Chumor 2.0: Towards Benchmarking Chinese Humor Understanding

TL;DR

This paper introduces Chumor 2.0, a large Chinese humor dataset, and evaluates how well current large language models understand Chinese humor, revealing significant challenges and gaps compared to human performance.

Contribution

The paper creates the first large Chinese humor explanation dataset, Chumor, and benchmarks multiple LLMs, highlighting their limitations in understanding culturally nuanced Chinese humor.

Findings

LLMs perform only slightly better than random on Chumor.

Human explanations significantly outperform GPT-4o and ERNIE-4-turbo.

Chumor presents substantial challenges for current humor understanding models.

Abstract

Existing humor datasets and evaluations predominantly focus on English, leaving limited resources for culturally nuanced humor in non-English languages like Chinese. To address this gap, we construct Chumor, the first Chinese humor explanation dataset that exceeds the size of existing humor datasets. Chumor is sourced from Ruo Zhi Ba, a Chinese Reddit-like platform known for sharing intellectually challenging and culturally specific jokes. We test ten LLMs through direct and chain-of-thought prompting, revealing that Chumor poses significant challenges to existing LLMs, with their accuracy slightly above random and far below human. In addition, our analysis highlights that human-annotated humor explanations are significantly better than those generated by GPT-4o and ERNIE-4-turbo. We release Chumor at https://huggingface.co/datasets/dnaihao/Chumor, our project page is at https://dnaihao.github.io/Chumor-dataset/, our leaderboard is at https://huggingface.co/spaces/dnaihao/Chumor, and our codebase is at https://github.com/dnaihao/Chumor-dataset.