Calculation graph processing method and device, readable medium and electronic equipment

Abstract

The invention relates to a computational graph processing method and device, a readable medium and electronic equipment, and the method comprises the steps: obtaining a target computational graph corresponding to a to-be-trained machine learning model, and carrying out the node fusion processing of the target computational graph, so as to obtain a fusion computational graph comprising a plurality of sub-graphs; according to multiple preset segmentation modes of each operator node, the operator nodes in the fusion calculation graph are effectively segmented to obtain multiple segmentation data flow graphs corresponding to the fusion calculation graph, and the segmentation data flow graphs are target calculation graphs with the effective segmentation modes; and then, according to the segmented data flow diagram, searching a parallel configuration strategy corresponding to the target calculation diagram, so that a model training strategy of mixed parallel of pipeline parallel, data parallel and model parallel can be automatically and efficiently generated, the model training efficiency of a machine learning model can be effectively improved, and the time for model training can be shortened.

Description

technical field [0001] The present disclosure relates to the field of data processing, and in particular, to a computing graph processing method, apparatus, readable medium, and electronic device. Background technique [0002] As a powerful tool for task automation, machine learning can automate tasks in various types of applications. Therefore, in recent years, machine learning, especially deep neural networks, have been widely used in various challenging tasks. Such as computer vision, natural language processing, speech recognition, etc., and have made major breakthroughs. However, with the further deepening of the application of neural network, the parameters and calculation amount of the neural network model are also increasing, and the training time of the neural network model is also prolonged. technical problem. SUMMARY OF THE INVENTION [0003] This Summary is provided to introduce concepts in a simplified form that are described in detail in the Detailed Descri...

AI Technical Summary

Problems solved by technology

However, with the further deepening of the application of neural networks, the parameters and calculations of the neural network model are also increasing, and the training time of the neural network model is also prolonged. Therefore, how to efficiently train the neural network model has become an urgent problem to be solved. technical problem

Images