Changelog
- New model: GR00T-N1.5 (
grootn15) _ NVIDIA's state-of-the-art vision-language-action (VLA) model for generalist robot control and dexterous manipulation, now published and available on AI Hub. Runs as a 4-component collection model (ViT, LLM, VLM projector, DiT) on the NPU. - Added GR00T-N1.5 performance data (QNN Context Binary and Precompiled QNN ONNX paths) for Dragonwing IQ-9075 EVK and other devices.