A data saving method and device, vehicle, electronic equipment and storage medium

By determining the computation order in the neural network and separating the output storage of key operators, the problem of excessive memory consumption in the inference engine is solved, achieving memory saving without affecting output accuracy.

CN117217256BActive Publication Date: 2026-06-19BEIJING CO WHEELS TECH CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
BEIJING CO WHEELS TECH CO LTD
Filing Date
2022-06-01
Publication Date
2026-06-19

AI Technical Summary

Technical Problem

Existing inference engines consume excessive memory when computing complex neural networks, and reducing the number of model parameters can affect the accuracy of the final output.

Method used

By determining the computation order of neural network operators, the outputs of M key operators are separated into different storage spaces, allowing the deletion of the outputs of non-key operators under specific conditions, thereby reducing memory usage.

Benefits of technology

Without reducing the number of parameters in the neural network model, memory usage was reduced while maintaining the model's computational efficiency and output accuracy.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117217256B_ABST
    Figure CN117217256B_ABST
Patent Text Reader

Abstract

This application relates to a data storage method, apparatus, vehicle, electronic device, and storage medium. The method includes: determining the computation order of N operators included in a neural network; determining M operators from the N operators based on the computation order; calculating the i-th operator among the N operators according to the computation order to obtain the output of the i-th operator; if the i-th operator is one of the M operators, then saving the output of the i-th operator to a first type of storage space, wherein the output of the i-th operator saved in the first type of storage space cannot be deleted after calculating the (i+1)-th operator but before calculating the k-th operator; if the i-th operator is not one of the M operators, then saving the output of the i-th operator to a second type of storage space, wherein the output of the i-th operator saved in the second type of storage space can be deleted after calculating the (i+1)-th operator. This method can solve the problem of excessive memory consumption during the computation of a neural network model.
Need to check novelty before this filing date? Find Prior Art