Disrepancy in architecture in paper vs implemented version
Froskekongen opened this issue · comments
Erlend Aune commented
To me it seems that the architecture in the referenced paper is quite different from the one that is implemented right now.
In particular:
- the use of relu and leakyrely activations.
- batchnormalisation between the layers
- striding instead of pooling
BTW: I really like this initiative (looking to use adversarial nets within keras quite soon.)
Brett Beaulieu-Jones commented
The paper also uses Adam instead of SGD