Model training sample verification method and device, equipment, medium and product

Abstract

The invention discloses a model training sample verification method and device, equipment, a medium and a product, and the method comprises the steps: responding to a category target annotation event acting on commodity query word samples and commodity title samples in a candidate sample pool, and obtaining a first annotation commodity category annotated by the event as the commodity query word samples and the commodity title samples; calling a first sample category annotation model, and annotating a commodity query word sample with a first annotation commodity category and a second annotation commodity category corresponding to the commodity title sample; pushing the commodity query word samples or commodity title samples with the same first marked commodity category and second marked commodity category to a training sample library; and according to a second sample class target injection model which is newly trained to be converged based on the training sample library, intervening in the first sample class target injection model for updating. According to the method, the training samples are automatically labeled and verified, the efficiency utilization of model resources is formed, and the reasoning ability of the model needing to be deployed online is effectively improved.

Description

technical field [0001] The present application relates to the field of neural network models, and in particular, to a method for verifying model training samples, and also to devices, equipment, non-volatile storage media and computer program products corresponding to the method. Background technique [0002] There are various types of online services based on neural network models in existing Internet platforms. For example, online services used to provide users with intelligent customer service generally include model services corresponding to neural network models used for semantic reasoning. In the online services that provide users with commodity classification or commodity query services, there are generally model services constructed based on neural network models. The neural network model in the model service is trained with new labeled sample data, thereby improving the business processing efficiency of online services built based on the neural network model and opt...

AI Technical Summary

Problems solved by technology

[0003] However, in the existing platforms, when the labeled sample data for training is pushed to the neural network model, the labeled sample data is often not verified. If the neural network model directly uses unverified labeled sample data, it is undoubtedly for the platform to pass The stability of the system operation effect of the online service provided by the neural network model

Images