Towards an Explorable Conceptual Map of Large Language Models

Lorenzo Bertetto, Francesca Bettinelli, Alessio Buda, Marco Da Mommio, Simone Di Bari, Claudio Savelli, Elena Baralis, Anna Bernasconi, Luca Cagliero, Stefano Ceri, Francesco Pierri

Abstract: Large Language Models (LLMs) have revolutionized the current landscape of Natural Language Processing, enabling unprecedented advances in text generation, translation, summarization, and more. Currently, limited efforts have been devoted to providing a high-level and systematic description of their properties. Today’s primary source of information is the Hugging Face (HF) catalog, a rich digital repository for researchers and developers. Although it hosts several models, datasets, and applications, its underlying data model supports limited exploration of linked information.