Aligners: Decoupling LLMs and Alignment

arXiv:2403.04224v1 Announce Type: cross
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 …

abstract alignment applications arxiv criterion cs.ai cs.cl cs.lg every human language language models large language large language models llm llms safety training type utility