Error for MNIST autoEncoder

[ir1979]

Your file produces Error:
RuntimeError: output with shape [1, 28, 28] doesn't match the broadcast shape [3, 28, 28]

It probably due to the gray-scale image downloaded automatically.

[thunfischtoast]

The normalize-operation in the transform expects 3 channels. For a gray-scale image with 1 channel change

transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))

to

transforms.Normalize((0.5,), (0.5,))

[Postroggy]

Lol，the Conv_autoencoder python code is way too too simply and way toto crude