Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Latam-GPT is new Latin America is developing and large language models are being developed. The project, led by the non -profit Chilean National Center for Artificial Intelligence (CNIA), is aiming to help the region gain technical independence by developing an open source AI model trained in the language and context of Latin America.
“This work cannot be adopted by only one group or a country in Latin America: this is a challenge that everyone needs to participate,” Senia’s director Alvaro Soto said in an interview to Ward N Espaol. “Latam-GPT is a project that wants to create an open, free and, above all partners.
The project stands for its associate consciousness. Soto explained, “We do not want to compete with Open A, DEPSEC or Google. We want a specific model with Latin America and Caribbean, conscious of cultural needs and challenges, such as understanding different dialects, the history of the region and the unique cultural aspects,” explained “Soto.
Thanks to the 33 strategic partnership with Latin America and the Caribbean organization, the project has collected a corpus equivalent to several million books than eight terabytes. This data base has enabled the development of a language model with 50 billion parameters, a scale that makes it comparable to GPT -1.5 and gives high power from a medium to perform complex tasks such as logic, translation and associations.
Latam-GPT is being trained in a regional database that compiled information from 20 Latin American countries and Spain, including a total of 2,645,3 documents. Data distribution shows a significant concentration in the largest countries in the region, including 6885,000 documents, including Mexico 385,000, Spain 325,000, Colombia 220,000 and Argentina 210,000 documents. The numbers reflect the size of these markets, their digital development and the availability of structural content.
“Initially, we will turn on a language model we
The first model is for scaling on a family development of more advanced technology in the future, including points, images and videos and on greater models. “It’s such an open project, we want other institutions to be able to use it. A team of Colombia can adapt to school education or adapt it for a health sector in Brazil.
