SeeingEye: Agentic Information Flow Unlocks Multimodal Reasoning In Text-only LLMs

Open in new window