| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "eval_steps": 500, | |
| "global_step": 600, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0, | |
| "loss": 1.5138, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.010299956639811e-07, | |
| "loss": 1.4586, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.771212547196623e-07, | |
| "loss": 1.481, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 6.020599913279622e-07, | |
| "loss": 1.4797, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 6.989700043360186e-07, | |
| "loss": 1.4761, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 7.781512503836435e-07, | |
| "loss": 1.4221, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.450980400142567e-07, | |
| "loss": 1.4175, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.030899869919433e-07, | |
| "loss": 1.3727, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.542425094393247e-07, | |
| "loss": 1.3055, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.999999999999997e-07, | |
| "loss": 1.3185, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1e-06, | |
| "loss": 1.2831, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1e-06, | |
| "loss": 1.3009, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1e-06, | |
| "loss": 1.2151, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1e-06, | |
| "loss": 1.1927, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1e-06, | |
| "loss": 1.1921, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1e-06, | |
| "loss": 1.2028, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1e-06, | |
| "loss": 1.1694, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1e-06, | |
| "loss": 1.1572, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1e-06, | |
| "loss": 1.1458, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1e-06, | |
| "loss": 1.1605, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1e-06, | |
| "loss": 1.1017, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1e-06, | |
| "loss": 1.0803, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1e-06, | |
| "loss": 1.0884, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1e-06, | |
| "loss": 1.0502, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1e-06, | |
| "loss": 1.0914, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1e-06, | |
| "loss": 1.0717, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1e-06, | |
| "loss": 1.0547, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1e-06, | |
| "loss": 1.061, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1e-06, | |
| "loss": 1.0401, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1e-06, | |
| "loss": 1.0505, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1e-06, | |
| "loss": 1.0518, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1e-06, | |
| "loss": 1.0305, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1e-06, | |
| "loss": 1.0261, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1e-06, | |
| "loss": 1.0229, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1e-06, | |
| "loss": 1.0206, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1e-06, | |
| "loss": 1.0073, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1e-06, | |
| "loss": 0.991, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1e-06, | |
| "loss": 1.0245, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1e-06, | |
| "loss": 1.0021, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9799, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9709, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9829, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9882, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9612, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9732, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9615, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9616, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9497, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1e-06, | |
| "loss": 0.934, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9355, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9532, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1e-06, | |
| "loss": 0.962, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9606, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9522, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9506, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9436, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9454, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9354, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9527, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9214, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8974, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9505, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9353, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9313, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9449, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1e-06, | |
| "loss": 0.929, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9263, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1e-06, | |
| "loss": 0.91, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9208, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9073, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9066, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9206, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9372, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9162, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9003, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9172, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9149, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8832, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8903, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9107, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9042, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9053, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8787, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8947, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9006, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9078, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8901, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8757, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9023, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8835, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1e-06, | |
| "loss": 0.9074, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8666, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8767, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8749, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8579, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8885, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1e-06, | |
| "loss": 0.876, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8948, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1e-06, | |
| "loss": 0.889, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1e-06, | |
| "loss": 0.866, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8952, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8506, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8773, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1e-06, | |
| "loss": 0.857, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8555, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8729, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8818, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8616, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8837, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8632, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8848, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8701, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8618, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8609, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8685, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8781, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8635, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8583, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8344, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8783, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8509, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8475, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8756, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1e-06, | |
| "loss": 0.845, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8522, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8606, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8543, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8581, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8627, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8432, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8599, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8712, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8328, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8365, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8607, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8607, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8604, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8266, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8404, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8411, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8383, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8654, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8448, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8535, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8447, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8389, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8263, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8416, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8436, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8445, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8572, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8369, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8479, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8494, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8259, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8449, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8278, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8516, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8362, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8322, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8462, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8368, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8577, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1e-06, | |
| "loss": 0.829, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8459, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8222, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8041, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8507, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8336, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8275, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8113, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1e-06, | |
| "loss": 0.819, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8531, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1e-06, | |
| "loss": 0.83, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8207, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8223, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8201, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8283, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8476, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8197, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8563, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8371, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8127, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8213, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8267, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8118, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8013, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7866, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8241, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8336, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8302, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8002, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8378, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8427, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1e-06, | |
| "loss": 0.821, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7903, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8261, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8048, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8286, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8197, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7974, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8129, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8058, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7958, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8115, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8232, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7933, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8153, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8075, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8091, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8141, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 1e-06, | |
| "loss": 0.796, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7917, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8102, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8002, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8028, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8043, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8075, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8312, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7926, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7859, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7925, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7928, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7939, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7946, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8156, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8056, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 1e-06, | |
| "loss": 0.788, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7959, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 1e-06, | |
| "loss": 0.814, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7935, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8152, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7957, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 1e-06, | |
| "loss": 0.815, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7919, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7817, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8076, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8103, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8206, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7933, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7738, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 1e-06, | |
| "loss": 0.811, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7855, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8138, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7972, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8069, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7916, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7953, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7852, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7929, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7714, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8024, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7821, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8007, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8025, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8009, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7841, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7919, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8099, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7922, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8012, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7799, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7971, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7935, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7861, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7786, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7909, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8193, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8008, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8199, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 1e-06, | |
| "loss": 0.775, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7691, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7936, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7969, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 1e-06, | |
| "loss": 0.786, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7858, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8229, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7838, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7811, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7697, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7968, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7775, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7847, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7975, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7821, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8118, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7737, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7877, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7816, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7847, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7893, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7925, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7858, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7924, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 1e-06, | |
| "loss": 0.796, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7717, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7792, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8346, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7714, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7925, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7689, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8011, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7842, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7947, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7657, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8022, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8032, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 1e-06, | |
| "loss": 0.77, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7843, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7919, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 1e-06, | |
| "loss": 0.782, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7761, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8237, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7919, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8057, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7756, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7594, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7884, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7874, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7712, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7676, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7996, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7767, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7658, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7946, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7649, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7556, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7811, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8021, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7654, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7793, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7699, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7841, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7731, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7914, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7727, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 1e-06, | |
| "loss": 0.78, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 1e-06, | |
| "loss": 0.788, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7791, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7715, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7832, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 1e-06, | |
| "loss": 0.785, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7753, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 1e-06, | |
| "loss": 0.778, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7855, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7736, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 1e-06, | |
| "loss": 0.765, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7637, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7955, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7623, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7788, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7686, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7825, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7796, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7725, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7919, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7804, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7626, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7698, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7979, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7862, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7501, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7845, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7945, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7806, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7874, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7769, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7704, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7833, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7956, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7716, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7794, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1e-06, | |
| "loss": 0.757, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7605, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7549, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 1e-06, | |
| "loss": 0.8044, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1e-06, | |
| "loss": 0.789, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1e-06, | |
| "loss": 0.802, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7887, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7616, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7599, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7982, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7661, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7997, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7933, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7717, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7621, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 1e-06, | |
| "loss": 0.756, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 1e-06, | |
| "loss": 0.78, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7822, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7736, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 1e-06, | |
| "loss": 0.761, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 1e-06, | |
| "loss": 0.759, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7832, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7747, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7815, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7623, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7887, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7727, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7722, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7385, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7442, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7539, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7645, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7679, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7462, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7607, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7404, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7576, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7584, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7355, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7615, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 1e-06, | |
| "loss": 0.748, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7535, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7458, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7444, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7413, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7398, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 1e-06, | |
| "loss": 0.757, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7271, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7429, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7587, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7565, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7523, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7496, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7543, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7597, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7589, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7631, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7485, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7546, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7702, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7586, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7477, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7522, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7763, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7603, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7639, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7546, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7607, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7359, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7355, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7459, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7544, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7563, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7527, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7338, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7509, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7247, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7513, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7538, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7469, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 1e-06, | |
| "loss": 0.756, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7463, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7333, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7365, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 1e-06, | |
| "loss": 0.765, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7468, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7554, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7419, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7528, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7606, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7509, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7521, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7596, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7712, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 1e-06, | |
| "loss": 0.744, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7527, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7349, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7157, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 1e-06, | |
| "loss": 0.761, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7515, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7596, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7455, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7568, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7594, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7511, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7625, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7618, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 1e-06, | |
| "loss": 0.741, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7537, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7453, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7515, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7654, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 1e-06, | |
| "loss": 0.756, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7412, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7661, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7361, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7352, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7722, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7402, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7457, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7612, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7689, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7388, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7461, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7721, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7355, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7552, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7333, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7734, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7609, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 1e-06, | |
| "loss": 0.748, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7444, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7351, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7557, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7449, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7404, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7381, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7492, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7545, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7498, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7366, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7691, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7473, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7826, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 1e-06, | |
| "loss": 0.754, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 1e-06, | |
| "loss": 0.751, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7424, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7435, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7578, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7354, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7471, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7471, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7441, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7578, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7633, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7601, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7492, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7427, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7628, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7342, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7524, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 1e-06, | |
| "loss": 0.752, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7388, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7767, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7173, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7709, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7458, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7727, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7531, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7735, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7246, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7532, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 1e-06, | |
| "loss": 0.739, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7493, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7193, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7454, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7612, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7726, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7388, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7448, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7472, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 1e-06, | |
| "loss": 0.767, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7487, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7444, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7391, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7496, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7472, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7313, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7525, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7587, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 1e-06, | |
| "loss": 0.759, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 1e-06, | |
| "loss": 0.738, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7558, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7308, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7406, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7367, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7466, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7283, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7194, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7565, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7184, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7271, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 1e-06, | |
| "loss": 0.735, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7698, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7418, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7545, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7394, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7254, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7572, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7406, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7418, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7375, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7553, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7367, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7159, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7487, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7273, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7512, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 1e-06, | |
| "loss": 0.737, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7362, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7245, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 1e-06, | |
| "loss": 0.744, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7471, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7449, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7491, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7462, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7393, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 1e-06, | |
| "loss": 0.7568, | |
| "step": 600 | |
| } | |
| ], | |
| "logging_steps": 1.0, | |
| "max_steps": 1000, | |
| "num_train_epochs": 5, | |
| "save_steps": 1000.0, | |
| "total_flos": 2.7983052785319936e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |