Migician: Revealing the Magic of Free-Form Multi-Image Grounding in Multimodal Large Language Models Paper • 2501.05767 • Published 10 days ago • 28