Unlock instant, AI-driven research and patent intelligence for your innovation.

Model parameter processing method, device, electronic equipment and storage medium

A technology of model parameters and processing methods, applied in the field of machine learning, can solve problems such as the influence of model size, and achieve the effect of reducing size and avoiding model parameter distortion

Active Publication Date: 2020-10-16
WUHAN DOUYU NETWORK TECH CO LTD
View PDF6 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] The object of the present invention is to provide a model parameter processing method, device, electronic equipment and storage medium to solve the problem that the size of the model affects the electronic equipment when the machine learning model is applied to the electronic equipment

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Model parameter processing method, device, electronic equipment and storage medium
  • Model parameter processing method, device, electronic equipment and storage medium
  • Model parameter processing method, device, electronic equipment and storage medium

Examples

Experimental program
Comparison scheme
Effect test

no. 1 example

[0026] Please refer to figure 2 , figure 2 It is a flowchart of steps of a method for processing model parameters provided by a preferred embodiment of the present invention. The method may include the steps of:

[0027] Step S101, according to the model to be processed, obtain a parameter set to be compressed corresponding to the model to be processed.

[0028] The aforementioned model to be processed may be a model parameter file generated after machine learning function training. The above-mentioned model parameter file may include multiple model parameters.

[0029] For a better compression effect, the above-mentioned model to be processed may be a model that has undergone model pruning after training.

[0030]In the embodiment of the present invention, the above-mentioned parameter set to be compressed is a model parameter set corresponding to the model to be processed. The above-mentioned set of parameters to be compressed may be one set or multiple sets. Each se...

Embodiment approach

[0033] As an implementation, the following query statement can be used:

[0034] f_max=A[0];

[0035] f_min=A[0];

[0036] For(inti=0; i

[0037] {

[0038] If(f_max

[0039] {

[0040] f_max=A[i];

[0041]}

[0042] Else if(f_min>A[i])

[0043] {

[0044] f_min=A[i];

[0045]}

[0046]}

[0047] The above-mentioned f_max represents the parameter value of the first model parameter. First, the parameter value of the first model parameter in the array is assigned to f_max, and f_max is compared with the parameter value of each model parameter in the array in turn. When there is a model parameter greater than the f_max, change the value of f_max to the model parameter, and continue to repeat the comparison until the comparison with the last model parameter in the array is completed, and the value of f_max is the parameter value of the first model parameter . The above-mentioned f_min represents the parameter value of the second model parameter. First, the ...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The embodiment of the invention provides a model parameter processing method and device, electronic equipment and a storage medium. The method comprises the steps that a parameter set to be compressedcorresponding to a model to be processed is acquired, wherein the parameter set to be compressed comprises multiple model parameters; a compression strategy is determined according to the model parameters in the parameter set to be compressed. Each model parameter in the parameter set to be compressed is compressed according to the compression strategy, and storage parameters of a data type corresponding to the compression strategy are obtained. According to the scheme, the compression strategy is flexibly selected according to the specific condition of the model parameters, the data type corresponding to the model parameters is converted to lower the size of a model, and the method is simple and efficient. Meanwhile, the proper compression strategy is selected, and distortion of the model parameters in the compression process is avoided.

Description

technical field [0001] The present invention relates to the technical field of machine learning, in particular to a model parameter processing method, device, electronic equipment and storage medium. Background technique [0002] Machine Learning (ML) is a multi-field interdisciplinary subject, involving probability theory, statistics, approximation theory, convex analysis, algorithm complexity theory and other disciplines. It is the core of artificial intelligence and the fundamental way to make electronic equipment intelligent, and its application pervades all fields of artificial intelligence. [0003] Machine learning algorithms are at the heart of machine learning. The accuracy of a machine learning algorithm depends on the corresponding model parameters. The model parameters are obtained by training a large number of samples, the number is large, and the storage space is also large. If the machine learning model is deployed on the server side, the size of the model ...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Patents(China)
IPC IPC(8): G06N20/00
CPCG06N20/00
Inventor 李亮张文明陈少杰
Owner WUHAN DOUYU NETWORK TECH CO LTD