Research behind GAISSA.

Our research explores how to make AI faster, use fewer resources and measure the difference.

Optimization

How to improve speed and resource use while preserving quality.

How to measure AI’s energy use and environmental impact.

How software design affects the resources AI needs.

Computing2025

Authors: Alexandra González, Joel Castaño, Xavier Franch, Silverio Martínez-Fernández.

Compares how optimization changes model accuracy, response time and energy use.

Journal of Systems and Software2025

Authors: Francisco Durán, Matias Martinez, Patricia Lago, Silverio Martínez-Fernández.

Compares ways to run language models for code and the computing resources they use.

ACM TOSEM2025

Authors: Luís Cruz, Xavier Franch, Silverio Martínez-Fernández.

Examines how software engineering can reduce the environmental impact of AI.

FSE Demo2024

Authors: Pau Duran, Joel Castaño, Cristina Gómez, Silverio Martínez-Fernández.

Introduces a tool that measures model energy use and presents it as an efficiency label.

CAIN at ICSE2024

Authors: Francisco Durán, Silverio Martínez-Fernández, Matias Martinez, Patricia Lago.

Studies how software architecture and deployment choices affect the efficiency of running models.

MSR2024

Authors: Joel Castaño, Silverio Martínez-Fernández, Xavier Franch, Justus Bogner.

Studies how developers update and maintain models on Hugging Face.

ESEM2023

Authors: Joel Castaño, Silverio Martínez-Fernández, Xavier Franch, Justus Bogner.

Analyzes the carbon-footprint information reported for models on Hugging Face.

SEAA2023

Authors: Silverio Martínez-Fernández, Xavier Franch, Francisco Durán.

Sets out GAISSA’s approach to reducing AI’s environmental impact through software design.

GAISSA Optimizer builds on our AI efficiency research at the Universitat Politècnica de Catalunya.