paper-with-me

홈 › Papers

Visual Instruction Tuning Aligns Modalities through Abstraction

2026-06-02 · Luis Palacios, Lorenzo Basile, Diego Doimo, Alberto Cazzaniga arxiv

Visual instruction tuning effectively adapts a pre-trained Large Language Model (LLM) to process image information alongside text. Yet, it remains unclear how visual features are embedded into the layer-wise hierarchy of abstractions of the LLM backbone. Across a diverse set of vision-language architectures, we show that instruction tuning primarily serves as a bridge, embedding visual features directly into the intermediate semantic layers of the LLM, bypassing the early layers devoted to unimodal processing. With probing analyses and causal interventions, we show that these intermediate layers are the semantic core of vision-language processing and play a critical role in the performance on a broad set of multimodal benchmarks. In addition, by comparing the geometry of semantically equivalent visual and textual representations, we find that fine-tuning extends and strengthens the existing abstraction phase, aligning visual features with pre-existing textual ones. Finally, we confirm the functional role of this localized alignment by restricting fine-tuning to intermediate layers alone: this strategy preserves the performance of full fine-tuning on vision-centric benchmarks while reducing training time. Our results suggest that multimodal integration is a localized phenomenon driven by the repurposing of the internal abstraction engine of the LLM.

📄 PDF Abstract BibTeX arXiv:2606.03871

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

LLaVA Steering: Visual Instruction Tuning with 500x Fewer Parameters through Modality Linear Representation-Steering

2024-12-16 · Jinhe Bi, Yujun Wang, Haokun Chen, Xun Xiao 외

Multimodal Large Language Models (MLLMs) have significantly advanced visual tasks by integrating visual representations into large language models (LLMs). The textual modality, inherited from LLMs, equips MLLMs with abil…

In-Context LearningInstruction Followingparameter-efficient fine-tuningQuestion Answering+1

Position-Enhanced Visual Instruction Tuning for Multimodal Large Language Models

2023-08-25 · Chi Chen, Ruoyu Qin, Fuwen Luo, Xiaoyue Mi 외

Recently, Multimodal Large Language Models (MLLMs) that enable Large Language Models (LLMs) to interpret images through visual instruction tuning have achieved significant success. However, existing visual instruction tu…

cross-modal alignmentPosition

ChatBridge: Bridging Modalities with Large Language Model as a Language Catalyst

2023-05-25 · Zijia Zhao, Longteng Guo, Tongtian Yue, Sihan Chen 외

Building general-purpose models that can perceive diverse real-world modalities and solve various tasks is an appealing target in artificial intelligence. In this paper, we present ChatBridge, a novel multimodal language…

Language ModelingLanguage ModellingLarge Language Model

MoAI: Mixture of All Intelligence for Large Language and Vision Models

2024-03-12 · Byung-Kwan Lee, Beomchan Park, Chae Won Kim, Yong Man Ro

The rise of large language models (LLMs) and instruction tuning has led to the current trend of instruction-tuned large language and vision models (LLVMs). This trend involves either meticulously curating numerous instru…

AllMixture-of-ExpertsScene UnderstandingVisual Question Answering

Generalize LMMs to Versatile Visual Modalities via Fabricated Modality Synthesis

2026-07-11 · Shihao Yuan, Yuanze Li, Ruyi Zhang, Ming Liu 외 arxiv

Despite the advancements of Large Multimodal Models (LMMs) in RGB vision, their ability to generalize to unseen visual modalities remains a largely unexplored challenge. We argue that different visual modalities are mere…