HKU NLP Group

HKU NLP Group is a research team in the Department of Computer Science at the University of Hong Kong. The group focuses on natural language processing, with current work spanning pretraining algorithms, semantic parsing, dialog systems, and machine translation. They publish papers, release code, and train students in these areas.

The group's research covers several core topics. They design novel pretraining objectives and neural network architectures, including distillation methods to make models smaller and faster. They work on semantic parsing to build natural language interfaces that let people interact with computers more intuitively. In dialog systems, they generate long, coherent text and build automatic evaluation metrics. Their machine translation work explores simultaneous translation (processing speech as it's spoken) and multimodal translation (using images or other data alongside text). All of this is made public through GitHub repositories and academic publications.

This group is most useful for researchers and graduate students interested in cutting-edge NLP. Practitioners building language interfaces or translation systems can follow their open-source work. The team's focus on both theory and implementation makes their research accessible to those looking to apply new techniques in real systems.