On August 27, Moore Thread officially announced that it completed the DAY-0 adaptation of the GLM-5.3-Flash smart spectrum multi-modal model based on the MTT S5000 intelligent computing card and the MUSA software stack. In response to the KDA mechanism used for linear attention, the R&D team completed customized operator tuning based on the MATE operator optimization engine and connected to the SGLang-musa cache system. This move transformed the traditional KV Cache into an incremental update of the state matrix, greatly relieved the pressure on video memory for long context inference, and enabled the model to operate efficiently on domestic computing power platforms.
Zhitongcaijing · 1d ago
Share
Listen to the news
On August 27, Moore Thread officially announced that it completed the DAY-0 adaptation of the GLM-5.3-Flash smart spectrum multi-modal model based on the MTT S5000 intelligent computing card and the MUSA software stack. In response to the KDA mechanism used for linear attention, the R&D team completed customized operator tuning based on the MATE operator optimization engine and connected to the SGLang-musa cache system. This move transformed the traditional KV Cache into an incremental update of the state matrix, greatly relieved the pressure on video memory for long context inference, and enabled the model to operate efficiently on domestic computing power platforms.
Disclaimer:Webull uses external vendor Google Translation Service for news translations where we endeavour to ensure these are correct, however, we recommend that you please double-check this information accordingly. Webull is not responsible for translation errors or issues.
What's Trending