T/CA 605.3-2026 Technical Specification for Comprehensive Evaluation of Large Models
T/CA 605.3-2026 Technical Specification for Comprehensive Evaluation of Large Models
Basic Information
Scope
This document specifies the evaluation of large-scale models in three core fundamental tasks of natural language processing, including text classification, information extraction, and text-based question answering; in basic visual tasks of image recognition, including static image classification, object detection, and static image segmentation; and in fundamental speech recognition tasks, including speech-to-text conversion, voiceprint recognition, and environmental sound classification. This document is applicable to large-scale model R&D institutions, application service providers, third-party evaluation agencies, and industry users, who can use it to evaluate and test the performance of large-scale models in the three fundamental tasks of natural language processing, image recognition, and speech recognition. This document is not applicable to evaluations beyond the scope of fundamental tasks, such as multi-modal complex interactions and industry-specific deep customization.