Query-Driven Service Model Selection with Large-Model Screening
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for selecting the most suitable expert model for input data in a mixture of experts (MoE) architecture are inaccurate and inefficient.
Innovation Solution
An information processing method that involves obtaining a query statement, determining model identifiers, generating prompt words, and using a pre-trained large model to output screening parameters for candidate service models, followed by selecting a target service model based on these parameters, and inputting the query statement to obtain feedback information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a gating network or weight allocation strategy is used to select the most suitable expert model, then the model selection process can be automated, but the accuracy and efficiency of selection cannot be improved
Solution Approach 1:
A large model is introduced as an intermediary component to evaluate and select expert models. The large model processes query statements and generates screening parameters for multiple expert models, enabling accurate and efficient selection without requiring complex gating networks or weight allocation strategies.
Solution Approach 2:
The selection process transitions from traditional gating mechanisms to a parameter-based evaluation system. The large model generates screening parameters (such as relevance scores, quality metrics) that dynamically change based on the query statement and expert model characteristics, enabling flexible and accurate selection.
2Productivity
If traditional gating networks are used for expert model selection, then the architecture remains simple, but the selection accuracy and efficiency are insufficient
Solution Approach 1:
The large model performs preliminary evaluation of expert models by generating screening parameters before the final selection is made. This preliminary action allows the system to pre-rank and filter expert models, improving selection efficiency while maintaining manageable architecture complexity through structured evaluation steps.
Solution Approach 2:
The expert model selection system is segmented into distinct functional components: the large model for evaluation, the screening parameter generation mechanism, and the selection logic. This segmentation allows each component to be optimized independently, improving overall efficiency without proportionally increasing overall system complexity.
Data Source
AI summary
An information processing method, an electronic device, and a storage medium. The method includes: obtaining a query statement of a user, determining at least one model identifier of at least one candidate service model based on the query statement; generating at least one first prompt word based on the query statement and the at least one model identifier, inputting the at least one first prompt word into a pre-trained target large model, and outputting, by the target large model, at least one screening parameter of the at least one candidate service model based on the at least one first prompt word; determining a target service model from the at least one candidate service model based on the at least one screening parameter; and inputting the query statement into the target service model, and obtaining feedback information corresponding to the query statement.