| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 840, | |
| "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": 5.3457, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4e-08, | |
| "loss": 5.2606, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7e-08, | |
| "loss": 5.0776, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1e-07, | |
| "loss": 5.149, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.4e-07, | |
| "loss": 5.3724, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.7000000000000001e-07, | |
| "loss": 5.2246, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.0999999999999997e-07, | |
| "loss": 5.032, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.5e-07, | |
| "loss": 5.0639, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.9e-07, | |
| "loss": 4.5604, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 3.3e-07, | |
| "loss": 4.1731, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 3.7e-07, | |
| "loss": 4.1152, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.0999999999999994e-07, | |
| "loss": 4.101, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.5e-07, | |
| "loss": 3.8602, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.9e-07, | |
| "loss": 3.7321, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 5.3e-07, | |
| "loss": 3.6759, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 5.699999999999999e-07, | |
| "loss": 3.6372, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 6.1e-07, | |
| "loss": 3.5755, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 6.5e-07, | |
| "loss": 3.415, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 6.9e-07, | |
| "loss": 3.215, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 7.3e-07, | |
| "loss": 2.8689, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 7.699999999999999e-07, | |
| "loss": 2.843, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 8.1e-07, | |
| "loss": 2.9887, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 8.499999999999999e-07, | |
| "loss": 2.6316, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 8.9e-07, | |
| "loss": 2.4875, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.3e-07, | |
| "loss": 2.4555, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.7e-07, | |
| "loss": 2.3976, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.99367088607595e-07, | |
| "loss": 2.2534, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.968354430379747e-07, | |
| "loss": 2.1527, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.943037974683544e-07, | |
| "loss": 2.0296, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.91772151898734e-07, | |
| "loss": 2.0501, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.89240506329114e-07, | |
| "loss": 2.0385, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.867088607594938e-07, | |
| "loss": 1.9757, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.841772151898733e-07, | |
| "loss": 1.7414, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.81645569620253e-07, | |
| "loss": 1.7658, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.791139240506329e-07, | |
| "loss": 1.6414, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.765822784810126e-07, | |
| "loss": 1.7797, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.740506329113924e-07, | |
| "loss": 1.5513, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.715189873417722e-07, | |
| "loss": 1.5515, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.689873417721517e-07, | |
| "loss": 1.471, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.664556962025317e-07, | |
| "loss": 1.3805, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.639240506329115e-07, | |
| "loss": 1.3442, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.61392405063291e-07, | |
| "loss": 1.2072, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.588607594936708e-07, | |
| "loss": 1.2704, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.563291139240506e-07, | |
| "loss": 1.3914, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.537974683544304e-07, | |
| "loss": 1.2708, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.5126582278481e-07, | |
| "loss": 1.2163, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.487341772151898e-07, | |
| "loss": 1.2604, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.462025316455697e-07, | |
| "loss": 1.0958, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.436708860759493e-07, | |
| "loss": 1.1216, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.411392405063291e-07, | |
| "loss": 1.1185, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.386075949367089e-07, | |
| "loss": 1.0496, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.360759493670885e-07, | |
| "loss": 0.9438, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.335443037974683e-07, | |
| "loss": 0.8912, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.31012658227848e-07, | |
| "loss": 0.8424, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.284810126582277e-07, | |
| "loss": 0.9812, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.259493670886076e-07, | |
| "loss": 0.8999, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.234177215189873e-07, | |
| "loss": 0.7623, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.208860759493671e-07, | |
| "loss": 0.6592, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.183544303797468e-07, | |
| "loss": 0.7013, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.158227848101265e-07, | |
| "loss": 0.5614, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.132911392405063e-07, | |
| "loss": 0.5246, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 9.10759493670886e-07, | |
| "loss": 0.3732, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 9.082278481012657e-07, | |
| "loss": 0.3963, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 9.056962025316456e-07, | |
| "loss": 0.3281, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 9.031645569620254e-07, | |
| "loss": 0.3707, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 9.00632911392405e-07, | |
| "loss": 0.3442, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 8.981012658227848e-07, | |
| "loss": 0.286, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 8.955696202531646e-07, | |
| "loss": 0.2778, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 8.930379746835442e-07, | |
| "loss": 0.3224, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 8.90506329113924e-07, | |
| "loss": 0.2294, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 8.879746835443038e-07, | |
| "loss": 0.3471, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 8.854430379746834e-07, | |
| "loss": 0.2537, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 8.829113924050633e-07, | |
| "loss": 0.3189, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 8.803797468354431e-07, | |
| "loss": 0.2704, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 8.778481012658227e-07, | |
| "loss": 0.2685, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 8.753164556962025e-07, | |
| "loss": 0.4016, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 8.727848101265822e-07, | |
| "loss": 0.2614, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 8.70253164556962e-07, | |
| "loss": 0.2208, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 8.677215189873417e-07, | |
| "loss": 0.3452, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 8.651898734177214e-07, | |
| "loss": 0.2481, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 8.626582278481013e-07, | |
| "loss": 0.2698, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 8.60126582278481e-07, | |
| "loss": 0.2386, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 8.575949367088607e-07, | |
| "loss": 0.27, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 8.550632911392405e-07, | |
| "loss": 0.4073, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 8.525316455696202e-07, | |
| "loss": 0.263, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 8.499999999999999e-07, | |
| "loss": 0.2443, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 8.474683544303797e-07, | |
| "loss": 0.1876, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 8.449367088607595e-07, | |
| "loss": 0.1893, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 8.424050632911392e-07, | |
| "loss": 0.356, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 8.39873417721519e-07, | |
| "loss": 0.3026, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 8.373417721518988e-07, | |
| "loss": 0.2265, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 8.348101265822784e-07, | |
| "loss": 0.3389, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 8.322784810126582e-07, | |
| "loss": 0.274, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 8.297468354430379e-07, | |
| "loss": 0.3028, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 8.272151898734176e-07, | |
| "loss": 0.2582, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 8.246835443037974e-07, | |
| "loss": 0.253, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 8.221518987341772e-07, | |
| "loss": 0.2579, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 8.19620253164557e-07, | |
| "loss": 0.3189, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 8.170886075949367e-07, | |
| "loss": 0.2599, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 8.145569620253164e-07, | |
| "loss": 0.2791, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 8.120253164556962e-07, | |
| "loss": 0.3284, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 8.094936708860759e-07, | |
| "loss": 0.3026, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 8.069620253164556e-07, | |
| "loss": 0.3195, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 8.044303797468354e-07, | |
| "loss": 0.2776, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 8.018987341772152e-07, | |
| "loss": 0.3193, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 7.993670886075949e-07, | |
| "loss": 0.2963, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 7.968354430379747e-07, | |
| "loss": 0.2865, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 7.943037974683544e-07, | |
| "loss": 0.3051, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 7.917721518987341e-07, | |
| "loss": 0.2691, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 7.892405063291139e-07, | |
| "loss": 0.2542, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 7.867088607594937e-07, | |
| "loss": 0.3001, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 7.841772151898733e-07, | |
| "loss": 0.2793, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 7.816455696202531e-07, | |
| "loss": 0.2546, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 7.79113924050633e-07, | |
| "loss": 0.2801, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 7.765822784810126e-07, | |
| "loss": 0.2635, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 7.740506329113924e-07, | |
| "loss": 0.2678, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 7.715189873417721e-07, | |
| "loss": 0.379, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 7.689873417721518e-07, | |
| "loss": 0.3077, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 7.664556962025316e-07, | |
| "loss": 0.1788, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 7.639240506329113e-07, | |
| "loss": 0.3147, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 7.61392405063291e-07, | |
| "loss": 0.3064, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 7.588607594936709e-07, | |
| "loss": 0.2644, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 7.563291139240506e-07, | |
| "loss": 0.1713, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 7.537974683544304e-07, | |
| "loss": 0.2332, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 7.512658227848101e-07, | |
| "loss": 0.3326, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 7.487341772151898e-07, | |
| "loss": 0.3293, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 7.462025316455696e-07, | |
| "loss": 0.1996, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 7.436708860759493e-07, | |
| "loss": 0.2282, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 7.41139240506329e-07, | |
| "loss": 0.2591, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 7.386075949367089e-07, | |
| "loss": 0.2376, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 7.360759493670887e-07, | |
| "loss": 0.2115, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 7.335443037974683e-07, | |
| "loss": 0.2515, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 7.310126582278481e-07, | |
| "loss": 0.3904, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 7.284810126582279e-07, | |
| "loss": 0.2693, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 7.259493670886075e-07, | |
| "loss": 0.2653, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 7.234177215189873e-07, | |
| "loss": 0.2901, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 7.20886075949367e-07, | |
| "loss": 0.3418, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 7.183544303797468e-07, | |
| "loss": 0.2681, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 7.158227848101266e-07, | |
| "loss": 0.2793, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 7.132911392405063e-07, | |
| "loss": 0.2777, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 7.10759493670886e-07, | |
| "loss": 0.1978, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 7.082278481012658e-07, | |
| "loss": 0.2681, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 7.056962025316455e-07, | |
| "loss": 0.2808, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 7.031645569620253e-07, | |
| "loss": 0.2718, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 7.00632911392405e-07, | |
| "loss": 0.2197, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 6.981012658227847e-07, | |
| "loss": 0.2063, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 6.955696202531646e-07, | |
| "loss": 0.2994, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 6.930379746835443e-07, | |
| "loss": 0.3452, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 6.90506329113924e-07, | |
| "loss": 0.2579, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 6.879746835443038e-07, | |
| "loss": 0.2208, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 6.854430379746835e-07, | |
| "loss": 0.2463, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 6.829113924050632e-07, | |
| "loss": 0.2835, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 6.80379746835443e-07, | |
| "loss": 0.2034, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 6.778481012658228e-07, | |
| "loss": 0.2464, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 6.753164556962025e-07, | |
| "loss": 0.2682, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 6.727848101265823e-07, | |
| "loss": 0.285, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 6.702531645569621e-07, | |
| "loss": 0.2276, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 6.677215189873417e-07, | |
| "loss": 0.2754, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 6.651898734177215e-07, | |
| "loss": 0.2306, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 6.626582278481012e-07, | |
| "loss": 0.2431, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 6.601265822784809e-07, | |
| "loss": 0.2079, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 6.575949367088607e-07, | |
| "loss": 0.2426, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 6.550632911392405e-07, | |
| "loss": 0.2854, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 6.525316455696203e-07, | |
| "loss": 0.3083, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 6.5e-07, | |
| "loss": 0.2305, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 6.474683544303797e-07, | |
| "loss": 0.2648, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 6.449367088607595e-07, | |
| "loss": 0.2992, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 6.424050632911392e-07, | |
| "loss": 0.3059, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 6.398734177215189e-07, | |
| "loss": 0.2373, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 6.373417721518987e-07, | |
| "loss": 0.2808, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 6.348101265822785e-07, | |
| "loss": 0.3125, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 6.322784810126582e-07, | |
| "loss": 0.1994, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 6.29746835443038e-07, | |
| "loss": 0.1612, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 6.272151898734178e-07, | |
| "loss": 0.2562, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 6.246835443037974e-07, | |
| "loss": 0.3455, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 6.221518987341772e-07, | |
| "loss": 0.2547, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 6.196202531645569e-07, | |
| "loss": 0.2416, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 6.170886075949366e-07, | |
| "loss": 0.2431, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 6.145569620253165e-07, | |
| "loss": 0.3004, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 6.120253164556962e-07, | |
| "loss": 0.378, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 6.094936708860759e-07, | |
| "loss": 0.2185, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 6.069620253164557e-07, | |
| "loss": 0.2064, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 6.044303797468354e-07, | |
| "loss": 0.2867, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 6.018987341772151e-07, | |
| "loss": 0.178, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 5.993670886075949e-07, | |
| "loss": 0.2377, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 5.968354430379746e-07, | |
| "loss": 0.2475, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 5.943037974683544e-07, | |
| "loss": 0.2546, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 5.917721518987342e-07, | |
| "loss": 0.246, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 5.892405063291139e-07, | |
| "loss": 0.1832, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 5.867088607594937e-07, | |
| "loss": 0.2818, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 5.841772151898734e-07, | |
| "loss": 0.3186, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 5.816455696202531e-07, | |
| "loss": 0.1874, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 5.791139240506329e-07, | |
| "loss": 0.2877, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 5.765822784810126e-07, | |
| "loss": 0.2143, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 5.740506329113923e-07, | |
| "loss": 0.1791, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 5.715189873417722e-07, | |
| "loss": 0.2619, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 5.68987341772152e-07, | |
| "loss": 0.2721, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 5.664556962025316e-07, | |
| "loss": 0.2182, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 5.639240506329114e-07, | |
| "loss": 0.3081, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 5.613924050632911e-07, | |
| "loss": 0.1964, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 5.588607594936708e-07, | |
| "loss": 0.221, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 5.563291139240506e-07, | |
| "loss": 0.2069, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 5.537974683544303e-07, | |
| "loss": 0.294, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 5.512658227848101e-07, | |
| "loss": 0.1952, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 5.487341772151899e-07, | |
| "loss": 0.2268, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 5.462025316455696e-07, | |
| "loss": 0.232, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 5.436708860759493e-07, | |
| "loss": 0.2639, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 5.411392405063291e-07, | |
| "loss": 0.1895, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 5.386075949367088e-07, | |
| "loss": 0.2694, | |
| "step": 836 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 5.360759493670886e-07, | |
| "loss": 0.2513, | |
| "step": 840 | |
| } | |
| ], | |
| "logging_steps": 4, | |
| "max_steps": 1680, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "save_steps": 200, | |
| "total_flos": 1.371080632762368e+19, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |