X-Former: Unifying Contrastive and Reconstruction Learning for MLLMs

TL;DR

X-Former is a novel transformer module that unifies contrastive and reconstruction learning to enhance visual representations in multimodal large language models, leading to improved detailed visual understanding.

Contribution

It introduces X-Former, a lightweight transformer that combines contrastive and masked image modeling for better visual features in MLLMs, with an innovative interaction mechanism.

Findings

Outperforms existing models on GQA visual reasoning tasks.

Achieves superior results on fine-grained visual perception benchmarks.

Enhances detailed visual understanding in multimodal models.

Abstract

Recent advancements in Multimodal Large Language Models (MLLMs) have revolutionized the field of vision-language understanding by integrating visual perception capabilities into Large Language Models (LLMs). The prevailing trend in this field involves the utilization of a vision encoder derived from vision-language contrastive learning (CL), showing expertise in capturing overall representations while facing difficulties in capturing detailed local patterns. In this work, we focus on enhancing the visual representations for MLLMs by combining high-frequency and detailed visual representations, obtained through masked image modeling (MIM), with semantically-enriched low-frequency representations captured by CL. To achieve this goal, we introduce X-Former which is a lightweight transformer module designed to exploit the complementary strengths of CL and MIM through an innovative interaction mechanism. Specifically, X-Former first bootstraps vision-language representation learning and multimodal-to-multimodal generative learning from two frozen vision encoders, i.e., CLIP-ViT (CL-based) and MAE-ViT (MIM-based). It further bootstraps vision-to-language generative learning from a frozen LLM to ensure visual features from X-Former can be interpreted by the LLM. To demonstrate the effectiveness of our approach, we assess its performance on tasks demanding detailed visual understanding. Extensive evaluations indicate that X-Former excels in visual reasoning tasks involving both structural and semantic categories in the GQA dataset. Assessment on fine-grained visual perception benchmark further confirms its superior capabilities in visual understanding.