CoMP: Continual Multimodal Pre-training for Vision Foundation Models Paper • 2503.18931 • Published 29 days ago • 30