zhongyu wei

Associate Professor@Fudan University·中国·OpenReview

研究方向

Text Generation · cross vision and language · computational argumentation · Computational Social Science · Natural language processing · Text Summarization

zhongyu wei

HardcoreLogic: Challenging Large Reasoning Models with Long-tail Logic Puzzle Games

Mixture-of-Visual-Thoughts: Exploring Context-Adaptive Reasoning Mode Selection for General Visual Reasoning

Not All Models Suit Expert Offloading: On Local Routing Consistency of Mixture-of-Expert Models

ExpertWeaver: Unlocking the Inherent MoE in Dense LLMs with GLU Activation Patterns

Autoregressive Semantic Visual Reconstruction Helps VLMs Understand Better

OViP: Online Vision-Language Preference Learning for VLM Hallucination

AutoJudger: An Agent-Driven Framework for Efficient Benchmarking of MLLMs

Incentivizing Multimodal Reasoning via Progressive Curriculum Reinforcement Learning

SpeechMedAssist: Efficiently and Effectively Adapting Speech Language Model for Medical Consultation

Towards Adaptive GUI Agents with Memory-Driven Knowledge Evolution

Think Out Loud, Pause in Silence: Confidence-Guided Reflect–Pause–Abort for Robust Audio Perceptual Understanding

Multi-agent KTO: Enhancing Strategic Interactions of Large Language Model in Language Game

How Jailbreak Defenses Work and Ensemble? A Mechanistic Investigation

ITFormer: Bridging Time Series and Natural Language for Multi-Modal QA with Large-Scale Multitask Dataset