The Ministry of Science, Technology, Knowledge and Innovation (CTCI) and the National Artificial Intelligence Center (CENIA) of Chile have announced the creation of LATAM GPT. This large language model (LLM), based on artificial intelligence (AI), seeks to accurately represent the history and culture of Latin America.
The algorithm has been in development since 2023 and its official launch is scheduled for June of this year. It is able to process 55 billion parameters and has three trillion data in Tokens. Its capacity is comparable to that of Openai chatgpt 3.5. Latam GPT is conceived as a public, inclusive and open model, with licenses for the creation of personalized applications based on their computer architecture.
The project has the support of experts, organizations and research centers of Mexico, Argentina, Colombia, Ecuador, United States, Spain, Peru and Uruguay. Among the participating Mexican institutions include the Center for Research in Geospatial Information Sciences, the National Institute of Astrophysics, Optics and Electronics and the Center for Research and Innovation in Information and Communication Technologies.
“The LLMS are quickly integrating into our daily lives. But what happens when we ask questions about Latin America? Although current models are of high quality, their understanding of the Latin American context can be improved. Therefore, we believe it is essential to develop a regional model that precisely reflects our history and culture, ”says the site web of the model.
Latam GPT was presented at the Summit for Action on the AI of Paris. Aisén Etcheverry, head of the CTCI, stressed that the development of this model is strategic and seeks to project Chile’s leadership and the rest of the region in the artificial intelligence sector. He said that it is a step towards democratization of access to this technology. “All countries must have the possibility of developing AI systems, implementing them in their governments and applying them in their industries. The capacities to achieve this cannot be concentrated in a few companies or nations, ”he said.
For his part Rodrigo Durán, general manager of CENIA, indicated that the main objective is to offer an open and accessible algorithm that fosters the research and development of AI in Latin America. “The spirit of the model is that it is the best to solve specific problems in Latin America and the Caribbean,” he emphasized.
Latam GPT continues its training process. The page web The project suggests that the general public can participate in this phase. Those interested should request an access key according to their country of origin or the one they wish to represent, through an email. The platform admits text tickets in Spanish and covers general issues related to Latin America, such as semantic and dialect differences between nations, historical facts, geographical and political references, as well as environmental and cultural issues. “It is not necessary for all questions to be formal or with perfect spelling. Both types of writing are valued and used to optimize the system”, Says the portal.
The CENIA has allocated approximately $ 550,000 in the development of Latam GPT, with almost half of the funds provided by the Andean Development Corporation. In addition, the University of Tarapacá has invested 5 million dollars in the installation of a supercomputer that has accelerated the training of the model.
#Mexico #Chile #Argentina #join #create #Latam #GPT #model #Latin #America
