Model training method and apparatus, and computing device
Through the training process of predicting the first model using the second model, the problem of being unable to effectively predict the model's optimal index data in the prior art is solved, and the effects of resource saving, trial and error cost reduction and training efficiency improvement are achieved.
CN119692417APending Publication Date: 2025-03-25HUAWEI CLOUD COMPUTING TECHNOLOGIES CO LTD
0 Cites 2 Cited by
Patent Information
- Application Number
- CN202311607988.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2023-09-22
- Filing Date
- 2023-11-27
- Publication Date
- 2025-03-25
AI Technical Summary
Technical Problem
Existing model training platforms cannot effectively predict and find the optimal metric data for the model during the training process, resulting in waste of resources and high trial and error costs.
Method used
The second model predicts the training process or training performance of the first model, helping users find the best indicator data, reduce trial and error costs and improve training efficiency.
Benefits of technology
It avoids waste of resources during model training, saves training resources, reduces trial and error costs, and improves the training speed and efficiency of the model.
✦ Generated by Eureka AI based on patent content.
Abstract
The invention provides a model training method and device and computing equipment, and the method comprises the steps: receiving a plurality of values of configuration parameters of a first model inputted by a user, the configuration parameters of the first model comprising training parameters and / or model parameters of the first model; predicting a plurality of training index data of the first model corresponding to a plurality of values of the configuration parameters according to a second model, wherein the training index data of the first model comprises training process data of the first model and / or hardware index data consumed by a server for training the first model; determining target training index data from the plurality of training index data of the first model, and sending a target value corresponding to the target training index data in the plurality of values of the configuration parameter to the user; and receiving the target value confirmed by the user, and training the first model based on the target value. The method can improve the training speed and training efficiency of the model.
Need to check novelty before this filing date? Find Prior Art