GoTo Launches Open Source Friends-AI, Local LLM Supporting Regional Language
JAKARTA - PT GoTo Gojek Tokopedia Tbk (GOTO) launched Sahabat-AI, an open-source Large Language Model (LLM) that is innovative for Indonesian and regional languages throughout the country.
Developed with PT Indosat Ooredoo Hutchison Tbk, this LLM is trained with large amounts of data, which can understand and produce natural languages, so that it can be used for various digital applications and services.
Through Friends-AI, GoTo President Director, Patrick Walujo, hopes that this LLM can help develop AI-based applications and services that are uniquely adapted to the needs of the Indonesian people.
"This is in line with our mission to build Indonesia's digital sovereignty, in accordance with the direction of the President of the Republic of Indonesia, Prabowo Subianto," Patrick said in his official statement received on Thursday, November 14.
Catherine Hindra Sutjahyo, President of the On Demand GoTo Group (Gojek) Service, also revealed the difference between Friends-AI and other NGOs, where LLM GoTo focuses on Indonesian and regional languages, such as Javanese and Sundanese, so that it has a deep understanding of the local context and cultural relevance.
Nevertheless, Catherine emphasized that in the future, Sahabat-AI will continue to be enriched with additional regional languages such as Batak and Bali.
Friends-AI are deliberately developed as an open-source ecosystem so that this model can be used inclusively and accessible, especially by local developers and engineers to develop AI-based solutions for various purposes, ranging from public services, customer services, data analysis, research and development, education to business development.
"Although it is still in the first development stage, we are proud to be able to convey that Friends-AI have higher average performance in Indonesian, Javanese and Sundanese compared to several other open-source models with the same number of parameters," added Catherine.
The Friends-AI LLM model can be downloaded for free on the Honggging Face page, a platform where the machine learning community collaborates to develop various models, data and digital applications.