Xiaoran Fan

Critique-RL: Training Language Models For Critiquing Through Two-Stage Reinforcement Learning

Why Reinforcement Fine-Tuning Enables MLLMs Preserve Prior Knowledge Better: A Data Perspective

MDAR: A Multi-scene Dynamic Audio Reasoning Benchmark

Have the VLMs Lost Confidence? A Study of Sycophancy in VLMs

Distill Visual Chart Reasoning Ability from LLMs to MLLMs