confusion of enc_cnn

[sbzgyxzm]

In the enc-cnn file, it is written "# TODO: Encrypt the weights and biases when training the model". But this file seems to be only testing the accuracy on the encrypted model, but not training the model.

[rayhankinan]

For now, I end up deleting the encrypted training algorithm in order to clean my testing script. Although, you still can train all of the layer using encrypted input as I have defined the backward method on every autograd function. I'll add the encrypted training example as soon I finished polishing all of the encrypted layer! I'll notify you when this is done.

[sbzgyxzm]

Okay, thank you!!!