nur-devcc-by-4.0

Robot Execution Validation

VLABench execution records for SmolVLA and Pi0 policies comprising 320 primary confirmation attempts, 60 family-matched validation extension attempts, and 37 development attempts, with trajectories evaluated at 60 and 200 actions.

Downloads0
Episodes417

Why This Matters for Physical AI

This dataset provides reproducible execution validation records for vision-language-action policies on robotic manipulation tasks, enabling rigorous evaluation and benchmarking of embodied AI models.

Technical Profile

Modalities
rgblanguage
Action Space
language
Environment
simulation
Task Types
manipulationobject_selection
Episodes
417
Data Format
json
Annotation Types
language_instructionsreward_labels
License
cc-by-4.0
Part of the VLABench family

Access

Need custom rgb data?

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

Request a Sample Pack

Related Datasets