This project is to develop a LLM-based platform, named RobRun, aiming to upgrade a device with promptable intelligence (i.e., the device can respond to prompts or instructions given by users, and adapt to diverse tasks). RobRun includes modules of multi-modality perception encoder, LLM-based Agent, LLM inference system, database, and the underlying hardware.
Related publications and tools:
- ACL’24 “BitDistiller: Unleashing the Potential of Sub-4-Bit LLMs via Self-Distillation” (opens in new tab)
https://github.com/DD-DuDa/BitDistiller (opens in new tab) - EuroSys’25 “T-MAC: CPU Renaissance via Table Lookup for Low-Bit LLM Deployment on Edge (opens in new tab)”
https://github.com/microsoft/T-MAC (opens in new tab) - arXiv “LUT TENSOR CORE: Lookup Table Enables Efficient Low-Bit LLM Inference Acceleration (opens in new tab)”
- arXiv “Advancing Multi-Modal Sensing Through Expandable Modality Alignment (opens in new tab)”
- arXiv “Making Every Frame Matter: Continuous Video Understanding for Large Models via Adaptive State Modeling (opens in new tab)”