Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

DL4J: MultiLayerNetwork/CompGraph rnnGetPreviousState + wrapper layers #7437

Closed
AlexDBlack opened this issue Apr 3, 2019 · 1 comment

Comments

@AlexDBlack
Copy link
Contributor

commented Apr 3, 2019

Wrapper layers like LastTimeStep and FrozenLayer aren't Recurrent layers, so they can't return the previous state in the current implementation - even though it exists and we could return it...
https://github.com/deeplearning4j/deeplearning4j/blob/ce5f648509f3cafe0ef1dc065755d9481db420a2/deeplearning4j/deeplearning4j-nn/src/main/java/org/deeplearning4j/nn/graph/ComputationGraph.java#L3445-L3450

@AlexDBlack AlexDBlack self-assigned this Apr 3, 2019

AlexDBlack added a commit that referenced this issue Apr 4, 2019
[WIP] Evaluation and RNN fixes for wrapper layers (#7439)
* #7435 Fix Evaluation reset - labels issue

* #7437 Fix rnnGetPreviousState/rnnTimeStep with wrapper layers

* Small fix for TBPTT + wrapper layers

* Small Evaluation fix
@lock

This comment has been minimized.

Copy link

commented May 4, 2019

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.

@lock lock bot locked and limited conversation to collaborators May 4, 2019

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Projects
None yet
1 participant
You can’t perform that action at this time.