Enhance model card for LeVo: Add metadata, links, usage, and full description
#1
by
nielsr
HF Staff
- opened
This PR significantly enhances the model card for the LeVo (SongGeneration) model by:
- Adding
pipeline_tag: text-to-audioto correctly categorize the model for song generation, improving discoverability on the Hugging Face Hub (e.g., via huggingface.co/models?pipeline_tag=text-to-audio). - Providing a comprehensive description with clear links to:
- The official Hugging Face paper page: LeVo: High-Quality Song Generation with Multi-Preference Alignment
- The project page: https://levo-demo.github.io
- The GitHub repository: https://github.com/tencent-ailab/songgeneration
- The Hugging Face Space demo: https://huggingface.co/spaces/waytan22/SongGeneration-LeVo
- Incorporating detailed sections from the project's GitHub README, including:
- The paper abstract and model overview.
- News and updates.
- Model versions with download links.
- Installation instructions.
- Comprehensive inference instructions (
sh generate.shcommands, input guide, and flags) along with necessary download steps, ensuring accurate and verifiable sample usage. (The problematic Python snippet from the original README, which appears to be for a different model, has been omitted to adhere to usage guidelines.) - Gradio UI launch command.
- Evaluation performance tables.
- Citation information.
- License details.
These updates provide a much richer and more actionable model card, enabling users to better understand and utilize the LeVo model.