In last week’s Amsterdam AI Meetup edition in collaboration with Institute for Logic, Language and Computation (ILLC) – University of Amsterdam (UvA) we delved into alternatives for ChatGPT: How good are open-source and in-house LLMs?
Harm de Vries of ServiceNow Research unravelled the advancements in the Big Code project, an ambitious venture to create an open-source dataset and generative AI model for code generation. Meanwhile, Matthieu Laneuville from SURF provided a comprehensive overview of GPT-NL, a community-driven Large Language Model designed for the Dutch language, spotlighting a significant financial endorsement from the Dutch government.
In a lively discussion, Konstantinos Papakostas (Zeta Alpha), Iva Gornishka (Gemeente Amsterdam), Jelke Bloem (Institute for Logic, Language and Computation (ILLC) – University of Amsterdam (UvA), Universiteit van Amsterdam) and Oskar van der Wal (ILLC, University of Amsterdam) shared their perspectives on how to evaluate such models and make sure they are useful, safe and reliable.
Read also
