AI Model Framework Conversion for Cross-Device Deployment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In AI applications, model training on terminal devices with insufficient computing power or capability can be hindered, and the deployment of AI models is adversely affected by framework differences between network and terminal devices, leading to impaired interaction effects.
Innovation Solution
A model interaction method for a heterogeneous AI framework that involves determining and exchanging framework information between network and terminal devices, allowing for the adaptation and conversion of AI models to ensure compatibility and accurate transmission.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If model training is performed on terminal devices with insufficient computing power, then training speed and efficiency deteriorate, but using network devices for training increases computing resource consumption
Solution Approach 1:
The patent implements dynamic model interaction where the terminal device can switch between different model frameworks (first framework for local training, second framework for network device training) based on real-time assessment of computing power availability and task requirements, enabling flexible allocation of training resources
Solution Approach 2:
The patent introduces a model framework conversion mechanism that acts as an intermediary between terminal device and network device frameworks, enabling seamless model transmission and execution across different computing environments without requiring complete retraining
2Reliability
If different AI model frameworks are used by network device and terminal device, then framework compatibility deteriorates, but unifying frameworks reduces system adaptability
Solution Approach 1:
The patent changes the framework parameter state by converting models between first framework (terminal device) and second framework (network device), allowing the same model to adapt to different framework environments through parameter transformation rather than requiring fixed framework compatibility
Solution Approach 2:
The patent creates a universal model interaction mechanism that can operate across multiple framework types, where the model can be executed on either terminal device or network device regardless of the specific framework used, achieving multi-framework compatibility while maintaining system adaptability
Data Source
AI summary
A model interaction method for a heterogeneous Artificial Intelligence (AI) framework includes determining first framework information supported by a terminal device, and determining a second AI model framework. The first framework information includes a first AI model framework supported by the terminal device.