The invention provides a neural network training method executed on an integrated circuit chip device and a related product. The neural network comprises multiple layers, the method comprises the following steps of receiving a training instruction, determining the first-layer input data and the first-layer weight data according to the training instruction, and using the computing device to executen-layer forward operation of the neural network through the first-layer input data and the first-layer weight data to obtain an nth output result; obtaining the nth output result gradient according to the nth output result, obtaining the nth reverse operation of the nth layer of reverse operation according to the training instruction, and obtaining the nth reverse operation complexity by the computing device according to the nth output result gradient, the nth layer of input data, the nth layer of weight group data and the nth reverse operation. The technical scheme provided by the inventionhas the advantages of small calculated amount and low power consumption.