Search
Deepseek Affiliate Unveils Patent for Large Language Model Deployment Strategy
18 hour ago /
Read about 0 minute
Author：小编
On August 1, the Tianyancha App disclosed that Hangzhou Deepseek AI Basic Technology Research Co., Ltd., a subsidiary of Deepseek, has recently announced a patent entitled "A Method and System for Deploying Large Language Models." This patent delves into the realm of artificial intelligence, aiming to streamline load tasks, maximize hardware utilization, minimize idle computing resources, reduce latency, boost throughput, and enhance system scalability and fault tolerance. This is achieved by deploying the pre-filling and decoding stages separately on high-performance machines equipped with ample memory.
-
C114 Communication Network -
Communication Home
7 X 24 Track global technological trends
