Alibaba Cloud unveils a serverless version of its PAI-Elastic Algorithm Service (EAS) for cost-efficient model deployment and inference. The platform integrates vector engine technology into products like Hologres and Elasticsearch, enabling easier access to large language models. The CTO emphasizes Alibaba Cloud’s commitment to AI innovation, announcing upgrades to MaxCompute MaxFrame and introducing PAI-Artlab for model training. The company’s vector engine technology enhances various database solutions, supporting global customers in their digital transformation. Collaborations with companies like Haleon and rinna demonstrate the real-world applications of Alibaba Cloud’s AI technologies.
30 January 2024 – Alibaba Cloud, the digital technology arm of Alibaba Group, has introduced a serverless version of its Platform for AI (PAI)-Elastic Algorithm Service (EAS) to streamline model deployment and inference for individuals and enterprises. The serverless offering aims to provide a cost-efficient solution by allowing users to access computing resources on-demand, reducing inference costs by 50% compared to traditional pricing models. Alibaba Cloud also announced the integration of its vector engine technology into products like Hologres, Elasticsearch, and OpenSearch, facilitating easier access to large language models (LLMs) for building custom generative AI applications.
The PAI-EAS platform is set to expand its serverless capabilities to support prominent open-source LLMs and Alibaba’s AI model community, ModelScope, in March 2024. This includes models for image segmentation, summary generation, and voice recognition. The integration of LLMs, training services, and vector engine technology enables Alibaba Cloud to support a Retrieval-Augmented Generation (RAG) process, enhancing LLMs with knowledge bases for improved accuracy and nuanced insights.
Alibaba Cloud’s CTO, Zhou Jingren, emphasized the company’s commitment to AI and cloud technology innovation during the AI & Big Data Summit in Singapore. The company also announced MaxCompute MaxFrame, an upgraded big data service, and PAI-Artlab, a platform for model training and image generation to foster creativity among designers. Additionally, Alibaba Cloud’s vector engine technology has been integrated into various database solutions, enhancing performance and capabilities.
Alibaba Cloud’s technologies continue to support global customers in their digital transformation, with a focus on open-sourcing proprietary language models to empower clients in developing customized generative AI applications. Companies like Haleon, a consumer health company, have utilized Alibaba Cloud’s Large Language Model and Retrieval-Augmented Generation technology to introduce an AI nutritionist for Chinese consumers. rinna, a Japanese startup, has launched the Nekomata models based on Alibaba Cloud’s Tongyi Qianwen LLMs, showcasing the collaborative innovation in the AI space.
