Method and device for deep learning task, and computer readable storage medium

摘要

To provide a processing method and device for a deep learning task, and a computer-readable recording medium for improving the robustness of a training process that ensure the speed of model training, can reduce or avoid the problem that data is lost due to the occurrence of an abnormal incident, and can avoid an excessive influence on the model training and the other nodes.SOLUTION: A processing method includes: having a local storage and a network storage mounted on an application container; introducing sufficient training data before every model training to perform training of a model based on the training data introduced to the local storage; in a model training process, performing both reading of the training data and storage of an intermediate model file in the local storage; and transferring the intermediate model file of the model training to a network storage device in a timely manner.SELECTED DRAWING: Figure 2