Description
LLMs4EU is a European project coordinated by the Alliance for Language Technologies (ALT-EDIC), aimed at preserving European linguistic and cultural diversity in the digital age. Several European languages risk being left out of the development of generative AI simply because the resources needed to train language models are not available for them.
The project brings together the main European actors in generative AI to make sure that European companies — small and medium-sized enterprises in particular — have access to the tools and resources needed to be competitive in language technologies, and in Large Language Models specifically. The goal is to release LLMs, and the tooling required to use them, as open data across all EU languages, building on existing European programmes and expertise. The tools made available will cover the whole pipeline, from training the models to ensuring their compliance with European legislation such as the AI Act and the GDPR. The consortium built around ALT-EDIC spans organisations in more than 20 countries, which guarantees broad geographic and linguistic coverage.
CNR-ILC contributes to the scientific use case, with a focus on data and model documentation and on evaluation. This includes work on data collection and infrastructure, the definition of requirements for language technology tools, the development of efficient model fine-tuning and adaptation techniques, and the design of robust evaluation methodologies for LLMs — human evaluation in particular. Through its involvement in ALT-EDIC, CNR-ILC also works on ensuring transparent and traceable data practices across the whole life cycle of the models.
Duration: March 2025 — February 2028 · Type: European project