Browse publications on Google Scholar (top-right) ↗
Resume (English only)
Academic Achievements
Co-first author of 'From Passive to Active Reasoning: Can Large Language Models Ask the Right Questions under Incomplete Information?', ICML 2025.
Co-first author of 'From Debate to Equilibrium: Belief-Driven Multi-Agent LLM Reasoning via Bayesian Nash Equilibrium', ICML 2025.
Co-first author of 'Landscape of Thoughts: Visualizing the Reasoning Process of Large Language Models', ICLR/ICML 2025 Workshop.
Author of 'Rethinking LLM Unlearning Objectives: A Gradient Perspective and Go Beyond', ICLR 2025.
Author of 'Noisy Test-Time Adaptation in Vision-Language Models', ICLR 2025.
Author of 'Can Language Models Perform Robust Reasoning in Chain-of-thought Prompting with Noisy Rationales?', NeurIPS 2024.
Led or contributed to multiple projects and benchmarks including AR-Bench, Landscape of Thoughts, ECON, NoRa, DeepInception, G-effect, NTTA, EOE, GRA, RGIB, Subgraph, Neural Atoms, AdaProp, and KGTuner.
Background
PhD student at the TMLR group, Hong Kong Baptist University, focusing on Trustworthy Machine Reasoning.
Currently a visiting student at STAIR Lab, Stanford University, working with Prof. Sanmi Koyejo.
Previously a visiting student at LARS group, Tsinghua University, working with Prof. Quanming Yao and Prof. Yongqi Zhang.
Research focuses on trustworthy machine reasoning with foundation models (LLMs, VLMs) to solve complex problems like mathematics and coding, and to accelerate scientific discovery in biology, chemistry, and healthcare.
Believes reasoning is the essential pathway to achieving AGI; trustworthy reasoning includes reasoning power, robustness, safety, and explainability.
Works on developing advanced reasoning systems (with RL and tool learning), comprehensive evaluation benchmarks, and trustworthy reasoning on (knowledge) graphs.