| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 4.0, | |
| "global_step": 28952, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.913650179607627e-05, | |
| "loss": 0.3611, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.827300359215253e-05, | |
| "loss": 0.2469, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.740950538822879e-05, | |
| "loss": 0.307, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.6546007184305056e-05, | |
| "loss": 0.3492, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.568250898038132e-05, | |
| "loss": 0.2257, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.4819010776457586e-05, | |
| "loss": 0.293, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.395551257253385e-05, | |
| "loss": 0.3023, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.3092014368610116e-05, | |
| "loss": 0.2834, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.2228516164686374e-05, | |
| "loss": 0.3275, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.136501796076264e-05, | |
| "loss": 0.2694, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 4.0501519756838904e-05, | |
| "loss": 0.2631, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 3.963802155291517e-05, | |
| "loss": 0.2773, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 3.8774523348991434e-05, | |
| "loss": 0.3141, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.79110251450677e-05, | |
| "loss": 0.2893, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 3.7047526941143964e-05, | |
| "loss": 0.2038, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 3.618402873722023e-05, | |
| "loss": 0.2157, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 3.5320530533296494e-05, | |
| "loss": 0.2771, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 3.445703232937276e-05, | |
| "loss": 0.2511, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 3.3593534125449024e-05, | |
| "loss": 0.2678, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 3.273003592152529e-05, | |
| "loss": 0.1795, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 3.186653771760155e-05, | |
| "loss": 0.2357, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 3.100303951367781e-05, | |
| "loss": 0.2187, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 3.0139541309754078e-05, | |
| "loss": 0.2315, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 2.9276043105830343e-05, | |
| "loss": 0.2478, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 2.8412544901906608e-05, | |
| "loss": 0.2444, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 2.7549046697982873e-05, | |
| "loss": 0.3056, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 2.668554849405913e-05, | |
| "loss": 0.2881, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 2.5822050290135396e-05, | |
| "loss": 0.3426, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 2.495855208621166e-05, | |
| "loss": 0.1894, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 2.4095053882287926e-05, | |
| "loss": 0.1381, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 2.3231555678364188e-05, | |
| "loss": 0.1359, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 2.2368057474440453e-05, | |
| "loss": 0.1179, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 2.1504559270516718e-05, | |
| "loss": 0.1048, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 2.0641061066592983e-05, | |
| "loss": 0.141, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 1.9777562862669248e-05, | |
| "loss": 0.1726, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 1.8914064658745513e-05, | |
| "loss": 0.1307, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 1.8050566454821775e-05, | |
| "loss": 0.1875, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 1.718706825089804e-05, | |
| "loss": 0.0803, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 1.6323570046974305e-05, | |
| "loss": 0.1028, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 1.5460071843050566e-05, | |
| "loss": 0.1484, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 1.4596573639126831e-05, | |
| "loss": 0.1042, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 1.3733075435203096e-05, | |
| "loss": 0.1325, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 1.286957723127936e-05, | |
| "loss": 0.1005, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 1.2006079027355625e-05, | |
| "loss": 0.0656, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 1.1142580823431888e-05, | |
| "loss": 0.0533, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 1.0279082619508151e-05, | |
| "loss": 0.0527, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 9.415584415584416e-06, | |
| "loss": 0.0241, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 8.55208621166068e-06, | |
| "loss": 0.0463, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 7.688588007736943e-06, | |
| "loss": 0.0746, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 6.825089803813209e-06, | |
| "loss": 0.0723, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 5.961591599889472e-06, | |
| "loss": 0.044, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 5.098093395965737e-06, | |
| "loss": 0.0451, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 4.2345951920420006e-06, | |
| "loss": 0.0695, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 3.371096988118265e-06, | |
| "loss": 0.0228, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 2.507598784194529e-06, | |
| "loss": 0.0257, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 1.644100580270793e-06, | |
| "loss": 0.0158, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 7.806023763470573e-07, | |
| "loss": 0.0432, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "step": 28952, | |
| "total_flos": 9.075809388129869e+16, | |
| "train_loss": 0.1776402507509504, | |
| "train_runtime": 7042.0183, | |
| "train_samples_per_second": 12.334, | |
| "train_steps_per_second": 4.111 | |
| } | |
| ], | |
| "max_steps": 28952, | |
| "num_train_epochs": 4, | |
| "total_flos": 9.075809388129869e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |