Simulation implementation method, neural network compiler and computer readable storage medium

An implementation method and neural network technology, applied in the fields of computer-readable storage media, simulation implementation, and neural network compilers, can solve problems such as the inability to perform precision tests on ten-thousand-person test sets, so as to avoid time-consuming quantification, save costs, and speed up The Effect of Simulation Efficiency

Active Publication Date: 2022-02-01
浙江芯劢微电子股份有限公司
View PDF15 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0006] The purpose of this application is to provide a simulation implementation method, a neural network compiler, and a computer-readable storage medium to solve the problems in the above-mentioned prior art that can only simulate each intermediate layer of a neural network model and cannot perform a 10,000-person test set. Technical Issues of Accuracy Testing

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
  • Simulation implementation method, neural network compiler and computer readable storage medium

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0058] Such as figure 1 As shown, this embodiment includes a simulation implementation method, including the following steps:

[0059] Build a neural network compiler to receive quantization set pictures, multiple different types of neural network models, and a test set of 10,000 people. After the accuracy verification of the neural network compiler, the neural network model is simulated layer by layer.

[0060] The quantization set image is quantized by the neural network compiler to generate an executable file for the neural network model, and the 10,000-person test set is generated by the neural network compiler to generate the first input data, the first fixed-point feature file and the floating-point feature file.

[0061] Compare the first fixed-point feature file with the floating-point feature file, and output the accuracy table used for statistical neural network models; if the statistical results of the accuracy table meet the preset accuracy range, read the executab...

Embodiment 2

[0114] This embodiment includes a neural network compiler, which is applied to the simulation implementation method of Embodiment 1, including: sequentially connected network analysis module, network quantization module, network merging module, network storage module and network forward execution mod.

[0115] The network analysis module is used to receive quantization set pictures, multiple different types of neural network models and ten-thousand test sets, analyze and reconstruct the structure of the neural network model layer by layer, and at least obtain the input layer, output layer and One of layer operation name, layer parameter information and layer association information of the middle layer.

[0116] Specifically, the network analysis module analyzes the structure of the original neural network model layer by layer, and at least obtains one of the layer operation name, layer parameter information, and layer association information of the input layer, output layer, a...

Embodiment 3

[0126] A computer-readable storage medium. Computer instructions are stored on the computer-readable storage medium. When the computer instructions are executed by a processor, the steps of the method in Embodiment 2 are implemented.

[0127] Those skilled in the art should understand that the embodiments of the present invention may be provided as methods, apparatuses, or computer program products. Accordingly, the present invention can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) having computer-usable program code embodied therein.

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 invention discloses a simulation implementation method, a neural network compiler and a computer readable storage medium, and relates to the technical field of deep learning. The simulation implementation method comprises the following steps that: a quantization set picture quantifies a neural network model through the neural network compiler to generate an executable file, a ten-thousand-person test set generates first input data, a first fixed-point feature file and a floating-point feature file through a neural network compiler, and if a statistical result of a precision table conforms to a preset precision range, the executable file and the first input data are read for simulation of a neural network model. The invention has the beneficial effects that batch simulation of a plurality of different types of neural network models is realized, the correctness of transplantation to a chip or an FPGA is ensured, simulation is performed layer by layer for different types of neural network models, more simulation verification points are covered, the risk of chip tape-out is prevented, and meanwhile, comprehensive precision verification is carried out on a precision table for counting the neural network model.

Description

technical field [0001] The present application belongs to the technical field of deep learning, and more specifically, the present application relates to a simulation implementation method, a neural network compiler, and a computer-readable storage medium. Background technique [0002] With the development of Internet technology, the massive data collected provides enough scenarios for deep learning training. The development of intelligent algorithms based on convolutional neural networks relies on massive data. In the fields of image classification and object recognition, intelligent algorithms The accuracy has exceeded the recognition accuracy of humans. [0003] If the neural network algorithm wants to be implemented in the security field, it is necessary to parse the algorithm model trained on the server into a computer language recognizable by the embedded chip, so as to facilitate the installation and monitoring of security cameras. [0004] The process of implementin...

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 Applications(China)
IPC IPC(8): G06N3/063G06N3/04G06K9/62
CPCG06N3/063G06N3/065G06N3/045G06F18/22
Inventor 朱旭东吴春选
Owner 浙江芯劢微电子股份有限公司
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap