The two standards have been successfully completed, with system capabilities for edge deployment of large models and the underlying operator capabilities of large language models, forming standardized links from edge-side software and hardware infrastructure, inference frameworks and tools to operator adaptation, development, services, interfaces and evaluation indicators, providing technical support for efficient deployment, stable operation, and cross-platform migration of large models in heterogeneous computing power environments. Relevant results help promote interoperability and collaborative optimization between chips, operator libraries, inference frameworks, edge computing platforms and upper level applications, reduce the cost of cross-platform deployment, adaptation and migration of large models, and promote large-scale implementation of large models in edge application scenarios such as industry, communications, transportation, and energy.

The two standards have been successfully completed, with system capabilities for edge deployment of large models and the underlying operator capabilities of large language models, forming standardized links from edge-side software and hardware infrastructure, inference frameworks and tools to operator adaptation, development, services, interfaces and evaluation indicators, providing technical support for efficient deployment, stable operation, and cross-platform migration of large models in heterogeneous computing power environments. Relevant results help promote interoperability and collaborative optimization between chips, operator libraries, inference frameworks, edge computing platforms and upper level applications, reduce the cost of cross-platform deployment, adaptation and migration of large models, and promote large-scale implementation of large models in edge application scenarios such as industry, communications, transportation, and energy.