Inquiry about the availability and accessibility of SUP dataset

[ares89]

Hi DriveVLM team,

First, congratulations on the impressive work of DriveVLM and DriveVLM-Dual! The integration of VLMs with traditional autonomous driving pipelines demonstrates remarkable innovation.

According to the paper and documentation， your team constructed the SUP-AD dataset through a comprehensive data mining and annotation process, which includes:
• Long-tail object mining (e.g., rare obstacles, non-standard traffic elements)
• Challenging scenario mining (complex road layouts, dynamic human behaviors)
• Keyframe selection and scene annotation (with structured labels for scene description, analysis, and planning)

This dataset plays a critical role in evaluating Scene Understanding and Planning (SUP) capabilities. However, we noticed that current resources (GitHub repo, paper appendix) do not mention the dataset's public availability. We would like to kindly ask:

Open-source plan:
Will the SUP-AD dataset be released to the research community? If yes, what is the expected timeline and release format (e.g., full dataset download, API access, or partial samples)?

This dataset would significantly benefit the autonomous driving research community by enabling further exploration of VLM-based scene understanding. We appreciate your efforts in clarifying these details!