About TaiwanEval
Independent, checkable AI evaluation for Taiwan.
TaiwanEval tests where large language models go wrong on Taiwan's institutions, language and everyday context, and publishes the questions, sources, scoring and known limits alongside, so that results can be reproduced and checked.
Maintainer

Calvin Ku
He is responsible for the methods, the code and corrections.
Funding
TaiwanEval currently operates independently. It has taken no grants, sponsorship or investment, and has no paid engagements.
If that changes, through new funding or any relationship that could affect its judgement, this page will say so.
Contact
Questions about the evaluation and reports of mistakes are both welcome.
Content updated