Aya (language model for AI)

2024: Product Announcement

On February 13, 2024, Cohere for AI, a non-profit research laboratory created by Cohere in 2022, unveiled an open large language model (LLM) called Aya. This is said to be the first solution in this class to support more than 100 languages.

Aya was created as part of a large-scale project that started in January 2023. It was attended by more than 3,000 researchers from 119 countries. The goal is to develop a generative artificial intelligence (GENI) model based on the contribution of people from all over the world. It is noted that by the beginning of 2024, most LLMs are focused primarily on English. However, only about 5% of the world's inhabitants speak English in their home countries.

Cohere

LLM and artificial intelligence in general have transformed the global technology landscape, but many communities around the world have been left unsupported by the language limitations of existing models. This gap hinders the introduction and use of Genia on a global scale and could potentially further aggravate the situation, project participants say.

Aya provides an opportunity to solve the problem. This model supports 101 languages. These include Arabic, Bulgarian, Estonian, French, Irish, Hebrew, Indonesian, Italian,, Japanese Chinese, Portuguese, Swedish, and Russian and English. It is claimed that Aya closely follows the instructions and generates significantly higher quality responses than some other open models.

Overall, as project participants emphasize, the new model and training dataset are designed to help "researchers unlock the powerful potential of LLM for dozens of languages and cultures that are largely ignored by most other models available on the market."^[1]

Notes