Building Open Source Malaysian Multimodality Large Language Model

In this keynote session, join Mr. Husein Zolkepli as he shares the journey of building Large Language Models (LLMs) for the Malay language. Mr. Husein will introduce how these models learn, why making them bigger matters, and the powerful hardware that supports the process. The talk also highlights the role of volunteers in collecting local data, the use of synthetic data to improve translation, and the importance of cultural relevance. It concludes with the effort to train a model fully from scratch, showing how technology and culture come together to make AI more useful for Malay speakers.