Zhiyi Lyu

Uncertainty-Aware Tree Search for Efficient LLM Reasoning

Adversarial Test Case Generation via Reinforcement Learning Extends Scaling Laws

Towards Efficient Online Tuning of VLM Agents via Counterfactual Soft Reinforcement Learning

Let's Revise Step-by-Step: A Unified Local Search Framework for Code Generation with LLMs

Efficient LLM Alignment via Hierarchical Coarse-to-Fine Refinement

Q*: Improving Multi-step Reasoning for LLMs with Deliberative Planning