paper-with-me

홈 › Papers

Mementos: A Comprehensive Benchmark for Multimodal Large Language Model Reasoning over Image Sequences

2024-01-19 · Xiyao Wang, YuHang Zhou, Xiaoyu Liu, Hongjin Lu, Yuancheng Xu, Feihong He, Jaehong Yoon, Taixi Lu, Gedas Bertasius, Mohit Bansal, Huaxiu Yao, Furong Huang

Multimodal Large Language Models (MLLMs) have demonstrated proficiency in handling a variety of visual-language tasks. However, current MLLM benchmarks are predominantly designed to evaluate reasoning based on static information about a single image, and the ability of modern MLLMs to extrapolate from image sequences, which is essential for understanding our ever-changing world, has been less investigated. To address this challenge, this paper introduces Mementos, a new benchmark designed to assess MLLMs' sequential image reasoning abilities. Mementos features 4,761 diverse image sequences with varying lengths. We also employ a GPT-4 assisted method to evaluate MLLM reasoning performance. Through a careful evaluation of nine recent MLLMs on Mementos, including GPT-4V and Gemini, we find that they struggle to accurately describe dynamic information about given image sequences, often leading to hallucinations/misrepresentations of objects and their corresponding behaviors. Our quantitative analysis and case studies identify three key factors impacting MLLMs' sequential image reasoning: the correlation between object and behavioral hallucinations, the influence of cooccurring behaviors, and the compounding impact of behavioral hallucinations. Our dataset is available at https://github.com/umd-huang-lab/Mementos.

📄 PDF Abstract BibTeX arXiv:2401.10529

Code (1)

umd-huang-lab/mementos 공식 구현

Tasks

Language ModelingLanguage ModellingLarge Language ModelMultimodal Large Language Model

Methods 이 논문이 사용한 방법론

Attention 설명 없음
Absolute Position Encodings Absolute Position Encodings are a type of position embeddings for [Transformer-based models] where positional encodings are…
Label Smoothing Label Smoothing is a regularization technique that introduces noise for the labels. This accounts for the fact that datasets may have mistakes in them, so maximizing the…
Layer Normalization Unlike batch normalization, Layer Normalization directly estimates the normalization statistics from the summed inputs…
Adam 설명 없음
Residual Connection 설명 없음
Dropout Dropout is a regularization technique for neural networks that drops a unit (along with connections) at training time with a specified probability $p$ (a common value is…
Linear Layer A Linear Layer is a projection $\mathbf{XW + b}$.

Similar Papers 제목 키워드 기반

MEMENTO: Teaching LLMs to Manage Their Own Context

2026-04-10 · Vasilis Kontonis, Yuchen Zeng, Shivam Garg, Lingjiao Chen 외 arxiv

Reasoning models think in long, unstructured streams with no mechanism for compressing or organizing their own intermediate state. We introduce MEMENTO: a method that teaches models to segment reasoning into blocks, comp…

MementoEmbed and Raintale for Web Archive Storytelling

2020-08-01 · Shawn M. Jones, Martin Klein, Michele C. Weigle, Michael L. Nelson

For traditional library collections, archivists can select a representative sample from a collection and display it in a featured physical or digital library space. Web archive collections may consist of thousands of arc…

The Off-Topic Memento Toolkit

2018-06-18 · Shawn M. Jones, Michele C. Weigle, Michael L. Nelson

Web archive collections are created with a particular purpose in mind. A curator selects seeds, or original resources, which are then captured by an archiving system and stored as archived web pages, or mementos. The sys…

Lightweight Visual Reasoning for Socially-Aware Robots

2026-03-04 · Alessio Galatolo, Ronald Cumbal, Alexandros Rouchitsas, Katie Winkle 외 arxiv

Robots operating in shared human environments must not only navigate, interact, and detect their surroundings, they must also interpret and respond to dynamic, and often unpredictable, human behaviours. Although recent a…

Visual Reasoning

MMC: Advancing Multimodal Chart Understanding with Large-scale Instruction Tuning

2023-11-15 · Fuxiao Liu, Xiaoyang Wang, Wenlin Yao, Jianshu Chen 외

With the rapid development of large language models (LLMs) and their integration into large multimodal models (LMMs), there has been impressive progress in zero-shot completion of user-oriented vision-language tasks. How…

Chart Understanding