Towards Faster and Greener Intelligent Code Generation

Abstract Large code models have shown unprecedented excellence in tasks such as code generation and code completion, largely due to their enormous parameter sizes. However, the large number of parameters also means higher computational costs, financial investments, and environmental impacts. This talk will focus on the issue of computational resource wastage and propose strategies to improve resource utilization and inference efficiency both before and during inference. Additionally, it will introduce the first AI-oriented code grammar design, which represents code semantics with fewer tokens, therefore reducing computational overhead. The research outcome is expected to advance the commercialization and deployment of large code models.

Bio Dr. Du Xiaoning is a Lecturer (a.k.a Assistant Professor in the US) at the Faculty of Information Technology, Monash University. She received her Ph.D. from Nanyang Technological University in 2020 and her Bachelor’s degree from Fudan University in 2014. Her research primarily focuses on responsible code intelligence, and she looks into problems that are critical for code intelligence systems and tools to be deployed and used in practice, including dataset quality issues, dataset copyright issues, model efficiency problems, robustness, and interpretability. Her research has been published in top-tier conferences and journals, including ICSE, ASE, FSE, NeurIPS, AAAI, S&P, USENIX Security, and TDSC. One of her works, which evaluated and improved the quality of code search datasets, was published at ICSE 2021 and nominated for the ACM SIGSOFT Distinguished Paper Award. She is also a recipient of the Google Research Scholar Award 2024.