CoMP: Continual Multimodal Pre-training for Vision Foundation Models Paper • 2503.18931 • Published 18 days ago • 29 • 1