Feature/batchnorm (#76) · phutoan31299/DeepLearningZeroToAll@255fac7 · GitHub

Commit

Feature/batchnorm (hunkim#76)

Browse files

* refactor: add a FC layer

1. Add a FC Layer
  - hunkim#73
2. Add a numpy doc style
3. Explicitly define an activation function in LSTMCell
  - default is tanh
4. Change to AdamOptimizer
5. Use `with` statement else the session must be closed explicitly
6. No linebreak for the 79 char limit

* refactor: nn.tanh -> tanh

* add: MNIST Batchnormalization layer

* fix: image

Loading branch information

kkweon authored and hunkim committed Mar 31, 2017

1 parent 293a77a commit 255fac7