The development of the first version of the modern language
model KazLLM has been completed in Astana. It understands not only the state
language but also Russian, English, and Turkish. This is a scientific project
by the Institute of Smart Systems and Artificial Intelligence (ISSAI) at
Nazarbayev University. Essentially, the large language model is the cornerstone
on which Kazakhstan's IT community can build future products and services.
«The model is a symbol of sovereignty in the field of
artificial intelligence. We have collected over 150 billion tokens. A token is
a unit of data. We can say that a token is essentially a word,» noted Madina
Abdrakhmanova, deputy director of the ISSAI at Nazarbayev University.
Kazakh scientists develop online translator
The country has also developed its first domestic
multifunctional application, Soyle App. It is based on a fundamental speech
model designed to meet the needs of both local and global markets. This product
also addresses a strategically important mission for Kazakhstan - ensuring the
information security of users who entrust their content to online translators.
«Soyle» is not only capable of translating between four
languages - Kazakh, Russian, English, and Turkish - but it can also convert
speech to text, text into another language, text into speech, and switch
between languages. However, it cannot do this in real-time yet; this feature is
still under development. Many of us currently use ChatGPT. The problem with ChatGPT
is that when you use it, especially the free version, all your data is leaked.
Therefore, it is especially important for many countries to develop domestic
language models. This is truly a matter of the country’s AI sovereignty,» Abdrakhmanova
added.
Related Articles
