Ruoyu Wang

FlowBP's innovative approach to reward backpropagation leads to improved alignment of text-to-image models with human preferences while managing memory and gradient complexities.

Forget slow, complex training: you can now distill diffusion models to just 4 steps and still beat the state-of-the-art in preference alignment, aesthetics, and composition.

LLM-based vulnerability repair can be significantly improved by focusing on root cause analysis, leading to more robust and less superficial patches than current methods.

Deobfuscation just got a whole lot easier: PUSHAN cracks virtualization-obfuscated binaries without relying on brittle trace analysis or expensive symbolic execution.

Instance-specific timestep schedules can significantly boost diffusion model performance, challenging the reliance on global discretization strategies.

Synthetic data from generative AI can mislead statistical inference if used naively, but this paper clarifies the assumptions and pitfalls to avoid, offering a roadmap for principled application.

LLMs can perfectly cluster speakers in overlapping multi-party conversations, enabling near-perfect Joint ASR-Clustering Error Rate in challenging CHiME-9 tasks.

LLMs can navigate more efficiently by retrieving relevant past trajectories and pruning irrelevant actions, even without fine-tuning.