Preprints and workshop papers

1. EBRL: Asynchronous Embodied RL by Multi-Grained Resource Management

Published in arXiv, 2026 | Download Paper

2. Zeva-Ego: Egocentric Mid-Training with In-Context Causal Learning for Robot Manipulation

Published in arXiv, 2026 | Download Paper

3. Zeva: In-Context Causal Learning for Generalizable Embodied Manipulation

Published in arXiv, 2026 | Download Paper

4. Zetta ζ: An Efficient Closed-Loop Embodied Harness for Self-Evolving Physical Intelligence

Published in arXiv, 2026 | Download Paper

5. Embodied.cpp: A Portable Inference Runtime of Embodied AI Models on Heterogeneous Robots

Published in arXiv, 2026 | Download Paper

6. ActProbe: Action-Space Probe for Early Failure Detection of Generative Robot Policies

Published in arXiv, 2026 | Download Paper

7. GRIP-VLM: Group-Relative Importance Pruning for Efficient Vision-Language Models

Published in arXiv, 2026 | Download Paper

8. EmbodiSkill: Skill-Aware Reflection for Self-Evolving Embodied Agents

Published in arXiv, 2026 | Download Paper

9. MemCompiler: Compile, Don’t Inject – State-Conditioned Memory for Embodied Agents

Published in arXiv, 2026 | Download Paper

10. SwarmThinkers: Learning Physically Consistent Atomic KMC Transitions at Scale

Published in arXiv, 2025 | Download Paper

11. Scaling Up On-Device LLMs via Active-Weight Swapping Between DRAM and Flash

Published in arXiv, 2025 | Download Paper

12. MVDRAM: Enabling GeMV Execution in Unmodified DRAM for Low-Bit LLM Acceleration

Published in arXiv, 2025 | Download Paper

13. PUDTune: Multi-Level Charging for High-Precision Calibration in Processing-Using-DRAM

Published in IEEE Computer Architecture Letters, 2025 | Download Paper

14. Bulk Bitwise Accumulation in Commercial DRAM

Published in NeurIPS 2024 Workshop Machine Learning with new Compute Paradigms, 2024 | Download Paper

15. Hybrid SLM and LLM for Edge-Cloud Collaborative Inference

Published in EdgeFM'24 Workshop (Colocated with MobiCom'24), 2024 | Download Paper