Chinese AI firm iFlyTek says its LLMs are trained completely on Huawei platform

Chinese artificial intelligence (AI) firm iFlyTek says it has the only large language models (LLMs) trained completely on a home-grown computing platform, co-developed with Huawei Technologies, marking another step in the industry's pursuit of self-sufficiency in the face of US restrictions.

While most of China's LLMs are underpinned by computing platforms that include US components such as graphics process units (GPUs) from Nvidia, iFlyTek is determined to train its LLMs on "self-developed, controllable" infrastructure at a time when Washington has largely cut off the country's access to the most advanced chips and AI know-how, according to Liu Qingfeng, the company's chairman and president.

"How far we can go hinges on whether we have self-developed, controllable foundational capabilities to support [AI development]," Liu said on Thursday at a ceremony where iFlyTek unveiled its updated Xinghuo model.

Do you have questions about the biggest topics and trends from around the world? Get the answers with SCMP Knowledge, our new platform of curated content with explainers, FAQs, analyses and infographics brought to you by our award-winning team.

iFlyTek chairman Liu Qingfeng speaks at the opening ceremony of the World Intelligence Expo 2024 in Tianjin on June 20, 2024. Photo: Xinhua alt=iFlyTek chairman Liu Qingfeng speaks at the opening ceremony of the World Intelligence Expo 2024 in Tianjin on June 20, 2024. Photo: Xinhua>

In his keynote speech, Liu noted that US restrictions on technologies that contribute to the development of LLMs - the underpinning of AI products like OpenAI's ChatGPT - emphasises the imperative for China to develop its own computing infrastructure for AI training. US sanctions have restricted exports to China of certain technologies, most notably the advanced Nvidia GPUs that have helped fuel the AI boom in recent years.

Liu told the audience that the firm's Xinghuo models were trained on a computing platform dubbed Feixing Yihao, which is based on Huawei's Ascend computing solutions.

Huawei's Ascend AI chips have been quickly gaining popularity in China. The performance of the Ascend 910B chip has been found in some tests to deliver between 80 and 120 per cent that of Nvidia's A100 when training LLMs, said Wang Tao, chief operating officer of Jiangsu Kunpeng Ecosystem Innovation Centre, on the sidelines of the Nanjing World Semiconductor Conference this month.

Analysts have said the Ascend solution is China's best shot so far in developing home-grown AI infrastructure. There is also increasing support from state-backed firms as they place orders for Ascend-powered projects, said Li Yangwei, a Beijing-based technical consultant working in intelligent computing.

Yet Huawei's hardware still lacks the vibrant developer community of Nvidia's Cuda platform, the software that supports AI compute processes on the company's GPUs, according to Li.

"The chance for Huawei to catch up with Nvidia is minimal if it doesn't solve the issue," Li said.

At the unveiling of Xinghuo 4, iFlyTek's Liu boasted that the new model's capabilities can match those of GPT-4 Turbo, one of the most advanced models from OpenAI, which was launched more than eight months ago.

Xinghuo 4 has already been put to practical use in various business scenarios, including healthcare, where it has assisted doctors and patients with diagnosing and treatment, according to a presentation at the event.

This article originally appeared in the South China Morning Post (SCMP), the most authoritative voice reporting on China and Asia for more than a century. For more SCMP stories, please explore the SCMP app or visit the SCMP's Facebook and Twitter pages. Copyright © 2024 South China Morning Post Publishers Ltd. All rights reserved.

Copyright (c) 2024. South China Morning Post Publishers Ltd. All rights reserved.