Datasets:

---

Xue0823

/

ReMindView-Bench

like 0

This PR significantly enhances the dataset card for ReMindView-Bench by:

• Adding task_categories: ['image-text-to-text'] to the metadata, which is crucial for discoverability, aligning with the dataset's nature as a Visual Question Answering benchmark for Vision-Language Models.
• Adding language: - en and relevant tags such as vlm, spatial-reasoning, multi-view, vqa, and cognition for richer indexing and searchability.
• Updating and formalizing the introductory section with a clear dataset title, direct links to the paper (Reasoning Path and Latent State Analysis for Multi-view Visual Spatial Reasoning: A Cognitive Science Perspective) and the GitHub repository (https://github.com/pittisl/ReMindView-Bench), and a brief description of the dataset's purpose.
• Incorporating a new "Dataset Generation" section, based on the GitHub repository's "Quickstart" instructions, to guide users through the process of creating scenes, renders, and QA CSVs for the benchmark.

These changes provide a more complete, discoverable, and user-friendly dataset card on the Hugging Face Hub.