Progetto: MeSAS – Models and Tools for Sustainable AI

Descrizione: The MeSAS project is one of the sub-projects of the larger STRIVE – Sciences for Industrial, Green and Energy Transitions. MeSAS aims to develop methodologies and algorithms that go in the direction of making the entire AI process more sustainable from a technological point of view. The research challenges concern the simplification/compression of large-scale language models in the field of Natural Language Processing, both in inference tasks, to obtain smaller and eco-friendly ones with performances comparable to the original ones, and in the training phase /fine tuning, to obtain less complex and more efficient architectural solutions both in terms of functional blocks and hyperparameters.

Finanziamento: Allocation of the Ordinary Fund for Research Institutions 2022

Costo ICAR: 84.000 Euro

Finanziamento ICAR: 84.000 Euro

Durata: 01/10/2022 – 30/09/2025

Ambito di riferimento: Foundational Artificial Intelligence

Responsabile Scientifico ICAR: Massimo Esposito

Gruppo di riferimento: Language and Knowledge Engineering

Partner: CNR ICAR (Institute for High Performance Computing and Networking), CNR IEIIT (Institute of Electronics, Computer and Telecommunication Engineering), CNR IIT (Institute for informatics and telematics)