PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Ji, Jiaming, Hong, Donghai, Zhang, Borong, Chen, Boyuan, Dai, Juntao, Zheng, Boren, Qiu, Tianyi, Zhou, Jiayi, Wang, Kaile, Li, Boxuan, Han, Sirui, Guo, Yike, Yang, Yaodong
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915342329053184
author Ji, Jiaming
Hong, Donghai
Zhang, Borong
Chen, Boyuan
Dai, Juntao
Zheng, Boren
Qiu, Tianyi
Zhou, Jiayi
Wang, Kaile
Li, Boxuan
Han, Sirui
Guo, Yike
Yang, Yaodong
author_facet Ji, Jiaming
Hong, Donghai
Zhang, Borong
Chen, Boyuan
Dai, Juntao
Zheng, Boren
Qiu, Tianyi
Zhou, Jiayi
Wang, Kaile
Li, Boxuan
Han, Sirui
Guo, Yike
Yang, Yaodong
contents In this study, we introduce the safety human preference dataset, PKU-SafeRLHF, designed to promote research on safety alignment in large language models (LLMs). As a sibling project to SafeRLHF and BeaverTails, we separate annotations of helpfulness and harmlessness for question-answering pairs, providing distinct perspectives on these coupled attributes. Overall, we provide 44.6k refined prompts and 265k question-answer pairs with safety meta-labels for 19 harm categories and three severity levels ranging from minor to severe, with answers generated by Llama-family models. Based on this, we collected 166.8k preference data, including dual-preference (helpfulness and harmlessness decoupled) and single-preference data (trade-off the helpfulness and harmlessness from scratch), respectively. Using the large-scale annotation data, we further train severity-sensitive moderation for the risk control of LLMs and safety-centric RLHF algorithms for the safety alignment of LLMs. We believe this dataset will be a valuable resource for the community, aiding in the safe deployment of LLMs. Data is available at https://huggingface.co/datasets/PKU-Alignment/PKU-SafeRLHF.
format Preprint
id arxiv_https___arxiv_org_abs_2406_15513
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference
Ji, Jiaming
Hong, Donghai
Zhang, Borong
Chen, Boyuan
Dai, Juntao
Zheng, Boren
Qiu, Tianyi
Zhou, Jiayi
Wang, Kaile
Li, Boxuan
Han, Sirui
Guo, Yike
Yang, Yaodong
Artificial Intelligence
Computation and Language
In this study, we introduce the safety human preference dataset, PKU-SafeRLHF, designed to promote research on safety alignment in large language models (LLMs). As a sibling project to SafeRLHF and BeaverTails, we separate annotations of helpfulness and harmlessness for question-answering pairs, providing distinct perspectives on these coupled attributes. Overall, we provide 44.6k refined prompts and 265k question-answer pairs with safety meta-labels for 19 harm categories and three severity levels ranging from minor to severe, with answers generated by Llama-family models. Based on this, we collected 166.8k preference data, including dual-preference (helpfulness and harmlessness decoupled) and single-preference data (trade-off the helpfulness and harmlessness from scratch), respectively. Using the large-scale annotation data, we further train severity-sensitive moderation for the risk control of LLMs and safety-centric RLHF algorithms for the safety alignment of LLMs. We believe this dataset will be a valuable resource for the community, aiding in the safe deployment of LLMs. Data is available at https://huggingface.co/datasets/PKU-Alignment/PKU-SafeRLHF.
title PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference
topic Artificial Intelligence
Computation and Language
url https://arxiv.org/abs/2406.15513