Homepage of Hai Hu

oops, where is the photo?

My name is 胡 (Hu) 海 (Hai), but in English people say it in the reverse way, which sounds more like: Hi! Who?

I am an Assistant Professor in the Department of Linguistics and Translation at City University of Hong Kong, starting Aug 2025.

My research interests:

  1. Benchmarking (L)LMs on Chinese (esp. all kinds of reasoning abilities)
  2. BabyLM: training small LMs in controlled settings to study the learning dynamics and their alignment with kids/human language learning
  3. Cognitive science: computational modeling for psycho- and neuro-linguistics
  4. Chinese linguistics, treebanking, translation studies, etc.

I like building datasets and training models for Chinese NLP. I am a founding member of CLUE, one of the first evaluation suites for Chinese NLU, which includes some benchmarks we made, for instance, a reasoning dataset in Chinese: OCNLI. More recently, we constructed the first benchmark on conversational implicature in Chinese, based on the sitcom 武林外传, SwordsmanImp, the most comprehensive minimal pair benchmark for Chinese syntax, ZhoBLiMP, among several others. We conducted the first comprehensive study in China to examine whether EFL (English as Foreign Language) instructors can detect GPT-written essays (paper), and released a sentence-level detector of AI-written argumentative essays. You can find more on Datasets.

For publications, see my Google Scholar.

Experience:

Education:

Chengdu is my hometown, which is also home to pandas :panda_face:, hot pot :stew:, all kinds of spicy and non-spicy food, and :mahjong:.

OUR LAB is constangly HIRING students who are interested in computational linguistics.

You can reach me at: hu.hai at cityu.edu.hk