Need tensorflow 2.2.0 for Python 3.6

[redsigma]

The version 2.4.3 of keras does not work with tensorflow 2.2.0 on python 3.7

File "/home/pi/.pyenv/versions/3.7.7/lib/python3.7/site-packages/tensorflow_core/python/keras/saving/save.py", line 146, in load_model
return hdf5_format.load_model_from_hdf5(filepath, custom_objects, compile)
File "/home/pi/.pyenv/versions/3.7.7/lib/python3.7/site-packages/tensorflow_core/python/keras/saving/hdf5_format.py", line 184, in load_model_from_hdf5
training_config, custom_objects))
File "/home/pi/.pyenv/versions/3.7.7/lib/python3.7/site-packages/tensorflow_core/python/keras/saving/saving_utils.py", line 242, in compile_args_from_training_config
sample_weight_mode = training_config['sample_weight_mode']
KeyError: 'sample_weight_mode'

Even if i pass a hdf5 checkpoint or a fully saved model .h5 , the error is the same.

I have tested another built of tensorflow 2.2.0 but for a amd64 linux machine with python 3.6.10 and it worked without any problems.

If you could post reupload the python3.6 builds it would be great.

[Qengineering]

Please find the Python3.6 wheel (tensorflow-2.2.0-cp36-cp36-linux_aarch64.whl) at our 64 bit section. https://github.com/Qengineering/TensorFlow-Raspberry-Pi_64-bit