After building a deep learning network architecture, I customized a tensor and passed it into the network. I wanted to test the results, and then the following error occurred:
Only some screenshots are taken here. From the error report, it can be roughly seen that there is a problem with the convolution operation during the forward propagation process. According to the error message, the root function that caused the error report was found:
It was found that the kernel and stride were not specified when defining the convolution operation:
Plus it can run normally, a very low-level error! ! !