Isaac 0.5
Isaac 0.5 is a 36 billion parameter sparse embodied foundation model that reads images, video, language instructions, robot state, and previous actions. It answers questions about video, points to objects and parts, tracks them over time, reports task progress, and generates robot actions for policy fine tuning or use inside a planner or controller.
- Type
- Multimodal
- Weights
- Closed
- Downloads
- 578
- Likes
- 52
- Pipeline
- robotics
- License
- apache-2.0
Model hub
PerceptronAI/Isaac-0.5
- Repo
- PerceptronAI/Isaac-0.5
- Access
- Public
Technical tags and languages
More from Perceptron
3
About Perceptron
1
A decentralised AI data network that lets contributors monetise idle bandwidth and data via a node network, and lets AI companies commission datasets through a data-questing platform.