| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "global_step": 105, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1e-05, | |
| "loss": 3.0768, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 2e-05, | |
| "loss": 3.076, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.9995348836233517e-05, | |
| "loss": 2.7773, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.998139967159894e-05, | |
| "loss": 2.4612, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.9958165482066094e-05, | |
| "loss": 2.3567, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.992566788083908e-05, | |
| "loss": 2.2825, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.9883937098250962e-05, | |
| "loss": 2.2337, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.9833011953642525e-05, | |
| "loss": 2.174, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.9772939819251247e-05, | |
| "loss": 2.1664, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.9703776576144106e-05, | |
| "loss": 2.1585, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.962558656223516e-05, | |
| "loss": 2.1401, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.953844251243633e-05, | |
| "loss": 2.1218, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.9442425490996987e-05, | |
| "loss": 2.1043, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.933762481609536e-05, | |
| "loss": 2.1185, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.9224137976751797e-05, | |
| "loss": 2.1042, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.910207054214133e-05, | |
| "loss": 2.0782, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.8971536063389745e-05, | |
| "loss": 2.0411, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.8832655967944607e-05, | |
| "loss": 2.0752, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.868555944661949e-05, | |
| "loss": 2.0272, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.853038333341642e-05, | |
| "loss": 2.0785, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.8367271978238422e-05, | |
| "loss": 2.0635, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.8196377112610524e-05, | |
| "loss": 2.0478, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.8017857708534107e-05, | |
| "loss": 2.0435, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.783187983060594e-05, | |
| "loss": 2.0279, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.763861648153945e-05, | |
| "loss": 2.0068, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.743824744123196e-05, | |
| "loss": 2.0045, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.7230959099527512e-05, | |
| "loss": 1.9743, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.7016944282830935e-05, | |
| "loss": 2.0074, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.6796402074734404e-05, | |
| "loss": 2.0146, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.6569537630823385e-05, | |
| "loss": 2.0052, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.6336561987834155e-05, | |
| "loss": 2.0006, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.6097691867340547e-05, | |
| "loss": 1.986, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.585314947415242e-05, | |
| "loss": 1.9663, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.5603162289613503e-05, | |
| "loss": 1.9941, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.5347962859990744e-05, | |
| "loss": 1.9706, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.5087788580152207e-05, | |
| "loss": 1.9814, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.4822881472734563e-05, | |
| "loss": 1.9713, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.4553487963005712e-05, | |
| "loss": 1.9883, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.427985864963193e-05, | |
| "loss": 1.9605, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.400224807156278e-05, | |
| "loss": 1.9615, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.3720914471250644e-05, | |
| "loss": 1.9548, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.3436119554425133e-05, | |
| "loss": 1.9549, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.314812824664585e-05, | |
| "loss": 1.9659, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.285720844685996e-05, | |
| "loss": 1.9628, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.2563630778193805e-05, | |
| "loss": 1.9431, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.2267668336210411e-05, | |
| "loss": 1.9727, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.1969596434867063e-05, | |
| "loss": 1.9604, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.1669692350409223e-05, | |
| "loss": 1.939, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.1368235063439103e-05, | |
| "loss": 1.9414, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.1065504999398762e-05, | |
| "loss": 1.9298, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.0761783767709182e-05, | |
| "loss": 1.9248, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.0457353899807947e-05, | |
| "loss": 1.9556, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.015249858632926e-05, | |
| "loss": 1.9553, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 9.847501413670742e-06, | |
| "loss": 1.9302, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.542646100192056e-06, | |
| "loss": 1.9398, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.238216232290821e-06, | |
| "loss": 1.9346, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 8.934495000601241e-06, | |
| "loss": 1.9306, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.6317649365609e-06, | |
| "loss": 1.9383, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.330307649590782e-06, | |
| "loss": 1.9141, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.030403565132942e-06, | |
| "loss": 1.9167, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 7.732331663789592e-06, | |
| "loss": 1.933, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 7.436369221806201e-06, | |
| "loss": 1.9143, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 7.142791553140045e-06, | |
| "loss": 1.9258, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 6.851871753354154e-06, | |
| "loss": 1.9299, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 6.563880445574873e-06, | |
| "loss": 1.9293, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 6.2790855287493605e-06, | |
| "loss": 1.9081, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 5.99775192843722e-06, | |
| "loss": 1.915, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5.720141350368072e-06, | |
| "loss": 1.9079, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5.446512036994287e-06, | |
| "loss": 1.9374, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 5.177118527265438e-06, | |
| "loss": 1.91, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.912211419847795e-06, | |
| "loss": 1.9288, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.652037140009259e-06, | |
| "loss": 1.9157, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.396837710386503e-06, | |
| "loss": 1.9313, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.1468505258475785e-06, | |
| "loss": 1.9036, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.902308132659457e-06, | |
| "loss": 1.9472, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.6634380121658484e-06, | |
| "loss": 1.9138, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.4304623691766193e-06, | |
| "loss": 1.939, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.203597925265598e-06, | |
| "loss": 1.9081, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 2.98305571716907e-06, | |
| "loss": 1.8986, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.7690409004724883e-06, | |
| "loss": 1.893, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.56175255876804e-06, | |
| "loss": 1.8973, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.3613835184605527e-06, | |
| "loss": 1.9217, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.1681201693940667e-06, | |
| "loss": 1.9373, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.982142291465896e-06, | |
| "loss": 1.8995, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.8036228873894745e-06, | |
| "loss": 1.9153, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.6327280217615793e-06, | |
| "loss": 1.8962, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.4696166665835853e-06, | |
| "loss": 1.9066, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.3144405533805138e-06, | |
| "loss": 1.9025, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.1673440320553941e-06, | |
| "loss": 1.893, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.02846393661026e-06, | |
| "loss": 1.9014, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 8.979294578586739e-07, | |
| "loss": 1.8953, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 7.758620232482083e-07, | |
| "loss": 1.9089, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 6.623751839046455e-07, | |
| "loss": 1.8885, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 5.575745090030138e-07, | |
| "loss": 1.909, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 4.61557487563673e-07, | |
| "loss": 1.9084, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 3.7441343776484116e-07, | |
| "loss": 1.9123, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 2.9622342385589256e-07, | |
| "loss": 1.9122, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 2.2706018074875046e-07, | |
| "loss": 1.9041, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.669880463574758e-07, | |
| "loss": 1.8988, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.160629017490389e-07, | |
| "loss": 1.9071, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 7.433211916092143e-08, | |
| "loss": 1.9057, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 4.183451793390747e-08, | |
| "loss": 1.9146, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.860032840106163e-08, | |
| "loss": 1.9051, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.651163766484779e-09, | |
| "loss": 1.9185, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.0, | |
| "loss": 1.8839, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 105, | |
| "total_flos": 6.475878851632169e+17, | |
| "train_loss": 2.0050100269771756, | |
| "train_runtime": 1593.9957, | |
| "train_samples_per_second": 62.742, | |
| "train_steps_per_second": 0.066 | |
| } | |
| ], | |
| "max_steps": 105, | |
| "num_train_epochs": 1, | |
| "total_flos": 6.475878851632169e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |