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

Migrate to TF 2.0/tf.keras #157

Closed
minimaxir opened this issue Dec 1, 2019 · 5 comments
Closed

Migrate to TF 2.0/tf.keras #157

minimaxir opened this issue Dec 1, 2019 · 5 comments

Comments

@minimaxir
Copy link
Owner

I had made textgenrnn with external Keras since native TF was missing features. Now that there is parity, I am OK with merging it back into native TF with TF 2.0 support. textgenrnn does not use much custom Keras code so it should be a relatively simply change; the concern is not breaking old models, which may be possible due to the SavedModel change.

TF 2.1 also has TPU/mixed precision support which will be very helpful for training performance.

@minimaxir
Copy link
Owner Author

Handling work in TF-2.0 branch.

@minimaxir
Copy link
Owner Author

dammit I forgot about the AttentionWeightedAverage layer. Keras now has a native Attention layer but I suspect they are not compatible.

@ZerxXxes
Copy link

Would it be hard to add support for the MultiWorkerMirroredStrategy so that I can train on multiple workers at the same time?

@walter-grace
Copy link

The google colab needs updated to TF 2.x as well ;
https://drive.google.com/file/d/1mMKGnVxirJnqDViH7BDJxFqWrsXlPSoK/view

@minimaxir
Copy link
Owner Author

Done in 2.0

@minimaxir minimaxir unpinned this issue Feb 3, 2020
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

No branches or pull requests

3 participants