This repository contains the Python code for finetuning an italian BERT model on an italian legal dataset. The dataset is based on the First Book of the Italian Penal Code. However, this code can be applied to all legal code types.
This project comes from a research project called UNI4JUSTICE, to encourage the use of ML/AI models in the legal context. The aim was to provide a helper tool based on a multi-label classifier capable of correctly classifying any legal text referring to the Italian Penal Code. Currently, the model is limited to the First Book (or in any case to a single Book).