Vitruvian-1 is an innovative language model developed by Asc27, an Italian company founded in 2020 and based in Rome, specialising in artificial intelligence and cybersecurity. This model was designed exclusively for the Italian language, with a focus on the fields of medicine, law and military strategy, where terminological accuracy is crucial.
Unlike other multilingual models that may suffer from semantic distortions, vitruvian-1 has been trained on carefully selected Italian datasets only, thus capturing the nuances and peculiarities of the language. this focus allows it to offer high linguistic accuracy, making it particularly suitable for applications that require deep domain expertise and precise language usage.
In addition to its linguistic specialisation, Vitruvian-1 has been optimised for energy efficiency and scalability. it utilises advanced techniques such as quantisation and pruning to reduce energy consumption while ensuring high performance. the system leverages high-performance gpu clusters to balance computational requirements with operational costs.
Asc27 positions Vitruvian-1 as a specialised alternative to other european solutions, with a strong emphasis on security and adherence to italian and european data regulations. the name ‘vitruvian’ pays homage to leonardo da vinci, symbolising the aspiration to achieve significant breakthroughs with simple but effective tools. Asc27’s goal is to establish Vitruvian-1 as the benchmark for ia in italy, offering a domestic alternative to the US and chinese technology giants. Currently, Vitruvian-1 has completed its fifth training phase and is available for the beta phase. interested parties can visit the official website for more information and to participate in the beta programme.
image source: vitruvian.asc.com
