RLinf

/

RLinf-OpenVLAOFT-LIBERO-130

like 3

Follow

RLinf 89

This PR significantly improves the model card for RLinf/RLinf-OpenVLAOFT-LIBERO-130 by:

• Updating the pipeline_tag from reinforcement-learning to robotics, which accurately reflects the model's application domain in Vision-Language-Action (VLA) tasks for robotics. reinforcement-learning has been moved to the general tags list.
• Adding library_name: transformers to the metadata. Analysis of the config.json and other configuration files shows compatibility with the Hugging Face transformers library (using trust_remote_code=True), which will enable an automated usage snippet on the Hub.
• Adding a prominent link to the paper $\pi_\texttt{RL}$: Online RL Fine-tuning for Flow-based Vision-Language-Action Models for easy access to the research.
• Integrating the comprehensive content from the project's GitHub README, providing richer details about RLinf's features, installation, results, roadmap, and citation information. This content is combined with the specific model description and evaluation results from the original model card.
• Updating all image paths to use either https://huggingface.co/.../resolve/main/ or https://raw.githubusercontent.com/.../main/ to ensure they render correctly on the Hugging Face Hub.
• No sample usage code snippet is included, as the source GitHub repository does not provide a direct, self-contained inference snippet for this specific model, adhering strictly to the guidelines. The existing "How to Use" section, which describes integration with the broader RLinf framework, has been retained.

These changes will greatly enhance the discoverability, clarity, and usability of the model for the community.