add tagging

This commit is contained in:
Tom Young 2024-06-17 14:03:54 +00:00
parent 7fd26ba54f
commit b59f5892dd

View file

@ -65,8 +65,13 @@ python -m tools.datasets.datautil \
--refine-llm-caption \
--remove-empty-caption \
--output ${ROOT_META}/meta_clips_caption_cleaned.csv
# 4.4 Optionally generate tags (e.g., objects) based on the captions. This should output your_output_prefix_{key}.csv
torchrun --nproc_per_node 8 --standalone -m tools.caption.caption_llama3 ${ROOT_META}/meta_clips_caption_cleaned.csv --key objects --output_prefix your_output_prefix
```
For more information, please refer to:
- [Dataset Management](../tools/datasets/README.md)
- [Scene Detection and Video Splitting](../tools/scene_cut/README.md)