UC San Diego

Despite achieving state-of-the-art performance on many NLP tasks, the high energy cost and long inference delay prevent Transformer-based language models (LMs) from seeing broader adoption including for edge and mobile computing. Our efficient NLP research aims to comprehensively consider computation, time and carbon emission for the entire life-cycle of NLP, including data preparation, model training and inference.

We demonstrate ways to promote computational efficiency in natural language processing, thus reducing hardware and software bottlenecks of training and inference, which is crucial in applying such models in production. Efficient NLP further facilitates democratization of language technology and allows language models to be accessible to more people.