XGQ123452026apache-2.0
GiTBench training data
A multi-source dataset and benchmark for temporal grounding in robotic manipulation, containing 9 tasks with 160 training episodes each in LeRobot v3 format.
Downloads85
Episodes1440
Likes1
Why This Matters for Physical AI
This dataset enables training vision-language-action models for temporal grounding in robotic manipulation, advancing the ability of embodied AI systems to understand and execute complex manipulation tasks grounded in time.
Technical Profile
- Modalities
- rgblanguage
- Robot Embodiments
- dual-arm
- Environment
- simulation
- Task Types
- manipulation
- Episodes
- 1440
- Data Format
- LeRobot
- Annotation Types
- language_instructionstemporal_grounding
- License
- apache-2.0
Access
Need custom rgb data?
Claru builds purpose-built datasets for simulation applications with dense human annotations and quality assurance.
Request a Sample Pack