| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 4.0, | |
| "global_step": 7340, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.8637602179836514e-05, | |
| "loss": 2.0167, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.7275204359673027e-05, | |
| "loss": 1.3794, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.591280653950954e-05, | |
| "loss": 1.2342, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 1.455040871934605e-05, | |
| "loss": 1.0997, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 1.3188010899182562e-05, | |
| "loss": 0.9627, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 1.1825613079019073e-05, | |
| "loss": 0.9457, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 1.0463215258855586e-05, | |
| "loss": 0.9263, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 9.1008174386921e-06, | |
| "loss": 0.7995, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 7.73841961852861e-06, | |
| "loss": 0.7548, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 6.376021798365123e-06, | |
| "loss": 0.7403, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 5.013623978201635e-06, | |
| "loss": 0.7507, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 3.6512261580381475e-06, | |
| "loss": 0.6354, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 2.2888283378746596e-06, | |
| "loss": 0.6395, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 9.264305177111717e-07, | |
| "loss": 0.6381, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "step": 7340, | |
| "total_flos": 1.0353544299880243e+17, | |
| "train_loss": 0.9501987717132152, | |
| "train_runtime": 4143.6655, | |
| "train_samples_per_second": 127.5, | |
| "train_steps_per_second": 1.771 | |
| } | |
| ], | |
| "max_steps": 7340, | |
| "num_train_epochs": 4, | |
| "total_flos": 1.0353544299880243e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |