bag100cc-by-4.0

Action Atlas visualization bundle

Precomputed visualization data for the Action Atlas web frontend, including per-layer clustering layouts, feature embeddings, and natural-language descriptions for mechanistic interpretability of vision-language-action models.

Downloads244

Why This Matters for Physical AI

This interpretability resource enables understanding of how vision-language-action models process and represent robotic behaviors, advancing mechanistic interpretability in embodied AI systems.

Technical Profile

Modalities
visionlanguageaction
Annotation Types
language_descriptions
License
cc-by-4.0
Part of the Action Atlas family

Access

Need custom vision data?

Claru builds purpose-built datasets for any environment applications with dense human annotations and quality assurance.

Request a Sample Pack

Related Datasets