.jpg)
In the various artificial intelligence-based service offerings, one of the things that plays a major role is the language model (LLM) that powers it. In Malaysia itself, there are several developers who have developed their own multilingual models, including one of them is Mesolitica.
Most recently, the Ministry of Science, Technology and Innovation (MOSTI) shared that it will take proactive steps and examine the development of its own language model in the Malay language. This is also expected to reduce dependence on technology from abroad in addition to taking into account local values and culture in its development.
The development of the language model will show collaboration between MOSTI, universities and local companies. It is said that MOSTI, MIMOS and Mesolitica will cooperate in improving the language model, and is also expected to involve DBP (Dewan Bahasa dan Pustaka) in the future.
Mesolitica today offers their own language model called MaLLaM, which is trained using various local datasets, and is offered openly to the public. This model itself offers support for the integration of local values and culture. It is likely that we will see this collaboration improve this vernacular model for wider local use.