| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 420, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0, | |
| "loss": 5.3442, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0, | |
| "loss": 5.2525, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2e-08, | |
| "loss": 5.0768, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 6e-08, | |
| "loss": 5.1381, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8e-08, | |
| "loss": 5.3956, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.2e-07, | |
| "loss": 5.2933, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.6e-07, | |
| "loss": 5.2991, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2e-07, | |
| "loss": 5.5463, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.4e-07, | |
| "loss": 5.3438, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.6e-07, | |
| "loss": 5.0126, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 3e-07, | |
| "loss": 4.8808, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 3.4000000000000003e-07, | |
| "loss": 4.5046, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 3.7999999999999996e-07, | |
| "loss": 4.3779, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.1999999999999995e-07, | |
| "loss": 4.2311, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.6e-07, | |
| "loss": 4.0877, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 5e-07, | |
| "loss": 4.0972, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 5.4e-07, | |
| "loss": 4.1511, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 5.8e-07, | |
| "loss": 3.9641, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 6.2e-07, | |
| "loss": 3.8411, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 6.6e-07, | |
| "loss": 3.5198, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 7e-07, | |
| "loss": 3.5286, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 7.4e-07, | |
| "loss": 3.8059, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 7.799999999999999e-07, | |
| "loss": 3.374, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 8.199999999999999e-07, | |
| "loss": 3.3654, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 8.599999999999999e-07, | |
| "loss": 3.3283, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9e-07, | |
| "loss": 3.3164, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.399999999999999e-07, | |
| "loss": 3.13, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.8e-07, | |
| "loss": 2.9915, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.987341772151898e-07, | |
| "loss": 2.7748, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.962025316455696e-07, | |
| "loss": 2.8378, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.936708860759494e-07, | |
| "loss": 2.7933, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.91139240506329e-07, | |
| "loss": 2.7108, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.88607594936709e-07, | |
| "loss": 2.4029, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.860759493670887e-07, | |
| "loss": 2.4199, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.835443037974683e-07, | |
| "loss": 2.3015, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.81012658227848e-07, | |
| "loss": 2.4849, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.784810126582278e-07, | |
| "loss": 2.265, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.759493670886076e-07, | |
| "loss": 2.2422, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.734177215189873e-07, | |
| "loss": 2.188, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.70886075949367e-07, | |
| "loss": 2.1488, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.683544303797469e-07, | |
| "loss": 2.0795, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.658227848101267e-07, | |
| "loss": 1.9767, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.632911392405062e-07, | |
| "loss": 1.9811, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.60759493670886e-07, | |
| "loss": 2.0459, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.582278481012658e-07, | |
| "loss": 1.936, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.556962025316455e-07, | |
| "loss": 1.9373, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.531645569620253e-07, | |
| "loss": 2.0272, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.50632911392405e-07, | |
| "loss": 1.8032, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.481012658227847e-07, | |
| "loss": 1.8799, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.455696202531646e-07, | |
| "loss": 1.9145, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.430379746835443e-07, | |
| "loss": 1.8476, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.40506329113924e-07, | |
| "loss": 1.7027, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.379746835443037e-07, | |
| "loss": 1.6791, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.354430379746835e-07, | |
| "loss": 1.6089, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.329113924050633e-07, | |
| "loss": 1.6315, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.303797468354429e-07, | |
| "loss": 1.6194, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.278481012658227e-07, | |
| "loss": 1.5367, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.253164556962026e-07, | |
| "loss": 1.4617, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.227848101265822e-07, | |
| "loss": 1.6019, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.20253164556962e-07, | |
| "loss": 1.4418, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.177215189873418e-07, | |
| "loss": 1.4306, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 9.151898734177214e-07, | |
| "loss": 1.3711, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 9.126582278481012e-07, | |
| "loss": 1.4102, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 9.10126582278481e-07, | |
| "loss": 1.348, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 9.075949367088606e-07, | |
| "loss": 1.374, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 9.050632911392405e-07, | |
| "loss": 1.2833, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 9.025316455696203e-07, | |
| "loss": 1.2588, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 9e-07, | |
| "loss": 1.2172, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 8.974683544303797e-07, | |
| "loss": 1.3046, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 8.949367088607594e-07, | |
| "loss": 1.0875, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 8.924050632911392e-07, | |
| "loss": 1.2346, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 8.898734177215189e-07, | |
| "loss": 1.0975, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 8.873417721518986e-07, | |
| "loss": 1.2068, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 8.848101265822785e-07, | |
| "loss": 1.1378, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 8.822784810126583e-07, | |
| "loss": 1.0359, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 8.797468354430379e-07, | |
| "loss": 1.2252, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 8.772151898734177e-07, | |
| "loss": 1.062, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 8.746835443037975e-07, | |
| "loss": 0.9852, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 8.721518987341771e-07, | |
| "loss": 1.0557, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 8.696202531645569e-07, | |
| "loss": 0.9546, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 8.670886075949367e-07, | |
| "loss": 0.9371, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 8.645569620253164e-07, | |
| "loss": 0.9128, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 8.620253164556962e-07, | |
| "loss": 0.9508, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 8.59493670886076e-07, | |
| "loss": 1.0656, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 8.569620253164556e-07, | |
| "loss": 0.8574, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 8.544303797468354e-07, | |
| "loss": 0.7957, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 8.518987341772152e-07, | |
| "loss": 0.8379, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 8.493670886075949e-07, | |
| "loss": 0.7316, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 8.468354430379746e-07, | |
| "loss": 0.8871, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 8.443037974683543e-07, | |
| "loss": 0.7857, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 8.417721518987342e-07, | |
| "loss": 0.6546, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 8.392405063291139e-07, | |
| "loss": 0.7563, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 8.367088607594936e-07, | |
| "loss": 0.6105, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 8.341772151898734e-07, | |
| "loss": 0.6173, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 8.316455696202531e-07, | |
| "loss": 0.536, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 8.291139240506328e-07, | |
| "loss": 0.4445, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 8.265822784810126e-07, | |
| "loss": 0.377, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 8.240506329113924e-07, | |
| "loss": 0.4066, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 8.215189873417721e-07, | |
| "loss": 0.3306, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 8.189873417721519e-07, | |
| "loss": 0.3437, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 8.164556962025317e-07, | |
| "loss": 0.3812, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 8.139240506329113e-07, | |
| "loss": 0.3661, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 8.113924050632911e-07, | |
| "loss": 0.3579, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 8.088607594936709e-07, | |
| "loss": 0.3228, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 8.063291139240505e-07, | |
| "loss": 0.3723, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 8.037974683544303e-07, | |
| "loss": 0.3276, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 8.012658227848102e-07, | |
| "loss": 0.3386, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 7.987341772151898e-07, | |
| "loss": 0.3415, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 7.962025316455696e-07, | |
| "loss": 0.3055, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 7.936708860759494e-07, | |
| "loss": 0.2819, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 7.911392405063291e-07, | |
| "loss": 0.3376, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 7.886075949367088e-07, | |
| "loss": 0.306, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 7.860759493670885e-07, | |
| "loss": 0.2942, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 7.835443037974683e-07, | |
| "loss": 0.3164, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 7.810126582278481e-07, | |
| "loss": 0.2918, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 7.784810126582278e-07, | |
| "loss": 0.3092, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 7.759493670886076e-07, | |
| "loss": 0.4132, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 7.734177215189873e-07, | |
| "loss": 0.3428, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 7.70886075949367e-07, | |
| "loss": 0.2002, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 7.683544303797468e-07, | |
| "loss": 0.333, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 7.658227848101266e-07, | |
| "loss": 0.3346, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 7.632911392405062e-07, | |
| "loss": 0.282, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 7.60759493670886e-07, | |
| "loss": 0.1892, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 7.582278481012659e-07, | |
| "loss": 0.2727, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 7.556962025316455e-07, | |
| "loss": 0.3528, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 7.531645569620253e-07, | |
| "loss": 0.3449, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 7.506329113924051e-07, | |
| "loss": 0.2044, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 7.481012658227847e-07, | |
| "loss": 0.2418, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 7.455696202531645e-07, | |
| "loss": 0.2712, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 7.430379746835443e-07, | |
| "loss": 0.2555, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 7.40506329113924e-07, | |
| "loss": 0.231, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 7.379746835443038e-07, | |
| "loss": 0.2705, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 7.354430379746836e-07, | |
| "loss": 0.4189, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 7.329113924050633e-07, | |
| "loss": 0.2864, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 7.30379746835443e-07, | |
| "loss": 0.278, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 7.278481012658227e-07, | |
| "loss": 0.308, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 7.253164556962025e-07, | |
| "loss": 0.3659, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 7.227848101265822e-07, | |
| "loss": 0.2763, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 7.202531645569619e-07, | |
| "loss": 0.3028, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 7.177215189873418e-07, | |
| "loss": 0.2959, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 7.151898734177216e-07, | |
| "loss": 0.2135, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 7.126582278481012e-07, | |
| "loss": 0.2805, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 7.10126582278481e-07, | |
| "loss": 0.2925, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 7.075949367088608e-07, | |
| "loss": 0.2914, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 7.050632911392404e-07, | |
| "loss": 0.2366, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 7.025316455696202e-07, | |
| "loss": 0.2246, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 7e-07, | |
| "loss": 0.3242, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 6.974683544303797e-07, | |
| "loss": 0.3683, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 6.949367088607595e-07, | |
| "loss": 0.2697, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 6.924050632911393e-07, | |
| "loss": 0.2314, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 6.898734177215189e-07, | |
| "loss": 0.268, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 6.873417721518987e-07, | |
| "loss": 0.3017, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 6.848101265822784e-07, | |
| "loss": 0.2147, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 6.822784810126582e-07, | |
| "loss": 0.2574, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 6.797468354430379e-07, | |
| "loss": 0.2828, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 6.772151898734176e-07, | |
| "loss": 0.2967, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 6.746835443037975e-07, | |
| "loss": 0.2444, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 6.721518987341772e-07, | |
| "loss": 0.2925, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 6.696202531645569e-07, | |
| "loss": 0.243, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 6.670886075949367e-07, | |
| "loss": 0.2559, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 6.645569620253164e-07, | |
| "loss": 0.2143, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 6.620253164556961e-07, | |
| "loss": 0.2637, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 6.594936708860759e-07, | |
| "loss": 0.2977, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 6.569620253164557e-07, | |
| "loss": 0.3284, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 6.544303797468354e-07, | |
| "loss": 0.2531, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 6.518987341772152e-07, | |
| "loss": 0.2793, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 6.49367088607595e-07, | |
| "loss": 0.3183, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 6.468354430379746e-07, | |
| "loss": 0.3329, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 6.443037974683544e-07, | |
| "loss": 0.2545, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 6.417721518987342e-07, | |
| "loss": 0.286, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 6.392405063291138e-07, | |
| "loss": 0.3286, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 6.367088607594936e-07, | |
| "loss": 0.2061, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 6.341772151898735e-07, | |
| "loss": 0.1816, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 6.316455696202532e-07, | |
| "loss": 0.262, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 6.291139240506329e-07, | |
| "loss": 0.3591, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 6.265822784810126e-07, | |
| "loss": 0.2565, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 6.240506329113924e-07, | |
| "loss": 0.2609, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 6.215189873417721e-07, | |
| "loss": 0.2728, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 6.189873417721518e-07, | |
| "loss": 0.3175, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 6.164556962025316e-07, | |
| "loss": 0.393, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 6.139240506329114e-07, | |
| "loss": 0.234, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 6.113924050632911e-07, | |
| "loss": 0.2264, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 6.088607594936709e-07, | |
| "loss": 0.3062, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 6.063291139240507e-07, | |
| "loss": 0.184, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 6.037974683544303e-07, | |
| "loss": 0.2396, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 6.012658227848101e-07, | |
| "loss": 0.2631, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 5.987341772151899e-07, | |
| "loss": 0.2776, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 5.962025316455695e-07, | |
| "loss": 0.2634, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 5.936708860759494e-07, | |
| "loss": 0.1984, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 5.911392405063292e-07, | |
| "loss": 0.2973, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 5.886075949367088e-07, | |
| "loss": 0.3261, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 5.860759493670886e-07, | |
| "loss": 0.1961, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 5.835443037974684e-07, | |
| "loss": 0.3002, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 5.81012658227848e-07, | |
| "loss": 0.2281, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 5.784810126582278e-07, | |
| "loss": 0.1912, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 5.759493670886075e-07, | |
| "loss": 0.2726, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 5.734177215189873e-07, | |
| "loss": 0.2855, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 5.708860759493671e-07, | |
| "loss": 0.2241, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 5.683544303797468e-07, | |
| "loss": 0.3273, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 5.658227848101266e-07, | |
| "loss": 0.209, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 5.632911392405063e-07, | |
| "loss": 0.2155, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 5.60759493670886e-07, | |
| "loss": 0.2218, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 5.582278481012658e-07, | |
| "loss": 0.3097, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 5.556962025316455e-07, | |
| "loss": 0.1995, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 5.531645569620252e-07, | |
| "loss": 0.2408, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 5.506329113924051e-07, | |
| "loss": 0.2443, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 5.481012658227849e-07, | |
| "loss": 0.2719, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 5.455696202531645e-07, | |
| "loss": 0.1995, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 5.430379746835443e-07, | |
| "loss": 0.2849, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 5.405063291139241e-07, | |
| "loss": 0.272, | |
| "step": 420 | |
| } | |
| ], | |
| "logging_steps": 2, | |
| "max_steps": 840, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "save_steps": 200, | |
| "total_flos": 1.371080632762368e+19, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |