Compatibility Issue with TensorFlow Addons and TensorFlow 2.14.0

[zapod838]

I encountered a compatibility issue while running AUCMEDI with TensorFlow 2.14.0. The TensorFlow Addons (TFA) library shows a warning that it supports TensorFlow versions from 2.11.0 up to but not including 2.14.0. Here is the warning message:

TensorFlow Addons (TFA) has ended development and introduction of new features. TFA has entered a minimal maintenance and release mode until a planned end of life in May 2024. Please modify downstream libraries to take dependencies from other repositories in our TensorFlow community (e.g., Keras, Keras-CV, and Keras-NLP).

TensorFlow Addons supports using Python ops for all TensorFlow versions above or equal to 2.11.0 and strictly below 2.14.0 (nightly versions are not supported). The version of TensorFlow you are currently using is 2.14.0 and is not supported.

Could the compatibility matrix or requirements be updated to address this issue?

[zapod838]

Its resolved thanks!

[muellerdo]

Hi @zapod838,

thanks for spotting & reporting this issue!

Sadly, TFA dropped support which is why a subpackage of AUCMEDI for Vision Transformers is not functional anymore.

I wrote a pull request to fix this issue:
faustomorales/vit-keras#45

In the meantime to get your AUCMEDI instance functional again:

• Install AUCMEDI via the requirements.txt from the source code
  or
• Downgrade the following packages:
  classification-models-3D==1.0.10 and tensorflow==2.14.0

Cheers,
Dominik

[zapod838]

Hi @muellerdo,

Sorry to bother you on this matter, I tried everything what you have suggested but still struggling with same issues!

Though pretrained models are available on https://zenodo.org/records/4573990, implementing them for now!

One more issue I am facing is to how should I use XAI functions from pretrained models? as in AUCMEDI package within xai folder there is a decoder file asking for xai_dict, which is unavailable at the moment.

Can you please either guide me how should I use XAI functions from pretrained models or how should I get xai_dict required in decoder.py.

Cheers,
Manish