problem in use batchNormalization in the first layer with tensorflow 2.2

30 views Asked by At

I need to use batchnormalization in the first layer and still use tensorflow 2.2, as the place where the trained network model will be sent has a tensorflow of 2.2 and cannot be updated. My code

There is no problem running the code for tf>=2.3.0, but it is incompatible elsewhere. Would anyone know how to normalize the first layer with BatchNormalization and using tensorflow 2.2? Problem with tensorflow 2.2.3

To TF>=2.3.0

1

There are 1 answers

0
Fernando Augusto On

My question has been solved in root cern forum. To run, I use HTCondor (a cluster) and I used: #!/bin/bash source /cvmfs/sft.cern.ch/lcg/views/LCG_96py3cu10/x86_64-centos7-gcc7-opt/setup.sh cd /afs/cern.ch/work/f/fassunca/private/

this use "ancient version" tensorflow (1.14), so I changed to: source /cvmfs/sft.cern.ch/lcg/views/LCG_98py3cu10/x86_64-centos7-gcc8-opt/setup.sh In this source, the tensorflow is 2 and compatibility with all versions . thank you to everybody