CoMP: Continual Multimodal Pre-training for Vision Foundation Models

YitongChen (SII), Lingchen Meng, Wujian Peng, Zuxuan Wu, Yu-Gang Jiang

CoMP: Continual Multimodal Pre-training for Vision Foundation Models: 29 upvotes on Hugging Face Daily Papers, #4 of 34 papers on 2025-03-26. Day-by-day upvote history.

Pre-trained Vision Foundation Models (VFMs) provide strong visual representations for a wide range of applications. In this paper, we continually pre-train prevailing VFMs in a multimodal manner such that they can effortlessly process visual inputs of varying sizes and produce visual representations that are more aligned with language representations, regardless of their original pre-training process. To this end, we introduce CoMP, a carefully designed multimodal pre-training pipeline. CoMP uses a Continual Rotary Position Embedding to support native resolution continual pre-training, and an Alignment Loss between visual and textual features through language prototypes to align multimodal representations. By three-stage training, our VFMs achieve remarkable improvements not only in multimodal understanding but also in other downstream tasks such as classification and segmentation. Remarkably, CoMP-SigLIP achieves scores of 66.7 on ChartQA and 75.9 on DocVQA with a 0.5B LLM, while maintaining an 87.4% accuracy on ImageNet-1K and a 49.5 mIoU on ADE20K under frozen chunk evaluation.

Paper page on Hugging Face · arXiv

Data: hysts-bot-data/daily-papers-stats and the Daily Papers API. Open data: tardellirs/paper-pulse-data. Sister project: Model Pulse, the download history of every model on the Hub.