TMMLU+: An Improved Traditional Chinese Evaluation Suite for Foundation Models

Zhi Rui Tam, Ya Ting Pai, Yen-Wei Lee, Hong-Han Shuai, Jun-Da Chen, Wei Min Chu, Sega Cheng

Research Area: Evaluation, LMs for everyone, LMs and the world, LMs on diverse modalities and novel applications

Keywords: Traditional Chinese evaluation, multi-choice question answering

TL;DR: We introduce TMMLU+, a comprehensive Traditional Chinese dataset covering over 66 subjects, with insights into tokenization effects and the impacts of Simplified versus Traditional Chinese on LLM performance.

Abstract: We present TMMLU+, a new benchmark designed for Traditional Chinese language understanding. TMMLU+ is a multi-choice question-answering dataset with 66 subjects from elementary to professional level. It is six times larger and boasts a more balanced subject distribution than its predecessor, Taiwan Massive Multitask Language Understanding (TMMLU). We also benchmark closed-source models and 26 open-weight Chinese large language models (LLMs) of parameters ranging from 1.8B to 72B on the proposed TMMLU+. Our findings reveal that (1.) Traditional Chinese models still trail behind their Simplified Chinese counterparts, highlighting a need for more focused advancements in LLMs catering to Traditional Chinese. (2.) Current LLMs still fall short of human performance in average scores, indicating a potential need for future research to delve deeper into social science and humanities subjects. (3.) Among all the tokenization compression metrics examined, we identify that only the fertility score uniquely demonstrates strong correlations with our benchmark results. We foresee that TMMLU+ will pinpoint areas for future model improvement, thereby narrowing the gap between machine and human linguistic capabilities and supporting researchers in developing Traditional Chinese LLMs. Our dataset, along with the benchmark source code, is accessible at huggingface.co/datasets/ikala/tmmluplus.

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the COLM Code of Ethics on https://colmweb.org/CoE.html

Author Guide: I certify that this submission complies with the submission instructions as described on https://colmweb.org/AuthorGuide.html

Submission Number: 275