Summary of Uoe: Unlearning One Expert Is Enough For Mixture-of-experts Llms, by Haomin Zhuang et al.
UOE: Unlearning One Expert Is Enough For Mixture-of-experts LLMS
by Haomin Zhuang, Yihua Zhang, Kehan Guo, Jinghan Jia, Gaowen Liu, Sijia Liu, Xiangliang Zhang
First submitted to arxiv on: 27 Nov 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
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 MoE-based large language models (LLMs) have gained popularity for their exceptional performance and efficient inference processes. However, these models’ unique characteristics introduce challenges when applying traditional unlearning methods, leading to significant utility drops. A novel single-expert unlearning framework (UOE) is proposed, concentrating on the most actively engaged expert for targeted knowledge while stabilizing its active state through an anchor loss. This framework enhances forget quality by up to 5% and model utility by 35%, while only unlearning 0.06% of model parameters. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary MoE-based large language models are a type of AI that’s really good at understanding and generating human-like text. But when we try to remove unwanted data, it gets tricky. The problem is that these models have many experts that work together, and removing the wrong ones can make them less useful. Scientists came up with a new way to do this called UOE, which focuses on the right expert and helps keep the model working well. |
Keywords
* Artificial intelligence * Inference
