STELAR-VISION: Self-Topology-Aware Efficient Learning for Aligned Reasoning in Vision

LRMVLMRe-assign community