foamliu/autoencoder

Convolutional Autoencoder with SetNet in PyTorch

autoencoder's Issues

pre_model is don't work well

Thank your code .However,I want to ask some questions.I have download pre-trained model weights into "models" folder then run: python demo.py .But , have error:RuntimeError :cuda runtime error (10): invalid device ordinal at torch/csrc/cuda/module.cpp:32?
I don't know why have error.Can you help me ?
Thx.

Leaking data in the max-pool indices

One RGBA pixel is 32 bits, so a 2x2 of pixels is 128 bits

Each max pool index stores 2 bits of data. The first convolutional block has 64 channels.
So there are 2*64 = 128 bits of data in the max pooling indices for that block. Those get passed straight to the end of the network.