T/SAIAS 0013.2-2023 Active

Scope: This document establishes the basic functional framework of general artificial intelligence (AI) large models in the education field. This document is applicable to the design and implementation of general AI large models in the education field; Main technical content: This document mainly specifies the basic functional framework of general AI large models in the education field. This part is applicable to the design and implementation of general AI large models in the education field. The education general AI large model framework includes five modules: the large model layer, the application layer, external tools, external agents, and the large model interface layer. 1. The large model layer is divided into the basic large model layer and the fine-tuning layer. The fine-tuning layer is used to fine-tune the large model into a large model specific to a particular education stage, and it includes two components: the fine-tuning strategy and the specific education large model. 2. The application layer completes various teaching tasks based on the functions provided by the education large model. External tools refer to external modules used to assist in solving teaching tasks by calling the model, while external agents are other intelligent systems that interact with the current education large model. 3. The large model interface layer is located between the application layer and the large model layer, and is responsible for data transmission between the application layer and the large model layer. 4. The deployment method of the large model is one of the key factors determining whether the large model can be used, and it will affect the model's inference speed, computing resources and costs, data privacy, and security, etc. According to the different application scenarios, public cloud, private cloud, and private deployment methods can be adopted.