AlignBench: Benchmarking Chinese Alignment of Large Language Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Liu, Xiao, Lei, Xuanyu, Wang, Shengyuan, Huang, Yue, Feng, Zhuoer, Wen, Bosi, Cheng, Jiale, Ke, Pei, Xu, Yifan, Tam, Weng Lam, Zhang, Xiaohan, Sun, Lichao, Gu, Xiaotao, Wang, Hongning, Zhang, Jing, Huang, Minlie, Dong, Yuxiao, Tang, Jie
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929471989219328
author Liu, Xiao
Lei, Xuanyu
Wang, Shengyuan
Huang, Yue
Feng, Zhuoer
Wen, Bosi
Cheng, Jiale
Ke, Pei
Xu, Yifan
Tam, Weng Lam
Zhang, Xiaohan
Sun, Lichao
Gu, Xiaotao
Wang, Hongning
Zhang, Jing
Huang, Minlie
Dong, Yuxiao
Tang, Jie
author_facet Liu, Xiao
Lei, Xuanyu
Wang, Shengyuan
Huang, Yue
Feng, Zhuoer
Wen, Bosi
Cheng, Jiale
Ke, Pei
Xu, Yifan
Tam, Weng Lam
Zhang, Xiaohan
Sun, Lichao
Gu, Xiaotao
Wang, Hongning
Zhang, Jing
Huang, Minlie
Dong, Yuxiao
Tang, Jie
contents Alignment has become a critical step for instruction-tuned Large Language Models (LLMs) to become helpful assistants. However, the effective evaluation of alignment for emerging Chinese LLMs is still largely unexplored. To fill in this gap, we introduce AlignBench, a comprehensive multi-dimensional benchmark for evaluating LLMs' alignment in Chinese. We design a human-in-the-loop data curation pipeline, containing eight main categories, 683 real-scenario rooted queries and corresponding human verified references. To ensure the correctness of references, each knowledge-intensive query is accompanied with evidences collected from reliable web sources (including URLs and quotations) by our annotators. For automatic evaluation, our benchmark employs a rule-calibrated multi-dimensional LLM-as-Judge~\cite{zheng2023judging} approach with Chain-of-Thought to generate explanations and final ratings, ensuring high reliability and interpretability. All evaluation code, data, and LLM generations are available at \url{https://github.com/THUDM/AlignBench}. Since its release, AlignBench has been adopted by top (Chinese) LLMs for evaluating their alignment capabilities in Chinese, including ChatGLM, Qwen, DeepSeek, Yi, Baichuan, and Abab.
format Preprint
id arxiv_https___arxiv_org_abs_2311_18743
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle AlignBench: Benchmarking Chinese Alignment of Large Language Models
Liu, Xiao
Lei, Xuanyu
Wang, Shengyuan
Huang, Yue
Feng, Zhuoer
Wen, Bosi
Cheng, Jiale
Ke, Pei
Xu, Yifan
Tam, Weng Lam
Zhang, Xiaohan
Sun, Lichao
Gu, Xiaotao
Wang, Hongning
Zhang, Jing
Huang, Minlie
Dong, Yuxiao
Tang, Jie
Computation and Language
Artificial Intelligence
Machine Learning
Alignment has become a critical step for instruction-tuned Large Language Models (LLMs) to become helpful assistants. However, the effective evaluation of alignment for emerging Chinese LLMs is still largely unexplored. To fill in this gap, we introduce AlignBench, a comprehensive multi-dimensional benchmark for evaluating LLMs' alignment in Chinese. We design a human-in-the-loop data curation pipeline, containing eight main categories, 683 real-scenario rooted queries and corresponding human verified references. To ensure the correctness of references, each knowledge-intensive query is accompanied with evidences collected from reliable web sources (including URLs and quotations) by our annotators. For automatic evaluation, our benchmark employs a rule-calibrated multi-dimensional LLM-as-Judge~\cite{zheng2023judging} approach with Chain-of-Thought to generate explanations and final ratings, ensuring high reliability and interpretability. All evaluation code, data, and LLM generations are available at \url{https://github.com/THUDM/AlignBench}. Since its release, AlignBench has been adopted by top (Chinese) LLMs for evaluating their alignment capabilities in Chinese, including ChatGLM, Qwen, DeepSeek, Yi, Baichuan, and Abab.
title AlignBench: Benchmarking Chinese Alignment of Large Language Models
topic Computation and Language
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2311.18743