Summary of Bapo: Base-anchored Preference Optimization For Overcoming Forgetting in Large Language Models Personalization, by Gihun Lee et al.
BAPO: Base-Anchored Preference Optimization for Overcoming Forgetting in Large Language Models Personalization
by Gihun Lee, Minchan Jeong, Yujin Kim, Hojung Jung, Jaehoon Oh, Sangmook Kim, Se-Young Yun
First submitted to arxiv on: 30 Jun 2024
Categories
- Main: Artificial Intelligence (cs.AI)
- Secondary: Computation and Language (cs.CL); 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 The paper investigates how to optimize Large Language Models (LLMs) for personalized preferences without sacrificing previous knowledge. It reveals that existing methods using KL constraints can lead to significant knowledge loss and misalignment when dealing with diverse user preferences. To address this issue, the authors propose Base-Anchored Preference Optimization (BAPO), a simple yet effective approach that leverages initial reference model responses to minimize forgetting while accommodating personalized alignment. The paper demonstrates the efficacy of BAPO in various setups. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper looks at how to make Large Language Models better match what people like. Right now, these models are good at learning from data, but they’re not very good at understanding what people want. This is a problem because different people have different preferences. The authors found that the methods we use to train these models aren’t working well when we try to make them personalized. To fix this, they came up with a new way of training called Base-Anchored Preference Optimization (BAPO). It’s designed to help the model learn from people’s preferences without forgetting what it already knows. The results show that BAPO works well in different situations. |
Keywords
» Artificial intelligence » Alignment » Optimization
