Model online migration training method and device based on resistive random access memory and chip

The invention discloses a model online migration training method and device based on a resistive random access memory and a chip. The method comprises the steps that neural network weights corresponding to all layers of a neural network model needing to be mapped to the resistive random access memory are obtained, the neural network weights are obtained on the basis of the offline training process of the neural network model, conductance offset functions of multiple conductance supported by the resistive random access memory are obtained, and therefore the resistive random access memory is obtained according to the conductance offset functions; on-line migration training is carried out on the neural network model through iterative training so as to update the weight of the neural network, and further, the conductance value of the memristor array is updated according to the updated weight of the neural network so as to obtain a trained resistive random access memory. Therefore, the weight of the neural network