| { | |
| "output_dir": "./outputs_phi9_5", | |
| "overwrite_output_dir": true, | |
| "evaluation_strategy": "epoch", | |
| "save_strategy": "epoch", | |
| "save_total_limit": 2, | |
| "logging_strategy": "steps", | |
| "logging_steps": 50, | |
| "learning_rate": 2e-5, | |
| "per_device_train_batch_size": 16, | |
| "per_device_eval_batch_size": 16, | |
| "num_train_epochs": 4, | |
| "warmup_ratio": 0.05, | |
| "weight_decay": 0.02, | |
| "gradient_checkpointing": true, | |
| "fp16": true, | |
| "push_to_hub": true, | |
| "hub_model_id": "antonypamo/ProSavantEngine_Phi9_5", | |
| "logging_dir": "./logs_phi9_5", | |
| "report_to": "tensorboard", | |
| "seed": 73 | |
| } | |