| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 1.0, |
| "global_step": 2065, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.1, |
| "learning_rate": 4.527845036319613e-05, |
| "loss": 1.6102, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 4.043583535108959e-05, |
| "loss": 1.0235, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 3.559322033898305e-05, |
| "loss": 0.9512, |
| "step": 600 |
| }, |
| { |
| "epoch": 0.39, |
| "learning_rate": 3.0750605326876516e-05, |
| "loss": 0.9026, |
| "step": 800 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 2.5907990314769975e-05, |
| "loss": 0.873, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.58, |
| "learning_rate": 2.106537530266344e-05, |
| "loss": 1.253, |
| "step": 1200 |
| }, |
| { |
| "epoch": 0.68, |
| "learning_rate": 1.6222760290556903e-05, |
| "loss": 0.8331, |
| "step": 1400 |
| }, |
| { |
| "epoch": 0.77, |
| "learning_rate": 1.1380145278450363e-05, |
| "loss": 1.5833, |
| "step": 1600 |
| }, |
| { |
| "epoch": 0.87, |
| "learning_rate": 6.537530266343826e-06, |
| "loss": 0.7863, |
| "step": 1800 |
| }, |
| { |
| "epoch": 0.97, |
| "learning_rate": 1.6949152542372882e-06, |
| "loss": 0.7779, |
| "step": 2000 |
| }, |
| { |
| "epoch": 1.0, |
| "eval_HasAns_exact": 78.34008097165992, |
| "eval_HasAns_f1": 84.67679499790667, |
| "eval_HasAns_total": 5928, |
| "eval_NoAns_exact": 79.47855340622371, |
| "eval_NoAns_f1": 79.47855340622371, |
| "eval_NoAns_total": 5945, |
| "eval_best_exact": 78.9101322327971, |
| "eval_best_exact_thresh": 0.0, |
| "eval_best_f1": 82.07395272867763, |
| "eval_best_f1_thresh": 0.0, |
| "eval_exact": 78.9101322327971, |
| "eval_f1": 82.07395272867774, |
| "eval_total": 11873, |
| "step": 2065 |
| }, |
| { |
| "epoch": 1.0, |
| "step": 2065, |
| "total_flos": 2.697374883446784e+16, |
| "train_loss": 1.0508381767365316, |
| "train_runtime": 1311.9498, |
| "train_samples_per_second": 100.713, |
| "train_steps_per_second": 1.574 |
| } |
| ], |
| "max_steps": 2065, |
| "num_train_epochs": 1, |
| "total_flos": 2.697374883446784e+16, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
