Bug: should not use cross entropy to train Mixed_NN_ADQC network

[ifsheldon]

Python-for-Tensor-Network-Tutorial/AD_Circuits/Mixed_NN_ADQC.py

Line 40 in 4c89b07

criterion = nn.CrossEntropyLoss()

This line has a minor bug in that the below line gives normalized probabilities, as you mentioned in your lecture.

Python-for-Tensor-Network-Tutorial/AD_Circuits/Mixed_NN_ADQC.py

Line 49 in 4c89b07

y = probabilities_adqc_classifier(psi1, 1, 2)

So, the criterion should not be cross entropy. Instead, it should be like this line

Python-for-Tensor-Network-Tutorial/Algorithms/ADQC_algo.py

Line 57 in 4c89b07

loss = criteria(tc.log(psi1), lbs)