This is Alibaba's first region-specific large language model AI, and Southeast Asia is seen as an important growth market, and Da Mo's new LLM is a sign that Chinese companies are still doubling down on their efforts to embrace the wave of generative AI brought about by OpenAI's ChatGPT.
The DAMO Academy, the research arm of Alibaba Group Holdings, has unveiled artificial intelligence (AI) large language models (LLMs) tailored for Southeast Asian languages, underscoring the company's ambition to expand the market in a wider region.
Southeast Asia LLMs (Seallm), pre-trained on Vietnamese, Indonesian, Thai, Malay, Khmer, Lao, Tagalog and Burmese datasets, outperformed other open-source models on language and security tasks, Alibaba Research said in a statement on Monday.
This is Alibaba's first region-specific LLM, and Southeast Asia is seen as an important growth market. For example, Alibaba's Southeast Asian e-commerce platform, Lazada, aims to achieve a turnover of $100 billion by 2030, with 300 million consumers in the region.
Seallm Chat is a fine-tuned chat assistant that comes with LLMs and is designed to help businesses using LLMs interact with the Southeast Asian market.
Da Mo's new LLM is a sign that Chinese companies are still doubling down on the wave of generative AI brought about by OpenAI's ChatGPT last year.
As of July this year, a total of 130 LLMs have been issued by Chinese companies and research institutions, and the so-called"The battle of the top 100 models"。
Alibaba's LLM "Tongyi Qianwen" was released in April this year, and as of Tuesday this week, it is on an open-source artificial intelligence platform"hugging face "Ranked 4th in the world out of all models tracked.
Hugging Face's rankings are based on a variety of factors, including scientific knowledge, multitasking accuracy, and common-sense reasoning.
Seallm outperforms other LLMs like ChatGPT in non-Latin tasks, with up to 9 times longer ability to interpret and process non-Latin text.
Seallm also achieved better results in translations between English and low-resource languages (with limited data for training AI dialogue systems), such as Lao and Khmer.
Bing Lidong, director of the Da Mo Language Technology Lab, says Seallm can"Embrace the rich culture of Southeast Asia", adding that innovation will inevitably empower communities that have historically been underrepresented in the digital space.
Analysts say that despite more than 100 master's degrees in linguistics, China's AI market still faces challenges such as U.S. chip restrictions and a lack of killer applications, making it difficult to attract more users in a rapidly changing market.