Adding optional parameter activation_dtype to models #327
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
As I mentioned on issue #323 , when using mixed_precision, the training does not converge when softmax conversion is also converted. So, to overcome this problem, I've create a parameter activation_dtype that allows user to set the data type of the activations.
I've modified all models and also the EfficientNet ones (I'll pull request the changes to the other repository).
I've tested mixed precision training with the proposed changes and all went well. I've also tested training the normal way and my changes did not break anything, also all unittests passed.
I've also updated the requirements to include noisy students weights to segmentation_models using EfficientNet.
Please feel free to suggest any changes to this pull request, I just want to contribute to this great project.