Report: DeepSeek V5 Plans to Use Ascend for Training
According to industry sources, DeepSeek's next-generation large model, V5, is planned to be trained on Huawei Ascend chips, which were previously used mainly for the inference tasks of DeepSeek models. In late September this year, DeepSeek open-sourced infrastructure components for the Ascend computing platform, including compiler tools, computing libraries, and distributed communication libraries, which correspond one-to-one with the components previously open-sourced for the NVIDIA platform.
According to sources in the industry chain, DeepSeek's next-generation large model, V5, is planned to be trained on Huawei Ascend chips. Previously, Ascend chips were primarily used for the inference tasks of DeepSeek models.
At the end of September this year, DeepSeek open-sourced infrastructure components for the Ascend computing platform, including compilation tools, computing libraries, and distributed communication libraries, which correspond one-to-one with the components previously open-sourced for the NVIDIA platform.
Sci-Tech Innovation Board Daily
Source: Original Article