Datasets:

---

Abdrah

/

warehouse-rgbd-smolRGPT

like 1

This PR significantly enhances the dataset card for the SmolRGPT dataset by adding:

• Metadata:
  • language: en to specify the language of the dataset and its documentation.
  • task_categories: ['image-text-to-text'] to accurately categorize the dataset for vision-language tasks.
  • Relevant tags such as robotics, spatial-reasoning, warehouse, depth-perception, multimodal, and vision-language-model to improve discoverability.
• Links: Direct links to the associated paper (SmolRGPT: Efficient Spatial Reasoning for Warehouse Environments with 600M Parameters) and the GitHub repository (https://github.com/abtraore/SmolRGPT) for easy access to related resources.
• Abstract: The abstract of the associated paper, providing a concise overview of the dataset's context and purpose.
• Dataset Overview: A section describing the dataset's role in spatial reasoning for warehouse environments.
• Data Download and Preparation: Specific instructions for cloning the dataset and preparing the RGB and Depth images, extracted directly from the GitHub repository.
• Sample Usage: Code snippets demonstrating how to generate results using the SmolRGPT model across its three stages, making it easier for users to get started.
• Citation and Acknowledgement: Sections for citation and acknowledgements, as found in the GitHub README.

These updates make the dataset card much more informative and accessible to the community.