BeaverTails: Towards Improved Safety Alignment of LLM via a Human-Preference Dataset
This dataset addresses safety concerns for LLM developers and users by providing a resource to improve harmlessness and helpfulness, though it is incremental as it builds on existing safety alignment efforts.
The paper tackles the problem of safety alignment in large language models by introducing the BeaverTails dataset, which provides 333,963 safety-labeled question-answer pairs and 361,903 expert comparisons for helpfulness and harmlessness, enabling applications in content moderation and RLHF.
In this paper, we introduce the BeaverTails dataset, aimed at fostering research on safety alignment in large language models (LLMs). This dataset uniquely separates annotations of helpfulness and harmlessness for question-answering pairs, thus offering distinct perspectives on these crucial attributes. In total, we have gathered safety meta-labels for 333,963 question-answer (QA) pairs and 361,903 pairs of expert comparison data for both the helpfulness and harmlessness metrics. We further showcase applications of BeaverTails in content moderation and reinforcement learning with human feedback (RLHF), emphasizing its potential for practical safety measures in LLMs. We believe this dataset provides vital resources for the community, contributing towards the safe development and deployment of LLMs. Our project page is available at the following URL: https://sites.google.com/view/pku-beavertails.