Ensuring Safe and High-Quality Outputs: A Guideline Library Approach for Language Models

arXiv:2403.11838v1 Announce Type: new
Abstract: Large Language Models (LLMs) exhibit impressive capabilities but also present risks such as biased content generation and privacy issues. One of the current alignment techniques includes principle-driven integration, but it faces challenges arising from the imprecision of manually crafted rules and inadequate risk perception in models without safety training. To address these, we introduce Guide-Align, a two-stage approach. Initially, a safety-trained model identifies potential risks and formulates specific guidelines for various inputs, thereby establishing a …

abstract alignment arxiv capabilities challenges content generation cs.ai cs.cl current integration language language models large language large language models library llms privacy quality risk risks rules type