SAILViT: Towards Robust and Generalizable Visual Backbones for MLLMs via Gradual Feature Refinement

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Yin, Weijie, Yang, Dingkang, Dong, Hongyuan, Kang, Zijian, Wang, Jiacong, Liang, Xiao, Feng, Chao, Ran, Jiao
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916822261956608
author Yin, Weijie
Yang, Dingkang
Dong, Hongyuan
Kang, Zijian
Wang, Jiacong
Liang, Xiao
Feng, Chao
Ran, Jiao
author_facet Yin, Weijie
Yang, Dingkang
Dong, Hongyuan
Kang, Zijian
Wang, Jiacong
Liang, Xiao
Feng, Chao
Ran, Jiao
contents Vision Transformers (ViTs) are essential as foundation backbones in establishing the visual comprehension capabilities of Multimodal Large Language Models (MLLMs). Although most ViTs achieve impressive performance through image-text pair-based contrastive learning or self-supervised mechanisms, they struggle to engage in connector-based co-training directly with LLMs due to potential parameter initialization conflicts and modality semantic gaps. To address the above challenges, this paper proposes SAILViT, a gradual feature learning-enhanced ViT for facilitating MLLMs to break through performance bottlenecks in complex multimodal interactions. SAILViT achieves coarse-to-fine-grained feature alignment and world knowledge infusion with gradual feature refinement, which better serves target training demands. We perform thorough empirical analyses to confirm the powerful robustness and generalizability of SAILViT across different dimensions, including parameter sizes, model architectures, training strategies, and data scales. Equipped with SAILViT, existing MLLMs show significant and consistent performance improvements on the OpenCompass benchmark across extensive downstream tasks. SAILViT series models are released at https://huggingface.co/BytedanceDouyinContent.
format Preprint
id arxiv_https___arxiv_org_abs_2507_01643
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle SAILViT: Towards Robust and Generalizable Visual Backbones for MLLMs via Gradual Feature Refinement
Yin, Weijie
Yang, Dingkang
Dong, Hongyuan
Kang, Zijian
Wang, Jiacong
Liang, Xiao
Feng, Chao
Ran, Jiao
Computer Vision and Pattern Recognition
Vision Transformers (ViTs) are essential as foundation backbones in establishing the visual comprehension capabilities of Multimodal Large Language Models (MLLMs). Although most ViTs achieve impressive performance through image-text pair-based contrastive learning or self-supervised mechanisms, they struggle to engage in connector-based co-training directly with LLMs due to potential parameter initialization conflicts and modality semantic gaps. To address the above challenges, this paper proposes SAILViT, a gradual feature learning-enhanced ViT for facilitating MLLMs to break through performance bottlenecks in complex multimodal interactions. SAILViT achieves coarse-to-fine-grained feature alignment and world knowledge infusion with gradual feature refinement, which better serves target training demands. We perform thorough empirical analyses to confirm the powerful robustness and generalizability of SAILViT across different dimensions, including parameter sizes, model architectures, training strategies, and data scales. Equipped with SAILViT, existing MLLMs show significant and consistent performance improvements on the OpenCompass benchmark across extensive downstream tasks. SAILViT series models are released at https://huggingface.co/BytedanceDouyinContent.
title SAILViT: Towards Robust and Generalizable Visual Backbones for MLLMs via Gradual Feature Refinement
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2507.01643