|
| 1 | + |
| 2 | +{ |
| 3 | + "output_dir": "T5-MSAizer-full", |
| 4 | + "dataset_name": "Murhaf/dialect_msa_silver_parallel", |
| 5 | + "model_name": "UBC-NLP/AraT5v2-base-1024", |
| 6 | + "model_revision": "main", |
| 7 | + "source_prefix": "ترجمة للفصحى: ", |
| 8 | + |
| 9 | + "source_column": "source", |
| 10 | + "target_column": "target", |
| 11 | + |
| 12 | + "overwrite_output_dir": true, |
| 13 | + "do_train": true, |
| 14 | + "do_eval": true, |
| 15 | + "do_predict": false, |
| 16 | + |
| 17 | + "max_source_length": 1024, |
| 18 | + "max_target_length": 1024, |
| 19 | + "generation_max_length": 512, |
| 20 | + |
| 21 | + "learning_rate": 2e-5, |
| 22 | + "weight_decay": 0.0001, |
| 23 | + "warmup_ratio": 0.05, |
| 24 | + "gradient_accumulation_steps": 1, |
| 25 | + "num_train_epochs": 1, |
| 26 | + "per_device_train_batch_size": 32, |
| 27 | + "per_device_eval_batch_size": 32, |
| 28 | + |
| 29 | + "evaluation_strategy": "steps", |
| 30 | + "eval_steps": 0.1, |
| 31 | + "save_total_limit": 3, |
| 32 | + "predict_with_generate": true, |
| 33 | + "hub_strategy": "end", |
| 34 | + "hub_private_repo": true, |
| 35 | + "report_to": "wandb", |
| 36 | + "logging_steps": 1, |
| 37 | + |
| 38 | + "push_to_hub": true, |
| 39 | + "token": "", |
| 40 | + "hub_token": "", |
| 41 | + "per_gpu_train_batch_size": null, |
| 42 | + "per_gpu_eval_batch_size": null |
| 43 | + } |
0 commit comments