BPO: Staying Close to the Behavior LLM Creates Better Online LLM Alignment

Wenda Xu, Jiachen Li, William Yang Wang, Lei Li


Abstract
Direct alignment from preferences (DAP) has emerged as a promising paradigm for aligning large language models (LLMs) to human desiderata from pre-collected, offline preference datasets. While recent studies indicate that existing offline DAP methods can directly benefit from online training samples, we highlight the need to develop specific online DAP algorithms to fully harness the power of online training. Specifically, we identify that the learned LLM should adhere to the proximity of the behavior LLM, which collects the training samples. To this end, we propose online Preference Optimization in proximity to the Behavior LLM (BPO), emphasizing the importance of constructing a proper trust region for LLM alignment.We conduct extensive experiments to validate the effectiveness and applicability of our approach by integrating it with various DAP methods, resulting in significant performance improvements across a wide range of tasks when training with the same amount of preference data. Even when only introducing one additional data collection phase, our online BPO improves its offline DAP baseline from 72.0% to 80.2% on TL;DR and from 82.2% to 89.1% on Anthropic Helpfulness in terms of win rate against human reference text.
Anthology ID:
2024.emnlp-main.623
Volume:
Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing
Month:
November
Year:
2024
Address:
Miami, Florida, USA
Editors:
Yaser Al-Onaizan, Mohit Bansal, Yun-Nung Chen
Venue:
EMNLP
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
11125–11139
Language:
URL:
https://aclanthology.org/2024.emnlp-main.623
DOI:
Bibkey:
Cite (ACL):
Wenda Xu, Jiachen Li, William Yang Wang, and Lei Li. 2024. BPO: Staying Close to the Behavior LLM Creates Better Online LLM Alignment. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 11125–11139, Miami, Florida, USA. Association for Computational Linguistics.
Cite (Informal):
BPO: Staying Close to the Behavior LLM Creates Better Online LLM Alignment (Xu et al., EMNLP 2024)
Copy Citation:
PDF:
https://aclanthology.org/2024.emnlp-main.623.pdf