Data recovery method and device, electronic equipment and storage medium

The invention provides a data recovery method and device, electronic equipment and a storage medium, and relates to the technical field of computers, in particular to the technical field of deep learning and model training. The specific implementation scheme is as follows: when a specified opportunity is reached, generating snapshot data of a to-be-trained sub-model in a training process; when a preset model data recovery condition is met, selecting target snapshot data to be utilized from the generated snapshot data; loading model intermediate result data represented by the target snapshot data to the to-be-trained sub-model to obtain the to-be-trained sub-model of which the data is recovered to target model data; according to the scheme, the model data of the federated model can be subjected to data recovery. 本公开提供了一种数据恢复方法、装置、电子设备及存储介质，涉及计算机技术领域，尤其涉及深度学习、模型训练技术领域。具体实现方案为：当达到指定时机时，生成处于训练过程的待训练子模型的快照数据；当满足预定的模型数据恢复条件时，从所生成的快照数据中，选取待利用的目标快照数据；将所述目标快照数据所表征的模型中间结果数据，加载在所述待训练子模型，得到数据恢复为目标模型数据的待训练子模型；通过本方案可以对联邦模型