Aligners: Decoupling LLMs and Alignment

Lilian Ngweta, Mayank Agarwal, Subha Maity, Alex Gittens, Yuekai Sun, Mikhail Yurochkin


Abstract
Large Language Models (LLMs) need to be aligned with human expectations to ensure their safety and utility in most applications. Alignment is challenging, costly, and needs to be repeated for every LLM and alignment criterion. We propose to decouple LLMs and alignment by training *aligner* models that can be used to align any LLM for a given criteria on an as-needed basis, thus also reducing the potential negative impacts of alignment on performance. Our recipe for training the aligner models solely relies on synthetic data generated with a (prompted) LLM and can be easily adjusted for a variety of alignment criteria. We use the same synthetic data to train *inspectors*, binary miss-alignment classification models to guide a *squad* of multiple aligners. Our empirical results demonstrate consistent improvements when applying aligner squad to various LLMs, including chat-aligned models, across several instruction-following and red-teaming datasets.
Anthology ID:
2024.findings-emnlp.808
Volume:
Findings of the Association for Computational Linguistics: EMNLP 2024
Month:
November
Year:
2024
Address:
Miami, Florida, USA
Editors:
Yaser Al-Onaizan, Mohit Bansal, Yun-Nung Chen
Venue:
Findings
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
13785–13802
Language:
URL:
https://aclanthology.org/2024.findings-emnlp.808
DOI:
Bibkey:
Cite (ACL):
Lilian Ngweta, Mayank Agarwal, Subha Maity, Alex Gittens, Yuekai Sun, and Mikhail Yurochkin. 2024. Aligners: Decoupling LLMs and Alignment. In Findings of the Association for Computational Linguistics: EMNLP 2024, pages 13785–13802, Miami, Florida, USA. Association for Computational Linguistics.
Cite (Informal):
Aligners: Decoupling LLMs and Alignment (Ngweta et al., Findings 2024)
Copy Citation:
PDF:
https://aclanthology.org/2024.findings-emnlp.808.pdf