English
Related papers

Related papers: Interpreting Attention Heads for Image-to-Text Inf…

200 papers

Recent advancements in dialogue systems have highlighted the significance of integrating multimodal responses, which enable conveying ideas through diverse modalities rather than solely relying on text-based interactions. This enrichment…

Computation and Language · Computer Science 2024-07-08 Chang-Sheng Kao , Yun-Nung Chen

While large text-to-image models are able to synthesize "novel" images, these images are necessarily a reflection of the training data. The problem of data attribution in such models -- which of the images in the training set are most…

Computer Vision and Pattern Recognition · Computer Science 2023-08-09 Sheng-Yu Wang , Alexei A. Efros , Jun-Yan Zhu , Richard Zhang

Vision-language models (VLMs) excel at extracting and reasoning about information from images. Yet, their capacity to leverage internal knowledge about specific entities remains underexplored. This work investigates the disparity in model…

Computation and Language · Computer Science 2026-01-06 Ido Cohen , Daniela Gottesman , Mor Geva , Raja Giryes

This paper presents VideoStreaming, an advanced vision-language large model (VLLM) for video understanding, that capably understands arbitrary-length video with a constant number of video tokens streamingly encoded and adaptively selected.…

Computer Vision and Pattern Recognition · Computer Science 2024-05-28 Rui Qian , Xiaoyi Dong , Pan Zhang , Yuhang Zang , Shuangrui Ding , Dahua Lin , Jiaqi Wang

Induction head mechanism is a part of the computational circuits for in-context learning (ICL) that enable large language models (LLMs) to adapt to new tasks without fine-tuning. Most existing work explains the training dynamics behind…

Computation and Language · Computer Science 2025-07-09 Shuo Wang , Issei Sato

Multi-modal Large Language Models (MLLMs) have demonstrated remarkable capabilities in understanding and generating content across various modalities, such as images and text. However, their interpretability remains a challenge, hindering…

Computer Vision and Pattern Recognition · Computer Science 2024-05-29 Loris Giulivi , Giacomo Boracchi

Large Language Models (LLMs) tend to attend heavily to the first token in the sequence -- creating a so-called attention sink. Many works have studied this phenomenon in detail, proposing various ways to either leverage or alleviate it.…

Visual attention plays an important role to understand images and demonstrates its effectiveness in generating natural language descriptions of images. On the other hand, recent studies show that language associated with an image can steer…

Computer Vision and Pattern Recognition · Computer Science 2016-12-13 Jonghwan Mun , Minsu Cho , Bohyung Han

Vision Large Language Models (VLLMs) usually take input as a concatenation of image token embeddings and text token embeddings and conduct causal modeling. However, their internal behaviors remain underexplored, raising the question of…

Computation and Language · Computer Science 2025-05-16 Houjing Wei , Yuting Shi , Naoya Inoue

Despite the impressive capabilities of Multimodal Large Language Models (MLLMs) in integrating text and image modalities, challenges remain in accurately interpreting detailed visual elements. Vision detection models excel at recognizing…

Computer Vision and Pattern Recognition · Computer Science 2024-12-20 Qirui Jiao , Daoyuan Chen , Yilun Huang , Yaliang Li , Ying Shen

Large-scale diffusion models have made significant advances in image generation, particularly through cross-attention mechanisms. While cross-attention has been well-studied in text-to-image tasks, their interpretability in image-to-image…

Computer Vision and Pattern Recognition · Computer Science 2025-03-21 Junseo Park , Hyeryung Jang

In the Transformer model, "self-attention" combines information from attended embeddings into the representation of the focal embedding in the next layer. Thus, across layers of the Transformer, information originating from different tokens…

Machine Learning · Computer Science 2020-06-02 Samira Abnar , Willem Zuidema

Vision Language Models (VLMs) excel at identifying and describing objects but often fail at spatial reasoning. We study why VLMs, such as LLaVA, underutilize spatial cues despite having positional encodings and spatially rich vision encoder…

Computer Vision and Pattern Recognition · Computer Science 2025-10-02 Jianing Qi , Jiawei Liu , Hao Tang , Zhigang Zhu

Multimodal Large Language Models (MLLMs) have achieved remarkable progress in multimodal reasoning. However, they often excessively rely on textual information during the later stages of inference, neglecting the crucial integration of…

Computer Vision and Pattern Recognition · Computer Science 2025-07-08 Shuo Yang , Yuwei Niu , Yuyang Liu , Yang Ye , Bin Lin , Li Yuan

Multimodal Large Language Models (MLLMs) have significantly improved performance across various image-language applications. Recently, there has been a growing interest in adapting image pre-trained MLLMs for video-related tasks. However,…

Computer Vision and Pattern Recognition · Computer Science 2024-09-06 Mingze Gao , Jingyu Liu , Mingda Li , Jiangtao Xie , Qingbin Liu , Bo Zhao , Xi Chen , Hui Xiong

Flowcharts are typically presented as images, driving the trend of using vision-language models (VLMs) for end-to-end flowchart understanding. However, two key challenges arise: (i) Limited controllability--users have minimal influence over…

Computer Vision and Pattern Recognition · Computer Science 2024-12-24 Junyi Ye , Ankan Dash , Wenpeng Yin , Guiling Wang

Large language models (LLMs) exhibit strong in-context learning capabilities, but how they track and retrieve information from context remains underexplored. Drawing on the free recall paradigm in cognitive science (where participants…

Computation and Language · Computer Science 2026-04-02 Anooshka Bajaj , Deven Mahesh Mistry , Sahaj Singh Maini , Yash Aggarwal , Billy Dickson , Zoran Tiganj

The current large language models are mainly based on decode-only structure transformers, which have great in-context learning (ICL) capabilities. It is generally believed that the important foundation of its ICL capability is the induction…

Computation and Language · Computer Science 2024-12-06 Mingyu Xu , Wei Cheng , Bingning Wang , Weipeng Chen

Vision-Language Models (VLMs) have been shown to be blind, often underutilizing their visual inputs even on tasks that require visual reasoning. In this work, we demonstrate that VLMs are selectively blind. They modulate the amount of…

Computer Vision and Pattern Recognition · Computer Science 2026-03-23 Wan-Cyuan Fan , Jiayun Luo , Declan Kutscher , Leonid Sigal , Ritwik Gupta

Large Language Models (LLMs) are becoming increasingly popular in pervasive computing due to their versatility and strong performance. However, despite their ubiquitous use, the exact mechanisms underlying their outstanding performance…

Computation and Language · Computer Science 2026-02-02 Alhassan Abdelhalim , Janick Edinger , Sören Laue , Michaela Regneri