| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.861003861003861, | |
| "eval_steps": 1000, | |
| "global_step": 2000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 7.239382239382239e-07, | |
| "loss": 2.3335, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 1.4478764478764478e-06, | |
| "loss": 2.322, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 2.171814671814672e-06, | |
| "loss": 2.268, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "eval_loss": 2.2363617420196533, | |
| "eval_runtime": 56.2328, | |
| "eval_samples_per_second": 51.98, | |
| "eval_steps_per_second": 3.254, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 2.8957528957528956e-06, | |
| "loss": 2.2427, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 3.61969111969112e-06, | |
| "loss": 2.2085, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 4.343629343629344e-06, | |
| "loss": 2.1902, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "eval_loss": 2.1604719161987305, | |
| "eval_runtime": 56.4281, | |
| "eval_samples_per_second": 51.8, | |
| "eval_steps_per_second": 3.243, | |
| "step": 2000 | |
| } | |
| ], | |
| "logging_steps": 300, | |
| "max_steps": 8288, | |
| "num_train_epochs": 16, | |
| "save_steps": 1000, | |
| "total_flos": 1.733060194257961e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |