This PR adds a comprehensive model card for the Guidance-Free Training (GFT) method, which enables efficient visual generation without classifier-free guidance.
The model card now includes:
- A clear description of the paper [Visual Generation Without Guidance](https://huggingface.co/papers/2501.15420).
- Relevant metadata: `pipeline_tag: text-to-image` (ensuring discoverability at https://huggingface.co/models?pipeline_tag=text-to-image) and `library_name: diffusers` (given its application to models like Stable Diffusion 1.5 and DiT).
- A link to the GitHub repository (https://github.com/thu-ml/GFT) for code and further usage examples.
- Visual examples directly from the paper's GitHub repository.
- The academic citation.
Please review and merge this PR to enhance the visibility and usability of this important artifact.
