kanashi6

/

UniLIP-1B

like 2

This PR significantly improves the model card for UniLiP by:

• Adding the pipeline_tag: any-to-any to accurately reflect its versatile multimodal capabilities (understanding, generation, and editing), enhancing discoverability on the Hub.
• Specifying library_name: transformers, as the model's config.json indicates compatibility with the Hugging Face Transformers library (e.g., transformers_version, Qwen2Tokenizer), enabling the "how to use" widget on the model page.
• Incorporating key sections from the GitHub README, including a detailed introduction, main results tables, and the BibTeX citation, to provide a more comprehensive overview.
• Adding a quick sample usage snippet for text-to-image generation, directly extracted from the GitHub README's "Quick Start" section, to make the model more immediately usable.

This update makes the model card more informative and user-friendly, without changing the existing arXiv paper link.