The new AI system is called Minerva7B, where 7B stands for 7 billion parameters, more than double the previous one.
It has been designed by a team led by Roberto Navigli of the Natural Language Processing of Sapienza.
Accessible to all here, Minerva 7B is a project developed by the Sapienza Nlp research group within the Future Artificial Intelligence Research.
The project implements the national strategy on artificial intelligence thanks to the EU-financed National Recovery and Resilience Plan funds in collaboration with the high-tech Cineca company, which made the Leonardo supercomputer available.
“Today we present the first pre-trained model in Italian style ChatGpt, therefore a linguistic model, which is called Minerva 7B,” said Navigli.
“It’s a model that is also trained to respond to questions, instructions and commands.
“It’s particularly interesting because we have done important work on the security of the model, the so-called safety.
“That is, the ability of the model to avoid and respond to questions that could lead to the generation of unwanted or malicious content.”
ANSA
