Towards Measuring Representational Similarity of Large Language Models

Abstract

Understanding the similarity of the numerous large language models released has many uses, e.g., simplifying model selection, detecting illegal model reuse, and advancing our understanding of what makes LLMs perform well. In this work, we measure the similarity of representations of a set of LLMs with 7B parameters. Our results suggest that some LLMs are substantially different from others. We identify challenges of using representational similarity measures that suggest the need of careful study of similarity scores to avoid false conclusions.

PDF NeurIPSW OpenReview Semantic Scholar

Cite

Text

Klabunde et al. "Towards Measuring Representational Similarity of Large Language Models." NeurIPS 2023 Workshops: UniReps, 2023.

Markdown

[Klabunde et al. "Towards Measuring Representational Similarity of Large Language Models." NeurIPS 2023 Workshops: UniReps, 2023.](https://mlanthology.org/neuripsw/2023/klabunde2023neuripsw-measuring/)

BibTeX

@inproceedings{klabunde2023neuripsw-measuring,
  title     = {{Towards Measuring Representational Similarity of Large Language Models}},
  author    = {Klabunde, Max and Amor, Mehdi Ben and Granitzer, Michael and Lemmerich, Florian},
  booktitle = {NeurIPS 2023 Workshops: UniReps},
  year      = {2023},
  url       = {https://mlanthology.org/neuripsw/2023/klabunde2023neuripsw-measuring/}
}