| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "global_step": 2305, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8915401301518446e-05, | |
| "loss": 0.001, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.7830802603036875e-05, | |
| "loss": 0.0003, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.674620390455532e-05, | |
| "loss": 0.0006, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.5661605206073755e-05, | |
| "loss": 0.0001, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.45770065075922e-05, | |
| "loss": 0.0003, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.349240780911063e-05, | |
| "loss": 0.0003, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.240780911062907e-05, | |
| "loss": 0.0, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.132321041214751e-05, | |
| "loss": 0.0, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.0238611713665944e-05, | |
| "loss": 0.0, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 3.915401301518438e-05, | |
| "loss": 0.0, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 3.8069414316702824e-05, | |
| "loss": 0.0, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 3.698481561822126e-05, | |
| "loss": 0.0, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 3.59002169197397e-05, | |
| "loss": 0.0, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.481561822125813e-05, | |
| "loss": 0.0002, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 3.3731019522776576e-05, | |
| "loss": 0.0, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 3.264642082429501e-05, | |
| "loss": 0.0001, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.156182212581345e-05, | |
| "loss": 0.0, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.0477223427331893e-05, | |
| "loss": 0.0, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 2.9392624728850326e-05, | |
| "loss": 0.0, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 2.8308026030368766e-05, | |
| "loss": 0.0, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 2.7223427331887202e-05, | |
| "loss": 0.0, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 2.6138828633405642e-05, | |
| "loss": 0.0, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.505422993492408e-05, | |
| "loss": 0.0, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.3969631236442515e-05, | |
| "loss": 0.0, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.2885032537960955e-05, | |
| "loss": 0.0, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 2.1800433839479395e-05, | |
| "loss": 0.0, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.0715835140997834e-05, | |
| "loss": 0.0, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.963123644251627e-05, | |
| "loss": 0.0002, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.8546637744034707e-05, | |
| "loss": 0.0002, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.7462039045553147e-05, | |
| "loss": 0.0, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.6377440347071584e-05, | |
| "loss": 0.0001, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.5292841648590023e-05, | |
| "loss": 0.0, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.420824295010846e-05, | |
| "loss": 0.0, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.3123644251626898e-05, | |
| "loss": 0.0004, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.2039045553145336e-05, | |
| "loss": 0.0001, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.0954446854663774e-05, | |
| "loss": 0.0, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 9.869848156182213e-06, | |
| "loss": 0.0001, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 8.78524945770065e-06, | |
| "loss": 0.0, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.70065075921909e-06, | |
| "loss": 0.0, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 6.616052060737528e-06, | |
| "loss": 0.0, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 5.531453362255966e-06, | |
| "loss": 0.0, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.446854663774403e-06, | |
| "loss": 0.0, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.362255965292842e-06, | |
| "loss": 0.0, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 2.27765726681128e-06, | |
| "loss": 0.0001, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.193058568329718e-06, | |
| "loss": 0.0, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.0845986984815619e-07, | |
| "loss": 0.0, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 2305, | |
| "total_flos": 1.828765247127552e+16, | |
| "train_loss": 9.778687784283113e-05, | |
| "train_runtime": 1221.5492, | |
| "train_samples_per_second": 120.738, | |
| "train_steps_per_second": 1.887 | |
| } | |
| ], | |
| "max_steps": 2305, | |
| "num_train_epochs": 1, | |
| "total_flos": 1.828765247127552e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |