| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 5.0, | |
| "global_step": 2500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4e-05, | |
| "loss": 0.2585, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.9453883495145631, | |
| "eval_loss": 0.15823796391487122, | |
| "eval_runtime": 7.4369, | |
| "eval_samples_per_second": 110.799, | |
| "eval_steps_per_second": 6.992, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1304, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.9550970873786407, | |
| "eval_loss": 0.13580937683582306, | |
| "eval_runtime": 7.422, | |
| "eval_samples_per_second": 111.021, | |
| "eval_steps_per_second": 7.006, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 2e-05, | |
| "loss": 0.0571, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.9526699029126213, | |
| "eval_loss": 0.2397720217704773, | |
| "eval_runtime": 7.3747, | |
| "eval_samples_per_second": 111.733, | |
| "eval_steps_per_second": 7.051, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 1e-05, | |
| "loss": 0.0261, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.9757281553398058, | |
| "eval_loss": 0.10911418497562408, | |
| "eval_runtime": 7.4257, | |
| "eval_samples_per_second": 110.966, | |
| "eval_steps_per_second": 7.003, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.008, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.9684466019417476, | |
| "eval_loss": 0.15623603761196136, | |
| "eval_runtime": 7.4414, | |
| "eval_samples_per_second": 110.732, | |
| "eval_steps_per_second": 6.988, | |
| "step": 2500 | |
| } | |
| ], | |
| "max_steps": 2500, | |
| "num_train_epochs": 5, | |
| "total_flos": 1.05244422144e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |