pytorch中data shape和卷积层的weight shape

# input_data shape
Input: (batch_size, in_channel, width, height)
# conv layer
class torch.nn.Conv2d(in_channels, out_channels, kernel_size, stride=1, padding=0, dilation=1, groups=1, bias=True)
input: (Batch_size, C_in, H_in, W_in)
output: (Batch_size, C_out, H_out, W_out)

weight(tensor): (out_channels, in_channels,kernel_size)
bias(tensor): (out_channel)