Add dataset card, task category, and paper/code links
#2
by nielsr HF Staff - opened
This PR improves the dataset card by adding:
- The
image-text-to-texttask category and license metadata. - Links to the research paper "Does Seeing More Mean Knowing More? Mono-Anchored Advantage Normalization for Multi-Source Visual Reasoning" and its official GitHub repository.
- A summary of the MARS framework and the multi-source reinforcement fine-tuning data it provides (Infrared, Depth, Multi-view, and Text-rich).
- Proper citation information.