Zi-Yang Rao

The Geometric Nature and a Free Proxy for Flow-Matching Uncertainty

Source-Lifted Flow Matching for Intervenable Multimodal Imitation

Flow-matching policies are promising for imitation learning because they model complex multimodal action distributions. However, their stochasticity is largely passive: repeated sampling may yield diverse behaviors, but users cannot directly choose among valid continuations from the same state. We propose Source-Lifted...

Towards Mechanistically Understanding Why Memorized Knowledge Fails to Generalize in Large Language Model Finetuning

Self-patching identifies activation locations where relocating representations substantially improves failed generalization cases and design a simple heuristic strategy which recovers 58--75\% of the oracle headroom in generalization failure.

How Should Vision-Language-Action Models Use Proprioceptive State?

Five representative interfaces are implemented -- discrete state prompt, VLM prefix, action prefix, state expert, and feature modulation -- under matched implementation details, and evaluated on 45 atomic tasks spanning three task families plus 20 composite tasks.