Package dependence reproducibility.

[Jon-Croz]

Hey folks,

This is a nice library, but I'm running into package conflicts with Python, TF-Keras and Numpy. It seems like this library assumes an older version of Keras from tensorflow (in contrast to current stand-alone keras), but it seems difficult to trace exactly what versions of Python / Keras / TF / Numpy this is intended to work with.

What are the exact Python / Keras / Numpy versions used when this package was developed and released? I'm looking at 2020-era Python / TF / Numpy releases but still haven't been able to resolve something that actually works without throwing AttributeErrors.

Thanks!

[nitko12]

Have you found versions that work?

[lucaskdc]

So far i remember i had success using it with python 3.9.1, numpy 1.19.5 and tensorflow 2.5.0

[Jon-Croz]

Interesting. I just tried your configuration, and others as well but still run into the same problems. Have any of you seen something like this from keras2c?

File ".../keras2c/check_model.py", line 118, in check_layer
activation = layer.get_config().get('activation').get('class_name')
AttributeError: 'str' object has no attribute 'get'

[anchal-physics]

I got keras2c running with the attached conda env. I
keras2c.yml.zip
You can create the environment by running conda env create -f keras2c.yml

To create this environment, in requirements.txt, I changed line 5 to tensorflow == 2.15.0 and created a conda environment with python=3.11.5 and ran python -m pip install requirements