Gradient Centralization TensorFlow Versions Save

Instantly improve your training performance of TensorFlow models with just 2 lines of code!

v0.0.3

3 years ago

This release includes some fixes and improvements

✅ Bug Fixes / Improvements

  • Allow wider versions for TensorFlow and Keras while installing the package (#14 )
  • Fixed incorrect usage example in docstrings and description for centralized_gradients_for_optimizer (#13 )
  • Add clear aims for each of the examples of using gctf (#15 )
  • Updates PyPi classifiers to clearly show the aims of this project. This should have no changes in the way you use this package (#18 )
  • Add clear instructions for using this with custom optimizers i.e. directly use get_centralized_gradients however a complete example has not been pushed due to the reasons mentioned in the issue (#16 )

v0.0.2

3 years ago

This release includes some fixes and improvements

✅ Bug Fixes / Improvements

  • Fix the issue of supporting multiple modules
  • Fix multiple typos.

v0.0.1

3 years ago

This is the initial version of the Gradient-Centralization-TensorFlow package.

Features:

  • Implement Gradient centralization for optimizers using tf.keras.optimizer.Optimizers base class
  • Supports custom optimizers
  • Pre-built optimizers implementing GC for testing purposes.

Thanks, @ialimustufa for his contributions to this package.