Kernel size can't be greater than actual input size solution

报错信息：RuntimeError: Calculated padded input size per channel: (49). Kernel size: (64). Kernel size can't be greater than actual input size

Solution: No need to modify the size of the convolution kernel, just modify the stride and padding

before fixing:

self.conv1 = nn.Conv1d(in_channels=3, out_channels=3, kernel_size=64, stride=16, padding=24)

After modification:

self.conv1 = nn.Conv1d(in_channels=3, out_channels=3, kernel_size=64, stride=2, padding=60)