KAZ-LLM

2024: Work on the creation of an AI model

In Kazakhstan, work has begun on the creation of a national model of artificial intelligence called KAZ-LLM. This became known in August 2024. This project is being implemented by a scientific working group led by the Institute of Smart Systems and Artificial Intelligence (ISSAI) at Nazarbayev University, with the aim of bridging the language gap in the field of AI and creating technologies adapted to the needs of Kazakhstanis.

As Profit.kz writes, most modern AI models are designed for languages ​ ​ with a high level of resources, such as English, Spanish or Russian, which puts countries with low-resource languages ​ ​ at a disadvantage. KAZ-LLM will be the first major language model in Kazakhstan, covering Kazakh, Russian and English. This will allow residents of Kazakhstan to use advanced digital technologies in their native language, as well as provide access to the national and cultural features that form the basis of the model.

Profit.kz

An important partner in this project is Beeline Kazakhstan and its IT subsidiary QazCode. In 2023, Beeline already launched and published in the public domain the Kaz-RoBERTA-conversational model, which is used to serve customers on digital platforms. Now Beeline and QazCode provide the KAZ-LLM project with the necessary infrastructure, including cloud computing power of 8 DGX H100, which has significantly increased the amount of training data and model training capabilities.

KAZ-LLM will play a key role in preserving the national cultural heritage, covering the historical context, specialized areas and conversational data specific to Kazakhstan. The project is also aimed at developing advanced specialists in the field of generative AI, which contributes to strengthening the scientific potential of the country.^[1]

Notes