Multi-task model training method and device and multi-task identification method and device

The invention provides a multi-task model training method and device and a multi-task recognition method and device, and relates to the technical field of deep learning and artificial intelligence. The method comprises the steps of obtaining a training sample set, wherein the training sample set comprises a plurality of sample elements and task labeling results corresponding to the sample elements; performing layer-by-layer embedding mapping on the sample elements based on a feature sharing network in the multi-task model to obtain shared feature representation of each layer; performing a multi-attention mechanism on each layer of shared feature representation based on multi-attention networks of different tasks in the multi-task model, and obtaining attention masks of each layer of different tasks; and for different tasks, based on the shared feature representation of each layer and the attention mask of each layer, obtaining a prediction result, and according to the task labeling result and the prediction r