.728, | |
| "eval_steps_per_second": 1.576, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 7.978395061728395e-05, | |
| "loss": 0.0829, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 7.962962962962964e-05, | |
| "loss": 0.0512, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 7.94753086419753e-05, | |
| "loss": 0.0201, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 7.932098765432099e-05, | |
| "loss": 0.0399, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 7.916666666666666e-05, | |
| "loss": 0.0316, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 7.901234567901235e-05, | |
| "loss": 0.1958, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 7.885802469135802e-05, | |
| "loss": 0.0638, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 7.870370370370372e-05, | |
| "loss": 0.0682, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 7.854938271604939e-05, | |
| "loss": 0.0606, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 7.839506172839507e-05, | |
| "loss": 0.0445, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 7.824074074074074e-05, | |
| "loss": 0.0633, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 7.808641975308643e-05, | |
| "loss": 0.1446, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 7.793209876543211e-05, | |
| "loss": 0.1459, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 7.777777777777778e-05, | |
| "loss": 0.0616, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 7.762345679012347e-05, | |
| "loss": 0.0508, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 7.746913580246914e-05, | |
| "loss": 0.0831, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 7.731481481481482e-05, | |
| "loss": 0.0416, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 7.716049382716049e-05, | |
| "loss": 0.0865, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 7.700617283950618e-05, | |
| "loss": 0.0472, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 7.685185185185185e-05, | |
| "loss": 0.0726, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 7.669753086419753e-05, | |
| "loss": 0.0214, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 7.65432098765432e-05, | |
| "loss": 0.0554, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 7.638888888888889e-05, | |
| "loss": 0.0857, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 7.623456790123457e-05, | |
| "loss": 0.0922, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 7.608024691358026e-05, | |
| "loss": 0.0391, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 7.592592592592593e-05, | |
| "loss": 0.0915, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 7.577160493827161e-05, | |
| "loss": 0.0738, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 7.56172839506173e-05, | |
| "loss": 0.2464, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 7.546296296296297e-05, | |
| "loss": 0.0579, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 7.530864197530865e-05, | |
| "loss": 0.0369, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 7.515432098765432e-05, | |
| "loss": 0.0187, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 7.500000000000001e-05, | |
| "loss": 0.0665, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 7.484567901234568e-05, | |
| "loss": 0.022, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 7.469135802469136e-05, | |
| "loss": 0.0811, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 7.453703703703703e-05, | |
| "loss": 0.025, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.438271604938272e-05, | |
| "loss": 0.1041, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.422839506172839e-05, | |
| "loss": 0.0691, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 7.407407407407407e-05, | |
| "loss": 0.0482, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 7.391975308641976e-05, | |
| "loss": 0.027, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 7.376543209876543e-05, | |
| "loss": 0.1431, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 7.361111111111111e-05, | |
| "loss": 0.0569, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 7.34567901234568e-05, | |
| "loss": 0.1425, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 7.330246913580248e-05, | |
| "loss": 0.0331, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 7.314814814814815e-05, | |
| "loss": 0.0367, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 7.299382716049384e-05, | |
| "loss": 0.141, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 7.283950617283951e-05, | |
| "loss": 0.0296, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 7.268518518518519e-05, | |
| "loss": 0.1589, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 7.253086419753086e-05, | |
| "loss": 0.1954, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.237654320987655e-05, | |
| "loss": 0.0294, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.222222222222222e-05, | |
| "loss": 0.0252, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 7.20679012345679e-05, | |
| "loss": 0.0735, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 7.191358024691359e-05, | |
| "loss": 0.0415, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.175925925925926e-05, | |
| "loss": 0.0541, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.160493827160494e-05, | |
| "loss": 0.0417, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 7.145061728395061e-05, | |
| "loss": 0.0174, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 7.12962962962963e-05, | |
| "loss": 0.0148, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 7.114197530864198e-05, | |
| "loss": 0.0867, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 7.098765432098767e-05, | |
| "loss": 0.0497, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 7.083333333333334e-05, | |
| "loss": 0.021, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 7.067901234567902e-05, | |
| "loss": 0.0174, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 7.05246913580247e-05, | |
| "loss": 0.0845, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 7.037037037037038e-05, | |
| "loss": 0.0829, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 7.021604938271605e-05, | |
| "loss": 0.0635, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 7.006172839506173e-05, | |
| "loss": 0.05, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 6.99074074074074e-05, | |
| "loss": 0.0283, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 6.975308641975309e-05, | |
| "loss": 0.0046, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 6.959876543209877e-05, | |
| "loss": 0.004, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 6.944444444444444e-05, | |
| "loss": 0.0743, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 6.929012345679013e-05, | |
| "loss": 0.0347, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 6.91358024691358e-05, | |
| "loss": 0.0943, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 6.898148148148148e-05, | |
| "loss": 0.0931, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.882716049382716e-05, | |
| "loss": 0.0865, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.867283950617285e-05, | |
| "loss": 0.0526, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.851851851851852e-05, | |
| "loss": 0.0852, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 6.836419753086421e-05, | |
| "loss": 0.3528, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 6.820987654320988e-05, | |
| "loss": 0.0508, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 6.805555555555556e-05, | |
| "loss": 0.1138, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 6.790123456790123e-05, | |
| "loss": 0.0263, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 6.774691358024692e-05, | |
| "loss": 0.0389, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 6.759259259259259e-05, | |
| "loss": 0.0753, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 6.743827160493827e-05, | |
| "loss": 0.1219, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 6.728395061728396e-05, | |
| "loss": 0.0235, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 6.712962962962963e-05, | |
| "loss": 0.0385, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 6.697530864197531e-05, | |
| "loss": 0.0623, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 6.682098765432099e-05, | |
| "loss": 0.0144, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 6.666666666666667e-05, | |
| "loss": 0.0202, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 6.651234567901234e-05, | |
| "loss": 0.0906, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 6.635802469135803e-05, | |
| "loss": 0.0589, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 6.620370370370371e-05, | |
| "loss": 0.0481, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 6.60493827160494e-05, | |
| "loss": 0.0506, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 6.589506172839507e-05, | |
| "loss": 0.0205, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 6.574074074074075e-05, | |
| "loss": 0.174, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 6.558641975308642e-05, | |
| "loss": 0.049, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 6.54320987654321e-05, | |
| "loss": 0.0608, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 6.527777777777778e-05, | |
| "loss": 0.0549, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 6.512345679012346e-05, | |
| "loss": 0.0745, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 6.496913580246914e-05, | |
| "loss": 0.0416, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 6.481481481481482e-05, | |
| "loss": 0.0426, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 6.46604938271605e-05, | |
| "loss": 0.0894, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 6.450617283950617e-05, | |
| "loss": 0.0432, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 6.435185185185186e-05, | |
| "loss": 0.0465, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 6.419753086419753e-05, | |
| "loss": 0.1222, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 6.404320987654321e-05, | |
| "loss": 0.0613, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 6.388888888888888e-05, | |
| "loss": 0.0243, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 6.373456790123458e-05, | |
| "loss": 0.0566, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 6.358024691358025e-05, | |
| "loss": 0.0601, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 6.342592592592594e-05, | |
| "loss": 0.0149, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 6.32716049382716e-05, | |
| "loss": 0.0261, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 6.311728395061729e-05, | |
| "loss": 0.1259, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 6.296296296296296e-05, | |
| "loss": 0.0551, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 6.280864197530865e-05, | |
| "loss": 0.0968, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 6.265432098765433e-05, | |
| "loss": 0.0455, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 6.25e-05, | |
| "loss": 0.0166, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 6.234567901234569e-05, | |
| "loss": 0.0432, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 6.219135802469136e-05, | |
| "loss": 0.046, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 6.203703703703704e-05, | |
| "loss": 0.0612, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 6.188271604938271e-05, | |
| "loss": 0.0583, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 6.17283950617284e-05, | |
| "loss": 0.1726, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 6.157407407407407e-05, | |
| "loss": 0.0148, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 6.141975308641975e-05, | |
| "loss": 0.0688, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 6.126543209876542e-05, | |
| "loss": 0.0999, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 6.111111111111112e-05, | |
| "loss": 0.0175, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 6.09567901234568e-05, | |
| "loss": 0.0761, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 6.0802469135802476e-05, | |
| "loss": 0.0926, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 6.0648148148148154e-05, | |
| "loss": 0.0252, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 6.049382716049383e-05, | |
| "loss": 0.0541, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 6.033950617283951e-05, | |
| "loss": 0.0114, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 6.018518518518519e-05, | |
| "loss": 0.0717, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 6.0030864197530865e-05, | |
| "loss": 0.0604, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 5.987654320987654e-05, | |
| "loss": 0.0256, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "eval_loss": 0.07697688043117523, | |
| "eval_runtime": 45.7451, | |
| "eval_samples_per_second": 4.722, | |
| "eval_steps_per_second": 1.574, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 5.972222222222223e-05, | |
| "loss": 0.0444, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 5.9567901234567904e-05, | |
| "loss": 0.0486, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 5.941358024691358e-05, | |
| "loss": 0.0555, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 5.925925925925926e-05, | |
| "loss": 0.032, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 5.910493827160494e-05, | |
| "loss": 0.0534, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 5.8950617283950615e-05, | |
| "loss": 0.0037, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 5.879629629629629e-05, | |
| "loss": 0.0123, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 5.8641975308641984e-05, | |
| "loss": 0.0237, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 5.848765432098766e-05, | |
| "loss": 0.036, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 5.833333333333334e-05, | |
| "loss": 0.1375, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 5.817901234567902e-05, | |
| "loss": 0.0378, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 5.8024691358024695e-05, | |
| "loss": 0.0117, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 5.787037037037037e-05, | |
| "loss": 0.0117, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 5.771604938271605e-05, | |
| "loss": 0.0414, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 5.7561728395061735e-05, | |
| "loss": 0.0875, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 5.740740740740741e-05, | |
| "loss": 0.0776, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 5.725308641975309e-05, | |
| "loss": 0.0979, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 5.709876543209877e-05, | |
| "loss": 0.1294, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 5.6944444444444445e-05, | |
| "loss": 0.0606, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 5.679012345679012e-05, | |
| "loss": 0.0217, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 5.66358024691358e-05, | |
| "loss": 0.0449, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 5.648148148148148e-05, | |
| "loss": 0.017, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 5.6327160493827156e-05, | |
| "loss": 0.0496, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 5.617283950617285e-05, | |
| "loss": 0.0408, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 5.6018518518518525e-05, | |
| "loss": 0.1131, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 5.58641975308642e-05, | |
| "loss": 0.0641, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 5.570987654320988e-05, | |
| "loss": 0.0808, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 5.555555555555556e-05, | |
| "loss": 0.0185, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 5.5401234567901236e-05, | |
| "loss": 0.0487, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 5.524691358024692e-05, | |
| "loss": 0.0266, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 5.50925925925926e-05, | |
| "loss": 0.1075, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 5.4938271604938276e-05, | |
| "loss": 0.0725, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 5.4783950617283953e-05, | |
| "loss": 0.0681, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 5.462962962962963e-05, | |
| "loss": 0.0367, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 5.447530864197531e-05, | |
| "loss": 0.0419, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 5.4320987654320986e-05, | |
| "loss": 0.076, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 5.4166666666666664e-05, | |
| "loss": 0.0528, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 5.401234567901234e-05, | |
| "loss": 0.0414, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 5.385802469135802e-05, | |
| "loss": 0.061, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 5.370370370370371e-05, | |
| "loss": 0.0264, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 5.354938271604939e-05, | |
| "loss": 0.0327, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 5.3395061728395066e-05, | |
| "loss": 0.0071, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 5.3240740740740744e-05, | |
| "loss": 0.0963, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 5.308641975308642e-05, | |
| "loss": 0.0371, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 5.2932098765432106e-05, | |
| "loss": 0.0676, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 5.2777777777777784e-05, | |
| "loss": 0.0143, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 5.262345679012346e-05, | |
| "loss": 0.0423, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 5.246913580246914e-05, | |
| "loss": 0.0108, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 5.231481481481482e-05, | |
| "loss": 0.0958, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 5.2160493827160494e-05, | |
| "loss": 0.068, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 5.200617283950617e-05, | |
| "loss": 0.1593, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 5.185185185185185e-05, | |
| "loss": 0.0249, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 5.169753086419753e-05, | |
| "loss": 0.0178, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 5.1543209876543205e-05, | |
| "loss": 0.0496, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 5.138888888888889e-05, | |
| "loss": 0.0374, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 5.1234567901234574e-05, | |
| "loss": 0.1266, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 5.108024691358025e-05, | |
| "loss": 0.0329, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 5.092592592592593e-05, | |
| "loss": 0.0852, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 5.0771604938271614e-05, | |
| "loss": 0.0317, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 5.061728395061729e-05, | |
| "loss": 0.0249, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 5.046296296296297e-05, | |
| "loss": 0.0356, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 5.030864197530865e-05, | |
| "loss": 0.0429, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 5.0154320987654325e-05, | |
| "loss": 0.0504, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0141, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 4.984567901234568e-05, | |
| "loss": 0.055, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 4.969135802469136e-05, | |
| "loss": 0.1841, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 4.9537037037037035e-05, | |
| "loss": 0.0501, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 4.938271604938271e-05, | |
| "loss": 0.0763, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 4.92283950617284e-05, | |
| "loss": 0.1532, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 4.9074074074074075e-05, | |
| "loss": 0.0163, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 4.891975308641975e-05, | |
| "loss": 0.0681, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 4.876543209876544e-05, | |
| "loss": 0.0324, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 4.8611111111111115e-05, | |
| "loss": 0.0874, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 4.845679012345679e-05, | |
| "loss": 0.039, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 4.830246913580247e-05, | |
| "loss": 0.0188, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 4.814814814814815e-05, | |
| "loss": 0.1394, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 4.799382716049383e-05, | |
| "loss": 0.0314, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 4.783950617283951e-05, | |
| "loss": 0.0669, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 4.768518518518519e-05, | |
| "loss": 0.1124, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 4.7530864197530866e-05, | |
| "loss": 0.0415, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 4.7376543209876543e-05, | |
| "loss": 0.02, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 4.722222222222222e-05, | |
| "loss": 0.0509, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 4.70679012345679e-05, | |
| "loss": 0.083, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 4.691358024691358e-05, | |
| "loss": 0.0836, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 4.675925925925926e-05, | |
| "loss": 0.1457, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 4.6604938271604945e-05, | |
| "loss": 0.0811, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 4.645061728395062e-05, | |
| "loss": 0.0169, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 4.62962962962963e-05, | |
| "loss": 0.0138, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 4.614197530864198e-05, | |
| "loss": 0.0262, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 4.5987654320987656e-05, | |
| "loss": 0.1558, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 4.5833333333333334e-05, | |
| "loss": 0.0507, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 4.567901234567901e-05, | |
| "loss": 0.0323, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 4.5524691358024696e-05, | |
| "loss": 0.0144, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 4.5370370370370374e-05, | |
| "loss": 0.0587, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 4.521604938271605e-05, | |
| "loss": 0.0302, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 4.506172839506173e-05, | |
| "loss": 0.0222, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 4.490740740740741e-05, | |
| "loss": 0.0283, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 4.4753086419753084e-05, | |
| "loss": 0.0394, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 4.459876543209877e-05, | |
| "loss": 0.0588, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 4.4444444444444447e-05, | |
| "loss": 0.03, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 4.429012345679013e-05, | |
| "loss": 0.0571, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 4.413580246913581e-05, | |
| "loss": 0.1115, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 4.3981481481481486e-05, | |
| "loss": 0.1039, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 4.3827160493827164e-05, | |
| "loss": 0.0164, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 4.367283950617284e-05, | |
| "loss": 0.0722, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 4.351851851851852e-05, | |
| "loss": 0.055, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 4.33641975308642e-05, | |
| "loss": 0.0733, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 4.3209876543209875e-05, | |
| "loss": 0.096, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 4.305555555555556e-05, | |
| "loss": 0.0373, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 4.290123456790124e-05, | |
| "loss": 0.0123, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 4.2746913580246915e-05, | |
| "loss": 0.1025, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 4.259259259259259e-05, | |
| "loss": 0.0416, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 4.243827160493827e-05, | |
| "loss": 0.059, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 4.2283950617283955e-05, | |
| "loss": 0.0721, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 4.212962962962963e-05, | |
| "loss": 0.082, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 4.197530864197531e-05, | |
| "loss": 0.0378, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 4.1820987654320994e-05, | |
| "loss": 0.0503, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 4.166666666666667e-05, | |
| "loss": 0.0475, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 4.151234567901235e-05, | |
| "loss": 0.0163, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 4.135802469135803e-05, | |
| "loss": 0.0191, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 4.1203703703703705e-05, | |
| "loss": 0.0127, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 4.104938271604938e-05, | |
| "loss": 0.0116, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 4.089506172839506e-05, | |
| "loss": 0.0617, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 4.074074074074074e-05, | |
| "loss": 0.031, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 4.058641975308642e-05, | |
| "loss": 0.0102, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 4.04320987654321e-05, | |
| "loss": 0.0659, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 4.027777777777778e-05, | |
| "loss": 0.0274, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 4.012345679012346e-05, | |
| "loss": 0.019, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 3.996913580246914e-05, | |
| "loss": 0.0305, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 3.981481481481482e-05, | |
| "loss": 0.0438, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "eval_loss": 0.06827498227357864, | |
| "eval_runtime": 45.7127, | |
| "eval_samples_per_second": 4.725, | |
| "eval_steps_per_second": 1.575, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 3.9660493827160496e-05, | |
| "loss": 0.0467, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 3.950617283950617e-05, | |
| "loss": 0.043, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 3.935185185185186e-05, | |
| "loss": 0.0319, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 3.9197530864197535e-05, | |
| "loss": 0.0074, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 3.904320987654321e-05, | |
| "loss": 0.0251, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 3.888888888888889e-05, | |
| "loss": 0.0158, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 3.873456790123457e-05, | |
| "loss": 0.0231, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 3.8580246913580246e-05, | |
| "loss": 0.0105, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 3.8425925925925924e-05, | |
| "loss": 0.0211, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 3.82716049382716e-05, | |
| "loss": 0.0449, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 3.8117283950617286e-05, | |
| "loss": 0.0453, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 3.7962962962962964e-05, | |
| "loss": 0.0081, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 3.780864197530865e-05, | |
| "loss": 0.0902, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 3.7654320987654326e-05, | |
| "loss": 0.0261, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 3.7500000000000003e-05, | |
| "loss": 0.0672, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 3.734567901234568e-05, | |
| "loss": 0.0239, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 3.719135802469136e-05, | |
| "loss": 0.0144, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 3.7037037037037037e-05, | |
| "loss": 0.0275, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 3.6882716049382714e-05, | |
| "loss": 0.0261, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 3.67283950617284e-05, | |
| "loss": 0.0064, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 3.6574074074074076e-05, | |
| "loss": 0.0116, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 3.6419753086419754e-05, | |
| "loss": 0.0236, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 3.626543209876543e-05, | |
| "loss": 0.0744, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 3.611111111111111e-05, | |
| "loss": 0.0161, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 3.5956790123456794e-05, | |
| "loss": 0.0378, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 3.580246913580247e-05, | |
| "loss": 0.0567, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 3.564814814814815e-05, | |
| "loss": 0.0702, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 3.5493827160493834e-05, | |
| "loss": 0.0402, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 3.533950617283951e-05, | |
| "loss": 0.0281, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 3.518518518518519e-05, | |
| "loss": 0.1717, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 3.503086419753087e-05, | |
| "loss": 0.0933, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 3.4876543209876545e-05, | |
| "loss": 0.1137, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 3.472222222222222e-05, | |
| "loss": 0.0322, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 3.45679012345679e-05, | |
| "loss": 0.0736, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 3.441358024691358e-05, | |
| "loss": 0.0031, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 3.425925925925926e-05, | |
| "loss": 0.0498, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 3.410493827160494e-05, | |
| "loss": 0.0933, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 3.395061728395062e-05, | |
| "loss": 0.1058, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 3.3796296296296295e-05, | |
| "loss": 0.0216, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 3.364197530864198e-05, | |
| "loss": 0.0332, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.348765432098766e-05, | |
| "loss": 0.0463, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.3333333333333335e-05, | |
| "loss": 0.1228, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.317901234567901e-05, | |
| "loss": 0.0206, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.30246913580247e-05, | |
| "loss": 0.071, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.2870370370370375e-05, | |
| "loss": 0.0812, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.271604938271605e-05, | |
| "loss": 0.0265, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.256172839506173e-05, | |
| "loss": 0.0686, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 3.240740740740741e-05, | |
| "loss": 0.0268, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 3.2253086419753086e-05, | |
| "loss": 0.0113, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 3.209876543209876e-05, | |
| "loss": 0.0422, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 3.194444444444444e-05, | |
| "loss": 0.015, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 3.1790123456790125e-05, | |
| "loss": 0.0167, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 3.16358024691358e-05, | |
| "loss": 0.0296, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 3.148148148148148e-05, | |
| "loss": 0.0227, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 3.1327160493827165e-05, | |
| "loss": 0.0192, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 3.117283950617284e-05, | |
| "loss": 0.0404, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 3.101851851851852e-05, | |
| "loss": 0.0512, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 3.08641975308642e-05, | |
| "loss": 0.0716, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 3.0709876543209876e-05, | |
| "loss": 0.0662, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 3.055555555555556e-05, | |
| "loss": 0.0171, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 3.0401234567901238e-05, | |
| "loss": 0.0349, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 3.0246913580246916e-05, | |
| "loss": 0.0301, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 3.0092592592592593e-05, | |
| "loss": 0.0227, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 2.993827160493827e-05, | |
| "loss": 0.012, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 2.9783950617283952e-05, | |
| "loss": 0.0499, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 2.962962962962963e-05, | |
| "loss": 0.0432, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 2.9475308641975308e-05, | |
| "loss": 0.0366, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 2.9320987654320992e-05, | |
| "loss": 0.0519, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 2.916666666666667e-05, | |
| "loss": 0.0114, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 2.9012345679012347e-05, | |
| "loss": 0.0142, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 2.8858024691358025e-05, | |
| "loss": 0.0183, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 2.8703703703703706e-05, | |
| "loss": 0.0502, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 2.8549382716049384e-05, | |
| "loss": 0.0352, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 2.839506172839506e-05, | |
| "loss": 0.0442, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 2.824074074074074e-05, | |
| "loss": 0.0165, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 2.8086419753086424e-05, | |
| "loss": 0.0666, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 2.79320987654321e-05, | |
| "loss": 0.0072, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 2.777777777777778e-05, | |
| "loss": 0.0296, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 2.762345679012346e-05, | |
| "loss": 0.0312, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 2.7469135802469138e-05, | |
| "loss": 0.0284, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 2.7314814814814816e-05, | |
| "loss": 0.0178, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 2.7160493827160493e-05, | |
| "loss": 0.0634, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 2.700617283950617e-05, | |
| "loss": 0.0281, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 2.6851851851851855e-05, | |
| "loss": 0.0334, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 2.6697530864197533e-05, | |
| "loss": 0.0355, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 2.654320987654321e-05, | |
| "loss": 0.0729, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 2.6388888888888892e-05, | |
| "loss": 0.0211, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 2.623456790123457e-05, | |
| "loss": 0.0295, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 2.6080246913580247e-05, | |
| "loss": 0.0461, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 2.5925925925925925e-05, | |
| "loss": 0.016, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 2.5771604938271603e-05, | |
| "loss": 0.0059, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 2.5617283950617287e-05, | |
| "loss": 0.0448, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 2.5462962962962965e-05, | |
| "loss": 0.0349, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 2.5308641975308646e-05, | |
| "loss": 0.0148, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 2.5154320987654324e-05, | |
| "loss": 0.0048, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 2.5e-05, | |
| "loss": 0.0242, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 2.484567901234568e-05, | |
| "loss": 0.0472, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 2.4691358024691357e-05, | |
| "loss": 0.0299, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 2.4537037037037038e-05, | |
| "loss": 0.0451, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 2.438271604938272e-05, | |
| "loss": 0.0096, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 2.4228395061728396e-05, | |
| "loss": 0.05, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 2.4074074074074074e-05, | |
| "loss": 0.0148, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 2.3919753086419755e-05, | |
| "loss": 0.0398, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 2.3765432098765433e-05, | |
| "loss": 0.0308, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 2.361111111111111e-05, | |
| "loss": 0.0104, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 2.345679012345679e-05, | |
| "loss": 0.005, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 2.3302469135802473e-05, | |
| "loss": 0.0048, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 2.314814814814815e-05, | |
| "loss": 0.0236, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 2.2993827160493828e-05, | |
| "loss": 0.0747, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 2.2839506172839506e-05, | |
| "loss": 0.0382, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 2.2685185185185187e-05, | |
| "loss": 0.0228, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 2.2530864197530865e-05, | |
| "loss": 0.0049, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 2.2376543209876542e-05, | |
| "loss": 0.0228, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 2.2222222222222223e-05, | |
| "loss": 0.0278, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 2.2067901234567904e-05, | |
| "loss": 0.0267, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 2.1913580246913582e-05, | |
| "loss": 0.0174, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 2.175925925925926e-05, | |
| "loss": 0.0808, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 2.1604938271604937e-05, | |
| "loss": 0.0534, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 2.145061728395062e-05, | |
| "loss": 0.054, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 2.1296296296296296e-05, | |
| "loss": 0.0396, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 2.1141975308641977e-05, | |
| "loss": 0.0532, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 2.0987654320987655e-05, | |
| "loss": 0.0616, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 2.0833333333333336e-05, | |
| "loss": 0.016, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 2.0679012345679014e-05, | |
| "loss": 0.0273, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 2.052469135802469e-05, | |
| "loss": 0.0128, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 2.037037037037037e-05, | |
| "loss": 0.0559, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 2.021604938271605e-05, | |
| "loss": 0.0218, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 2.006172839506173e-05, | |
| "loss": 0.0496, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 1.990740740740741e-05, | |
| "loss": 0.0086, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 1.9753086419753087e-05, | |
| "loss": 0.0452, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "eval_loss": 0.06977373361587524, | |
| "eval_runtime": 45.3906, | |
| "eval_samples_per_second": 4.759, | |
| "eval_steps_per_second": 1.586, | |
| "step": 520 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 648, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 130, | |
| "total_flos": 1.345583649712128e+16, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |