Multi-task model training method, information recommendation method, device and equipment

• Main Authors: ZHANG WENHUI, WU ZHIHUA, YU DIANHAI, WANG ZHEN
• Format: Patent
• Language: Chinese; English
• Published: 20.09.2022
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; PHYSICS

The invention provides a multi-task model training method, and relates to the technical field of artificial intelligence, in particular to the technical field of deep learning, cloud computing, multi-task parallel processing and data search. According to the specific implementation scheme, sample behavior data of a sample object is input into a shared sub-model, behavior feature information of the sample object is obtained, and the behavior feature information comprises multiple pieces of behavior feature sub-information; according to the behavior feature sub-information related to the task processing sub-model, obtaining a sub-loss value of the task processing sub-model; determining a target gradient value corresponding to the sub-loss value according to the sub-loss value of the task processing sub-model; determining a weight value corresponding to the sub-loss value according to the plurality of target gradient values; and training the multi-task model according to the plurality of sub-loss values and the