| uv run python tools/preprocess_multiproc.py ^ | |
| --manifest dataset/thaitts_emilia_dataset/forbpe.jsonl ^ | |
| --output-dir omniman2_data ^ | |
| --tokenizer checkpoints/thai3normalbpe.model ^ | |
| --config checkpoints/config.yaml ^ | |
| --gpt-checkpoint checkpoints/gpt.pth ^ | |
| --language th ^ | |
| --device cuda ^ | |
| --batch-size 1 ^ | |
| --workers 1 ^ | |
| --num-processes 1 ^ | |
| --launch-delay 20 ^ | |
| --skip-existing ^ | |
| --val-ratio 0.01 | |