Meta joined this Friday to the race of major technologies in the field artificial intelligence (AI) with a tool for researchers with prior permission only, named LLAMA (Large Language Model Meta AI).
It is not a “chat bot” like those launched by Microsoft or Google.but a program for those developing this type of technology and hopes to contribute to “limiting or eliminating” existing bias, “toxicity” or “misinformation” problems that weigh on these initial results.
LLaMA is “a major model designed for the next generation of core language. help researchers advance their work in this subfield of artificial intelligenceMeta said in a statement that it was trying to “democratize access” and “for those who don’t have the adequate infrastructure to study it.”
“Despite all these recent advances in major language models,full access to your research remains limited “for the resources needed to train and operate this technology that hinders understanding and solving problems.”
calls can produce text and is trained in the 20 most spoken languagesespecially those with the Latin and Cyrillic alphabets are indicated in the note.
Meta noted that it has released this tool under a research-oriented, non-commercial license. “Avoid abuse” After examining the demands of relevant researchers, educational institutions, the state and civil society one by one, it will allow its use.