Alibaba's Research Division Launches AI Large Language Model SeaLLM Tailored for Southeast Asia
-
DAMO Academy, the research arm of Alibaba Group, has introduced an AI large language model (LLM) specifically designed for Southeast Asian languages, highlighting the company's ambition to expand its market presence in the region.
Alibaba's research division stated that the Southeast Asian LLM (SeaLLM) has been pre-trained on datasets in Vietnamese, Indonesian, Thai, Malay, Khmer, Lao, Tagalog, and Burmese, and outperforms other open-source models in language and security tasks.
This is Alibaba's first region-specific LLM, with Southeast Asia being viewed as a crucial growth market. For instance, Lazada, Alibaba's e-commerce platform in Southeast Asia, aims to achieve $100 billion in turnover by 2030, serving 300 million consumers in the region.
SeaLLM chat is a fine-tuned chatbot assistant accompanying the LLM, designed to help businesses utilizing the LLM engage with the Southeast Asian market.
It is reported that Alibaba's LLM 'Tongyi Qianwen' was released in April this year and, as of this Tuesday, ranks fourth globally among all models tracked by the open-source AI platform 'Hugging Face'.
SeaLLM outperforms other LLMs like ChatGPT in non-Latin language tasks, with its ability to interpret and process non-Latin texts extended by nine times. SeaLLM also achieves better results in translations between English and low-resource languages (where data for training AI dialogue systems is limited), such as Lao and Khmer.