| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 4.979384651289432, | |
| "global_step": 16000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.805477746654217e-05, | |
| "loss": 3.0433, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.610955493308435e-05, | |
| "loss": 2.9881, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.416433239962652e-05, | |
| "loss": 2.9623, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.221910986616869e-05, | |
| "loss": 2.9484, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.027388733271086e-05, | |
| "loss": 2.9376, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.832866479925304e-05, | |
| "loss": 2.9268, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 3.638344226579521e-05, | |
| "loss": 2.9135, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 3.443821973233738e-05, | |
| "loss": 2.9008, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 3.2492997198879555e-05, | |
| "loss": 2.899, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 3.0547774665421726e-05, | |
| "loss": 2.8906, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 2.86025521319639e-05, | |
| "loss": 2.8878, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 2.665732959850607e-05, | |
| "loss": 2.8841, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 2.4712107065048243e-05, | |
| "loss": 2.879, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 2.2766884531590417e-05, | |
| "loss": 2.8676, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 2.0821661998132587e-05, | |
| "loss": 2.8657, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 1.8876439464674758e-05, | |
| "loss": 2.8636, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 1.693121693121693e-05, | |
| "loss": 2.8664, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 1.4985994397759103e-05, | |
| "loss": 2.8641, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 1.3040771864301277e-05, | |
| "loss": 2.8596, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 1.1095549330843449e-05, | |
| "loss": 2.8559, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 9.150326797385621e-06, | |
| "loss": 2.8521, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 7.205104263927794e-06, | |
| "loss": 2.8517, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 5.259881730469966e-06, | |
| "loss": 2.8496, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 3.3146591970121383e-06, | |
| "loss": 2.8498, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 1.3694366635543105e-06, | |
| "loss": 2.8509, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 9.539371304077187e-06, | |
| "loss": 2.8453, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 7.983193277310924e-06, | |
| "loss": 2.8481, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "learning_rate": 6.427015250544663e-06, | |
| "loss": 2.8479, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 4.51, | |
| "learning_rate": 4.8708372237784005e-06, | |
| "loss": 2.8431, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 4.67, | |
| "learning_rate": 3.3146591970121383e-06, | |
| "loss": 2.8463, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 4.82, | |
| "learning_rate": 1.7584811702458762e-06, | |
| "loss": 2.8432, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 2.023031434796141e-07, | |
| "loss": 2.8453, | |
| "step": 16000 | |
| } | |
| ], | |
| "max_steps": 16065, | |
| "num_train_epochs": 5, | |
| "total_flos": 1.550918552519762e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |