Deep learning model evaluation method and system, storage medium and equipment

Abstract

The invention provides a deep learning model evaluation method and system, a storage medium and equipment. The method comprises the following steps: extracting common characteristics of a plurality of deep learning models in a data preprocessing process in respective model training processes to obtain a common base class; taking other independent features in the data preprocessing process as respective inheritance classes; in response to evaluation on one deep learning model in the multiple deep learning models, obtaining an original training data set, and performing base class-based preprocessing on the original training data set in the deep learning model to obtain a base class data set; carrying out inheritance class preprocessing on the base class data set to obtain sample data; sequentially performing forward calculation, loss degree calculation and reverse calculation on the deep learning model based on the sample data to obtain an output result; and comparing the output result with the original training data set to obtain an evaluation result of the deep learning model. According to the invention, the evaluation efficiency of the deep learning model is improved.

Description

technical field [0001] The present invention relates to the technical field of artificial intelligence, in particular to a deep learning model evaluation method, system, storage medium and equipment. Background technique [0002] With the rapid development of artificial intelligence technology, more and more deep learning models have been applied to industrial deployment through training, such as resnet50 (a deep residual network) for image classification, ssd (deep learning target recognition algorithm), dlrm (deep learning recommendation model) for intelligent recommendation, bert (a pre-trained language representation model) for machine translation, etc. New models are constantly being proposed, and with the continuous rise of training chips, various manufacturers and enterprises are paying more and more attention to the performance efficiency of deep learning models on running software and hardware, such as the resource utilization of each model when training on chips. ...

AI Technical Summary

Problems solved by technology

[0006] In view of this, the purpose of the present invention is to propose a deep learning model evaluation method, system, storage medium and equipment to solve the problem of poor multi-domain model extension support, evaluation software and evaluation problems existing in the AI ​​benchmark evaluation system in the prior art. Inconsistent standards and low evaluation efficiency

Images