SRPO: Enhancing Multimodal LLM Reasoning via Reflection-Aware Reinforcement Learning

LRMRe-assign community