CoMP: Continual Multimodal Pre-training for Vision Foundation Models Paper • 2503.18931 • Published 13 days ago • 29