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.

Sign up for GitHub

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

Jump to bottom

Add support for CNTK #10

Closed
cesarsouza opened this issue Oct 1, 2017 · 1 comment
Closed

Add support for CNTK #10

cesarsouza opened this issue Oct 1, 2017 · 1 comment

Comments

@cesarsouza
Copy link
Owner

cesarsouza commented Oct 1, 2017 •
edited
Loading

While TensorFlow gradient's issue is pending, we can start adding support for CNTK to exercise the frontend.

From accord-net/framework#904
cesarsouza added a commit that referenced this issue Oct 1, 2017
@cesarsouza
Initial work to add support for CNTK as a new backend:
74adba1 
 - Separating TensorFlowBackend into a dedicated project;
 - Adding a new separate project for a CTNK CPU-Only backend;
 - Adding dynamic backend selection based on referenced assemblies;

Updates GH-10: Add support for CNTK
cesarsouza added a commit that referenced this issue Oct 2, 2017
@cesarsouza
Continuing to work on #10
d46eecd 
 - Adding unit tests for CNTK;
 - Adding a mechanism to repeat common unit tests for both TensorFlow and CNTK backends.
cesarsouza added a commit that referenced this issue Oct 3, 2017
@cesarsouza
Continuing to work on #10
78fccbe 
 - Continuing to add support for training;
 - Adding more CNTK backend implementations;
 - Updates #1: Contributing to Keras-Sharp;
cesarsouza added a commit that referenced this issue Oct 4, 2017
@cesarsouza
Adding more passing unit tests for CNTK (#10)
0c42c44 
 - Updates #1
@cesarsouza
Copy link
Owner Author

Backends have been added for CNTK. However, a few issues on CNTKs side currently prevent full models from being eventually learned. For instance, using the CNTK backend might result in multiple AccessViolationExceptions which are not present when using the TensorFlowSharp backend, and have thus nothing to do with the main engine (since the engine does not use any unmanaged code at all at this moment).

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
1 participant
@cesarsouza