Unicom Cloud launches DeepSeek-R1 series models

Unicom Cloud launches DeepSeek-R1 series models

IT Home reported on February 5 that China Unicom announced today that China Unicom has implemented the domestic and mainstream computing power based on the Xingluo platform to adapt to the multi-spec DeepSeek-R1 model, taking into account both privatization and publicization scenarios, and providing full-scale operation service guarantees. .

IT Home learned from the official introduction that Unicom Cloud is based on A800, H800, L40S and other mainstream computing power cards.Preset DeepSeek-R1 Multi-Size Modelusers can flexibly choose and quickly expand on demand, and quickly build a DeepSeek-R1 reasoning and fine-tuning environment. Through the computing network convergence scheduling and tenant-level model preset functions, customers can be quickly pulled up and used directly through SDK or standard restful APIs to meet the agile delivery needs of AI applications in various industries.

Unicom Cloud Backbone Cloud PoolHas been deployed in 270+ cities across the countrywith the high-performance computing power layout of “one city, one pool”, China Unicom Cloud version 7.5 has been launched on the DeepSeek-R1 mirror to form pre-deployment capabilities.

Unicom Cloud Programming Assistant deeply integrates the DeepSeek-R1 big model, adapts to multi-end plug-ins such as VSCode and Jetbrains.Supports over 20 mainstream programming languages. In addition, by building a local industry knowledge base, the assistant also supports advanced features such as local engineering knowledge Q&A and multi-file collaborative editing. Unicom cloud computer supports DeepSeek access to achieve intelligent upgrades.

Comments

Leave a Reply

Your email address will not be published. Required fields are marked *