AI Model Fine-Tuning via Configuration and Mode Information
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current model training methods in communication technologies, particularly in AI modules, face challenges with poor training effects, which negatively impact communication performance.
Innovation Solution
A model fine-tuning method and apparatus that involves obtaining and utilizing fine-tuning configuration-related information and mode information of an AI model to improve the training efficiency and effectiveness.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If transfer learning manner is used for model training, then training process can be simplified, but model training effect is poor
Solution Approach 1:
The patent applies preliminary action by pre-processing training data to generate fine-tuning configuration information and fine-tuning mode information before the actual fine-tuning process. This pre-computed information is then used during fine-tuning to guide the process, improving training effectiveness while maintaining simplicity. The fine-tuning configuration-related information and fine-tuning mode information are prepared in advance and transmitted to the terminal device for efficient model adjustment.
2Manufacturing precision
If fine-tuning configuration information and mode information are transmitted and utilized, then model training effect is improved, but information transmission and processing complexity increases
Solution Approach 1:
The patent extracts and separates fine-tuning configuration-related information and fine-tuning mode information from the overall training process. These extracted information elements are independently generated, transmitted, and processed. By taking out these critical information components, the system can efficiently transmit only the necessary data to the terminal device without overwhelming it with complete training datasets, thus improving training effect while managing complexity.
Solution Approach 2:
The patent introduces fine-tuning configuration information and fine-tuning mode information as intermediary elements between the training system and the terminal device. These intermediaries carry essential training guidance without requiring direct transmission of large training datasets. The intermediary information acts as a bridge, enabling efficient information exchange and reducing the complexity of direct data transmission while maintaining high training effectiveness.
Data Source
AI summary
This application discloses a model fine-tuning method and apparatus, and a device. The model fine-tuning method includes: obtaining, by a first device, first target information; fine-tuning, by the first device, the first Artificial Intelligence (AI) model based on the first information or the second information. The first target information includes first information and/or second information, the first information at least includes fine-tuning configuration-related information of a first AI model, and the second information at least includes fine-tuning mode information of the first AI model.