Weight data processing method, system and device and readable storage medium

The invention provides a weight data processing method, system and device and a readable storage medium, and the method is applied to a learning engine, and comprises the steps: firstly obtaining target sample data; in response to a received target sampling request, sampling target sample data according to a target sampling strategy to obtain sampled data, forming a to-be-trained data set by using the sampled data, the to-be-trained data set comprising multiple pieces of sampled data, and performing training on the to-be-trained data set; multiple times of weight updating iteration of the machine learning model is carried out based on the multiple pieces of sampled data in the to-be-trained data set, model weight data after each time of weight updating iteration is obtained, and after the model weight data after each time of weight updating iteration is obtained, if a pushing request is received, the machine learning model is pushed to the to-be-trained data set. And pushing the model weight data to a predict