Inception and MBZUAI Launch SHERKALA: A Groundbreaking Kazakh Large Language Model Empowering 13 Million Kazakh Speakers
Inception and MBZUAI Launch SHERKALA: A New Era for Kazakh Language AI
Abu Dhabi — In a groundbreaking collaboration, Inception, a G42 company, and the Mohamed bin Zayed University of Artificial Intelligence (MBZUAI) have unveiled SHERKALA, a revolutionary Kazakh Large Language Model (LLM) aimed at empowering over 13 million Kazakh speakers. This innovative model, boasting 8 billion parameters, is designed to enhance the accessibility and efficiency of generative AI for the Kazakh language, while also incorporating English, Russian, and Turkish.
SHERKALA is trained on a staggering 45 billion words, utilizing the advanced Llama 3.1 architecture with a unique 25% tokenizer expansion for improved Kazakh language processing. The model was developed on Condor Galaxy, one of the world’s most powerful AI supercomputers, a testament to the cutting-edge technology behind its creation.
Dr. Andrew Jackson, CEO of Inception, emphasized the significance of SHERKALA in addressing the needs of underserved linguistic communities. “This model not only empowers Kazakh speakers but also redefines the LLM landscape with scalable and inclusive AI solutions,” he stated. The launch follows the success of similar models like JAIS for Arabic and NANDA for Hindi, reinforcing Inception’s commitment to AI inclusivity.
Professor Preslav Nakov from MBZUAI echoed this sentiment, highlighting the partnership’s goal of democratizing AI access and preserving linguistic heritage. “SHERKALA represents a significant leap forward in empowering communities to thrive in the digital era,” he remarked.
With its superior performance in Kazakh understanding and generative capabilities, SHERKALA sets a new benchmark for language models, promising to transform the digital landscape for Kazakh speakers and beyond. As technology continues to evolve, SHERKALA stands as a beacon of hope for linguistic representation in the AI ecosystem.
