Datasets:

---

aimagelab

/

ReT-M2KR

like 2

Follow

AImageLab 23

This PR significantly updates the dataset card for aimagelab/ReT-M2KR to align it with the latest research paper, Recurrence Meets Transformers for Universal Multimodal Retrieval (ReT-2), and its corresponding GitHub repository.

Key changes include:

• Adding visual-document-retrieval as a task_category and relevant tags to the metadata for better discoverability.
• Updating the introductory description to clearly reference ReT-2.
• Replacing the outdated paper and code repository links with the correct ones for ReT-2:
  • Paper: Recurrence Meets Transformers for Universal Multimodal Retrieval
  • Code: https://github.com/aimagelab/ReT-2
• Incorporating a "Sample Usage" section with a Python code snippet directly from the ReT-2 GitHub README, demonstrating how to use ReT-2 with Hugging Face's Transformers.
• Updating the BibTeX citation to reflect the ReT-2 paper.

These improvements provide comprehensive and up-to-date information, making the dataset more accessible and useful for the community.