"With the continuous enhancement of AI technology, a more generalized AI platform has been formed, enabling AI-powered businesses. Ultimately, all technologies need to be empowered through physical entities, so AI must move towards productization," said Chen Ning, Chairman and CEO of Shenzhen Intellifusion Technology Co., Ltd.
The high-quality development-themed media delegation visited Shenzhen Intellifusion Technology Co., Ltd. on May 20, exploring the practice of AI large models in Shenzhen.
Product of Intellifusion (Photo: Nina)
Chen pointed out that the mismatch between input and output for low-frequency algorithms has posed significant challenges for many scenarios in achieving commercial value.
"In March this year, Intellifusion launched the 'DeepEye AiBox,' a thousand-yuan-level large model hardware product. We hope to apply multimodal large model technology and Intellifusion's algorithm chip technology to a standardized product, quickly enabling various scenarios," said Chen.
Currently, the 'DeepEye AiBox' allows users to train practical algorithms based on their specific scene requirements, solving the final challenge of AI implementation.
Chen introduced that the "DeepEye AiBox" is equipped with Intellifusion's self-developed large model edge training and inference chip, DeepEdge10 Max, and self-developed multimodal large model "SmartBook." It possesses online learning capabilities on the algorithm's edge side, effectively fitting a large model into a small box.
At present, Intellifusion's business covers fields such as smart transportation, smart security, urban governance, and smart commercial, promoting the construction of smart cities.
Reporter | Nina
Editor | Nan, James
