| { | |
| "best_metric": 0.016997970640659332, | |
| "best_model_checkpoint": "output_qwen_7b_gt_only_new/checkpoint-1850", | |
| "epoch": 4.852459016393443, | |
| "eval_steps": 50, | |
| "global_step": 1850, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.282051282051282e-06, | |
| "loss": 3.7144, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.564102564102564e-06, | |
| "loss": 3.7383, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.846153846153847e-06, | |
| "loss": 3.7315, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.128205128205128e-06, | |
| "loss": 3.832, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.41025641025641e-06, | |
| "loss": 3.9406, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 7.692307692307694e-06, | |
| "loss": 3.9295, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.974358974358976e-06, | |
| "loss": 3.7117, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.0256410256410256e-05, | |
| "loss": 4.1448, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.153846153846154e-05, | |
| "loss": 3.8611, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.282051282051282e-05, | |
| "loss": 3.7229, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.4102564102564104e-05, | |
| "loss": 3.4408, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.5384615384615387e-05, | |
| "loss": 3.3446, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.6666666666666667e-05, | |
| "loss": 3.7562, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.794871794871795e-05, | |
| "loss": 3.4758, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.923076923076923e-05, | |
| "loss": 3.6359, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.0512820512820512e-05, | |
| "loss": 3.5937, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.1794871794871795e-05, | |
| "loss": 3.5099, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.307692307692308e-05, | |
| "loss": 3.7363, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.435897435897436e-05, | |
| "loss": 3.4163, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.564102564102564e-05, | |
| "loss": 3.2308, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2.6923076923076923e-05, | |
| "loss": 3.316, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2.8205128205128207e-05, | |
| "loss": 3.0125, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2.948717948717949e-05, | |
| "loss": 3.1502, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 3.0769230769230774e-05, | |
| "loss": 3.0518, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.205128205128206e-05, | |
| "loss": 2.8666, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.3333333333333335e-05, | |
| "loss": 2.5435, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.461538461538462e-05, | |
| "loss": 2.7219, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.58974358974359e-05, | |
| "loss": 2.534, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3.717948717948718e-05, | |
| "loss": 2.5938, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3.846153846153846e-05, | |
| "loss": 2.2513, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3.974358974358974e-05, | |
| "loss": 2.258, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.1025641025641023e-05, | |
| "loss": 2.0368, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.230769230769231e-05, | |
| "loss": 1.8706, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.358974358974359e-05, | |
| "loss": 1.9183, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.4871794871794874e-05, | |
| "loss": 1.7306, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.615384615384616e-05, | |
| "loss": 1.6126, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.7435897435897435e-05, | |
| "loss": 1.5315, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.871794871794872e-05, | |
| "loss": 1.3916, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 5e-05, | |
| "loss": 1.3437, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.9999991324456964e-05, | |
| "loss": 1.1718, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.999996529783388e-05, | |
| "loss": 1.1439, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.9999921920148804e-05, | |
| "loss": 1.2322, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.999986119143186e-05, | |
| "loss": 1.106, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.999978311172517e-05, | |
| "loss": 1.0669, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.999968768108295e-05, | |
| "loss": 1.0308, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.999957489957142e-05, | |
| "loss": 1.1053, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.999944476726885e-05, | |
| "loss": 0.9832, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.999929728426558e-05, | |
| "loss": 0.9761, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.999913245066394e-05, | |
| "loss": 0.933, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.999895026657835e-05, | |
| "loss": 0.7558, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "eval_loss": 0.9146232008934021, | |
| "eval_runtime": 26.9094, | |
| "eval_samples_per_second": 22.297, | |
| "eval_steps_per_second": 2.787, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.9998750732135255e-05, | |
| "loss": 0.8559, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.999853384747313e-05, | |
| "loss": 0.8866, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.9998299612742505e-05, | |
| "loss": 0.9498, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.999804802810596e-05, | |
| "loss": 0.8116, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.999777909373809e-05, | |
| "loss": 0.9047, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.9997492809825556e-05, | |
| "loss": 0.898, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.9997189176567046e-05, | |
| "loss": 0.8543, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.9996868194173306e-05, | |
| "loss": 0.7572, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.999652986286711e-05, | |
| "loss": 0.79, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.999617418288326e-05, | |
| "loss": 0.7742, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.9995801154468624e-05, | |
| "loss": 0.8558, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.99954107778821e-05, | |
| "loss": 0.7483, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.999500305339462e-05, | |
| "loss": 0.8833, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.999457798128917e-05, | |
| "loss": 0.7535, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.9994135561860764e-05, | |
| "loss": 0.7344, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.999367579541647e-05, | |
| "loss": 0.8373, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.9993198682275365e-05, | |
| "loss": 0.6668, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.9992704222768606e-05, | |
| "loss": 0.7379, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.999219241723937e-05, | |
| "loss": 0.7731, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.9991663266042854e-05, | |
| "loss": 0.6745, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.999111676954632e-05, | |
| "loss": 0.6995, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.999055292812908e-05, | |
| "loss": 0.7747, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.9989971742182426e-05, | |
| "loss": 0.7372, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.998937321210976e-05, | |
| "loss": 0.7959, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.9988757338326474e-05, | |
| "loss": 0.7453, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.998812412126001e-05, | |
| "loss": 0.652, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.998747356134985e-05, | |
| "loss": 0.7298, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.998680565904752e-05, | |
| "loss": 0.7309, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.9986120414816565e-05, | |
| "loss": 0.7156, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.9985417829132563e-05, | |
| "loss": 0.7668, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.998469790248316e-05, | |
| "loss": 0.6943, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.9983960635368e-05, | |
| "loss": 0.5747, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.998320602829879e-05, | |
| "loss": 0.725, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.998243408179925e-05, | |
| "loss": 0.6572, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.9981644796405154e-05, | |
| "loss": 0.637, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.998083817266429e-05, | |
| "loss": 0.6354, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.9980014211136505e-05, | |
| "loss": 0.6864, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.997917291239365e-05, | |
| "loss": 0.5963, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.9978314277019625e-05, | |
| "loss": 0.6064, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.9977438305610366e-05, | |
| "loss": 0.6303, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.9976544998773837e-05, | |
| "loss": 0.6072, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.997563435713002e-05, | |
| "loss": 0.5951, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.997470638131094e-05, | |
| "loss": 0.6087, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.997376107196067e-05, | |
| "loss": 0.6461, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.997279842973529e-05, | |
| "loss": 0.6479, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.99718184553029e-05, | |
| "loss": 0.6044, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.997082114934366e-05, | |
| "loss": 0.6382, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.996980651254974e-05, | |
| "loss": 0.6603, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.9968774545625344e-05, | |
| "loss": 0.5042, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.996772524928669e-05, | |
| "loss": 0.6829, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "eval_loss": 0.639582097530365, | |
| "eval_runtime": 27.0017, | |
| "eval_samples_per_second": 22.221, | |
| "eval_steps_per_second": 2.778, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.9966658624262044e-05, | |
| "loss": 0.518, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.99655746712917e-05, | |
| "loss": 0.6625, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.996447339112795e-05, | |
| "loss": 0.5904, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.996335478453514e-05, | |
| "loss": 0.6856, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.996221885228964e-05, | |
| "loss": 0.5244, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.996106559517982e-05, | |
| "loss": 0.581, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.99598950140061e-05, | |
| "loss": 0.5233, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.99587071095809e-05, | |
| "loss": 0.6757, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.9957501882728705e-05, | |
| "loss": 0.6409, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.995627933428597e-05, | |
| "loss": 0.6027, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.99550394651012e-05, | |
| "loss": 0.576, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.9953782276034946e-05, | |
| "loss": 0.5882, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.995250776795971e-05, | |
| "loss": 0.5466, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.9951215941760075e-05, | |
| "loss": 0.6233, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.9949906798332636e-05, | |
| "loss": 0.6017, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.9948580338585974e-05, | |
| "loss": 0.6036, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.9947236563440726e-05, | |
| "loss": 0.4385, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.994587547382952e-05, | |
| "loss": 0.5811, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.994449707069702e-05, | |
| "loss": 0.6104, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.994310135499989e-05, | |
| "loss": 0.5636, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.9941688327706826e-05, | |
| "loss": 0.5811, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.9940257989798524e-05, | |
| "loss": 0.5375, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.9938810342267695e-05, | |
| "loss": 0.5879, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.9937345386119085e-05, | |
| "loss": 0.4887, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.9935863122369416e-05, | |
| "loss": 0.5587, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.993436355204746e-05, | |
| "loss": 0.5162, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.993284667619399e-05, | |
| "loss": 0.5298, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.993131249586176e-05, | |
| "loss": 0.57, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.992976101211558e-05, | |
| "loss": 0.5001, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.992819222603223e-05, | |
| "loss": 0.4982, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.992660613870053e-05, | |
| "loss": 0.545, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.992500275122128e-05, | |
| "loss": 0.4539, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.9923382064707315e-05, | |
| "loss": 0.5473, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.992174408028345e-05, | |
| "loss": 0.5759, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.992008879908653e-05, | |
| "loss": 0.5563, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.991841622226537e-05, | |
| "loss": 0.4783, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.991672635098084e-05, | |
| "loss": 0.503, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.9915019186405766e-05, | |
| "loss": 0.5458, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.991329472972499e-05, | |
| "loss": 0.4992, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.991155298213538e-05, | |
| "loss": 0.5645, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.990979394484576e-05, | |
| "loss": 0.4729, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.9908017619077e-05, | |
| "loss": 0.4912, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.990622400606193e-05, | |
| "loss": 0.4898, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.99044131070454e-05, | |
| "loss": 0.5889, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.9902584923284264e-05, | |
| "loss": 0.4933, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.9900739456047345e-05, | |
| "loss": 0.4551, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.989887670661549e-05, | |
| "loss": 0.4577, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.989699667628152e-05, | |
| "loss": 0.4432, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.989509936635026e-05, | |
| "loss": 0.465, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.989318477813853e-05, | |
| "loss": 0.4931, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "eval_loss": 0.5148335099220276, | |
| "eval_runtime": 26.8859, | |
| "eval_samples_per_second": 22.317, | |
| "eval_steps_per_second": 2.79, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.989125291297513e-05, | |
| "loss": 0.5112, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.988930377220087e-05, | |
| "loss": 0.4956, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.988733735716852e-05, | |
| "loss": 0.4955, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.9885353669242866e-05, | |
| "loss": 0.5444, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.988335270980068e-05, | |
| "loss": 0.4508, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.9881334480230716e-05, | |
| "loss": 0.4569, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.9879298981933694e-05, | |
| "loss": 0.3779, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.987724621632236e-05, | |
| "loss": 0.4522, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.987517618482142e-05, | |
| "loss": 0.3876, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.987308888886755e-05, | |
| "loss": 0.4106, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.987098432990943e-05, | |
| "loss": 0.4362, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.9868862509407724e-05, | |
| "loss": 0.4347, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.9866723428835066e-05, | |
| "loss": 0.4685, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.986456708967606e-05, | |
| "loss": 0.5076, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.986239349342732e-05, | |
| "loss": 0.4966, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.9860202641597384e-05, | |
| "loss": 0.4583, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.9857994535706834e-05, | |
| "loss": 0.5343, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.985576917728817e-05, | |
| "loss": 0.5018, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.985352656788589e-05, | |
| "loss": 0.3912, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.985126670905646e-05, | |
| "loss": 0.3894, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.984898960236832e-05, | |
| "loss": 0.458, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.98466952494019e-05, | |
| "loss": 0.434, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.984438365174956e-05, | |
| "loss": 0.4502, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.984205481101565e-05, | |
| "loss": 0.3437, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.98397087288165e-05, | |
| "loss": 0.4711, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.983734540678038e-05, | |
| "loss": 0.4713, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.983496484654755e-05, | |
| "loss": 0.4109, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.983256704977021e-05, | |
| "loss": 0.3906, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.983015201811254e-05, | |
| "loss": 0.3827, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.9827719753250675e-05, | |
| "loss": 0.4998, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.982527025687271e-05, | |
| "loss": 0.5082, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.982280353067872e-05, | |
| "loss": 0.4098, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.9820319576380706e-05, | |
| "loss": 0.3899, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.9817818395702634e-05, | |
| "loss": 0.4118, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.981529999038044e-05, | |
| "loss": 0.4021, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.9812764362162e-05, | |
| "loss": 0.4317, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.9810211512807166e-05, | |
| "loss": 0.3808, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.980764144408771e-05, | |
| "loss": 0.4484, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.980505415778738e-05, | |
| "loss": 0.3865, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.980244965570186e-05, | |
| "loss": 0.3782, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.979982793963879e-05, | |
| "loss": 0.3953, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.979718901141776e-05, | |
| "loss": 0.341, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.979453287287029e-05, | |
| "loss": 0.4293, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.979185952583987e-05, | |
| "loss": 0.3317, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.97891689721819e-05, | |
| "loss": 0.5515, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.978646121376376e-05, | |
| "loss": 0.4385, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.9783736252464744e-05, | |
| "loss": 0.4524, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.978099409017609e-05, | |
| "loss": 0.384, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.9778234728800975e-05, | |
| "loss": 0.3622, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.977545817025453e-05, | |
| "loss": 0.4031, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "eval_loss": 0.4207773804664612, | |
| "eval_runtime": 27.0247, | |
| "eval_samples_per_second": 22.202, | |
| "eval_steps_per_second": 2.775, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.97726644164638e-05, | |
| "loss": 0.4301, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.976985346936776e-05, | |
| "loss": 0.3188, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.976702533091735e-05, | |
| "loss": 0.3839, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.97641800030754e-05, | |
| "loss": 0.331, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.976131748781671e-05, | |
| "loss": 0.4465, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.9758437787127974e-05, | |
| "loss": 0.3405, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.975554090300785e-05, | |
| "loss": 0.3813, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.9752626837466875e-05, | |
| "loss": 0.343, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.9749695592527555e-05, | |
| "loss": 0.4073, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.9746747170224306e-05, | |
| "loss": 0.4009, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.9743781572603445e-05, | |
| "loss": 0.3842, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.9740798801723235e-05, | |
| "loss": 0.3068, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.973779885965384e-05, | |
| "loss": 0.3705, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.9734781748477355e-05, | |
| "loss": 0.4452, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.9731747470287795e-05, | |
| "loss": 0.2932, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.972869602719107e-05, | |
| "loss": 0.3706, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.972562742130502e-05, | |
| "loss": 0.4109, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.972254165475938e-05, | |
| "loss": 0.3886, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.971943872969582e-05, | |
| "loss": 0.3941, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.9716318648267904e-05, | |
| "loss": 0.3429, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.971318141264109e-05, | |
| "loss": 0.3953, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.9710027024992775e-05, | |
| "loss": 0.3593, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.970685548751222e-05, | |
| "loss": 0.3499, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.970366680240063e-05, | |
| "loss": 0.2849, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.9700460971871076e-05, | |
| "loss": 0.3127, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.969723799814855e-05, | |
| "loss": 0.3821, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.969399788346994e-05, | |
| "loss": 0.3157, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.969074063008402e-05, | |
| "loss": 0.4177, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.9687466240251465e-05, | |
| "loss": 0.3918, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.968417471624485e-05, | |
| "loss": 0.3505, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.968086606034862e-05, | |
| "loss": 0.3169, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.9677540274859145e-05, | |
| "loss": 0.3477, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.967419736208466e-05, | |
| "loss": 0.3797, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.967083732434529e-05, | |
| "loss": 0.2928, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.966746016397303e-05, | |
| "loss": 0.3138, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.966406588331181e-05, | |
| "loss": 0.3771, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.966065448471739e-05, | |
| "loss": 0.3108, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.965722597055742e-05, | |
| "loss": 0.3732, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.965378034321144e-05, | |
| "loss": 0.3365, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.9650317605070886e-05, | |
| "loss": 0.333, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.9646837758539026e-05, | |
| "loss": 0.36, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.964334080603103e-05, | |
| "loss": 0.2805, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.963982674997395e-05, | |
| "loss": 0.3268, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.963629559280668e-05, | |
| "loss": 0.3853, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.963274733697999e-05, | |
| "loss": 0.3204, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 4.962918198495654e-05, | |
| "loss": 0.3035, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 4.9625599539210825e-05, | |
| "loss": 0.2801, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 4.962200000222923e-05, | |
| "loss": 0.2999, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 4.961838337650997e-05, | |
| "loss": 0.2818, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.961474966456317e-05, | |
| "loss": 0.3822, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "eval_loss": 0.3398887515068054, | |
| "eval_runtime": 26.9438, | |
| "eval_samples_per_second": 22.269, | |
| "eval_steps_per_second": 2.784, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.961109886891076e-05, | |
| "loss": 0.2777, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.9607430992086564e-05, | |
| "loss": 0.2942, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.960374603663624e-05, | |
| "loss": 0.27, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.9600044005117305e-05, | |
| "loss": 0.3825, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.9596324900099145e-05, | |
| "loss": 0.2462, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.959258872416296e-05, | |
| "loss": 0.3602, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.958883547990184e-05, | |
| "loss": 0.3178, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.958506516992069e-05, | |
| "loss": 0.2902, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.958127779683625e-05, | |
| "loss": 0.2936, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.957747336327715e-05, | |
| "loss": 0.3336, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.957365187188382e-05, | |
| "loss": 0.3285, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.956981332530854e-05, | |
| "loss": 0.2905, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.956595772621543e-05, | |
| "loss": 0.2634, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.9562085077280443e-05, | |
| "loss": 0.2824, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.9558195381191364e-05, | |
| "loss": 0.23, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.9554288640647815e-05, | |
| "loss": 0.3001, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.955036485836124e-05, | |
| "loss": 0.3123, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.954642403705492e-05, | |
| "loss": 0.2238, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.9542466179463956e-05, | |
| "loss": 0.2913, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.953849128833526e-05, | |
| "loss": 0.38, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.953449936642759e-05, | |
| "loss": 0.269, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.953049041651151e-05, | |
| "loss": 0.3593, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.952646444136942e-05, | |
| "loss": 0.2931, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.9522421443795496e-05, | |
| "loss": 0.2661, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.951836142659577e-05, | |
| "loss": 0.2888, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.951428439258806e-05, | |
| "loss": 0.2596, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 4.951019034460202e-05, | |
| "loss": 0.2613, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 4.9506079285479085e-05, | |
| "loss": 0.287, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 4.950195121807251e-05, | |
| "loss": 0.3136, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 4.949780614524736e-05, | |
| "loss": 0.3299, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 4.949364406988049e-05, | |
| "loss": 0.3073, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 4.948946499486056e-05, | |
| "loss": 0.2867, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 4.948526892308803e-05, | |
| "loss": 0.297, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 4.948105585747517e-05, | |
| "loss": 0.244, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 4.947682580094601e-05, | |
| "loss": 0.278, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 4.94725787564364e-05, | |
| "loss": 0.2499, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 4.9468314726893975e-05, | |
| "loss": 0.2954, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 4.9464033715278154e-05, | |
| "loss": 0.2678, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 4.9459735724560154e-05, | |
| "loss": 0.2308, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 4.945542075772296e-05, | |
| "loss": 0.2117, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 4.945108881776135e-05, | |
| "loss": 0.2704, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.944673990768187e-05, | |
| "loss": 0.2982, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.944237403050286e-05, | |
| "loss": 0.296, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.943799118925443e-05, | |
| "loss": 0.2265, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.943359138697845e-05, | |
| "loss": 0.33, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.942917462672859e-05, | |
| "loss": 0.2858, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.942474091157027e-05, | |
| "loss": 0.244, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.9420290244580666e-05, | |
| "loss": 0.2833, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.9415822628848755e-05, | |
| "loss": 0.2337, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.9411338067475244e-05, | |
| "loss": 0.2662, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "eval_loss": 0.27772608399391174, | |
| "eval_runtime": 27.1665, | |
| "eval_samples_per_second": 22.086, | |
| "eval_steps_per_second": 2.761, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.940683656357262e-05, | |
| "loss": 0.2667, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.9402318120265115e-05, | |
| "loss": 0.2497, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.9397782740688734e-05, | |
| "loss": 0.2302, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 4.939323042799121e-05, | |
| "loss": 0.246, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 4.9388661185332066e-05, | |
| "loss": 0.2551, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 4.9384075015882544e-05, | |
| "loss": 0.2779, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.9379471922825656e-05, | |
| "loss": 0.2132, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.9374851909356134e-05, | |
| "loss": 0.2284, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.937021497868047e-05, | |
| "loss": 0.241, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.936556113401691e-05, | |
| "loss": 0.2694, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.9360890378595406e-05, | |
| "loss": 0.2252, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.9356202715657675e-05, | |
| "loss": 0.26, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.935149814845715e-05, | |
| "loss": 0.2273, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.934677668025902e-05, | |
| "loss": 0.2259, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.934203831434019e-05, | |
| "loss": 0.2122, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.933728305398927e-05, | |
| "loss": 0.2042, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.933251090250663e-05, | |
| "loss": 0.2797, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.9327721863204354e-05, | |
| "loss": 0.235, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 4.9322915939406236e-05, | |
| "loss": 0.2502, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 4.9318093134447804e-05, | |
| "loss": 0.2573, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 4.9313253451676286e-05, | |
| "loss": 0.254, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 4.930839689445064e-05, | |
| "loss": 0.2346, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 4.930352346614152e-05, | |
| "loss": 0.2592, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 4.9298633170131304e-05, | |
| "loss": 0.2032, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 4.9293726009814066e-05, | |
| "loss": 0.2311, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 4.928880198859558e-05, | |
| "loss": 0.2085, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 4.9283861109893354e-05, | |
| "loss": 0.2458, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 4.927890337713655e-05, | |
| "loss": 0.2159, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 4.927392879376605e-05, | |
| "loss": 0.2127, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 4.926893736323446e-05, | |
| "loss": 0.2244, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 4.926392908900601e-05, | |
| "loss": 0.1844, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 4.925890397455668e-05, | |
| "loss": 0.1856, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 4.9253862023374134e-05, | |
| "loss": 0.1896, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 4.924880323895766e-05, | |
| "loss": 0.2515, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 4.9243727624818324e-05, | |
| "loss": 0.2378, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 4.9238635184478784e-05, | |
| "loss": 0.2079, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 4.9233525921473446e-05, | |
| "loss": 0.1777, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 4.922839983934834e-05, | |
| "loss": 0.2403, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 4.922325694166119e-05, | |
| "loss": 0.1945, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 4.92180972319814e-05, | |
| "loss": 0.2083, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 4.921292071389002e-05, | |
| "loss": 0.1945, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 4.92077273909798e-05, | |
| "loss": 0.1997, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 4.920251726685511e-05, | |
| "loss": 0.2585, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 4.9197290345132014e-05, | |
| "loss": 0.217, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 4.919204662943822e-05, | |
| "loss": 0.2329, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.918678612341309e-05, | |
| "loss": 0.2231, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.9181508830707646e-05, | |
| "loss": 0.1705, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.9176214754984566e-05, | |
| "loss": 0.191, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 4.917090389991816e-05, | |
| "loss": 0.2344, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 4.91655762691944e-05, | |
| "loss": 0.1979, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "eval_loss": 0.22052711248397827, | |
| "eval_runtime": 26.9339, | |
| "eval_samples_per_second": 22.277, | |
| "eval_steps_per_second": 2.785, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 4.9160231866510885e-05, | |
| "loss": 0.2315, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 4.915487069557686e-05, | |
| "loss": 0.2132, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 4.9149492760113224e-05, | |
| "loss": 0.1852, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 4.9144098063852485e-05, | |
| "loss": 0.1515, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 4.913868661053881e-05, | |
| "loss": 0.2082, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 4.913325840392796e-05, | |
| "loss": 0.2033, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.912781344778737e-05, | |
| "loss": 0.2223, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.912235174589607e-05, | |
| "loss": 0.2051, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.911687330204471e-05, | |
| "loss": 0.1809, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.9111378120035565e-05, | |
| "loss": 0.2251, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 4.910586620368255e-05, | |
| "loss": 0.1981, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 4.910033755681116e-05, | |
| "loss": 0.173, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 4.909479218325852e-05, | |
| "loss": 0.211, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 4.908923008687335e-05, | |
| "loss": 0.1992, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.908365127151601e-05, | |
| "loss": 0.1609, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.9078055741058416e-05, | |
| "loss": 0.201, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.9072443499384116e-05, | |
| "loss": 0.2366, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 4.9066814550388254e-05, | |
| "loss": 0.1868, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 4.9061168897977564e-05, | |
| "loss": 0.1688, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 4.905550654607038e-05, | |
| "loss": 0.167, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 4.9049827498596606e-05, | |
| "loss": 0.2089, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 4.9044131759497755e-05, | |
| "loss": 0.1676, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 4.903841933272693e-05, | |
| "loss": 0.1566, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 4.9032690222248775e-05, | |
| "loss": 0.1515, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 4.902694443203957e-05, | |
| "loss": 0.1876, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.902118196608711e-05, | |
| "loss": 0.2092, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.901540282839085e-05, | |
| "loss": 0.151, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.90096070229617e-05, | |
| "loss": 0.1907, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.9003794553822244e-05, | |
| "loss": 0.1577, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.899796542500657e-05, | |
| "loss": 0.168, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.899211964056036e-05, | |
| "loss": 0.2188, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.898625720454083e-05, | |
| "loss": 0.1593, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.8980378121016765e-05, | |
| "loss": 0.1225, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8974482394068514e-05, | |
| "loss": 0.1646, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8968570027787955e-05, | |
| "loss": 0.1843, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.8962641026278536e-05, | |
| "loss": 0.1291, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.895669539365525e-05, | |
| "loss": 0.1243, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.89507331340446e-05, | |
| "loss": 0.125, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.8944754251584676e-05, | |
| "loss": 0.114, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.893875875042507e-05, | |
| "loss": 0.1571, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 4.893274663472692e-05, | |
| "loss": 0.1226, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 4.892671790866291e-05, | |
| "loss": 0.1415, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 4.8920672576417214e-05, | |
| "loss": 0.1283, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 4.891461064218558e-05, | |
| "loss": 0.1273, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 4.890853211017523e-05, | |
| "loss": 0.1068, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 4.890243698460495e-05, | |
| "loss": 0.1221, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 4.889632526970501e-05, | |
| "loss": 0.1485, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 4.889019696971721e-05, | |
| "loss": 0.1474, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 4.888405208889486e-05, | |
| "loss": 0.1306, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 4.887789063150276e-05, | |
| "loss": 0.1329, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "eval_loss": 0.16800016164779663, | |
| "eval_runtime": 26.9666, | |
| "eval_samples_per_second": 22.25, | |
| "eval_steps_per_second": 2.781, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 4.887171260181724e-05, | |
| "loss": 0.1142, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 4.886551800412612e-05, | |
| "loss": 0.1273, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 4.8859306842728724e-05, | |
| "loss": 0.1353, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 4.8853079121935865e-05, | |
| "loss": 0.1464, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 4.8846834846069845e-05, | |
| "loss": 0.1229, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 4.884057401946446e-05, | |
| "loss": 0.127, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 4.8834296646465015e-05, | |
| "loss": 0.1538, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 4.882800273142827e-05, | |
| "loss": 0.1199, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 4.8821692278722465e-05, | |
| "loss": 0.1144, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 4.881536529272734e-05, | |
| "loss": 0.109, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 4.880902177783409e-05, | |
| "loss": 0.1064, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 4.8802661738445396e-05, | |
| "loss": 0.121, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 4.879628517897541e-05, | |
| "loss": 0.117, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 4.878989210384972e-05, | |
| "loss": 0.1178, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 4.878348251750542e-05, | |
| "loss": 0.1234, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 4.877705642439103e-05, | |
| "loss": 0.1497, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 4.8770613828966526e-05, | |
| "loss": 0.1085, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 4.876415473570337e-05, | |
| "loss": 0.1063, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 4.875767914908445e-05, | |
| "loss": 0.0911, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 4.87511870736041e-05, | |
| "loss": 0.0866, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 4.874467851376809e-05, | |
| "loss": 0.1127, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 4.873815347409367e-05, | |
| "loss": 0.1103, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 4.8731611959109477e-05, | |
| "loss": 0.111, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 4.8725053973355616e-05, | |
| "loss": 0.1244, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 4.871847952138362e-05, | |
| "loss": 0.137, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 4.871188860775643e-05, | |
| "loss": 0.1005, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 4.8705281237048436e-05, | |
| "loss": 0.0917, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 4.869865741384544e-05, | |
| "loss": 0.1539, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 4.869201714274467e-05, | |
| "loss": 0.112, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 4.868536042835474e-05, | |
| "loss": 0.1237, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 4.867868727529573e-05, | |
| "loss": 0.0852, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 4.867199768819907e-05, | |
| "loss": 0.1352, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 4.866529167170764e-05, | |
| "loss": 0.1161, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 4.8658569230475704e-05, | |
| "loss": 0.1532, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 4.8651830369168925e-05, | |
| "loss": 0.0956, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 4.8645075092464364e-05, | |
| "loss": 0.123, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 4.863830340505048e-05, | |
| "loss": 0.0959, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 4.863151531162712e-05, | |
| "loss": 0.0932, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 4.862471081690552e-05, | |
| "loss": 0.1418, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 4.861788992560828e-05, | |
| "loss": 0.1089, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 4.86110526424694e-05, | |
| "loss": 0.1027, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 4.860419897223427e-05, | |
| "loss": 0.0984, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 4.8597328919659605e-05, | |
| "loss": 0.1141, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 4.8590442489513543e-05, | |
| "loss": 0.104, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 4.858353968657555e-05, | |
| "loss": 0.1402, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 4.857662051563649e-05, | |
| "loss": 0.1181, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 4.856968498149855e-05, | |
| "loss": 0.0688, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 4.85627330889753e-05, | |
| "loss": 0.1549, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 4.855576484289166e-05, | |
| "loss": 0.1017, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 4.854878024808388e-05, | |
| "loss": 0.0746, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "eval_loss": 0.1376887559890747, | |
| "eval_runtime": 27.1573, | |
| "eval_samples_per_second": 22.094, | |
| "eval_steps_per_second": 2.762, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 4.8541779309399586e-05, | |
| "loss": 0.1044, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 4.853476203169773e-05, | |
| "loss": 0.1151, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 4.852772841984861e-05, | |
| "loss": 0.1172, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 4.852067847873385e-05, | |
| "loss": 0.1063, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 4.851361221324643e-05, | |
| "loss": 0.0837, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 4.8506529628290625e-05, | |
| "loss": 0.1099, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 4.8499430728782066e-05, | |
| "loss": 0.0857, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 4.849231551964771e-05, | |
| "loss": 0.0893, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 4.8485184005825815e-05, | |
| "loss": 0.1143, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 4.847803619226594e-05, | |
| "loss": 0.1008, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 4.8470872083929005e-05, | |
| "loss": 0.0968, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 4.8463691685787205e-05, | |
| "loss": 0.1305, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 4.8456495002824034e-05, | |
| "loss": 0.1634, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 4.844928204003433e-05, | |
| "loss": 0.091, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 4.844205280242418e-05, | |
| "loss": 0.0934, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 4.843480729501099e-05, | |
| "loss": 0.0847, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 4.842754552282347e-05, | |
| "loss": 0.0897, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 4.8420267490901595e-05, | |
| "loss": 0.0928, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 4.841297320429664e-05, | |
| "loss": 0.0844, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 4.840566266807115e-05, | |
| "loss": 0.1003, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 4.8398335887298965e-05, | |
| "loss": 0.0999, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 4.839099286706519e-05, | |
| "loss": 0.0724, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 4.8383633612466186e-05, | |
| "loss": 0.0851, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 4.837625812860961e-05, | |
| "loss": 0.0983, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 4.836886642061435e-05, | |
| "loss": 0.1012, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 4.83614584936106e-05, | |
| "loss": 0.1237, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 4.8354034352739754e-05, | |
| "loss": 0.0622, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 4.834659400315451e-05, | |
| "loss": 0.0703, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 4.833913745001878e-05, | |
| "loss": 0.0912, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 4.833166469850775e-05, | |
| "loss": 0.1437, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 4.8324175753807816e-05, | |
| "loss": 0.1134, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 4.8316670621116645e-05, | |
| "loss": 0.0911, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 4.830914930564312e-05, | |
| "loss": 0.0974, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 4.830161181260736e-05, | |
| "loss": 0.1338, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 4.829405814724072e-05, | |
| "loss": 0.1153, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 4.828648831478576e-05, | |
| "loss": 0.0677, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 4.8278902320496276e-05, | |
| "loss": 0.1033, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 4.827130016963729e-05, | |
| "loss": 0.12, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 4.8263681867485e-05, | |
| "loss": 0.1481, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 4.825604741932687e-05, | |
| "loss": 0.0802, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 4.824839683046152e-05, | |
| "loss": 0.0675, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 4.8240730106198784e-05, | |
| "loss": 0.0738, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 4.8233047251859716e-05, | |
| "loss": 0.0777, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 4.8225348272776544e-05, | |
| "loss": 0.0782, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 4.8217633174292697e-05, | |
| "loss": 0.1094, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 4.820990196176279e-05, | |
| "loss": 0.0894, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 4.820215464055261e-05, | |
| "loss": 0.0893, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 4.819439121603915e-05, | |
| "loss": 0.0912, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 4.818661169361055e-05, | |
| "loss": 0.065, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 4.817881607866614e-05, | |
| "loss": 0.0898, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "eval_loss": 0.11340118199586868, | |
| "eval_runtime": 27.0787, | |
| "eval_samples_per_second": 22.158, | |
| "eval_steps_per_second": 2.77, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 4.817100437661643e-05, | |
| "loss": 0.0784, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 4.8163176592883064e-05, | |
| "loss": 0.0954, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 4.815533273289886e-05, | |
| "loss": 0.0851, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 4.814747280210782e-05, | |
| "loss": 0.0842, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 4.8139596805965054e-05, | |
| "loss": 0.1074, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 4.813170474993686e-05, | |
| "loss": 0.0737, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 4.812379663950066e-05, | |
| "loss": 0.0892, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 4.811587248014503e-05, | |
| "loss": 0.0909, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 4.8107932277369675e-05, | |
| "loss": 0.078, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 4.809997603668545e-05, | |
| "loss": 0.0695, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 4.8092003763614325e-05, | |
| "loss": 0.0712, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 4.80840154636894e-05, | |
| "loss": 0.0866, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 4.8076011142454925e-05, | |
| "loss": 0.0918, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 4.8067990805466215e-05, | |
| "loss": 0.0867, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 4.8059954458289756e-05, | |
| "loss": 0.0913, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 4.805190210650311e-05, | |
| "loss": 0.1037, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 4.804383375569496e-05, | |
| "loss": 0.0726, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 4.803574941146509e-05, | |
| "loss": 0.0797, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 4.80276490794244e-05, | |
| "loss": 0.0836, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 4.801953276519486e-05, | |
| "loss": 0.0806, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 4.801140047440954e-05, | |
| "loss": 0.0705, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 4.8003252212712616e-05, | |
| "loss": 0.0963, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 4.7995087985759334e-05, | |
| "loss": 0.0649, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 4.798690779921601e-05, | |
| "loss": 0.0743, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 4.7978711658760066e-05, | |
| "loss": 0.0767, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 4.797049957007996e-05, | |
| "loss": 0.0983, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 4.796227153887526e-05, | |
| "loss": 0.0788, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 4.7954027570856556e-05, | |
| "loss": 0.0864, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 4.794576767174553e-05, | |
| "loss": 0.0839, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 4.7937491847274916e-05, | |
| "loss": 0.0835, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 4.792920010318849e-05, | |
| "loss": 0.0833, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 4.792089244524108e-05, | |
| "loss": 0.0795, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 4.7912568879198575e-05, | |
| "loss": 0.07, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 4.790422941083786e-05, | |
| "loss": 0.0625, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 4.789587404594693e-05, | |
| "loss": 0.0599, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 4.788750279032473e-05, | |
| "loss": 0.0804, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 4.787911564978132e-05, | |
| "loss": 0.0719, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 4.7870712630137695e-05, | |
| "loss": 0.0592, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 4.786229373722595e-05, | |
| "loss": 0.0473, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 4.785385897688914e-05, | |
| "loss": 0.0767, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 4.784540835498136e-05, | |
| "loss": 0.076, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 4.7836941877367726e-05, | |
| "loss": 0.0825, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 4.7828459549924315e-05, | |
| "loss": 0.0579, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 4.7819961378538244e-05, | |
| "loss": 0.0569, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 4.781144736910762e-05, | |
| "loss": 0.0624, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 4.780291752754153e-05, | |
| "loss": 0.087, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 4.779437185976004e-05, | |
| "loss": 0.0871, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 4.778581037169424e-05, | |
| "loss": 0.0889, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 4.7777233069286154e-05, | |
| "loss": 0.0688, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 4.776863995848882e-05, | |
| "loss": 0.0607, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "eval_loss": 0.09115689992904663, | |
| "eval_runtime": 27.0579, | |
| "eval_samples_per_second": 22.175, | |
| "eval_steps_per_second": 2.772, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 4.776003104526621e-05, | |
| "loss": 0.0763, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 4.775140633559329e-05, | |
| "loss": 0.06, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 4.7742765835455996e-05, | |
| "loss": 0.0565, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 4.77341095508512e-05, | |
| "loss": 0.0654, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 4.772543748778674e-05, | |
| "loss": 0.0764, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 4.7716749652281415e-05, | |
| "loss": 0.084, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 4.770804605036494e-05, | |
| "loss": 0.0475, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 4.769932668807801e-05, | |
| "loss": 0.0681, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 4.7690591571472234e-05, | |
| "loss": 0.062, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 4.768184070661016e-05, | |
| "loss": 0.0718, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 4.7673074099565286e-05, | |
| "loss": 0.0505, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 4.766429175642198e-05, | |
| "loss": 0.0811, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 4.765549368327562e-05, | |
| "loss": 0.0587, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 4.7646679886232414e-05, | |
| "loss": 0.0511, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 4.763785037140953e-05, | |
| "loss": 0.0622, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 4.7629005144935036e-05, | |
| "loss": 0.0623, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 4.762014421294791e-05, | |
| "loss": 0.0466, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 4.761126758159801e-05, | |
| "loss": 0.0743, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 4.760237525704611e-05, | |
| "loss": 0.08, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 4.759346724546386e-05, | |
| "loss": 0.047, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 4.758454355303383e-05, | |
| "loss": 0.0501, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 4.757560418594944e-05, | |
| "loss": 0.0524, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 4.7566649150415e-05, | |
| "loss": 0.0496, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 4.755767845264568e-05, | |
| "loss": 0.0538, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 4.7548692098867555e-05, | |
| "loss": 0.0828, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 4.7539690095317526e-05, | |
| "loss": 0.056, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 4.753067244824338e-05, | |
| "loss": 0.0577, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 4.7521639163903765e-05, | |
| "loss": 0.058, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 4.7512590248568163e-05, | |
| "loss": 0.0668, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 4.7503525708516916e-05, | |
| "loss": 0.0605, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 4.7494445550041214e-05, | |
| "loss": 0.0567, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 4.748534977944308e-05, | |
| "loss": 0.0617, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 4.747623840303537e-05, | |
| "loss": 0.0588, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 4.7467111427141776e-05, | |
| "loss": 0.0873, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 4.7457968858096815e-05, | |
| "loss": 0.0478, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 4.744881070224583e-05, | |
| "loss": 0.0656, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 4.743963696594498e-05, | |
| "loss": 0.0587, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 4.743044765556123e-05, | |
| "loss": 0.0546, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 4.7421242777472366e-05, | |
| "loss": 0.0895, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 4.7412022338066976e-05, | |
| "loss": 0.0667, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 4.7402786343744437e-05, | |
| "loss": 0.0581, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 4.739353480091494e-05, | |
| "loss": 0.0799, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 4.738426771599945e-05, | |
| "loss": 0.0617, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 4.7374985095429725e-05, | |
| "loss": 0.0631, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 4.736568694564831e-05, | |
| "loss": 0.0639, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 4.7356373273108534e-05, | |
| "loss": 0.0571, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 4.7347044084274486e-05, | |
| "loss": 0.0552, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 4.733769938562101e-05, | |
| "loss": 0.059, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 4.7328339183633753e-05, | |
| "loss": 0.0601, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 4.7318963484809097e-05, | |
| "loss": 0.0491, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "eval_loss": 0.07235535979270935, | |
| "eval_runtime": 26.886, | |
| "eval_samples_per_second": 22.316, | |
| "eval_steps_per_second": 2.79, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 4.730957229565419e-05, | |
| "loss": 0.0739, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 4.730016562268691e-05, | |
| "loss": 0.0543, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 4.729074347243591e-05, | |
| "loss": 0.0732, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 4.728130585144056e-05, | |
| "loss": 0.0597, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 4.727185276625099e-05, | |
| "loss": 0.0596, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 4.726238422342805e-05, | |
| "loss": 0.0682, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 4.7252900229543316e-05, | |
| "loss": 0.0577, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 4.7243400791179096e-05, | |
| "loss": 0.0789, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 4.723388591492841e-05, | |
| "loss": 0.0431, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 4.7224355607395e-05, | |
| "loss": 0.061, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 4.72148098751933e-05, | |
| "loss": 0.032, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 4.720524872494848e-05, | |
| "loss": 0.0622, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 4.719567216329638e-05, | |
| "loss": 0.0569, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 4.718608019688356e-05, | |
| "loss": 0.0516, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 4.717647283236726e-05, | |
| "loss": 0.0483, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 4.7166850076415396e-05, | |
| "loss": 0.0463, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 4.715721193570659e-05, | |
| "loss": 0.053, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 4.7147558416930136e-05, | |
| "loss": 0.0491, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 4.713788952678598e-05, | |
| "loss": 0.0486, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 4.7128205271984774e-05, | |
| "loss": 0.0393, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 4.711850565924778e-05, | |
| "loss": 0.0655, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 4.710879069530698e-05, | |
| "loss": 0.0632, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 4.709906038690496e-05, | |
| "loss": 0.0408, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 4.708931474079499e-05, | |
| "loss": 0.0829, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 4.7079553763740966e-05, | |
| "loss": 0.039, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 4.706977746251743e-05, | |
| "loss": 0.0502, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 4.705998584390957e-05, | |
| "loss": 0.0364, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 4.7050178914713176e-05, | |
| "loss": 0.0529, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 4.7040356681734695e-05, | |
| "loss": 0.0433, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 4.7030519151791186e-05, | |
| "loss": 0.0473, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 4.7020666331710315e-05, | |
| "loss": 0.0482, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 4.701079822833037e-05, | |
| "loss": 0.0467, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 4.700091484850024e-05, | |
| "loss": 0.0358, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 4.699101619907943e-05, | |
| "loss": 0.0364, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 4.698110228693802e-05, | |
| "loss": 0.0474, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 4.69711731189567e-05, | |
| "loss": 0.0713, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 4.696122870202675e-05, | |
| "loss": 0.0589, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 4.695126904305002e-05, | |
| "loss": 0.054, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 4.6941294148938954e-05, | |
| "loss": 0.0379, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 4.6931304026616553e-05, | |
| "loss": 0.0516, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 4.692129868301639e-05, | |
| "loss": 0.0369, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 4.691127812508263e-05, | |
| "loss": 0.0343, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 4.690124235976996e-05, | |
| "loss": 0.0343, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 4.689119139404364e-05, | |
| "loss": 0.0509, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 4.6881125234879474e-05, | |
| "loss": 0.0343, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 4.6871043889263825e-05, | |
| "loss": 0.0469, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 4.686094736419357e-05, | |
| "loss": 0.027, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 4.6850835666676144e-05, | |
| "loss": 0.0367, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 4.684070880372951e-05, | |
| "loss": 0.0526, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 4.683056678238213e-05, | |
| "loss": 0.0429, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "eval_loss": 0.06248383969068527, | |
| "eval_runtime": 27.0153, | |
| "eval_samples_per_second": 22.21, | |
| "eval_steps_per_second": 2.776, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 4.682040960967303e-05, | |
| "loss": 0.0452, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 4.6810237292651715e-05, | |
| "loss": 0.054, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 4.680004983837823e-05, | |
| "loss": 0.0371, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 4.678984725392309e-05, | |
| "loss": 0.0283, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 4.6779629546367355e-05, | |
| "loss": 0.0631, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 4.676939672280254e-05, | |
| "loss": 0.0411, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 4.6759148790330673e-05, | |
| "loss": 0.0314, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 4.6748885756064266e-05, | |
| "loss": 0.0305, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 4.673860762712632e-05, | |
| "loss": 0.0329, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 4.672831441065028e-05, | |
| "loss": 0.0589, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 4.671800611378011e-05, | |
| "loss": 0.0652, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 4.67076827436702e-05, | |
| "loss": 0.0484, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 4.6697344307485426e-05, | |
| "loss": 0.0309, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 4.668699081240111e-05, | |
| "loss": 0.0322, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 4.667662226560302e-05, | |
| "loss": 0.0318, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 4.666623867428739e-05, | |
| "loss": 0.0362, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 4.665584004566087e-05, | |
| "loss": 0.0573, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 4.664542638694057e-05, | |
| "loss": 0.0446, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 4.6634997705354024e-05, | |
| "loss": 0.0291, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 4.6624554008139175e-05, | |
| "loss": 0.0498, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 4.6614095302544425e-05, | |
| "loss": 0.0269, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 4.660362159582855e-05, | |
| "loss": 0.044, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 4.659313289526076e-05, | |
| "loss": 0.0401, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 4.658262920812069e-05, | |
| "loss": 0.0586, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 4.657211054169832e-05, | |
| "loss": 0.0351, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 4.656157690329409e-05, | |
| "loss": 0.0464, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 4.655102830021879e-05, | |
| "loss": 0.0396, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 4.6540464739793624e-05, | |
| "loss": 0.0423, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 4.6529886229350134e-05, | |
| "loss": 0.0472, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 4.651929277623029e-05, | |
| "loss": 0.0337, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 4.6508684387786395e-05, | |
| "loss": 0.043, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 4.649806107138114e-05, | |
| "loss": 0.0477, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 4.6487422834387564e-05, | |
| "loss": 0.0453, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 4.6476769684189065e-05, | |
| "loss": 0.056, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 4.6466101628179395e-05, | |
| "loss": 0.0358, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 4.6455418673762646e-05, | |
| "loss": 0.0394, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 4.6444720828353256e-05, | |
| "loss": 0.0468, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 4.6434008099375984e-05, | |
| "loss": 0.0348, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 4.642328049426595e-05, | |
| "loss": 0.0374, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 4.641253802046855e-05, | |
| "loss": 0.0327, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 4.640178068543956e-05, | |
| "loss": 0.0412, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 4.639100849664501e-05, | |
| "loss": 0.0426, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 4.638022146156127e-05, | |
| "loss": 0.0355, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 4.6369419587675025e-05, | |
| "loss": 0.0354, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 4.635860288248324e-05, | |
| "loss": 0.0292, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 4.634777135349317e-05, | |
| "loss": 0.0304, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 4.633692500822239e-05, | |
| "loss": 0.0272, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 4.63260638541987e-05, | |
| "loss": 0.0364, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 4.631518789896023e-05, | |
| "loss": 0.0374, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 4.630429715005538e-05, | |
| "loss": 0.0424, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "eval_loss": 0.05031120032072067, | |
| "eval_runtime": 27.0122, | |
| "eval_samples_per_second": 22.212, | |
| "eval_steps_per_second": 2.777, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 4.629339161504278e-05, | |
| "loss": 0.0294, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 4.628247130149135e-05, | |
| "loss": 0.0411, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 4.6271536216980274e-05, | |
| "loss": 0.0336, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 4.626058636909897e-05, | |
| "loss": 0.0262, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 4.6249621765447096e-05, | |
| "loss": 0.0554, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 4.623864241363458e-05, | |
| "loss": 0.041, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 4.622764832128156e-05, | |
| "loss": 0.0497, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 4.621663949601841e-05, | |
| "loss": 0.0263, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 4.620561594548574e-05, | |
| "loss": 0.0435, | |
| "step": 709 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 4.6194577677334386e-05, | |
| "loss": 0.0238, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 4.6183524699225354e-05, | |
| "loss": 0.0339, | |
| "step": 711 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 4.6172457018829916e-05, | |
| "loss": 0.0366, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 4.616137464382952e-05, | |
| "loss": 0.0474, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 4.6150277581915804e-05, | |
| "loss": 0.0582, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 4.613916584079062e-05, | |
| "loss": 0.057, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 4.612803942816599e-05, | |
| "loss": 0.0479, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 4.611689835176414e-05, | |
| "loss": 0.0401, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 4.610574261931746e-05, | |
| "loss": 0.0415, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 4.609457223856851e-05, | |
| "loss": 0.0384, | |
| "step": 719 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 4.6083387217270013e-05, | |
| "loss": 0.0257, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 4.607218756318487e-05, | |
| "loss": 0.0329, | |
| "step": 721 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 4.606097328408612e-05, | |
| "loss": 0.0312, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 4.6049744387756966e-05, | |
| "loss": 0.045, | |
| "step": 723 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 4.6038500881990746e-05, | |
| "loss": 0.0504, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 4.6027242774590936e-05, | |
| "loss": 0.0503, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 4.601597007337116e-05, | |
| "loss": 0.0346, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 4.600468278615516e-05, | |
| "loss": 0.0462, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 4.59933809207768e-05, | |
| "loss": 0.0375, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 4.598206448508007e-05, | |
| "loss": 0.037, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 4.597073348691907e-05, | |
| "loss": 0.0411, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 4.595938793415801e-05, | |
| "loss": 0.0266, | |
| "step": 731 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 4.594802783467117e-05, | |
| "loss": 0.0472, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 4.5936653196342984e-05, | |
| "loss": 0.0274, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 4.592526402706793e-05, | |
| "loss": 0.0342, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 4.5913860334750583e-05, | |
| "loss": 0.0215, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 4.5902442127305614e-05, | |
| "loss": 0.0384, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 4.589100941265775e-05, | |
| "loss": 0.037, | |
| "step": 737 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 4.5879562198741776e-05, | |
| "loss": 0.0397, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 4.586810049350257e-05, | |
| "loss": 0.03, | |
| "step": 739 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 4.585662430489507e-05, | |
| "loss": 0.0348, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 4.584513364088422e-05, | |
| "loss": 0.0272, | |
| "step": 741 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 4.583362850944505e-05, | |
| "loss": 0.0469, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 4.582210891856262e-05, | |
| "loss": 0.0407, | |
| "step": 743 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 4.581057487623204e-05, | |
| "loss": 0.0421, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 4.579902639045841e-05, | |
| "loss": 0.035, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 4.578746346925691e-05, | |
| "loss": 0.0239, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 4.5775886120652686e-05, | |
| "loss": 0.0403, | |
| "step": 747 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 4.576429435268094e-05, | |
| "loss": 0.0221, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 4.5752688173386845e-05, | |
| "loss": 0.0221, | |
| "step": 749 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 4.5741067590825604e-05, | |
| "loss": 0.039, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "eval_loss": 0.03871891275048256, | |
| "eval_runtime": 27.0151, | |
| "eval_samples_per_second": 22.21, | |
| "eval_steps_per_second": 2.776, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 4.572943261306241e-05, | |
| "loss": 0.0313, | |
| "step": 751 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 4.571778324817243e-05, | |
| "loss": 0.0459, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 4.570611950424084e-05, | |
| "loss": 0.0318, | |
| "step": 753 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 4.5694441389362783e-05, | |
| "loss": 0.0338, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 4.5682748911643374e-05, | |
| "loss": 0.0316, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 4.56710420791977e-05, | |
| "loss": 0.0202, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 4.5659320900150824e-05, | |
| "loss": 0.0422, | |
| "step": 757 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 4.5647585382637734e-05, | |
| "loss": 0.0239, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 4.5635835534803406e-05, | |
| "loss": 0.0318, | |
| "step": 759 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 4.5624071364802735e-05, | |
| "loss": 0.0212, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.561229288080057e-05, | |
| "loss": 0.0394, | |
| "step": 761 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.560050009097169e-05, | |
| "loss": 0.0277, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.55886930035008e-05, | |
| "loss": 0.0302, | |
| "step": 763 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.5576871626582534e-05, | |
| "loss": 0.0241, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.5565035968421446e-05, | |
| "loss": 0.0222, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.555318603723199e-05, | |
| "loss": 0.0185, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.5541321841238525e-05, | |
| "loss": 0.0164, | |
| "step": 767 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.552944338867534e-05, | |
| "loss": 0.0225, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.551755068778659e-05, | |
| "loss": 0.0209, | |
| "step": 769 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.550564374682632e-05, | |
| "loss": 0.0218, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.5493722574058455e-05, | |
| "loss": 0.0179, | |
| "step": 771 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.548178717775683e-05, | |
| "loss": 0.0208, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 4.546983756620511e-05, | |
| "loss": 0.0224, | |
| "step": 773 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 4.545787374769686e-05, | |
| "loss": 0.0239, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 4.544589573053547e-05, | |
| "loss": 0.0258, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 4.5433903523034226e-05, | |
| "loss": 0.014, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 4.542189713351622e-05, | |
| "loss": 0.0165, | |
| "step": 777 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 4.540987657031443e-05, | |
| "loss": 0.0215, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 4.5397841841771626e-05, | |
| "loss": 0.0104, | |
| "step": 779 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 4.5385792956240444e-05, | |
| "loss": 0.019, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 4.537372992208334e-05, | |
| "loss": 0.0192, | |
| "step": 781 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 4.536165274767258e-05, | |
| "loss": 0.0141, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 4.534956144139024e-05, | |
| "loss": 0.0174, | |
| "step": 783 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 4.5337456011628214e-05, | |
| "loss": 0.012, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 4.5325336466788205e-05, | |
| "loss": 0.0195, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 4.53132028152817e-05, | |
| "loss": 0.0144, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 4.5301055065529975e-05, | |
| "loss": 0.0134, | |
| "step": 787 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 4.528889322596409e-05, | |
| "loss": 0.0113, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 4.527671730502491e-05, | |
| "loss": 0.0152, | |
| "step": 789 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 4.526452731116304e-05, | |
| "loss": 0.013, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 4.525232325283887e-05, | |
| "loss": 0.0231, | |
| "step": 791 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 4.5240105138522536e-05, | |
| "loss": 0.0276, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 4.522787297669396e-05, | |
| "loss": 0.014, | |
| "step": 793 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 4.5215626775842775e-05, | |
| "loss": 0.0186, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 4.520336654446838e-05, | |
| "loss": 0.0319, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 4.519109229107992e-05, | |
| "loss": 0.0174, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 4.517880402419624e-05, | |
| "loss": 0.0115, | |
| "step": 797 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 4.516650175234596e-05, | |
| "loss": 0.0304, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 4.515418548406736e-05, | |
| "loss": 0.0321, | |
| "step": 799 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 4.514185522790848e-05, | |
| "loss": 0.0173, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "eval_loss": 0.034719910472631454, | |
| "eval_runtime": 26.9947, | |
| "eval_samples_per_second": 22.227, | |
| "eval_steps_per_second": 2.778, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 4.512951099242706e-05, | |
| "loss": 0.019, | |
| "step": 801 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 4.511715278619052e-05, | |
| "loss": 0.0207, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 4.5104780617776013e-05, | |
| "loss": 0.0127, | |
| "step": 803 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 4.5092394495770335e-05, | |
| "loss": 0.012, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 4.5079994428770015e-05, | |
| "loss": 0.0156, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 4.506758042538122e-05, | |
| "loss": 0.0149, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 4.505515249421982e-05, | |
| "loss": 0.0156, | |
| "step": 807 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 4.5042710643911345e-05, | |
| "loss": 0.0141, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 4.503025488309096e-05, | |
| "loss": 0.0163, | |
| "step": 809 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 4.501778522040353e-05, | |
| "loss": 0.0209, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 4.500530166450351e-05, | |
| "loss": 0.0238, | |
| "step": 811 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 4.499280422405506e-05, | |
| "loss": 0.0117, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 4.498029290773193e-05, | |
| "loss": 0.0168, | |
| "step": 813 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 4.496776772421752e-05, | |
| "loss": 0.0158, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 4.495522868220485e-05, | |
| "loss": 0.0267, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 4.4942675790396575e-05, | |
| "loss": 0.0128, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 4.493010905750493e-05, | |
| "loss": 0.0265, | |
| "step": 817 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 4.491752849225177e-05, | |
| "loss": 0.0187, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 4.490493410336857e-05, | |
| "loss": 0.0169, | |
| "step": 819 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 4.4892325899596375e-05, | |
| "loss": 0.0086, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 4.487970388968583e-05, | |
| "loss": 0.0158, | |
| "step": 821 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 4.4867068082397144e-05, | |
| "loss": 0.0183, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 4.485441848650014e-05, | |
| "loss": 0.0148, | |
| "step": 823 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 4.484175511077416e-05, | |
| "loss": 0.0173, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 4.482907796400817e-05, | |
| "loss": 0.0097, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 4.481638705500063e-05, | |
| "loss": 0.0172, | |
| "step": 826 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 4.480368239255959e-05, | |
| "loss": 0.0195, | |
| "step": 827 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 4.479096398550265e-05, | |
| "loss": 0.0166, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 4.477823184265693e-05, | |
| "loss": 0.0149, | |
| "step": 829 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 4.476548597285908e-05, | |
| "loss": 0.0262, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 4.47527263849553e-05, | |
| "loss": 0.0144, | |
| "step": 831 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 4.473995308780129e-05, | |
| "loss": 0.0113, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 4.4727166090262285e-05, | |
| "loss": 0.0092, | |
| "step": 833 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 4.4714365401213e-05, | |
| "loss": 0.0213, | |
| "step": 834 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 4.470155102953769e-05, | |
| "loss": 0.0186, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 4.468872298413007e-05, | |
| "loss": 0.0144, | |
| "step": 836 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 4.467588127389336e-05, | |
| "loss": 0.012, | |
| "step": 837 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 4.466302590774028e-05, | |
| "loss": 0.0175, | |
| "step": 838 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 4.4650156894592995e-05, | |
| "loss": 0.0151, | |
| "step": 839 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 4.463727424338316e-05, | |
| "loss": 0.0135, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 4.4624377963051914e-05, | |
| "loss": 0.0151, | |
| "step": 841 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 4.461146806254982e-05, | |
| "loss": 0.0149, | |
| "step": 842 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 4.459854455083691e-05, | |
| "loss": 0.0207, | |
| "step": 843 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 4.458560743688266e-05, | |
| "loss": 0.0243, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 4.457265672966601e-05, | |
| "loss": 0.0147, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 4.4559692438175284e-05, | |
| "loss": 0.0171, | |
| "step": 846 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 4.454671457140829e-05, | |
| "loss": 0.0114, | |
| "step": 847 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 4.453372313837221e-05, | |
| "loss": 0.0217, | |
| "step": 848 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 4.4520718148083665e-05, | |
| "loss": 0.0181, | |
| "step": 849 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 4.4507699609568695e-05, | |
| "loss": 0.0188, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "eval_loss": 0.031473103910684586, | |
| "eval_runtime": 27.0103, | |
| "eval_samples_per_second": 22.214, | |
| "eval_steps_per_second": 2.777, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 4.449466753186273e-05, | |
| "loss": 0.0153, | |
| "step": 851 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 4.44816219240106e-05, | |
| "loss": 0.0134, | |
| "step": 852 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 4.446856279506651e-05, | |
| "loss": 0.0118, | |
| "step": 853 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 4.445549015409407e-05, | |
| "loss": 0.021, | |
| "step": 854 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 4.444240401016626e-05, | |
| "loss": 0.0257, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 4.4429304372365444e-05, | |
| "loss": 0.0094, | |
| "step": 856 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 4.441619124978332e-05, | |
| "loss": 0.0131, | |
| "step": 857 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 4.4403064651520975e-05, | |
| "loss": 0.0257, | |
| "step": 858 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 4.438992458668884e-05, | |
| "loss": 0.0122, | |
| "step": 859 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 4.4376771064406684e-05, | |
| "loss": 0.0172, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 4.4363604093803635e-05, | |
| "loss": 0.0179, | |
| "step": 861 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 4.435042368401813e-05, | |
| "loss": 0.0183, | |
| "step": 862 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 4.4337229844197954e-05, | |
| "loss": 0.0148, | |
| "step": 863 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 4.43240225835002e-05, | |
| "loss": 0.0227, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 4.431080191109128e-05, | |
| "loss": 0.0167, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 4.4297567836146924e-05, | |
| "loss": 0.0163, | |
| "step": 866 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 4.4284320367852145e-05, | |
| "loss": 0.0131, | |
| "step": 867 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 4.427105951540127e-05, | |
| "loss": 0.0231, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 4.425778528799789e-05, | |
| "loss": 0.0166, | |
| "step": 869 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 4.424449769485492e-05, | |
| "loss": 0.0179, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 4.423119674519451e-05, | |
| "loss": 0.0134, | |
| "step": 871 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 4.421788244824811e-05, | |
| "loss": 0.0276, | |
| "step": 872 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 4.420455481325639e-05, | |
| "loss": 0.0189, | |
| "step": 873 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 4.4191213849469346e-05, | |
| "loss": 0.0301, | |
| "step": 874 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 4.417785956614616e-05, | |
| "loss": 0.0123, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 4.41644919725553e-05, | |
| "loss": 0.0294, | |
| "step": 876 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 4.415111107797445e-05, | |
| "loss": 0.0162, | |
| "step": 877 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 4.4137716891690536e-05, | |
| "loss": 0.0171, | |
| "step": 878 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 4.41243094229997e-05, | |
| "loss": 0.0133, | |
| "step": 879 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 4.4110888681207314e-05, | |
| "loss": 0.0127, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 4.4097454675627946e-05, | |
| "loss": 0.0163, | |
| "step": 881 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 4.408400741558539e-05, | |
| "loss": 0.0149, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 4.4070546910412625e-05, | |
| "loss": 0.011, | |
| "step": 883 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 4.405707316945182e-05, | |
| "loss": 0.0174, | |
| "step": 884 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 4.404358620205435e-05, | |
| "loss": 0.016, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 4.4030086017580744e-05, | |
| "loss": 0.0152, | |
| "step": 886 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 4.4016572625400723e-05, | |
| "loss": 0.0121, | |
| "step": 887 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 4.400304603489316e-05, | |
| "loss": 0.0121, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 4.398950625544611e-05, | |
| "loss": 0.0178, | |
| "step": 889 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 4.397595329645675e-05, | |
| "loss": 0.0136, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 4.396238716733144e-05, | |
| "loss": 0.0263, | |
| "step": 891 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 4.3948807877485666e-05, | |
| "loss": 0.016, | |
| "step": 892 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 4.393521543634402e-05, | |
| "loss": 0.0166, | |
| "step": 893 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 4.392160985334027e-05, | |
| "loss": 0.0089, | |
| "step": 894 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 4.390799113791727e-05, | |
| "loss": 0.0148, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 4.3894359299527e-05, | |
| "loss": 0.0157, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 4.388071434763056e-05, | |
| "loss": 0.0112, | |
| "step": 897 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 4.386705629169813e-05, | |
| "loss": 0.0066, | |
| "step": 898 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 4.385338514120899e-05, | |
| "loss": 0.0157, | |
| "step": 899 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 4.3839700905651517e-05, | |
| "loss": 0.0134, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "eval_loss": 0.03026873990893364, | |
| "eval_runtime": 26.9629, | |
| "eval_samples_per_second": 22.253, | |
| "eval_steps_per_second": 2.782, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 4.382600359452317e-05, | |
| "loss": 0.0202, | |
| "step": 901 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 4.381229321733047e-05, | |
| "loss": 0.0127, | |
| "step": 902 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 4.3798569783589014e-05, | |
| "loss": 0.014, | |
| "step": 903 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 4.378483330282348e-05, | |
| "loss": 0.0104, | |
| "step": 904 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 4.377108378456756e-05, | |
| "loss": 0.0083, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 4.3757321238364015e-05, | |
| "loss": 0.0112, | |
| "step": 906 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 4.374354567376467e-05, | |
| "loss": 0.0159, | |
| "step": 907 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 4.3729757100330346e-05, | |
| "loss": 0.0157, | |
| "step": 908 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 4.371595552763093e-05, | |
| "loss": 0.0199, | |
| "step": 909 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 4.37021409652453e-05, | |
| "loss": 0.0211, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 4.368831342276136e-05, | |
| "loss": 0.0173, | |
| "step": 911 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 4.3674472909776046e-05, | |
| "loss": 0.0061, | |
| "step": 912 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 4.366061943589524e-05, | |
| "loss": 0.0149, | |
| "step": 913 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 4.3646753010733895e-05, | |
| "loss": 0.0095, | |
| "step": 914 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 4.363287364391588e-05, | |
| "loss": 0.0201, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 4.36189813450741e-05, | |
| "loss": 0.0138, | |
| "step": 916 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 4.36050761238504e-05, | |
| "loss": 0.0154, | |
| "step": 917 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 4.359115798989562e-05, | |
| "loss": 0.0134, | |
| "step": 918 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 4.357722695286953e-05, | |
| "loss": 0.014, | |
| "step": 919 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 4.3563283022440895e-05, | |
| "loss": 0.0148, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 4.35493262082874e-05, | |
| "loss": 0.0114, | |
| "step": 921 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 4.353535652009568e-05, | |
| "loss": 0.0166, | |
| "step": 922 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 4.352137396756131e-05, | |
| "loss": 0.0112, | |
| "step": 923 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 4.350737856038878e-05, | |
| "loss": 0.0172, | |
| "step": 924 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 4.3493370308291516e-05, | |
| "loss": 0.0219, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 4.347934922099186e-05, | |
| "loss": 0.0163, | |
| "step": 926 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 4.3465315308221046e-05, | |
| "loss": 0.0195, | |
| "step": 927 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 4.3451268579719216e-05, | |
| "loss": 0.0179, | |
| "step": 928 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 4.3437209045235425e-05, | |
| "loss": 0.0134, | |
| "step": 929 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 4.342313671452759e-05, | |
| "loss": 0.0175, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 4.340905159736253e-05, | |
| "loss": 0.0106, | |
| "step": 931 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 4.3394953703515905e-05, | |
| "loss": 0.017, | |
| "step": 932 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 4.338084304277229e-05, | |
| "loss": 0.0129, | |
| "step": 933 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 4.3366719624925084e-05, | |
| "loss": 0.0128, | |
| "step": 934 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 4.3352583459776554e-05, | |
| "loss": 0.0102, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 4.3338434557137816e-05, | |
| "loss": 0.0129, | |
| "step": 936 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 4.332427292682882e-05, | |
| "loss": 0.015, | |
| "step": 937 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 4.331009857867836e-05, | |
| "loss": 0.015, | |
| "step": 938 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 4.3295911522524044e-05, | |
| "loss": 0.0077, | |
| "step": 939 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 4.32817117682123e-05, | |
| "loss": 0.0225, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 4.326749932559838e-05, | |
| "loss": 0.0125, | |
| "step": 941 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 4.3253274204546344e-05, | |
| "loss": 0.0098, | |
| "step": 942 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 4.323903641492903e-05, | |
| "loss": 0.0063, | |
| "step": 943 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 4.322478596662809e-05, | |
| "loss": 0.0083, | |
| "step": 944 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 4.3210522869533946e-05, | |
| "loss": 0.0131, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 4.319624713354582e-05, | |
| "loss": 0.0079, | |
| "step": 946 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 4.318195876857169e-05, | |
| "loss": 0.0143, | |
| "step": 947 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 4.316765778452829e-05, | |
| "loss": 0.023, | |
| "step": 948 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 4.315334419134114e-05, | |
| "loss": 0.0126, | |
| "step": 949 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 4.3139017998944486e-05, | |
| "loss": 0.0113, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "eval_loss": 0.027030887082219124, | |
| "eval_runtime": 26.8967, | |
| "eval_samples_per_second": 22.308, | |
| "eval_steps_per_second": 2.788, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 4.312467921728133e-05, | |
| "loss": 0.0222, | |
| "step": 951 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 4.311032785630341e-05, | |
| "loss": 0.0131, | |
| "step": 952 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 4.3095963925971194e-05, | |
| "loss": 0.024, | |
| "step": 953 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 4.308158743625388e-05, | |
| "loss": 0.0095, | |
| "step": 954 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 4.306719839712936e-05, | |
| "loss": 0.0142, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 4.305279681858426e-05, | |
| "loss": 0.0151, | |
| "step": 956 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 4.303838271061391e-05, | |
| "loss": 0.0157, | |
| "step": 957 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 4.302395608322233e-05, | |
| "loss": 0.016, | |
| "step": 958 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 4.30095169464222e-05, | |
| "loss": 0.0152, | |
| "step": 959 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 4.2995065310234926e-05, | |
| "loss": 0.0108, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 4.2980601184690576e-05, | |
| "loss": 0.0138, | |
| "step": 961 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 4.2966124579827874e-05, | |
| "loss": 0.0113, | |
| "step": 962 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 4.2951635505694214e-05, | |
| "loss": 0.014, | |
| "step": 963 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 4.293713397234565e-05, | |
| "loss": 0.0071, | |
| "step": 964 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 4.2922619989846856e-05, | |
| "loss": 0.0091, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 4.290809356827119e-05, | |
| "loss": 0.0108, | |
| "step": 966 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 4.2893554717700604e-05, | |
| "loss": 0.0141, | |
| "step": 967 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 4.2879003448225694e-05, | |
| "loss": 0.0119, | |
| "step": 968 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 4.286443976994569e-05, | |
| "loss": 0.0119, | |
| "step": 969 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 4.284986369296839e-05, | |
| "loss": 0.0157, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 4.2835275227410245e-05, | |
| "loss": 0.0129, | |
| "step": 971 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 4.2820674383396274e-05, | |
| "loss": 0.015, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 4.28060611710601e-05, | |
| "loss": 0.015, | |
| "step": 973 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 4.2791435600543934e-05, | |
| "loss": 0.0105, | |
| "step": 974 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 4.277679768199855e-05, | |
| "loss": 0.0131, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 4.2762147425583285e-05, | |
| "loss": 0.0126, | |
| "step": 976 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 4.274748484146609e-05, | |
| "loss": 0.0281, | |
| "step": 977 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 4.2732809939823404e-05, | |
| "loss": 0.0082, | |
| "step": 978 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 4.2718122730840254e-05, | |
| "loss": 0.0066, | |
| "step": 979 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 4.27034232247102e-05, | |
| "loss": 0.018, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 4.268871143163535e-05, | |
| "loss": 0.0127, | |
| "step": 981 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 4.2673987361826315e-05, | |
| "loss": 0.0251, | |
| "step": 982 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 4.2659251025502244e-05, | |
| "loss": 0.012, | |
| "step": 983 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 4.264450243289079e-05, | |
| "loss": 0.0223, | |
| "step": 984 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 4.2629741594228126e-05, | |
| "loss": 0.0182, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 4.261496851975891e-05, | |
| "loss": 0.0073, | |
| "step": 986 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 4.260018321973629e-05, | |
| "loss": 0.0084, | |
| "step": 987 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 4.258538570442192e-05, | |
| "loss": 0.0124, | |
| "step": 988 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 4.257057598408591e-05, | |
| "loss": 0.0127, | |
| "step": 989 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 4.255575406900685e-05, | |
| "loss": 0.0135, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 4.2540919969471793e-05, | |
| "loss": 0.0098, | |
| "step": 991 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 4.252607369577625e-05, | |
| "loss": 0.0164, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 4.251121525822419e-05, | |
| "loss": 0.0117, | |
| "step": 993 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 4.249634466712799e-05, | |
| "loss": 0.0238, | |
| "step": 994 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 4.248146193280851e-05, | |
| "loss": 0.0134, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 4.2466567065595e-05, | |
| "loss": 0.0207, | |
| "step": 996 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 4.245166007582515e-05, | |
| "loss": 0.0118, | |
| "step": 997 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 4.243674097384506e-05, | |
| "loss": 0.018, | |
| "step": 998 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 4.2421809770009225e-05, | |
| "loss": 0.014, | |
| "step": 999 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 4.240686647468056e-05, | |
| "loss": 0.0146, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "eval_loss": 0.024349618703126907, | |
| "eval_runtime": 26.998, | |
| "eval_samples_per_second": 22.224, | |
| "eval_steps_per_second": 2.778, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 4.239191109823036e-05, | |
| "loss": 0.0126, | |
| "step": 1001 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 4.237694365103831e-05, | |
| "loss": 0.0142, | |
| "step": 1002 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 4.236196414349246e-05, | |
| "loss": 0.0133, | |
| "step": 1003 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 4.2346972585989234e-05, | |
| "loss": 0.0096, | |
| "step": 1004 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 4.2331968988933436e-05, | |
| "loss": 0.009, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 4.231695336273821e-05, | |
| "loss": 0.0099, | |
| "step": 1006 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 4.2301925717825055e-05, | |
| "loss": 0.0123, | |
| "step": 1007 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 4.22868860646238e-05, | |
| "loss": 0.0119, | |
| "step": 1008 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 4.227183441357263e-05, | |
| "loss": 0.0101, | |
| "step": 1009 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 4.225677077511804e-05, | |
| "loss": 0.0084, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 4.2241695159714844e-05, | |
| "loss": 0.0204, | |
| "step": 1011 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 4.2226607577826184e-05, | |
| "loss": 0.0072, | |
| "step": 1012 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 4.22115080399235e-05, | |
| "loss": 0.0132, | |
| "step": 1013 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 4.219639655648651e-05, | |
| "loss": 0.0255, | |
| "step": 1014 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 4.2181273138003244e-05, | |
| "loss": 0.0065, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 4.216613779497003e-05, | |
| "loss": 0.0104, | |
| "step": 1016 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 4.215099053789143e-05, | |
| "loss": 0.0096, | |
| "step": 1017 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 4.213583137728031e-05, | |
| "loss": 0.0147, | |
| "step": 1018 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 4.212066032365779e-05, | |
| "loss": 0.0175, | |
| "step": 1019 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 4.210547738755323e-05, | |
| "loss": 0.0175, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 4.209028257950426e-05, | |
| "loss": 0.0141, | |
| "step": 1021 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 4.207507591005673e-05, | |
| "loss": 0.0145, | |
| "step": 1022 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 4.205985738976472e-05, | |
| "loss": 0.0162, | |
| "step": 1023 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 4.204462702919055e-05, | |
| "loss": 0.011, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 4.2029384838904776e-05, | |
| "loss": 0.0127, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 4.20141308294861e-05, | |
| "loss": 0.0124, | |
| "step": 1026 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 4.199886501152149e-05, | |
| "loss": 0.0166, | |
| "step": 1027 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 4.198358739560608e-05, | |
| "loss": 0.0114, | |
| "step": 1028 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 4.196829799234321e-05, | |
| "loss": 0.0079, | |
| "step": 1029 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 4.195299681234438e-05, | |
| "loss": 0.0167, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 4.1937683866229286e-05, | |
| "loss": 0.0065, | |
| "step": 1031 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 4.1922359164625744e-05, | |
| "loss": 0.028, | |
| "step": 1032 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 4.190702271816981e-05, | |
| "loss": 0.0122, | |
| "step": 1033 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 4.189167453750561e-05, | |
| "loss": 0.0054, | |
| "step": 1034 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 4.187631463328545e-05, | |
| "loss": 0.0129, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 4.186094301616979e-05, | |
| "loss": 0.0145, | |
| "step": 1036 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 4.184555969682719e-05, | |
| "loss": 0.0224, | |
| "step": 1037 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 4.183016468593434e-05, | |
| "loss": 0.0116, | |
| "step": 1038 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 4.181475799417605e-05, | |
| "loss": 0.0198, | |
| "step": 1039 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 4.1799339632245224e-05, | |
| "loss": 0.0127, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 4.178390961084289e-05, | |
| "loss": 0.0132, | |
| "step": 1041 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 4.1768467940678147e-05, | |
| "loss": 0.0133, | |
| "step": 1042 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 4.175301463246818e-05, | |
| "loss": 0.0108, | |
| "step": 1043 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 4.173754969693826e-05, | |
| "loss": 0.0064, | |
| "step": 1044 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 4.172207314482171e-05, | |
| "loss": 0.0127, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 4.170658498685996e-05, | |
| "loss": 0.0192, | |
| "step": 1046 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 4.169108523380245e-05, | |
| "loss": 0.0082, | |
| "step": 1047 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 4.167557389640667e-05, | |
| "loss": 0.0091, | |
| "step": 1048 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 4.166005098543818e-05, | |
| "loss": 0.011, | |
| "step": 1049 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 4.164451651167054e-05, | |
| "loss": 0.0148, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "eval_loss": 0.0241435207426548, | |
| "eval_runtime": 26.9627, | |
| "eval_samples_per_second": 22.253, | |
| "eval_steps_per_second": 2.782, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 4.1628970485885355e-05, | |
| "loss": 0.0103, | |
| "step": 1051 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 4.161341291887224e-05, | |
| "loss": 0.0058, | |
| "step": 1052 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 4.1597843821428836e-05, | |
| "loss": 0.0245, | |
| "step": 1053 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 4.1582263204360755e-05, | |
| "loss": 0.0246, | |
| "step": 1054 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 4.156667107848163e-05, | |
| "loss": 0.0139, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 4.155106745461308e-05, | |
| "loss": 0.0213, | |
| "step": 1056 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 4.153545234358469e-05, | |
| "loss": 0.0119, | |
| "step": 1057 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 4.151982575623402e-05, | |
| "loss": 0.0109, | |
| "step": 1058 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 4.1504187703406604e-05, | |
| "loss": 0.0233, | |
| "step": 1059 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 4.148853819595595e-05, | |
| "loss": 0.0194, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 4.1472877244743466e-05, | |
| "loss": 0.016, | |
| "step": 1061 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 4.1457204860638544e-05, | |
| "loss": 0.0075, | |
| "step": 1062 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 4.144152105451851e-05, | |
| "loss": 0.0102, | |
| "step": 1063 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 4.1425825837268596e-05, | |
| "loss": 0.0108, | |
| "step": 1064 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 4.141011921978196e-05, | |
| "loss": 0.012, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 4.139440121295969e-05, | |
| "loss": 0.0101, | |
| "step": 1066 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 4.1378671827710755e-05, | |
| "loss": 0.0079, | |
| "step": 1067 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 4.136293107495203e-05, | |
| "loss": 0.0137, | |
| "step": 1068 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 4.13471789656083e-05, | |
| "loss": 0.0099, | |
| "step": 1069 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 4.133141551061219e-05, | |
| "loss": 0.0108, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 4.1315640720904236e-05, | |
| "loss": 0.0172, | |
| "step": 1071 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 4.129985460743283e-05, | |
| "loss": 0.0105, | |
| "step": 1072 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 4.128405718115421e-05, | |
| "loss": 0.0051, | |
| "step": 1073 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 4.126824845303248e-05, | |
| "loss": 0.0076, | |
| "step": 1074 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 4.125242843403958e-05, | |
| "loss": 0.0079, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 4.12365971351553e-05, | |
| "loss": 0.0148, | |
| "step": 1076 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 4.122075456736725e-05, | |
| "loss": 0.0112, | |
| "step": 1077 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 4.120490074167085e-05, | |
| "loss": 0.0121, | |
| "step": 1078 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 4.1189035669069344e-05, | |
| "loss": 0.0096, | |
| "step": 1079 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 4.117315936057379e-05, | |
| "loss": 0.0129, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 4.1157271827203026e-05, | |
| "loss": 0.0131, | |
| "step": 1081 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 4.11413730799837e-05, | |
| "loss": 0.0049, | |
| "step": 1082 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 4.112546312995024e-05, | |
| "loss": 0.014, | |
| "step": 1083 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 4.110954198814482e-05, | |
| "loss": 0.0062, | |
| "step": 1084 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 4.1093609665617416e-05, | |
| "loss": 0.0109, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 4.1077666173425756e-05, | |
| "loss": 0.0118, | |
| "step": 1086 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 4.106171152263532e-05, | |
| "loss": 0.0121, | |
| "step": 1087 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 4.10457457243193e-05, | |
| "loss": 0.011, | |
| "step": 1088 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 4.102976878955869e-05, | |
| "loss": 0.0104, | |
| "step": 1089 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 4.101378072944216e-05, | |
| "loss": 0.0157, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 4.0997781555066115e-05, | |
| "loss": 0.0166, | |
| "step": 1091 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 4.0981771277534684e-05, | |
| "loss": 0.0164, | |
| "step": 1092 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 4.0965749907959696e-05, | |
| "loss": 0.0102, | |
| "step": 1093 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 4.0949717457460674e-05, | |
| "loss": 0.0117, | |
| "step": 1094 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 4.093367393716483e-05, | |
| "loss": 0.0114, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 4.091761935820708e-05, | |
| "loss": 0.0197, | |
| "step": 1096 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 4.090155373172998e-05, | |
| "loss": 0.0105, | |
| "step": 1097 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 4.088547706888378e-05, | |
| "loss": 0.0127, | |
| "step": 1098 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 4.0869389380826395e-05, | |
| "loss": 0.0083, | |
| "step": 1099 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 4.0853290678723364e-05, | |
| "loss": 0.017, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "eval_loss": 0.021086279302835464, | |
| "eval_runtime": 27.0777, | |
| "eval_samples_per_second": 22.158, | |
| "eval_steps_per_second": 2.77, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 4.083718097374788e-05, | |
| "loss": 0.0132, | |
| "step": 1101 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 4.08210602770808e-05, | |
| "loss": 0.0088, | |
| "step": 1102 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 4.080492859991057e-05, | |
| "loss": 0.0094, | |
| "step": 1103 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 4.0788785953433286e-05, | |
| "loss": 0.0091, | |
| "step": 1104 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 4.077263234885264e-05, | |
| "loss": 0.0126, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 4.075646779737993e-05, | |
| "loss": 0.0135, | |
| "step": 1106 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 4.074029231023406e-05, | |
| "loss": 0.0047, | |
| "step": 1107 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 4.072410589864154e-05, | |
| "loss": 0.0157, | |
| "step": 1108 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 4.0707908573836415e-05, | |
| "loss": 0.0123, | |
| "step": 1109 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 4.069170034706035e-05, | |
| "loss": 0.012, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 4.067548122956254e-05, | |
| "loss": 0.0088, | |
| "step": 1111 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 4.065925123259978e-05, | |
| "loss": 0.0115, | |
| "step": 1112 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 4.064301036743638e-05, | |
| "loss": 0.0116, | |
| "step": 1113 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 4.0626758645344205e-05, | |
| "loss": 0.0059, | |
| "step": 1114 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 4.061049607760266e-05, | |
| "loss": 0.0101, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 4.059422267549868e-05, | |
| "loss": 0.015, | |
| "step": 1116 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 4.05779384503267e-05, | |
| "loss": 0.0114, | |
| "step": 1117 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 4.056164341338868e-05, | |
| "loss": 0.0061, | |
| "step": 1118 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 4.05453375759941e-05, | |
| "loss": 0.0113, | |
| "step": 1119 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 4.0529020949459904e-05, | |
| "loss": 0.0203, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 4.051269354511055e-05, | |
| "loss": 0.0085, | |
| "step": 1121 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 4.049635537427795e-05, | |
| "loss": 0.0121, | |
| "step": 1122 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 4.048000644830152e-05, | |
| "loss": 0.0158, | |
| "step": 1123 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 4.046364677852812e-05, | |
| "loss": 0.0037, | |
| "step": 1124 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 4.044727637631207e-05, | |
| "loss": 0.0101, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 4.0430895253015144e-05, | |
| "loss": 0.0181, | |
| "step": 1126 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 4.041450342000655e-05, | |
| "loss": 0.0146, | |
| "step": 1127 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 4.039810088866293e-05, | |
| "loss": 0.0089, | |
| "step": 1128 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 4.038168767036836e-05, | |
| "loss": 0.0124, | |
| "step": 1129 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 4.036526377651432e-05, | |
| "loss": 0.0182, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 4.0348829218499706e-05, | |
| "loss": 0.0116, | |
| "step": 1131 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 4.0332384007730826e-05, | |
| "loss": 0.0213, | |
| "step": 1132 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 4.031592815562135e-05, | |
| "loss": 0.0076, | |
| "step": 1133 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 4.0299461673592376e-05, | |
| "loss": 0.0145, | |
| "step": 1134 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 4.028298457307235e-05, | |
| "loss": 0.012, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 4.026649686549708e-05, | |
| "loss": 0.0149, | |
| "step": 1136 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 4.0249998562309785e-05, | |
| "loss": 0.0097, | |
| "step": 1137 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 4.023348967496098e-05, | |
| "loss": 0.0171, | |
| "step": 1138 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 4.0216970214908545e-05, | |
| "loss": 0.0165, | |
| "step": 1139 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 4.020044019361772e-05, | |
| "loss": 0.0098, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 4.018389962256104e-05, | |
| "loss": 0.0068, | |
| "step": 1141 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 4.01673485132184e-05, | |
| "loss": 0.0153, | |
| "step": 1142 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 4.015078687707697e-05, | |
| "loss": 0.0062, | |
| "step": 1143 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 4.013421472563127e-05, | |
| "loss": 0.0093, | |
| "step": 1144 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 4.011763207038307e-05, | |
| "loss": 0.0062, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 4.010103892284146e-05, | |
| "loss": 0.0096, | |
| "step": 1146 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 4.00844352945228e-05, | |
| "loss": 0.0094, | |
| "step": 1147 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 4.006782119695074e-05, | |
| "loss": 0.0073, | |
| "step": 1148 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 4.0051196641656185e-05, | |
| "loss": 0.0061, | |
| "step": 1149 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 4.003456164017729e-05, | |
| "loss": 0.0038, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "eval_loss": 0.020303795114159584, | |
| "eval_runtime": 26.893, | |
| "eval_samples_per_second": 22.311, | |
| "eval_steps_per_second": 2.789, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 4.001791620405947e-05, | |
| "loss": 0.0121, | |
| "step": 1151 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 4.000126034485539e-05, | |
| "loss": 0.0061, | |
| "step": 1152 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 3.9984594074124924e-05, | |
| "loss": 0.0033, | |
| "step": 1153 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 3.99679174034352e-05, | |
| "loss": 0.0103, | |
| "step": 1154 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 3.995123034436056e-05, | |
| "loss": 0.0043, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 3.993453290848252e-05, | |
| "loss": 0.0101, | |
| "step": 1156 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 3.9917825107389853e-05, | |
| "loss": 0.0055, | |
| "step": 1157 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 3.9901106952678494e-05, | |
| "loss": 0.0034, | |
| "step": 1158 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 3.9884378455951554e-05, | |
| "loss": 0.005, | |
| "step": 1159 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 3.986763962881935e-05, | |
| "loss": 0.0038, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 3.985089048289935e-05, | |
| "loss": 0.0044, | |
| "step": 1161 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 3.983413102981619e-05, | |
| "loss": 0.0143, | |
| "step": 1162 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 3.981736128120166e-05, | |
| "loss": 0.0064, | |
| "step": 1163 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 3.980058124869469e-05, | |
| "loss": 0.007, | |
| "step": 1164 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 3.978379094394135e-05, | |
| "loss": 0.0128, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 3.9766990378594854e-05, | |
| "loss": 0.0083, | |
| "step": 1166 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 3.9750179564315515e-05, | |
| "loss": 0.0108, | |
| "step": 1167 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 3.973335851277077e-05, | |
| "loss": 0.0064, | |
| "step": 1168 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 3.9716527235635146e-05, | |
| "loss": 0.008, | |
| "step": 1169 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 3.9699685744590304e-05, | |
| "loss": 0.0092, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 3.968283405132496e-05, | |
| "loss": 0.0075, | |
| "step": 1171 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 3.966597216753492e-05, | |
| "loss": 0.0108, | |
| "step": 1172 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 3.9649100104923055e-05, | |
| "loss": 0.0088, | |
| "step": 1173 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 3.963221787519933e-05, | |
| "loss": 0.0062, | |
| "step": 1174 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 3.961532549008072e-05, | |
| "loss": 0.0033, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 3.9598422961291306e-05, | |
| "loss": 0.0038, | |
| "step": 1176 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 3.9581510300562155e-05, | |
| "loss": 0.0055, | |
| "step": 1177 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 3.9564587519631384e-05, | |
| "loss": 0.0058, | |
| "step": 1178 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 3.9547654630244156e-05, | |
| "loss": 0.0075, | |
| "step": 1179 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 3.953071164415262e-05, | |
| "loss": 0.0041, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 3.951375857311595e-05, | |
| "loss": 0.0076, | |
| "step": 1181 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 3.949679542890031e-05, | |
| "loss": 0.0051, | |
| "step": 1182 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 3.947982222327886e-05, | |
| "loss": 0.0038, | |
| "step": 1183 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 3.946283896803175e-05, | |
| "loss": 0.0016, | |
| "step": 1184 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 3.944584567494608e-05, | |
| "loss": 0.0086, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 3.942884235581594e-05, | |
| "loss": 0.0022, | |
| "step": 1186 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 3.941182902244238e-05, | |
| "loss": 0.0064, | |
| "step": 1187 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 3.93948056866334e-05, | |
| "loss": 0.0048, | |
| "step": 1188 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 3.937777236020391e-05, | |
| "loss": 0.005, | |
| "step": 1189 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 3.936072905497579e-05, | |
| "loss": 0.0049, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 3.934367578277783e-05, | |
| "loss": 0.0055, | |
| "step": 1191 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 3.932661255544576e-05, | |
| "loss": 0.0057, | |
| "step": 1192 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 3.930953938482218e-05, | |
| "loss": 0.0029, | |
| "step": 1193 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 3.929245628275662e-05, | |
| "loss": 0.0063, | |
| "step": 1194 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 3.9275363261105494e-05, | |
| "loss": 0.0081, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 3.925826033173211e-05, | |
| "loss": 0.0046, | |
| "step": 1196 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 3.924114750650663e-05, | |
| "loss": 0.0108, | |
| "step": 1197 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 3.922402479730611e-05, | |
| "loss": 0.0114, | |
| "step": 1198 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 3.9206892216014446e-05, | |
| "loss": 0.0062, | |
| "step": 1199 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 3.9189749774522396e-05, | |
| "loss": 0.0045, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "eval_loss": 0.02036808617413044, | |
| "eval_runtime": 26.9559, | |
| "eval_samples_per_second": 22.259, | |
| "eval_steps_per_second": 2.782, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 3.917259748472756e-05, | |
| "loss": 0.0016, | |
| "step": 1201 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 3.915543535853437e-05, | |
| "loss": 0.0026, | |
| "step": 1202 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 3.9138263407854085e-05, | |
| "loss": 0.0058, | |
| "step": 1203 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 3.9121081644604794e-05, | |
| "loss": 0.0043, | |
| "step": 1204 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 3.910389008071139e-05, | |
| "loss": 0.0079, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 3.9086688728105544e-05, | |
| "loss": 0.0046, | |
| "step": 1206 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 3.9069477598725753e-05, | |
| "loss": 0.0088, | |
| "step": 1207 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 3.90522567045173e-05, | |
| "loss": 0.0048, | |
| "step": 1208 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 3.903502605743222e-05, | |
| "loss": 0.008, | |
| "step": 1209 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 3.901778566942934e-05, | |
| "loss": 0.0106, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 3.900053555247423e-05, | |
| "loss": 0.0076, | |
| "step": 1211 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 3.8983275718539216e-05, | |
| "loss": 0.0094, | |
| "step": 1212 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 3.896600617960339e-05, | |
| "loss": 0.0078, | |
| "step": 1213 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 3.894872694765255e-05, | |
| "loss": 0.005, | |
| "step": 1214 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 3.893143803467924e-05, | |
| "loss": 0.0057, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 3.8914139452682705e-05, | |
| "loss": 0.0051, | |
| "step": 1216 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 3.889683121366893e-05, | |
| "loss": 0.0029, | |
| "step": 1217 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 3.887951332965056e-05, | |
| "loss": 0.0089, | |
| "step": 1218 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 3.886218581264699e-05, | |
| "loss": 0.0051, | |
| "step": 1219 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 3.884484867468424e-05, | |
| "loss": 0.0077, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 3.8827501927795055e-05, | |
| "loss": 0.005, | |
| "step": 1221 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 3.881014558401883e-05, | |
| "loss": 0.0051, | |
| "step": 1222 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 3.879277965540161e-05, | |
| "loss": 0.0084, | |
| "step": 1223 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 3.877540415399612e-05, | |
| "loss": 0.0108, | |
| "step": 1224 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 3.875801909186171e-05, | |
| "loss": 0.0037, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 3.8740624481064345e-05, | |
| "loss": 0.0083, | |
| "step": 1226 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 3.8723220333676666e-05, | |
| "loss": 0.0099, | |
| "step": 1227 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 3.870580666177791e-05, | |
| "loss": 0.0074, | |
| "step": 1228 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 3.86883834774539e-05, | |
| "loss": 0.0073, | |
| "step": 1229 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 3.86709507927971e-05, | |
| "loss": 0.0118, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 3.8653508619906545e-05, | |
| "loss": 0.0106, | |
| "step": 1231 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 3.863605697088786e-05, | |
| "loss": 0.0092, | |
| "step": 1232 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 3.8618595857853234e-05, | |
| "loss": 0.0071, | |
| "step": 1233 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 3.860112529292147e-05, | |
| "loss": 0.0091, | |
| "step": 1234 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 3.858364528821788e-05, | |
| "loss": 0.0088, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 3.856615585587434e-05, | |
| "loss": 0.0081, | |
| "step": 1236 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 3.8548657008029284e-05, | |
| "loss": 0.0185, | |
| "step": 1237 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 3.8531148756827685e-05, | |
| "loss": 0.0085, | |
| "step": 1238 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 3.851363111442101e-05, | |
| "loss": 0.0054, | |
| "step": 1239 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 3.849610409296727e-05, | |
| "loss": 0.0087, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 3.847856770463097e-05, | |
| "loss": 0.0083, | |
| "step": 1241 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 3.846102196158315e-05, | |
| "loss": 0.0058, | |
| "step": 1242 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 3.84434668760013e-05, | |
| "loss": 0.0083, | |
| "step": 1243 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 3.842590246006942e-05, | |
| "loss": 0.0049, | |
| "step": 1244 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 3.8408328725977966e-05, | |
| "loss": 0.0054, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 3.839074568592388e-05, | |
| "loss": 0.0024, | |
| "step": 1246 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 3.837315335211056e-05, | |
| "loss": 0.0104, | |
| "step": 1247 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 3.835555173674784e-05, | |
| "loss": 0.0031, | |
| "step": 1248 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 3.8337940852052024e-05, | |
| "loss": 0.0059, | |
| "step": 1249 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 3.832032071024581e-05, | |
| "loss": 0.0057, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "eval_loss": 0.02174745313823223, | |
| "eval_runtime": 26.9569, | |
| "eval_samples_per_second": 22.258, | |
| "eval_steps_per_second": 2.782, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 3.830269132355835e-05, | |
| "loss": 0.009, | |
| "step": 1251 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 3.828505270422521e-05, | |
| "loss": 0.006, | |
| "step": 1252 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 3.8267404864488355e-05, | |
| "loss": 0.0043, | |
| "step": 1253 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 3.8249747816596136e-05, | |
| "loss": 0.0034, | |
| "step": 1254 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 3.823208157280334e-05, | |
| "loss": 0.0079, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 3.821440614537108e-05, | |
| "loss": 0.0039, | |
| "step": 1256 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 3.81967215465669e-05, | |
| "loss": 0.005, | |
| "step": 1257 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 3.8179027788664654e-05, | |
| "loss": 0.0116, | |
| "step": 1258 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 3.816132488394459e-05, | |
| "loss": 0.0069, | |
| "step": 1259 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 3.814361284469329e-05, | |
| "loss": 0.004, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 3.8125891683203686e-05, | |
| "loss": 0.0058, | |
| "step": 1261 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 3.810816141177502e-05, | |
| "loss": 0.011, | |
| "step": 1262 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 3.809042204271288e-05, | |
| "loss": 0.0048, | |
| "step": 1263 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 3.807267358832916e-05, | |
| "loss": 0.0045, | |
| "step": 1264 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 3.805491606094206e-05, | |
| "loss": 0.0043, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 3.803714947287607e-05, | |
| "loss": 0.0101, | |
| "step": 1266 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 3.8019373836461966e-05, | |
| "loss": 0.0096, | |
| "step": 1267 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 3.800158916403682e-05, | |
| "loss": 0.0059, | |
| "step": 1268 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 3.7983795467943975e-05, | |
| "loss": 0.0075, | |
| "step": 1269 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 3.796599276053303e-05, | |
| "loss": 0.0056, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 3.7948181054159814e-05, | |
| "loss": 0.0073, | |
| "step": 1271 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 3.793036036118645e-05, | |
| "loss": 0.0128, | |
| "step": 1272 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 3.7912530693981265e-05, | |
| "loss": 0.0101, | |
| "step": 1273 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 3.789469206491882e-05, | |
| "loss": 0.0066, | |
| "step": 1274 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 3.7876844486379895e-05, | |
| "loss": 0.0041, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 3.785898797075149e-05, | |
| "loss": 0.0103, | |
| "step": 1276 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 3.784112253042681e-05, | |
| "loss": 0.0134, | |
| "step": 1277 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 3.782324817780524e-05, | |
| "loss": 0.0045, | |
| "step": 1278 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 3.7805364925292344e-05, | |
| "loss": 0.0052, | |
| "step": 1279 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 3.77874727852999e-05, | |
| "loss": 0.0076, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 3.776957177024582e-05, | |
| "loss": 0.0114, | |
| "step": 1281 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 3.775166189255417e-05, | |
| "loss": 0.0045, | |
| "step": 1282 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 3.7733743164655214e-05, | |
| "loss": 0.0074, | |
| "step": 1283 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 3.77158155989853e-05, | |
| "loss": 0.0053, | |
| "step": 1284 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 3.769787920798696e-05, | |
| "loss": 0.0078, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 3.7679934004108806e-05, | |
| "loss": 0.006, | |
| "step": 1286 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 3.766197999980562e-05, | |
| "loss": 0.0064, | |
| "step": 1287 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 3.7644017207538224e-05, | |
| "loss": 0.0053, | |
| "step": 1288 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 3.76260456397736e-05, | |
| "loss": 0.0089, | |
| "step": 1289 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 3.7608065308984786e-05, | |
| "loss": 0.0053, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 3.7590076227650925e-05, | |
| "loss": 0.008, | |
| "step": 1291 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 3.757207840825721e-05, | |
| "loss": 0.0054, | |
| "step": 1292 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 3.755407186329492e-05, | |
| "loss": 0.0055, | |
| "step": 1293 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 3.753605660526136e-05, | |
| "loss": 0.0042, | |
| "step": 1294 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 3.751803264665992e-05, | |
| "loss": 0.002, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 3.7500000000000003e-05, | |
| "loss": 0.0075, | |
| "step": 1296 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 3.7481958677797055e-05, | |
| "loss": 0.005, | |
| "step": 1297 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 3.7463908692572526e-05, | |
| "loss": 0.009, | |
| "step": 1298 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 3.74458500568539e-05, | |
| "loss": 0.0062, | |
| "step": 1299 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 3.742778278317465e-05, | |
| "loss": 0.0061, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "eval_loss": 0.02023252286016941, | |
| "eval_runtime": 26.9716, | |
| "eval_samples_per_second": 22.246, | |
| "eval_steps_per_second": 2.781, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 3.740970688407425e-05, | |
| "loss": 0.0031, | |
| "step": 1301 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 3.739162237209817e-05, | |
| "loss": 0.0115, | |
| "step": 1302 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 3.737352925979782e-05, | |
| "loss": 0.0072, | |
| "step": 1303 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 3.735542755973064e-05, | |
| "loss": 0.0074, | |
| "step": 1304 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 3.7337317284459975e-05, | |
| "loss": 0.0078, | |
| "step": 1305 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 3.7319198446555134e-05, | |
| "loss": 0.0126, | |
| "step": 1306 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 3.73010710585914e-05, | |
| "loss": 0.0102, | |
| "step": 1307 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 3.728293513314995e-05, | |
| "loss": 0.0035, | |
| "step": 1308 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 3.7264790682817926e-05, | |
| "loss": 0.0044, | |
| "step": 1309 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 3.724663772018834e-05, | |
| "loss": 0.0052, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 3.7228476257860145e-05, | |
| "loss": 0.006, | |
| "step": 1311 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 3.721030630843819e-05, | |
| "loss": 0.012, | |
| "step": 1312 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 3.719212788453321e-05, | |
| "loss": 0.0032, | |
| "step": 1313 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 3.717394099876182e-05, | |
| "loss": 0.0048, | |
| "step": 1314 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 3.71557456637465e-05, | |
| "loss": 0.0051, | |
| "step": 1315 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 3.713754189211561e-05, | |
| "loss": 0.0042, | |
| "step": 1316 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 3.711932969650337e-05, | |
| "loss": 0.0024, | |
| "step": 1317 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 3.7101109089549815e-05, | |
| "loss": 0.0082, | |
| "step": 1318 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 3.708288008390083e-05, | |
| "loss": 0.0049, | |
| "step": 1319 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 3.706464269220817e-05, | |
| "loss": 0.0041, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 3.704639692712936e-05, | |
| "loss": 0.0022, | |
| "step": 1321 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 3.702814280132775e-05, | |
| "loss": 0.0036, | |
| "step": 1322 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 3.700988032747249e-05, | |
| "loss": 0.0096, | |
| "step": 1323 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 3.6991609518238547e-05, | |
| "loss": 0.0061, | |
| "step": 1324 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 3.697333038630665e-05, | |
| "loss": 0.0112, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 3.6955042944363304e-05, | |
| "loss": 0.0027, | |
| "step": 1326 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 3.69367472051008e-05, | |
| "loss": 0.015, | |
| "step": 1327 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 3.691844318121717e-05, | |
| "loss": 0.0055, | |
| "step": 1328 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 3.690013088541619e-05, | |
| "loss": 0.0032, | |
| "step": 1329 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 3.688181033040741e-05, | |
| "loss": 0.0033, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 3.686348152890608e-05, | |
| "loss": 0.0104, | |
| "step": 1331 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 3.684514449363318e-05, | |
| "loss": 0.0032, | |
| "step": 1332 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 3.6826799237315425e-05, | |
| "loss": 0.0085, | |
| "step": 1333 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 3.680844577268521e-05, | |
| "loss": 0.0044, | |
| "step": 1334 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 3.679008411248063e-05, | |
| "loss": 0.0073, | |
| "step": 1335 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 3.677171426944548e-05, | |
| "loss": 0.0053, | |
| "step": 1336 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 3.675333625632923e-05, | |
| "loss": 0.0046, | |
| "step": 1337 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 3.673495008588702e-05, | |
| "loss": 0.0075, | |
| "step": 1338 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 3.6716555770879654e-05, | |
| "loss": 0.0065, | |
| "step": 1339 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 3.669815332407358e-05, | |
| "loss": 0.0089, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 3.66797427582409e-05, | |
| "loss": 0.0107, | |
| "step": 1341 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 3.666132408615935e-05, | |
| "loss": 0.0029, | |
| "step": 1342 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 3.664289732061228e-05, | |
| "loss": 0.0095, | |
| "step": 1343 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 3.662446247438867e-05, | |
| "loss": 0.0029, | |
| "step": 1344 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 3.660601956028309e-05, | |
| "loss": 0.0071, | |
| "step": 1345 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 3.658756859109574e-05, | |
| "loss": 0.0084, | |
| "step": 1346 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 3.656910957963241e-05, | |
| "loss": 0.0038, | |
| "step": 1347 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 3.655064253870441e-05, | |
| "loss": 0.0077, | |
| "step": 1348 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 3.653216748112872e-05, | |
| "loss": 0.0098, | |
| "step": 1349 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 3.65136844197278e-05, | |
| "loss": 0.0066, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "eval_loss": 0.017896093428134918, | |
| "eval_runtime": 26.9269, | |
| "eval_samples_per_second": 22.283, | |
| "eval_steps_per_second": 2.785, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 3.6495193367329705e-05, | |
| "loss": 0.0075, | |
| "step": 1351 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 3.647669433676803e-05, | |
| "loss": 0.0054, | |
| "step": 1352 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 3.64581873408819e-05, | |
| "loss": 0.0053, | |
| "step": 1353 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 3.6439672392515975e-05, | |
| "loss": 0.0087, | |
| "step": 1354 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 3.642114950452046e-05, | |
| "loss": 0.0066, | |
| "step": 1355 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 3.6402618689751e-05, | |
| "loss": 0.0052, | |
| "step": 1356 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 3.6384079961068815e-05, | |
| "loss": 0.0104, | |
| "step": 1357 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 3.636553333134058e-05, | |
| "loss": 0.0023, | |
| "step": 1358 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 3.6346978813438464e-05, | |
| "loss": 0.0065, | |
| "step": 1359 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 3.63284164202401e-05, | |
| "loss": 0.009, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 3.630984616462861e-05, | |
| "loss": 0.0051, | |
| "step": 1361 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 3.629126805949255e-05, | |
| "loss": 0.0096, | |
| "step": 1362 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 3.627268211772592e-05, | |
| "loss": 0.0103, | |
| "step": 1363 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 3.625408835222819e-05, | |
| "loss": 0.0047, | |
| "step": 1364 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 3.623548677590424e-05, | |
| "loss": 0.0057, | |
| "step": 1365 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 3.621687740166436e-05, | |
| "loss": 0.0049, | |
| "step": 1366 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 3.619826024242427e-05, | |
| "loss": 0.0095, | |
| "step": 1367 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 3.6179635311105095e-05, | |
| "loss": 0.003, | |
| "step": 1368 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 3.6161002620633326e-05, | |
| "loss": 0.0056, | |
| "step": 1369 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 3.614236218394088e-05, | |
| "loss": 0.0024, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 3.612371401396502e-05, | |
| "loss": 0.0073, | |
| "step": 1371 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 3.61050581236484e-05, | |
| "loss": 0.0057, | |
| "step": 1372 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 3.608639452593899e-05, | |
| "loss": 0.0027, | |
| "step": 1373 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 3.606772323379017e-05, | |
| "loss": 0.0098, | |
| "step": 1374 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 3.60490442601606e-05, | |
| "loss": 0.0062, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 3.6030357618014316e-05, | |
| "loss": 0.0064, | |
| "step": 1376 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 3.601166332032066e-05, | |
| "loss": 0.0078, | |
| "step": 1377 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 3.599296138005428e-05, | |
| "loss": 0.0041, | |
| "step": 1378 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 3.597425181019513e-05, | |
| "loss": 0.02, | |
| "step": 1379 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 3.595553462372848e-05, | |
| "loss": 0.0048, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 3.5936809833644855e-05, | |
| "loss": 0.0078, | |
| "step": 1381 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 3.591807745294008e-05, | |
| "loss": 0.007, | |
| "step": 1382 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 3.5899337494615245e-05, | |
| "loss": 0.0068, | |
| "step": 1383 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 3.5880589971676685e-05, | |
| "loss": 0.0056, | |
| "step": 1384 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 3.5861834897136006e-05, | |
| "loss": 0.0039, | |
| "step": 1385 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 3.584307228401004e-05, | |
| "loss": 0.0066, | |
| "step": 1386 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 3.582430214532085e-05, | |
| "loss": 0.0091, | |
| "step": 1387 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 3.580552449409573e-05, | |
| "loss": 0.0056, | |
| "step": 1388 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 3.57867393433672e-05, | |
| "loss": 0.0029, | |
| "step": 1389 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 3.576794670617295e-05, | |
| "loss": 0.0132, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 3.57491465955559e-05, | |
| "loss": 0.003, | |
| "step": 1391 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 3.573033902456414e-05, | |
| "loss": 0.0065, | |
| "step": 1392 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 3.571152400625094e-05, | |
| "loss": 0.0053, | |
| "step": 1393 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 3.5692701553674735e-05, | |
| "loss": 0.0117, | |
| "step": 1394 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 3.567387167989913e-05, | |
| "loss": 0.0065, | |
| "step": 1395 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 3.565503439799289e-05, | |
| "loss": 0.0032, | |
| "step": 1396 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 3.563618972102988e-05, | |
| "loss": 0.0055, | |
| "step": 1397 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 3.561733766208915e-05, | |
| "loss": 0.0058, | |
| "step": 1398 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 3.5598478234254817e-05, | |
| "loss": 0.0046, | |
| "step": 1399 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 3.5579611450616165e-05, | |
| "loss": 0.0087, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "eval_loss": 0.018370576202869415, | |
| "eval_runtime": 27.0067, | |
| "eval_samples_per_second": 22.217, | |
| "eval_steps_per_second": 2.777, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 3.556073732426757e-05, | |
| "loss": 0.0067, | |
| "step": 1401 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "learning_rate": 3.554185586830848e-05, | |
| "loss": 0.0091, | |
| "step": 1402 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "learning_rate": 3.5522967095843447e-05, | |
| "loss": 0.0023, | |
| "step": 1403 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "learning_rate": 3.55040710199821e-05, | |
| "loss": 0.0098, | |
| "step": 1404 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 3.548516765383914e-05, | |
| "loss": 0.0046, | |
| "step": 1405 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 3.5466257010534324e-05, | |
| "loss": 0.0076, | |
| "step": 1406 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 3.5447339103192454e-05, | |
| "loss": 0.0069, | |
| "step": 1407 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 3.5428413944943386e-05, | |
| "loss": 0.0071, | |
| "step": 1408 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 3.540948154892201e-05, | |
| "loss": 0.0057, | |
| "step": 1409 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 3.53905419282682e-05, | |
| "loss": 0.0069, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 3.537159509612693e-05, | |
| "loss": 0.0087, | |
| "step": 1411 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 3.5352641065648065e-05, | |
| "loss": 0.0089, | |
| "step": 1412 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 3.5333679849986564e-05, | |
| "loss": 0.0055, | |
| "step": 1413 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 3.5314711462302315e-05, | |
| "loss": 0.0041, | |
| "step": 1414 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 3.529573591576022e-05, | |
| "loss": 0.0047, | |
| "step": 1415 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 3.527675322353011e-05, | |
| "loss": 0.0047, | |
| "step": 1416 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 3.5257763398786824e-05, | |
| "loss": 0.0136, | |
| "step": 1417 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 3.523876645471011e-05, | |
| "loss": 0.0034, | |
| "step": 1418 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 3.521976240448468e-05, | |
| "loss": 0.0048, | |
| "step": 1419 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 3.520075126130016e-05, | |
| "loss": 0.0043, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 3.518173303835111e-05, | |
| "loss": 0.0119, | |
| "step": 1421 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 3.5162707748837014e-05, | |
| "loss": 0.0059, | |
| "step": 1422 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 3.514367540596224e-05, | |
| "loss": 0.0089, | |
| "step": 1423 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 3.512463602293608e-05, | |
| "loss": 0.0123, | |
| "step": 1424 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 3.510558961297265e-05, | |
| "loss": 0.0036, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 3.508653618929103e-05, | |
| "loss": 0.0084, | |
| "step": 1426 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 3.5067475765115104e-05, | |
| "loss": 0.0088, | |
| "step": 1427 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 3.5048408353673643e-05, | |
| "loss": 0.0056, | |
| "step": 1428 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 3.502933396820025e-05, | |
| "loss": 0.0053, | |
| "step": 1429 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 3.5010252621933374e-05, | |
| "loss": 0.0022, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 3.499116432811631e-05, | |
| "loss": 0.0057, | |
| "step": 1431 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "learning_rate": 3.4972069099997164e-05, | |
| "loss": 0.0051, | |
| "step": 1432 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "learning_rate": 3.4952966950828835e-05, | |
| "loss": 0.0185, | |
| "step": 1433 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "learning_rate": 3.493385789386906e-05, | |
| "loss": 0.004, | |
| "step": 1434 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "learning_rate": 3.4914741942380355e-05, | |
| "loss": 0.0042, | |
| "step": 1435 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 3.4895619109630014e-05, | |
| "loss": 0.0039, | |
| "step": 1436 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 3.4876489408890126e-05, | |
| "loss": 0.0089, | |
| "step": 1437 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 3.485735285343752e-05, | |
| "loss": 0.0048, | |
| "step": 1438 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 3.48382094565538e-05, | |
| "loss": 0.0121, | |
| "step": 1439 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 3.481905923152532e-05, | |
| "loss": 0.0073, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 3.479990219164316e-05, | |
| "loss": 0.0054, | |
| "step": 1441 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 3.478073835020315e-05, | |
| "loss": 0.007, | |
| "step": 1442 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 3.476156772050582e-05, | |
| "loss": 0.0022, | |
| "step": 1443 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 3.474239031585642e-05, | |
| "loss": 0.0069, | |
| "step": 1444 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 3.4723206149564913e-05, | |
| "loss": 0.0012, | |
| "step": 1445 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 3.4704015234945926e-05, | |
| "loss": 0.004, | |
| "step": 1446 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 3.46848175853188e-05, | |
| "loss": 0.0152, | |
| "step": 1447 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 3.466561321400753e-05, | |
| "loss": 0.0083, | |
| "step": 1448 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 3.464640213434079e-05, | |
| "loss": 0.0082, | |
| "step": 1449 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 3.46271843596519e-05, | |
| "loss": 0.0074, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "eval_loss": 0.018056336790323257, | |
| "eval_runtime": 26.9975, | |
| "eval_samples_per_second": 22.224, | |
| "eval_steps_per_second": 2.778, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 3.460795990327883e-05, | |
| "loss": 0.0028, | |
| "step": 1451 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 3.45887287785642e-05, | |
| "loss": 0.0061, | |
| "step": 1452 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 3.456949099885522e-05, | |
| "loss": 0.0045, | |
| "step": 1453 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 3.4550246577503776e-05, | |
| "loss": 0.0057, | |
| "step": 1454 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "learning_rate": 3.453099552786631e-05, | |
| "loss": 0.0026, | |
| "step": 1455 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "learning_rate": 3.451173786330389e-05, | |
| "loss": 0.0109, | |
| "step": 1456 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "learning_rate": 3.4492473597182184e-05, | |
| "loss": 0.0085, | |
| "step": 1457 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "learning_rate": 3.447320274287142e-05, | |
| "loss": 0.0028, | |
| "step": 1458 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 3.4453925313746405e-05, | |
| "loss": 0.0106, | |
| "step": 1459 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 3.443464132318653e-05, | |
| "loss": 0.0076, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 3.44153507845757e-05, | |
| "loss": 0.0041, | |
| "step": 1461 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 3.4396053711302395e-05, | |
| "loss": 0.0089, | |
| "step": 1462 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 3.4376750116759625e-05, | |
| "loss": 0.0057, | |
| "step": 1463 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 3.435744001434492e-05, | |
| "loss": 0.0045, | |
| "step": 1464 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 3.4338123417460336e-05, | |
| "loss": 0.0054, | |
| "step": 1465 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 3.431880033951242e-05, | |
| "loss": 0.0089, | |
| "step": 1466 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 3.4299470793912235e-05, | |
| "loss": 0.0024, | |
| "step": 1467 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 3.4280134794075326e-05, | |
| "loss": 0.0133, | |
| "step": 1468 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 3.4260792353421706e-05, | |
| "loss": 0.0093, | |
| "step": 1469 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 3.424144348537589e-05, | |
| "loss": 0.0157, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 3.4222088203366816e-05, | |
| "loss": 0.0074, | |
| "step": 1471 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 3.420272652082789e-05, | |
| "loss": 0.0065, | |
| "step": 1472 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 3.418335845119697e-05, | |
| "loss": 0.0052, | |
| "step": 1473 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 3.416398400791633e-05, | |
| "loss": 0.0058, | |
| "step": 1474 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 3.414460320443269e-05, | |
| "loss": 0.0061, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 3.412521605419715e-05, | |
| "loss": 0.0065, | |
| "step": 1476 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 3.410582257066524e-05, | |
| "loss": 0.0041, | |
| "step": 1477 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 3.408642276729689e-05, | |
| "loss": 0.0029, | |
| "step": 1478 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 3.40670166575564e-05, | |
| "loss": 0.0077, | |
| "step": 1479 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 3.4047604254912445e-05, | |
| "loss": 0.0082, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 3.402818557283809e-05, | |
| "loss": 0.0059, | |
| "step": 1481 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 3.400876062481074e-05, | |
| "loss": 0.0035, | |
| "step": 1482 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 3.3989329424312145e-05, | |
| "loss": 0.0081, | |
| "step": 1483 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 3.3969891984828406e-05, | |
| "loss": 0.0043, | |
| "step": 1484 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 3.395044831984996e-05, | |
| "loss": 0.0128, | |
| "step": 1485 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 3.3930998442871534e-05, | |
| "loss": 0.0035, | |
| "step": 1486 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 3.391154236739221e-05, | |
| "loss": 0.0044, | |
| "step": 1487 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 3.389208010691535e-05, | |
| "loss": 0.0034, | |
| "step": 1488 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 3.387261167494858e-05, | |
| "loss": 0.0149, | |
| "step": 1489 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 3.385313708500386e-05, | |
| "loss": 0.0063, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 3.383365635059741e-05, | |
| "loss": 0.0048, | |
| "step": 1491 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 3.381416948524968e-05, | |
| "loss": 0.0049, | |
| "step": 1492 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 3.379467650248542e-05, | |
| "loss": 0.0093, | |
| "step": 1493 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 3.3775177415833605e-05, | |
| "loss": 0.0066, | |
| "step": 1494 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 3.375567223882744e-05, | |
| "loss": 0.0048, | |
| "step": 1495 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 3.373616098500437e-05, | |
| "loss": 0.0063, | |
| "step": 1496 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 3.371664366790606e-05, | |
| "loss": 0.0061, | |
| "step": 1497 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 3.369712030107836e-05, | |
| "loss": 0.0131, | |
| "step": 1498 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 3.367759089807134e-05, | |
| "loss": 0.0087, | |
| "step": 1499 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 3.365805547243927e-05, | |
| "loss": 0.0057, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "eval_loss": 0.016999540850520134, | |
| "eval_runtime": 27.3863, | |
| "eval_samples_per_second": 21.909, | |
| "eval_steps_per_second": 2.739, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 3.363851403774057e-05, | |
| "loss": 0.0049, | |
| "step": 1501 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 3.3618966607537835e-05, | |
| "loss": 0.0053, | |
| "step": 1502 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 3.359941319539785e-05, | |
| "loss": 0.0099, | |
| "step": 1503 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 3.357985381489153e-05, | |
| "loss": 0.0086, | |
| "step": 1504 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 3.356028847959392e-05, | |
| "loss": 0.0027, | |
| "step": 1505 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 3.354071720308423e-05, | |
| "loss": 0.019, | |
| "step": 1506 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 3.352113999894576e-05, | |
| "loss": 0.0058, | |
| "step": 1507 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 3.3501556880765944e-05, | |
| "loss": 0.0087, | |
| "step": 1508 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 3.348196786213633e-05, | |
| "loss": 0.0071, | |
| "step": 1509 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 3.3462372956652546e-05, | |
| "loss": 0.0052, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 3.3442772177914284e-05, | |
| "loss": 0.0075, | |
| "step": 1511 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 3.342316553952536e-05, | |
| "loss": 0.0046, | |
| "step": 1512 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 3.340355305509363e-05, | |
| "loss": 0.0041, | |
| "step": 1513 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 3.3383934738231e-05, | |
| "loss": 0.004, | |
| "step": 1514 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 3.336431060255344e-05, | |
| "loss": 0.0061, | |
| "step": 1515 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 3.334468066168095e-05, | |
| "loss": 0.0115, | |
| "step": 1516 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 3.332504492923757e-05, | |
| "loss": 0.0092, | |
| "step": 1517 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 3.3305403418851344e-05, | |
| "loss": 0.0095, | |
| "step": 1518 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 3.3285756144154334e-05, | |
| "loss": 0.0035, | |
| "step": 1519 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "learning_rate": 3.3266103118782606e-05, | |
| "loss": 0.0118, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "learning_rate": 3.3246444356376205e-05, | |
| "loss": 0.0047, | |
| "step": 1521 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "learning_rate": 3.3226779870579183e-05, | |
| "loss": 0.0047, | |
| "step": 1522 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "learning_rate": 3.3207109675039527e-05, | |
| "loss": 0.006, | |
| "step": 1523 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 3.3187433783409216e-05, | |
| "loss": 0.01, | |
| "step": 1524 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 3.316775220934417e-05, | |
| "loss": 0.0064, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 3.314806496650427e-05, | |
| "loss": 0.0043, | |
| "step": 1526 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 3.312837206855331e-05, | |
| "loss": 0.0027, | |
| "step": 1527 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 3.310867352915901e-05, | |
| "loss": 0.0024, | |
| "step": 1528 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 3.3088969361993004e-05, | |
| "loss": 0.0027, | |
| "step": 1529 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 3.306925958073087e-05, | |
| "loss": 0.0025, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 3.3049544199052034e-05, | |
| "loss": 0.0056, | |
| "step": 1531 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 3.3029823230639824e-05, | |
| "loss": 0.0048, | |
| "step": 1532 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 3.301009668918145e-05, | |
| "loss": 0.0033, | |
| "step": 1533 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 3.299036458836801e-05, | |
| "loss": 0.0052, | |
| "step": 1534 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 3.29706269418944e-05, | |
| "loss": 0.0073, | |
| "step": 1535 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 3.295088376345944e-05, | |
| "loss": 0.0031, | |
| "step": 1536 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 3.293113506676573e-05, | |
| "loss": 0.0046, | |
| "step": 1537 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 3.2911380865519736e-05, | |
| "loss": 0.0071, | |
| "step": 1538 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 3.289162117343173e-05, | |
| "loss": 0.0039, | |
| "step": 1539 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 3.28718560042158e-05, | |
| "loss": 0.0078, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 3.285208537158982e-05, | |
| "loss": 0.0034, | |
| "step": 1541 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 3.283230928927548e-05, | |
| "loss": 0.0018, | |
| "step": 1542 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 3.2812527770998234e-05, | |
| "loss": 0.006, | |
| "step": 1543 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 3.279274083048731e-05, | |
| "loss": 0.0045, | |
| "step": 1544 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 3.277294848147572e-05, | |
| "loss": 0.006, | |
| "step": 1545 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 3.27531507377002e-05, | |
| "loss": 0.0074, | |
| "step": 1546 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 3.273334761290125e-05, | |
| "loss": 0.005, | |
| "step": 1547 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 3.2713539120823095e-05, | |
| "loss": 0.0047, | |
| "step": 1548 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 3.269372527521369e-05, | |
| "loss": 0.0049, | |
| "step": 1549 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 3.2673906089824716e-05, | |
| "loss": 0.0053, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "eval_loss": 0.019064469262957573, | |
| "eval_runtime": 27.2336, | |
| "eval_samples_per_second": 22.032, | |
| "eval_steps_per_second": 2.754, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 3.265408157841153e-05, | |
| "loss": 0.0027, | |
| "step": 1551 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 3.263425175473323e-05, | |
| "loss": 0.0076, | |
| "step": 1552 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 3.2614416632552544e-05, | |
| "loss": 0.005, | |
| "step": 1553 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 3.259457622563593e-05, | |
| "loss": 0.0031, | |
| "step": 1554 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 3.257473054775348e-05, | |
| "loss": 0.0021, | |
| "step": 1555 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 3.2554879612678966e-05, | |
| "loss": 0.0081, | |
| "step": 1556 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 3.2535023434189796e-05, | |
| "loss": 0.0099, | |
| "step": 1557 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 3.251516202606701e-05, | |
| "loss": 0.0057, | |
| "step": 1558 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 3.249529540209531e-05, | |
| "loss": 0.0017, | |
| "step": 1559 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 3.247542357606299e-05, | |
| "loss": 0.0034, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 3.245554656176194e-05, | |
| "loss": 0.0028, | |
| "step": 1561 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 3.243566437298769e-05, | |
| "loss": 0.0049, | |
| "step": 1562 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 3.241577702353934e-05, | |
| "loss": 0.0081, | |
| "step": 1563 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 3.239588452721958e-05, | |
| "loss": 0.0017, | |
| "step": 1564 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 3.237598689783465e-05, | |
| "loss": 0.0027, | |
| "step": 1565 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 3.2356084149194374e-05, | |
| "loss": 0.0008, | |
| "step": 1566 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 3.233617629511214e-05, | |
| "loss": 0.0041, | |
| "step": 1567 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 3.2316263349404844e-05, | |
| "loss": 0.0007, | |
| "step": 1568 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 3.229634532589296e-05, | |
| "loss": 0.0066, | |
| "step": 1569 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 3.227642223840043e-05, | |
| "loss": 0.0006, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 3.225649410075476e-05, | |
| "loss": 0.0075, | |
| "step": 1571 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 3.2236560926786953e-05, | |
| "loss": 0.0025, | |
| "step": 1572 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 3.221662273033148e-05, | |
| "loss": 0.0067, | |
| "step": 1573 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 3.219667952522632e-05, | |
| "loss": 0.0006, | |
| "step": 1574 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 3.217673132531293e-05, | |
| "loss": 0.0045, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 3.215677814443622e-05, | |
| "loss": 0.0013, | |
| "step": 1576 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 3.2136819996444564e-05, | |
| "loss": 0.0011, | |
| "step": 1577 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 3.211685689518978e-05, | |
| "loss": 0.0042, | |
| "step": 1578 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 3.209688885452715e-05, | |
| "loss": 0.0027, | |
| "step": 1579 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 3.207691588831534e-05, | |
| "loss": 0.0051, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 3.205693801041646e-05, | |
| "loss": 0.0045, | |
| "step": 1581 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 3.203695523469603e-05, | |
| "loss": 0.0042, | |
| "step": 1582 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 3.201696757502296e-05, | |
| "loss": 0.0041, | |
| "step": 1583 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 3.199697504526955e-05, | |
| "loss": 0.0016, | |
| "step": 1584 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "learning_rate": 3.19769776593115e-05, | |
| "loss": 0.0034, | |
| "step": 1585 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "learning_rate": 3.195697543102785e-05, | |
| "loss": 0.0028, | |
| "step": 1586 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "learning_rate": 3.193696837430102e-05, | |
| "loss": 0.0078, | |
| "step": 1587 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 3.191695650301678e-05, | |
| "loss": 0.0091, | |
| "step": 1588 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 3.189693983106423e-05, | |
| "loss": 0.0057, | |
| "step": 1589 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 3.1876918372335825e-05, | |
| "loss": 0.0047, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 3.185689214072731e-05, | |
| "loss": 0.0057, | |
| "step": 1591 | |
| }, | |
| { | |
| "epoch": 4.18, | |
| "learning_rate": 3.183686115013776e-05, | |
| "loss": 0.0083, | |
| "step": 1592 | |
| }, | |
| { | |
| "epoch": 4.18, | |
| "learning_rate": 3.1816825414469564e-05, | |
| "loss": 0.0049, | |
| "step": 1593 | |
| }, | |
| { | |
| "epoch": 4.18, | |
| "learning_rate": 3.179678494762839e-05, | |
| "loss": 0.0094, | |
| "step": 1594 | |
| }, | |
| { | |
| "epoch": 4.18, | |
| "learning_rate": 3.1776739763523187e-05, | |
| "loss": 0.0057, | |
| "step": 1595 | |
| }, | |
| { | |
| "epoch": 4.19, | |
| "learning_rate": 3.175668987606619e-05, | |
| "loss": 0.0031, | |
| "step": 1596 | |
| }, | |
| { | |
| "epoch": 4.19, | |
| "learning_rate": 3.173663529917289e-05, | |
| "loss": 0.0039, | |
| "step": 1597 | |
| }, | |
| { | |
| "epoch": 4.19, | |
| "learning_rate": 3.1716576046762034e-05, | |
| "loss": 0.0025, | |
| "step": 1598 | |
| }, | |
| { | |
| "epoch": 4.19, | |
| "learning_rate": 3.169651213275562e-05, | |
| "loss": 0.0014, | |
| "step": 1599 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 3.167644357107886e-05, | |
| "loss": 0.0051, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "eval_loss": 0.017935335636138916, | |
| "eval_runtime": 27.0809, | |
| "eval_samples_per_second": 22.156, | |
| "eval_steps_per_second": 2.769, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 3.165637037566022e-05, | |
| "loss": 0.0028, | |
| "step": 1601 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 3.1636292560431366e-05, | |
| "loss": 0.0024, | |
| "step": 1602 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 3.161621013932719e-05, | |
| "loss": 0.0034, | |
| "step": 1603 | |
| }, | |
| { | |
| "epoch": 4.21, | |
| "learning_rate": 3.159612312628574e-05, | |
| "loss": 0.0092, | |
| "step": 1604 | |
| }, | |
| { | |
| "epoch": 4.21, | |
| "learning_rate": 3.157603153524828e-05, | |
| "loss": 0.0101, | |
| "step": 1605 | |
| }, | |
| { | |
| "epoch": 4.21, | |
| "learning_rate": 3.155593538015926e-05, | |
| "loss": 0.0052, | |
| "step": 1606 | |
| }, | |
| { | |
| "epoch": 4.22, | |
| "learning_rate": 3.153583467496628e-05, | |
| "loss": 0.0022, | |
| "step": 1607 | |
| }, | |
| { | |
| "epoch": 4.22, | |
| "learning_rate": 3.151572943362009e-05, | |
| "loss": 0.0018, | |
| "step": 1608 | |
| }, | |
| { | |
| "epoch": 4.22, | |
| "learning_rate": 3.149561967007462e-05, | |
| "loss": 0.0024, | |
| "step": 1609 | |
| }, | |
| { | |
| "epoch": 4.22, | |
| "learning_rate": 3.147550539828691e-05, | |
| "loss": 0.0042, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "learning_rate": 3.1455386632217144e-05, | |
| "loss": 0.0039, | |
| "step": 1611 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "learning_rate": 3.143526338582861e-05, | |
| "loss": 0.0036, | |
| "step": 1612 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "learning_rate": 3.141513567308772e-05, | |
| "loss": 0.002, | |
| "step": 1613 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "learning_rate": 3.139500350796397e-05, | |
| "loss": 0.0035, | |
| "step": 1614 | |
| }, | |
| { | |
| "epoch": 4.24, | |
| "learning_rate": 3.137486690442998e-05, | |
| "loss": 0.0019, | |
| "step": 1615 | |
| }, | |
| { | |
| "epoch": 4.24, | |
| "learning_rate": 3.135472587646142e-05, | |
| "loss": 0.0027, | |
| "step": 1616 | |
| }, | |
| { | |
| "epoch": 4.24, | |
| "learning_rate": 3.133458043803703e-05, | |
| "loss": 0.0041, | |
| "step": 1617 | |
| }, | |
| { | |
| "epoch": 4.24, | |
| "learning_rate": 3.131443060313862e-05, | |
| "loss": 0.0066, | |
| "step": 1618 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "learning_rate": 3.129427638575106e-05, | |
| "loss": 0.0076, | |
| "step": 1619 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "learning_rate": 3.1274117799862244e-05, | |
| "loss": 0.0048, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "learning_rate": 3.125395485946311e-05, | |
| "loss": 0.0074, | |
| "step": 1621 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "learning_rate": 3.1233787578547616e-05, | |
| "loss": 0.0043, | |
| "step": 1622 | |
| }, | |
| { | |
| "epoch": 4.26, | |
| "learning_rate": 3.1213615971112727e-05, | |
| "loss": 0.0046, | |
| "step": 1623 | |
| }, | |
| { | |
| "epoch": 4.26, | |
| "learning_rate": 3.1193440051158426e-05, | |
| "loss": 0.0027, | |
| "step": 1624 | |
| }, | |
| { | |
| "epoch": 4.26, | |
| "learning_rate": 3.1173259832687655e-05, | |
| "loss": 0.0045, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 4.26, | |
| "learning_rate": 3.1153075329706385e-05, | |
| "loss": 0.0084, | |
| "step": 1626 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "learning_rate": 3.113288655622352e-05, | |
| "loss": 0.0009, | |
| "step": 1627 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "learning_rate": 3.111269352625097e-05, | |
| "loss": 0.0029, | |
| "step": 1628 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "learning_rate": 3.1092496253803546e-05, | |
| "loss": 0.0014, | |
| "step": 1629 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 3.107229475289905e-05, | |
| "loss": 0.0058, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 3.10520890375582e-05, | |
| "loss": 0.0025, | |
| "step": 1631 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 3.103187912180464e-05, | |
| "loss": 0.0031, | |
| "step": 1632 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 3.1011665019664924e-05, | |
| "loss": 0.0067, | |
| "step": 1633 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "learning_rate": 3.0991446745168516e-05, | |
| "loss": 0.0007, | |
| "step": 1634 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "learning_rate": 3.097122431234779e-05, | |
| "loss": 0.0121, | |
| "step": 1635 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "learning_rate": 3.095099773523798e-05, | |
| "loss": 0.0041, | |
| "step": 1636 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "learning_rate": 3.093076702787722e-05, | |
| "loss": 0.0051, | |
| "step": 1637 | |
| }, | |
| { | |
| "epoch": 4.3, | |
| "learning_rate": 3.0910532204306484e-05, | |
| "loss": 0.0077, | |
| "step": 1638 | |
| }, | |
| { | |
| "epoch": 4.3, | |
| "learning_rate": 3.089029327856963e-05, | |
| "loss": 0.0017, | |
| "step": 1639 | |
| }, | |
| { | |
| "epoch": 4.3, | |
| "learning_rate": 3.087005026471335e-05, | |
| "loss": 0.007, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 4.3, | |
| "learning_rate": 3.0849803176787186e-05, | |
| "loss": 0.0007, | |
| "step": 1641 | |
| }, | |
| { | |
| "epoch": 4.31, | |
| "learning_rate": 3.082955202884347e-05, | |
| "loss": 0.0042, | |
| "step": 1642 | |
| }, | |
| { | |
| "epoch": 4.31, | |
| "learning_rate": 3.0809296834937404e-05, | |
| "loss": 0.0086, | |
| "step": 1643 | |
| }, | |
| { | |
| "epoch": 4.31, | |
| "learning_rate": 3.078903760912695e-05, | |
| "loss": 0.0098, | |
| "step": 1644 | |
| }, | |
| { | |
| "epoch": 4.31, | |
| "learning_rate": 3.0768774365472916e-05, | |
| "loss": 0.0047, | |
| "step": 1645 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "learning_rate": 3.074850711803886e-05, | |
| "loss": 0.0065, | |
| "step": 1646 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "learning_rate": 3.072823588089112e-05, | |
| "loss": 0.0065, | |
| "step": 1647 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "learning_rate": 3.070796066809883e-05, | |
| "loss": 0.0054, | |
| "step": 1648 | |
| }, | |
| { | |
| "epoch": 4.33, | |
| "learning_rate": 3.0687681493733874e-05, | |
| "loss": 0.0067, | |
| "step": 1649 | |
| }, | |
| { | |
| "epoch": 4.33, | |
| "learning_rate": 3.066739837187088e-05, | |
| "loss": 0.0051, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 4.33, | |
| "eval_loss": 0.018013320863246918, | |
| "eval_runtime": 26.893, | |
| "eval_samples_per_second": 22.311, | |
| "eval_steps_per_second": 2.789, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 4.33, | |
| "learning_rate": 3.0647111316587194e-05, | |
| "loss": 0.0043, | |
| "step": 1651 | |
| }, | |
| { | |
| "epoch": 4.33, | |
| "learning_rate": 3.062682034196293e-05, | |
| "loss": 0.0057, | |
| "step": 1652 | |
| }, | |
| { | |
| "epoch": 4.34, | |
| "learning_rate": 3.060652546208091e-05, | |
| "loss": 0.004, | |
| "step": 1653 | |
| }, | |
| { | |
| "epoch": 4.34, | |
| "learning_rate": 3.0586226691026656e-05, | |
| "loss": 0.001, | |
| "step": 1654 | |
| }, | |
| { | |
| "epoch": 4.34, | |
| "learning_rate": 3.0565924042888386e-05, | |
| "loss": 0.0034, | |
| "step": 1655 | |
| }, | |
| { | |
| "epoch": 4.34, | |
| "learning_rate": 3.054561753175704e-05, | |
| "loss": 0.0055, | |
| "step": 1656 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 3.05253071717262e-05, | |
| "loss": 0.0052, | |
| "step": 1657 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 3.0504992976892166e-05, | |
| "loss": 0.0046, | |
| "step": 1658 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 3.048467496135384e-05, | |
| "loss": 0.0019, | |
| "step": 1659 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 3.046435313921282e-05, | |
| "loss": 0.0025, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "learning_rate": 3.0444027524573338e-05, | |
| "loss": 0.0086, | |
| "step": 1661 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "learning_rate": 3.042369813154225e-05, | |
| "loss": 0.0089, | |
| "step": 1662 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "learning_rate": 3.040336497422904e-05, | |
| "loss": 0.0048, | |
| "step": 1663 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "learning_rate": 3.03830280667458e-05, | |
| "loss": 0.0034, | |
| "step": 1664 | |
| }, | |
| { | |
| "epoch": 4.37, | |
| "learning_rate": 3.036268742320722e-05, | |
| "loss": 0.005, | |
| "step": 1665 | |
| }, | |
| { | |
| "epoch": 4.37, | |
| "learning_rate": 3.0342343057730605e-05, | |
| "loss": 0.0029, | |
| "step": 1666 | |
| }, | |
| { | |
| "epoch": 4.37, | |
| "learning_rate": 3.0321994984435826e-05, | |
| "loss": 0.0045, | |
| "step": 1667 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "learning_rate": 3.030164321744532e-05, | |
| "loss": 0.0028, | |
| "step": 1668 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "learning_rate": 3.0281287770884098e-05, | |
| "loss": 0.0031, | |
| "step": 1669 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "learning_rate": 3.0260928658879727e-05, | |
| "loss": 0.0039, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "learning_rate": 3.0240565895562335e-05, | |
| "loss": 0.0027, | |
| "step": 1671 | |
| }, | |
| { | |
| "epoch": 4.39, | |
| "learning_rate": 3.0220199495064526e-05, | |
| "loss": 0.0065, | |
| "step": 1672 | |
| }, | |
| { | |
| "epoch": 4.39, | |
| "learning_rate": 3.0199829471521494e-05, | |
| "loss": 0.0027, | |
| "step": 1673 | |
| }, | |
| { | |
| "epoch": 4.39, | |
| "learning_rate": 3.017945583907092e-05, | |
| "loss": 0.0013, | |
| "step": 1674 | |
| }, | |
| { | |
| "epoch": 4.39, | |
| "learning_rate": 3.0159078611852983e-05, | |
| "loss": 0.0021, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 3.0138697804010357e-05, | |
| "loss": 0.006, | |
| "step": 1676 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 3.0118313429688215e-05, | |
| "loss": 0.0087, | |
| "step": 1677 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 3.0097925503034202e-05, | |
| "loss": 0.0054, | |
| "step": 1678 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 3.007753403819842e-05, | |
| "loss": 0.0041, | |
| "step": 1679 | |
| }, | |
| { | |
| "epoch": 4.41, | |
| "learning_rate": 3.005713904933344e-05, | |
| "loss": 0.0051, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 4.41, | |
| "learning_rate": 3.0036740550594254e-05, | |
| "loss": 0.0072, | |
| "step": 1681 | |
| }, | |
| { | |
| "epoch": 4.41, | |
| "learning_rate": 3.001633855613832e-05, | |
| "loss": 0.0029, | |
| "step": 1682 | |
| }, | |
| { | |
| "epoch": 4.41, | |
| "learning_rate": 2.9995933080125504e-05, | |
| "loss": 0.0052, | |
| "step": 1683 | |
| }, | |
| { | |
| "epoch": 4.42, | |
| "learning_rate": 2.9975524136718097e-05, | |
| "loss": 0.0021, | |
| "step": 1684 | |
| }, | |
| { | |
| "epoch": 4.42, | |
| "learning_rate": 2.995511174008078e-05, | |
| "loss": 0.0037, | |
| "step": 1685 | |
| }, | |
| { | |
| "epoch": 4.42, | |
| "learning_rate": 2.9934695904380655e-05, | |
| "loss": 0.0031, | |
| "step": 1686 | |
| }, | |
| { | |
| "epoch": 4.42, | |
| "learning_rate": 2.9914276643787192e-05, | |
| "loss": 0.0061, | |
| "step": 1687 | |
| }, | |
| { | |
| "epoch": 4.43, | |
| "learning_rate": 2.989385397247226e-05, | |
| "loss": 0.0053, | |
| "step": 1688 | |
| }, | |
| { | |
| "epoch": 4.43, | |
| "learning_rate": 2.9873427904610057e-05, | |
| "loss": 0.0024, | |
| "step": 1689 | |
| }, | |
| { | |
| "epoch": 4.43, | |
| "learning_rate": 2.9852998454377172e-05, | |
| "loss": 0.0032, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "learning_rate": 2.9832565635952537e-05, | |
| "loss": 0.004, | |
| "step": 1691 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "learning_rate": 2.9812129463517413e-05, | |
| "loss": 0.0046, | |
| "step": 1692 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "learning_rate": 2.979168995125538e-05, | |
| "loss": 0.0023, | |
| "step": 1693 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "learning_rate": 2.9771247113352357e-05, | |
| "loss": 0.0051, | |
| "step": 1694 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "learning_rate": 2.9750800963996566e-05, | |
| "loss": 0.0043, | |
| "step": 1695 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "learning_rate": 2.9730351517378514e-05, | |
| "loss": 0.0019, | |
| "step": 1696 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "learning_rate": 2.9709898787691014e-05, | |
| "loss": 0.0029, | |
| "step": 1697 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "learning_rate": 2.968944278912914e-05, | |
| "loss": 0.0052, | |
| "step": 1698 | |
| }, | |
| { | |
| "epoch": 4.46, | |
| "learning_rate": 2.9668983535890248e-05, | |
| "loss": 0.0018, | |
| "step": 1699 | |
| }, | |
| { | |
| "epoch": 4.46, | |
| "learning_rate": 2.964852104217395e-05, | |
| "loss": 0.0008, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 4.46, | |
| "eval_loss": 0.01821320876479149, | |
| "eval_runtime": 27.1196, | |
| "eval_samples_per_second": 22.124, | |
| "eval_steps_per_second": 2.766, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 4.46, | |
| "learning_rate": 2.9628055322182102e-05, | |
| "loss": 0.0023, | |
| "step": 1701 | |
| }, | |
| { | |
| "epoch": 4.46, | |
| "learning_rate": 2.9607586390118807e-05, | |
| "loss": 0.0142, | |
| "step": 1702 | |
| }, | |
| { | |
| "epoch": 4.47, | |
| "learning_rate": 2.9587114260190386e-05, | |
| "loss": 0.0073, | |
| "step": 1703 | |
| }, | |
| { | |
| "epoch": 4.47, | |
| "learning_rate": 2.956663894660539e-05, | |
| "loss": 0.0025, | |
| "step": 1704 | |
| }, | |
| { | |
| "epoch": 4.47, | |
| "learning_rate": 2.9546160463574592e-05, | |
| "loss": 0.002, | |
| "step": 1705 | |
| }, | |
| { | |
| "epoch": 4.47, | |
| "learning_rate": 2.952567882531092e-05, | |
| "loss": 0.0022, | |
| "step": 1706 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "learning_rate": 2.950519404602954e-05, | |
| "loss": 0.002, | |
| "step": 1707 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "learning_rate": 2.9484706139947765e-05, | |
| "loss": 0.0044, | |
| "step": 1708 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "learning_rate": 2.946421512128511e-05, | |
| "loss": 0.0024, | |
| "step": 1709 | |
| }, | |
| { | |
| "epoch": 4.49, | |
| "learning_rate": 2.9443721004263223e-05, | |
| "loss": 0.0009, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 4.49, | |
| "learning_rate": 2.9423223803105903e-05, | |
| "loss": 0.0042, | |
| "step": 1711 | |
| }, | |
| { | |
| "epoch": 4.49, | |
| "learning_rate": 2.9402723532039113e-05, | |
| "loss": 0.0034, | |
| "step": 1712 | |
| }, | |
| { | |
| "epoch": 4.49, | |
| "learning_rate": 2.9382220205290923e-05, | |
| "loss": 0.0105, | |
| "step": 1713 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "learning_rate": 2.936171383709154e-05, | |
| "loss": 0.0011, | |
| "step": 1714 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "learning_rate": 2.9341204441673266e-05, | |
| "loss": 0.0064, | |
| "step": 1715 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "learning_rate": 2.932069203327051e-05, | |
| "loss": 0.0078, | |
| "step": 1716 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "learning_rate": 2.930017662611978e-05, | |
| "loss": 0.007, | |
| "step": 1717 | |
| }, | |
| { | |
| "epoch": 4.51, | |
| "learning_rate": 2.9279658234459667e-05, | |
| "loss": 0.0023, | |
| "step": 1718 | |
| }, | |
| { | |
| "epoch": 4.51, | |
| "learning_rate": 2.9259136872530812e-05, | |
| "loss": 0.0099, | |
| "step": 1719 | |
| }, | |
| { | |
| "epoch": 4.51, | |
| "learning_rate": 2.9238612554575938e-05, | |
| "loss": 0.0072, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 4.51, | |
| "learning_rate": 2.921808529483982e-05, | |
| "loss": 0.0012, | |
| "step": 1721 | |
| }, | |
| { | |
| "epoch": 4.52, | |
| "learning_rate": 2.9197555107569264e-05, | |
| "loss": 0.0046, | |
| "step": 1722 | |
| }, | |
| { | |
| "epoch": 4.52, | |
| "learning_rate": 2.91770220070131e-05, | |
| "loss": 0.0008, | |
| "step": 1723 | |
| }, | |
| { | |
| "epoch": 4.52, | |
| "learning_rate": 2.91564860074222e-05, | |
| "loss": 0.0063, | |
| "step": 1724 | |
| }, | |
| { | |
| "epoch": 4.52, | |
| "learning_rate": 2.913594712304945e-05, | |
| "loss": 0.0055, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 4.53, | |
| "learning_rate": 2.9115405368149717e-05, | |
| "loss": 0.0026, | |
| "step": 1726 | |
| }, | |
| { | |
| "epoch": 4.53, | |
| "learning_rate": 2.9094860756979876e-05, | |
| "loss": 0.0017, | |
| "step": 1727 | |
| }, | |
| { | |
| "epoch": 4.53, | |
| "learning_rate": 2.907431330379877e-05, | |
| "loss": 0.0021, | |
| "step": 1728 | |
| }, | |
| { | |
| "epoch": 4.54, | |
| "learning_rate": 2.9053763022867235e-05, | |
| "loss": 0.0065, | |
| "step": 1729 | |
| }, | |
| { | |
| "epoch": 4.54, | |
| "learning_rate": 2.9033209928448056e-05, | |
| "loss": 0.0078, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 4.54, | |
| "learning_rate": 2.9012654034805975e-05, | |
| "loss": 0.0014, | |
| "step": 1731 | |
| }, | |
| { | |
| "epoch": 4.54, | |
| "learning_rate": 2.899209535620767e-05, | |
| "loss": 0.0056, | |
| "step": 1732 | |
| }, | |
| { | |
| "epoch": 4.55, | |
| "learning_rate": 2.897153390692176e-05, | |
| "loss": 0.002, | |
| "step": 1733 | |
| }, | |
| { | |
| "epoch": 4.55, | |
| "learning_rate": 2.8950969701218783e-05, | |
| "loss": 0.0025, | |
| "step": 1734 | |
| }, | |
| { | |
| "epoch": 4.55, | |
| "learning_rate": 2.8930402753371194e-05, | |
| "loss": 0.0037, | |
| "step": 1735 | |
| }, | |
| { | |
| "epoch": 4.55, | |
| "learning_rate": 2.8909833077653347e-05, | |
| "loss": 0.0109, | |
| "step": 1736 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "learning_rate": 2.8889260688341486e-05, | |
| "loss": 0.0015, | |
| "step": 1737 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "learning_rate": 2.8868685599713746e-05, | |
| "loss": 0.0037, | |
| "step": 1738 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "learning_rate": 2.884810782605014e-05, | |
| "loss": 0.0038, | |
| "step": 1739 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "learning_rate": 2.8827527381632523e-05, | |
| "loss": 0.0096, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "learning_rate": 2.880694428074463e-05, | |
| "loss": 0.0049, | |
| "step": 1741 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "learning_rate": 2.878635853767202e-05, | |
| "loss": 0.0011, | |
| "step": 1742 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "learning_rate": 2.8765770166702087e-05, | |
| "loss": 0.0049, | |
| "step": 1743 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "learning_rate": 2.8745179182124077e-05, | |
| "loss": 0.0062, | |
| "step": 1744 | |
| }, | |
| { | |
| "epoch": 4.58, | |
| "learning_rate": 2.8724585598229005e-05, | |
| "loss": 0.0057, | |
| "step": 1745 | |
| }, | |
| { | |
| "epoch": 4.58, | |
| "learning_rate": 2.8703989429309723e-05, | |
| "loss": 0.0076, | |
| "step": 1746 | |
| }, | |
| { | |
| "epoch": 4.58, | |
| "learning_rate": 2.868339068966086e-05, | |
| "loss": 0.0008, | |
| "step": 1747 | |
| }, | |
| { | |
| "epoch": 4.58, | |
| "learning_rate": 2.8662789393578852e-05, | |
| "loss": 0.0009, | |
| "step": 1748 | |
| }, | |
| { | |
| "epoch": 4.59, | |
| "learning_rate": 2.864218555536188e-05, | |
| "loss": 0.003, | |
| "step": 1749 | |
| }, | |
| { | |
| "epoch": 4.59, | |
| "learning_rate": 2.86215791893099e-05, | |
| "loss": 0.0052, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 4.59, | |
| "eval_loss": 0.01769648864865303, | |
| "eval_runtime": 27.02, | |
| "eval_samples_per_second": 22.206, | |
| "eval_steps_per_second": 2.776, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 4.59, | |
| "learning_rate": 2.8600970309724628e-05, | |
| "loss": 0.0095, | |
| "step": 1751 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "learning_rate": 2.8580358930909535e-05, | |
| "loss": 0.0029, | |
| "step": 1752 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "learning_rate": 2.8559745067169798e-05, | |
| "loss": 0.0071, | |
| "step": 1753 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "learning_rate": 2.8539128732812336e-05, | |
| "loss": 0.0044, | |
| "step": 1754 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "learning_rate": 2.851850994214579e-05, | |
| "loss": 0.0079, | |
| "step": 1755 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "learning_rate": 2.8497888709480493e-05, | |
| "loss": 0.001, | |
| "step": 1756 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "learning_rate": 2.8477265049128467e-05, | |
| "loss": 0.0043, | |
| "step": 1757 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "learning_rate": 2.8456638975403428e-05, | |
| "loss": 0.0037, | |
| "step": 1758 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "learning_rate": 2.843601050262078e-05, | |
| "loss": 0.0063, | |
| "step": 1759 | |
| }, | |
| { | |
| "epoch": 4.62, | |
| "learning_rate": 2.8415379645097574e-05, | |
| "loss": 0.0054, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 4.62, | |
| "learning_rate": 2.8394746417152522e-05, | |
| "loss": 0.0037, | |
| "step": 1761 | |
| }, | |
| { | |
| "epoch": 4.62, | |
| "learning_rate": 2.837411083310598e-05, | |
| "loss": 0.0023, | |
| "step": 1762 | |
| }, | |
| { | |
| "epoch": 4.62, | |
| "learning_rate": 2.8353472907279933e-05, | |
| "loss": 0.0046, | |
| "step": 1763 | |
| }, | |
| { | |
| "epoch": 4.63, | |
| "learning_rate": 2.833283265399801e-05, | |
| "loss": 0.0054, | |
| "step": 1764 | |
| }, | |
| { | |
| "epoch": 4.63, | |
| "learning_rate": 2.831219008758544e-05, | |
| "loss": 0.0081, | |
| "step": 1765 | |
| }, | |
| { | |
| "epoch": 4.63, | |
| "learning_rate": 2.8291545222369047e-05, | |
| "loss": 0.0061, | |
| "step": 1766 | |
| }, | |
| { | |
| "epoch": 4.63, | |
| "learning_rate": 2.8270898072677282e-05, | |
| "loss": 0.003, | |
| "step": 1767 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "learning_rate": 2.8250248652840155e-05, | |
| "loss": 0.0015, | |
| "step": 1768 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "learning_rate": 2.822959697718927e-05, | |
| "loss": 0.0024, | |
| "step": 1769 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "learning_rate": 2.8208943060057775e-05, | |
| "loss": 0.006, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "learning_rate": 2.8188286915780392e-05, | |
| "loss": 0.0043, | |
| "step": 1771 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "learning_rate": 2.8167628558693375e-05, | |
| "loss": 0.0066, | |
| "step": 1772 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "learning_rate": 2.814696800313454e-05, | |
| "loss": 0.0058, | |
| "step": 1773 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "learning_rate": 2.8126305263443187e-05, | |
| "loss": 0.0026, | |
| "step": 1774 | |
| }, | |
| { | |
| "epoch": 4.66, | |
| "learning_rate": 2.810564035396017e-05, | |
| "loss": 0.0019, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 4.66, | |
| "learning_rate": 2.8084973289027827e-05, | |
| "loss": 0.0042, | |
| "step": 1776 | |
| }, | |
| { | |
| "epoch": 4.66, | |
| "learning_rate": 2.806430408299e-05, | |
| "loss": 0.0061, | |
| "step": 1777 | |
| }, | |
| { | |
| "epoch": 4.66, | |
| "learning_rate": 2.8043632750192027e-05, | |
| "loss": 0.0027, | |
| "step": 1778 | |
| }, | |
| { | |
| "epoch": 4.67, | |
| "learning_rate": 2.8022959304980695e-05, | |
| "loss": 0.0054, | |
| "step": 1779 | |
| }, | |
| { | |
| "epoch": 4.67, | |
| "learning_rate": 2.8002283761704273e-05, | |
| "loss": 0.0028, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 4.67, | |
| "learning_rate": 2.7981606134712502e-05, | |
| "loss": 0.0049, | |
| "step": 1781 | |
| }, | |
| { | |
| "epoch": 4.67, | |
| "learning_rate": 2.7960926438356545e-05, | |
| "loss": 0.0037, | |
| "step": 1782 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 2.7940244686989004e-05, | |
| "loss": 0.0033, | |
| "step": 1783 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 2.7919560894963913e-05, | |
| "loss": 0.0058, | |
| "step": 1784 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 2.7898875076636732e-05, | |
| "loss": 0.0074, | |
| "step": 1785 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 2.7878187246364318e-05, | |
| "loss": 0.0038, | |
| "step": 1786 | |
| }, | |
| { | |
| "epoch": 4.69, | |
| "learning_rate": 2.7857497418504912e-05, | |
| "loss": 0.0057, | |
| "step": 1787 | |
| }, | |
| { | |
| "epoch": 4.69, | |
| "learning_rate": 2.783680560741816e-05, | |
| "loss": 0.0042, | |
| "step": 1788 | |
| }, | |
| { | |
| "epoch": 4.69, | |
| "learning_rate": 2.7816111827465073e-05, | |
| "loss": 0.0026, | |
| "step": 1789 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "learning_rate": 2.7795416093008048e-05, | |
| "loss": 0.0073, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "learning_rate": 2.7774718418410813e-05, | |
| "loss": 0.0083, | |
| "step": 1791 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "learning_rate": 2.7754018818038445e-05, | |
| "loss": 0.0077, | |
| "step": 1792 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "learning_rate": 2.7733317306257377e-05, | |
| "loss": 0.0064, | |
| "step": 1793 | |
| }, | |
| { | |
| "epoch": 4.71, | |
| "learning_rate": 2.7712613897435357e-05, | |
| "loss": 0.0043, | |
| "step": 1794 | |
| }, | |
| { | |
| "epoch": 4.71, | |
| "learning_rate": 2.7691908605941458e-05, | |
| "loss": 0.0023, | |
| "step": 1795 | |
| }, | |
| { | |
| "epoch": 4.71, | |
| "learning_rate": 2.7671201446146033e-05, | |
| "loss": 0.0025, | |
| "step": 1796 | |
| }, | |
| { | |
| "epoch": 4.71, | |
| "learning_rate": 2.765049243242076e-05, | |
| "loss": 0.0121, | |
| "step": 1797 | |
| }, | |
| { | |
| "epoch": 4.72, | |
| "learning_rate": 2.762978157913859e-05, | |
| "loss": 0.0101, | |
| "step": 1798 | |
| }, | |
| { | |
| "epoch": 4.72, | |
| "learning_rate": 2.7609068900673767e-05, | |
| "loss": 0.0025, | |
| "step": 1799 | |
| }, | |
| { | |
| "epoch": 4.72, | |
| "learning_rate": 2.7588354411401777e-05, | |
| "loss": 0.0017, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 4.72, | |
| "eval_loss": 0.017755404114723206, | |
| "eval_runtime": 27.037, | |
| "eval_samples_per_second": 22.192, | |
| "eval_steps_per_second": 2.774, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 4.72, | |
| "learning_rate": 2.756763812569938e-05, | |
| "loss": 0.0054, | |
| "step": 1801 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "learning_rate": 2.7546920057944582e-05, | |
| "loss": 0.0008, | |
| "step": 1802 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "learning_rate": 2.7526200222516617e-05, | |
| "loss": 0.0053, | |
| "step": 1803 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "learning_rate": 2.750547863379595e-05, | |
| "loss": 0.0084, | |
| "step": 1804 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "learning_rate": 2.748475530616427e-05, | |
| "loss": 0.0023, | |
| "step": 1805 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 2.7464030254004458e-05, | |
| "loss": 0.0043, | |
| "step": 1806 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 2.7443303491700607e-05, | |
| "loss": 0.0047, | |
| "step": 1807 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 2.742257503363799e-05, | |
| "loss": 0.0022, | |
| "step": 1808 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 2.7401844894203056e-05, | |
| "loss": 0.0093, | |
| "step": 1809 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "learning_rate": 2.738111308778342e-05, | |
| "loss": 0.0093, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "learning_rate": 2.7360379628767857e-05, | |
| "loss": 0.002, | |
| "step": 1811 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "learning_rate": 2.73396445315463e-05, | |
| "loss": 0.004, | |
| "step": 1812 | |
| }, | |
| { | |
| "epoch": 4.76, | |
| "learning_rate": 2.731890781050979e-05, | |
| "loss": 0.0051, | |
| "step": 1813 | |
| }, | |
| { | |
| "epoch": 4.76, | |
| "learning_rate": 2.7298169480050516e-05, | |
| "loss": 0.0054, | |
| "step": 1814 | |
| }, | |
| { | |
| "epoch": 4.76, | |
| "learning_rate": 2.727742955456179e-05, | |
| "loss": 0.0146, | |
| "step": 1815 | |
| }, | |
| { | |
| "epoch": 4.76, | |
| "learning_rate": 2.7256688048438016e-05, | |
| "loss": 0.0082, | |
| "step": 1816 | |
| }, | |
| { | |
| "epoch": 4.77, | |
| "learning_rate": 2.7235944976074695e-05, | |
| "loss": 0.0039, | |
| "step": 1817 | |
| }, | |
| { | |
| "epoch": 4.77, | |
| "learning_rate": 2.721520035186843e-05, | |
| "loss": 0.0016, | |
| "step": 1818 | |
| }, | |
| { | |
| "epoch": 4.77, | |
| "learning_rate": 2.7194454190216877e-05, | |
| "loss": 0.0014, | |
| "step": 1819 | |
| }, | |
| { | |
| "epoch": 4.77, | |
| "learning_rate": 2.7173706505518797e-05, | |
| "loss": 0.0034, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "learning_rate": 2.715295731217396e-05, | |
| "loss": 0.0015, | |
| "step": 1821 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "learning_rate": 2.7132206624583217e-05, | |
| "loss": 0.0042, | |
| "step": 1822 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "learning_rate": 2.711145445714845e-05, | |
| "loss": 0.003, | |
| "step": 1823 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "learning_rate": 2.7090700824272557e-05, | |
| "loss": 0.001, | |
| "step": 1824 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "learning_rate": 2.7069945740359478e-05, | |
| "loss": 0.0029, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "learning_rate": 2.7049189219814125e-05, | |
| "loss": 0.006, | |
| "step": 1826 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "learning_rate": 2.7028431277042425e-05, | |
| "loss": 0.0035, | |
| "step": 1827 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "learning_rate": 2.7007671926451305e-05, | |
| "loss": 0.0018, | |
| "step": 1828 | |
| }, | |
| { | |
| "epoch": 4.8, | |
| "learning_rate": 2.6986911182448648e-05, | |
| "loss": 0.0025, | |
| "step": 1829 | |
| }, | |
| { | |
| "epoch": 4.8, | |
| "learning_rate": 2.6966149059443312e-05, | |
| "loss": 0.0013, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 4.8, | |
| "learning_rate": 2.6945385571845113e-05, | |
| "loss": 0.0074, | |
| "step": 1831 | |
| }, | |
| { | |
| "epoch": 4.81, | |
| "learning_rate": 2.6924620734064815e-05, | |
| "loss": 0.003, | |
| "step": 1832 | |
| }, | |
| { | |
| "epoch": 4.81, | |
| "learning_rate": 2.6903854560514118e-05, | |
| "loss": 0.0058, | |
| "step": 1833 | |
| }, | |
| { | |
| "epoch": 4.81, | |
| "learning_rate": 2.6883087065605644e-05, | |
| "loss": 0.0067, | |
| "step": 1834 | |
| }, | |
| { | |
| "epoch": 4.81, | |
| "learning_rate": 2.6862318263752946e-05, | |
| "loss": 0.0024, | |
| "step": 1835 | |
| }, | |
| { | |
| "epoch": 4.82, | |
| "learning_rate": 2.6841548169370457e-05, | |
| "loss": 0.0046, | |
| "step": 1836 | |
| }, | |
| { | |
| "epoch": 4.82, | |
| "learning_rate": 2.6820776796873543e-05, | |
| "loss": 0.0048, | |
| "step": 1837 | |
| }, | |
| { | |
| "epoch": 4.82, | |
| "learning_rate": 2.680000416067844e-05, | |
| "loss": 0.0032, | |
| "step": 1838 | |
| }, | |
| { | |
| "epoch": 4.82, | |
| "learning_rate": 2.6779230275202243e-05, | |
| "loss": 0.0069, | |
| "step": 1839 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 2.6758455154862937e-05, | |
| "loss": 0.0041, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 2.673767881407937e-05, | |
| "loss": 0.0076, | |
| "step": 1841 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 2.6716901267271215e-05, | |
| "loss": 0.0033, | |
| "step": 1842 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 2.669612252885899e-05, | |
| "loss": 0.0044, | |
| "step": 1843 | |
| }, | |
| { | |
| "epoch": 4.84, | |
| "learning_rate": 2.6675342613264048e-05, | |
| "loss": 0.0027, | |
| "step": 1844 | |
| }, | |
| { | |
| "epoch": 4.84, | |
| "learning_rate": 2.6654561534908544e-05, | |
| "loss": 0.0028, | |
| "step": 1845 | |
| }, | |
| { | |
| "epoch": 4.84, | |
| "learning_rate": 2.6633779308215467e-05, | |
| "loss": 0.004, | |
| "step": 1846 | |
| }, | |
| { | |
| "epoch": 4.84, | |
| "learning_rate": 2.6612995947608566e-05, | |
| "loss": 0.0031, | |
| "step": 1847 | |
| }, | |
| { | |
| "epoch": 4.85, | |
| "learning_rate": 2.6592211467512408e-05, | |
| "loss": 0.005, | |
| "step": 1848 | |
| }, | |
| { | |
| "epoch": 4.85, | |
| "learning_rate": 2.657142588235232e-05, | |
| "loss": 0.0021, | |
| "step": 1849 | |
| }, | |
| { | |
| "epoch": 4.85, | |
| "learning_rate": 2.6550639206554405e-05, | |
| "loss": 0.0098, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 4.85, | |
| "eval_loss": 0.016997970640659332, | |
| "eval_runtime": 26.9876, | |
| "eval_samples_per_second": 22.232, | |
| "eval_steps_per_second": 2.779, | |
| "step": 1850 | |
| } | |
| ], | |
| "logging_steps": 1.0, | |
| "max_steps": 3810, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 10, | |
| "save_steps": 50, | |
| "total_flos": 2.1215423471026176e+18, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |