HellaSwag-Pro: A Large-Scale Bilingual Benchmark for Evaluating the Robustness of LLMs in Commonsense Reasoning
Abstract
Large language models (LLMs) have shown remarkable capabilities in commonsense reasoning; however, some variations in questions can trigger incorrect responses. Do these models truly understand commonsense knowledge, or just memorize expression patterns? To investigate this question, we present the first extensive robustness evaluation of LLMs in commonsense reasoning. We introduce HellaSwag-Pro, a large-scale bilingual benchmark consisting of 11,200 cases, by designing and compiling seven types of question variants. To construct this benchmark, we propose a two-stage method to develop Chinese HellaSwag, a finely annotated dataset comprising 12,000 instances across 56 categories. We conduct extensive experiments on 41 representative LLMs, revealing that these LLMs are far from robust in commonsense reasoning. Furthermore, this robustness varies depending on the language in which the LLM is tested. This work establishes a high-quality evaluation benchmark, with extensive experiments offering valuable insights to the community in commonsense reasoning for LLMs.
Citation:
@inproceedings{li2025hellaswag,
title={HellaSwag-Pro: A Large-Scale Bilingual Benchmark for Evaluating the Robustness of LLMs in Commonsense Reasoning},
author={Li, Xiaoyuan and Li, Moxin and Men, Rui and Zhang, Yichang and Bao, Keqin and Wang, Wenjie and Feng, Fuli and Liu, Dayiheng and Lin, Junyang},
booktitle = {The 63rd Annual Meeting of the Association for Computational Linguistics},
year = {2025}
}
Xiaoyuan Li
李晓媛
Keqin Bao
鲍克勤
Wenjie Wang
王文杰 教授
