Energy-Efficient Models for Sustainable Intelligent Software Engineering. This project aims to tackle the high operational costs and environmental impacts associated with software engineering tools po

Description

Energy-Efficient Models for Sustainable Intelligent Software Engineering. This project aims to tackle the high operational costs and environmental impacts associated with software engineering tools powered by large language models. The project expects to develop novel and effective techniques to reduce energy consumption when the models are deployed for service. Expected outcomes of this project include new insights in software engineering with the creation of the first compact code representations, a domain-specific model distillation framework, and an adaptive dynamic inference management strategy. This should provide significant benefits to reduce both environmental impact and operational expenses, making the techniques more accessible, sustainable, and beneficial for scholarly, public, and commercial sectors.. Scheme: Discovery Early Career Researcher Award. Field: 4612 - Software Engineering. Lead: Dr Xiaoning Du