In the context of the integrated development of artificial intelligence and the power industry,the State Grid has carried out in-depth research on the research and development and application of the Guangming Power mo...In the context of the integrated development of artificial intelligence and the power industry,the State Grid has carried out in-depth research on the research and development and application of the Guangming Power model.The model is based on the knowledge of the power industry,cognitive computing as the core,and knowledge service as the goal,and has the characteristics of large sample,large calculation,large parameters,large knowledge,and large tasks,providing important support for the intelligent transformation of the power industry.This study focuses on the efficient operation of Guangming Power large model,builds a trinity operation system of“model iterative optimization,sample full-process governance,and computing power resource collaboration”,and drives the continuous improvement of model capabilities and compliance development through a two-level collaboration mechanism.With the goal of“building a strong and excellent Bright Power Large Model”,the study clearly states that it is necessary to accelerate the construction of a two-level collaborative operation system,and promote model iterative optimization,capability evaluation,service monitoring and compliance review on a regular basis.At the same time,focusing on the whole process of R&D and application of large and small models,starting from four aspects:computing power planning and layout,allocation and scheduling,adaptation and optimization,monitoring and analysis,strengthening the application monitoring and analysis of two-level intelligent computing centers,and building an efficient computing power resource application and supply system to continuously improve its operation and service capabilities.展开更多
文摘In the context of the integrated development of artificial intelligence and the power industry,the State Grid has carried out in-depth research on the research and development and application of the Guangming Power model.The model is based on the knowledge of the power industry,cognitive computing as the core,and knowledge service as the goal,and has the characteristics of large sample,large calculation,large parameters,large knowledge,and large tasks,providing important support for the intelligent transformation of the power industry.This study focuses on the efficient operation of Guangming Power large model,builds a trinity operation system of“model iterative optimization,sample full-process governance,and computing power resource collaboration”,and drives the continuous improvement of model capabilities and compliance development through a two-level collaboration mechanism.With the goal of“building a strong and excellent Bright Power Large Model”,the study clearly states that it is necessary to accelerate the construction of a two-level collaborative operation system,and promote model iterative optimization,capability evaluation,service monitoring and compliance review on a regular basis.At the same time,focusing on the whole process of R&D and application of large and small models,starting from four aspects:computing power planning and layout,allocation and scheduling,adaptation and optimization,monitoring and analysis,strengthening the application monitoring and analysis of two-level intelligent computing centers,and building an efficient computing power resource application and supply system to continuously improve its operation and service capabilities.
