Hui Liu

How Far Are LLMs from Professional Poker Players? Revisiting Game-Theoretic Reasoning with Agentic Tool Use

Unlocking the Power of Multi-Agent LLM for Reasoning: From Lazy Agents to Deliberation

Agent-REINFORCE: Searching Compute-Optimal Multi-LLM Collaboration Graph for Test-Time Scaling

TRAJECT-Bench:A Trajectory-Aware Benchmark for Evaluating Agentic Tool Use

Bradley-Terry and Multi-Objective Reward Modeling Are Complementary

Seeing but Not Believing: Probing the Disconnect Between Visual Attention and Answer Correctness in VLMs

Synthesizing High-Quality Visual Question Answering from Medical Documents with Generator-Verifier LMMs

All You Need Are Random Visual Tokens? Demystifying Token Pruning in VLLMs

Where in the World? A Vision-Language Benchmark for Probing Model Geolocation Skills Across Scales

Adaptive Test-Time Reasoning via Reward-Guided Dual-Phase Search

Moralise: A Structured Benchmark for Moral Alignment in Visual Language Models

Keeping an Eye on LLM Unlearning: The Hidden Risk and Remedy

AgentTTS: Large Language Model Agent for Test-time Compute-optimal Scaling Strategy in Complex Tasks

Catastrophic Failure of LLM Unlearning via Quantization

Mitigating Heterogeneous Token Overfitting in LLM Knowledge Editing

Unlocking Efficient, Scalable, and Continual Knowledge Editing with Basis-Level Representation Fine-Tuning

Automatic Task-aware Instruction Optimizer for Black-box LLMs

Divide-Verify-Refine: Aligning LLM Responses with Complex Instructions

Robust Watermarking for Diffusion Models: A Unified Multi-Dimensional Recipe