search

vineetjohn / linguistic-style-transfer

Neural network parametrized objective to disentangle and transfer style and content in text
Apache License 2.0
138 stars 33 forks source link

Selective backprop #1

Closed vineetjohn closed 6 years ago

vineetjohn commented 6 years ago

The gradients should be propagated through separate sets of differentiable variables for both the Adversarial component and the Reconstruction component.

The sets are not mutually exclusive. They will have some common variables to update.

vineetjohn commented 6 years ago

Already handled by default in Tensorflow based on the dependency graph.