Summary of Knowledge Editing on Black-box Large Language Models, by Xiaoshuai Song et al.
Knowledge Editing on Black-box Large Language Models
by Xiaoshuai Song, Zhengyang Wang, Keqing He, Guanting Dong, Yutao Mou, Jinxu Zhao, Weiran Xu
First submitted to arxiv on: 13 Feb 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper introduces knowledge editing (KE) for black-box large language models (LLMs), which is critical for updating specific knowledge without affecting other aspects of the model. The researchers propose a comprehensive evaluation framework to overcome limitations in existing evaluations, which are not applicable to black-box LLMs editing and lack comprehensiveness. To address privacy concerns and style over-editing issues, they introduce a novel postEdit framework that uses downstream post-processing for resolving privacy leaks and fine-grained editing for maintaining textual style consistency. The experimental results demonstrate that postEdit outperforms all baselines and achieves strong generalization, with significant improvements in style retention (average +20.82%). |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper makes it possible to change specific information stored in big language models without messing up other things they know. It’s like updating a memory without changing everything else. The researchers created a new way to test this called postEdit, which helps keep the original writing style and keeps personal information private. |
Keywords
* Artificial intelligence * Generalization
