| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 5.0, | |
| "global_step": 3310, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.063444108761328e-06, | |
| "loss": 5.5777, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.8126888217522657e-05, | |
| "loss": 2.7803, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 2.719033232628399e-05, | |
| "loss": 2.5334, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.9305135951661632e-05, | |
| "loss": 2.4384, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.8298086606243708e-05, | |
| "loss": 2.3651, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 2.729103726082578e-05, | |
| "loss": 2.3344, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 2.6283987915407858e-05, | |
| "loss": 2.2344, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 2.527693856998993e-05, | |
| "loss": 2.1806, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.4269889224572005e-05, | |
| "loss": 2.1647, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 2.3262839879154077e-05, | |
| "loss": 2.1694, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 2.2255790533736152e-05, | |
| "loss": 2.1443, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 2.124874118831823e-05, | |
| "loss": 2.1277, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 2.0241691842900302e-05, | |
| "loss": 2.1174, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 1.9234642497482378e-05, | |
| "loss": 2.0509, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 1.822759315206445e-05, | |
| "loss": 2.0195, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 1.7220543806646525e-05, | |
| "loss": 1.9997, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 1.6213494461228603e-05, | |
| "loss": 2.015, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 1.5206445115810675e-05, | |
| "loss": 2.024, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 1.419939577039275e-05, | |
| "loss": 2.0084, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 1.3192346424974824e-05, | |
| "loss": 1.9815, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 1.2185297079556899e-05, | |
| "loss": 1.915, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 1.1178247734138972e-05, | |
| "loss": 1.9194, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 1.0171198388721048e-05, | |
| "loss": 1.9254, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 9.164149043303123e-06, | |
| "loss": 1.928, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 8.157099697885196e-06, | |
| "loss": 1.9235, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 7.1500503524672715e-06, | |
| "loss": 1.9297, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 6.143001007049345e-06, | |
| "loss": 1.8951, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "learning_rate": 5.13595166163142e-06, | |
| "loss": 1.8589, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "learning_rate": 4.1289023162134946e-06, | |
| "loss": 1.8632, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 4.53, | |
| "learning_rate": 3.121852970795569e-06, | |
| "loss": 1.8684, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 2.1148036253776437e-06, | |
| "loss": 1.8738, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 1.107754279959718e-06, | |
| "loss": 1.876, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 1.0070493454179255e-07, | |
| "loss": 1.8678, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "step": 3310, | |
| "total_flos": 5.435050739908608e+16, | |
| "train_loss": 2.1781476905338715, | |
| "train_runtime": 3749.3942, | |
| "train_samples_per_second": 56.498, | |
| "train_steps_per_second": 0.883 | |
| } | |
| ], | |
| "max_steps": 3310, | |
| "num_train_epochs": 5, | |
| "total_flos": 5.435050739908608e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |