Towards Understanding How Knowledge Evolves in Large Vision-Language Models

• URL: http://arxiv.org/abs/2504.02862v2
• Date: Mon, 07 Apr 2025 11:16:51 GMT
• Title: Towards Understanding How Knowledge Evolves in Large Vision-Language Models
• Authors: Sudong Wang, Yunjian Zhang, Yao Zhu, Jianing Li, Zizhe Wang, Yanwei Liu, Xiangyang Ji,
• Abstract summary: We investigate how multimodal knowledge evolves and eventually induces natural languages in Large Vision-Language Models (LVLMs)<n>We identify two key nodes in knowledge evolution: the critical layers and the mutation layers, dividing the evolution process into three stages: rapid evolution, stabilization, and mutation.<n>Our research is the first to reveal the trajectory of knowledge evolution in LVLMs, providing a fresh perspective for understanding their underlying mechanisms.
• Score: 55.82918299608732
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Large Vision-Language Models (LVLMs) are gradually becoming the foundation for many artificial intelligence applications. However, understanding their internal working mechanisms has continued to puzzle researchers, which in turn limits the further enhancement of their capabilities. In this paper, we seek to investigate how multimodal knowledge evolves and eventually induces natural languages in LVLMs. We design a series of novel strategies for analyzing internal knowledge within LVLMs, and delve into the evolution of multimodal knowledge from three levels, including single token probabilities, token probability distributions, and feature encodings. In this process, we identify two key nodes in knowledge evolution: the critical layers and the mutation layers, dividing the evolution process into three stages: rapid evolution, stabilization, and mutation. Our research is the first to reveal the trajectory of knowledge evolution in LVLMs, providing a fresh perspective for understanding their underlying mechanisms. Our codes are available at https://github.com/XIAO4579/Vlm-interpretability.

Related papers

• Open Problems and a Hypothetical Path Forward in LLM Knowledge Paradigms [14.222205207889543]
  This blog post highlights three critical open problems limiting model capabilities. We propose a hypothetical paradigm based on Contextual Knowledge Scaling. Evidence suggests this approach holds potential to address current shortcomings, serving as our vision for future model paradigms.
  arXiv  Detail & Related papers  (2025-04-09T12:31:25Z)
• How LLMs Learn: Tracing Internal Representations with Sparse Autoencoders [30.36521888592164]
  Large Language Models (LLMs) demonstrate remarkable multilingual capabilities and broad knowledge.<n>We analyze how the information encoded in LLMs' internal representations evolves during the training process.
  arXiv  Detail & Related papers  (2025-03-09T02:13:44Z)
• How Do LLMs Acquire New Knowledge? A Knowledge Circuits Perspective on Continual Pre-Training [92.88889953768455]
  Large Language Models (LLMs) face a critical gap in understanding how they internalize new knowledge.<n>We identify computational subgraphs that facilitate knowledge storage and processing.
  arXiv  Detail & Related papers  (2025-02-16T16:55:43Z)
• EvoWiki: Evaluating LLMs on Evolving Knowledge [72.92365627254063]
  EvoWiki is an evolving dataset designed to reflect knowledge evolution by categorizing information into stable, evolved, and uncharted states. Our results indicate that current models often struggle with evolved knowledge, frequently providing outdated or incorrect responses. EvoWiki provides a robust benchmark for advancing future research on the knowledge evolution capabilities of large language models.
  arXiv  Detail & Related papers  (2024-12-18T08:04:57Z)
• Knowledge Mechanisms in Large Language Models: A Survey and Perspective [88.51320482620679]
  This paper reviews knowledge mechanism analysis from a novel taxonomy including knowledge utilization and evolution.<n>We discuss what knowledge LLMs have learned, the reasons for the fragility of parametric knowledge, and the potential dark knowledge (hypothesis) that will be challenging to address.
  arXiv  Detail & Related papers  (2024-07-22T06:15:59Z)
• A Survey on Self-Evolution of Large Language Models [116.54238664264928]
  Large language models (LLMs) have significantly advanced in various fields and intelligent agent applications. To address this issue, self-evolution approaches that enable LLMs to autonomously acquire, refine, and learn from experiences generated by the model itself are rapidly growing.
  arXiv  Detail & Related papers  (2024-04-22T17:43:23Z)
• When Large Language Models Meet Evolutionary Algorithms: Potential Enhancements and Challenges [50.280704114978384]
  Pre-trained large language models (LLMs) exhibit powerful capabilities for generating natural text. Evolutionary algorithms (EAs) can discover diverse solutions to complex real-world problems.
  arXiv  Detail & Related papers  (2024-01-19T05:58:30Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.