Distributed training method, device and computer equipment for machine learning model

Abstract

This application relates to a distributed training method of a machine learning model, which is executed by working nodes of a distributed node cluster. The distributed node cluster can be realized through a cloud server. The method includes: obtaining more than one set of training samples, and according to the machine learning model The current training parameters process each group of training samples separately to obtain the corresponding parameter gradient; based on the parameter gradient corresponding to each group of training samples, determine the local gradient of the current time; transmit the local gradient of the current time to the parameter node; transmit The local gradient of is used to instruct the parameter node to determine the current global gradient based on the local gradient transmitted by each working node in the distributed node cluster, and update the training parameters according to the current global gradient; obtain the updated Training parameters, based on the updated training parameters, continue to perform model training on the machine learning model until the training ends when the training end conditions are met. This method can improve the efficiency of model training.

Description

technical field [0001] The present application relates to the technical field of artificial intelligence, in particular to a distributed training method, device and computer equipment of a machine learning model. Background technique [0002] With the development of computer technology, the use of machine learning for data processing has become more and more common. The process of machine learning usually refers to computer equipment building an initial model, inputting sample data into the initial model, analyzing the input sample data through a series of algorithms, and updating the model parameters of the initial model through iterative training to obtain the final suitable model. model. [0003] Since the machine learning model needs to perform feature learning on a large amount of sample data during the training process, in order to improve the learning efficiency, in the traditional solution, multiple working nodes are used to undertake the training task. In the trad...

AI Technical Summary

Problems solved by technology

[0005] Based on this, it is necessary to provide a distributed training method, device and computer equipment for a machine learning model that can improve the efficiency of model training for the above-mentioned technical problem of low model training efficiency

Images