The alignment of large language models is usually done by model providers to add or control behaviors that are common or universally understood across use cases and contexts. In contrast, in this article, we present an approach and architecture that empowers application developers to tune a model to their particular values, social norms, laws and other regulations, and orchestrate between potentially conflicting requirements in context. We lay out three main components of such an Alignment Studio architecture: Framers, Instructors, and Auditors that work in concert to control the behavior of a language model. We illustrate this approach with a running example of aligning a company's internal-facing enterprise chatbot to its business conduct guidelines.
@article{arxiv.2403.09704,
title = {Alignment Studio: Aligning Large Language Models to Particular Contextual Regulations},
author = {Swapnaja Achintalwar and Ioana Baldini and Djallel Bouneffouf and Joan Byamugisha and Maria Chang and Pierre Dognin and Eitan Farchi and Ndivhuwo Makondo and Aleksandra Mojsilovic and Manish Nagireddy and Karthikeyan Natesan Ramamurthy and Inkit Padhi and Orna Raz and Jesus Rios and Prasanna Sattigeri and Moninder Singh and Siphiwe Thwala and Rosario A. Uceda-Sosa and Kush R. Varshney},
journal= {arXiv preprint arXiv:2403.09704},
year = {2024}
}