This error is quite strange and gets me crazy because it does not prompt in Tensorflow 1.15. I create a data generator that works perfectly when using Tensorflow 1.15, but because I have some Out of Memory problems that I could not manage in TF1, y upgrade to TF2 and I get this message at the beginning of the training.
I am using a HPC that has a Tesla V100-PCIE-16GB. It seems as if the GPU could not load the python interpreter, python 3.7.
Any thoughts? As I said, TF1 works perfectly.
2021-05-20 17:15:31.744926: W tensorflow/core/kernels/data/generator_dataset_op.cc:107] Error occurred when finalizing GeneratorDataset iterator: Failed precondition: Python interpreter state is not initialized. The process may be terminated.
[[{{node PyFunc}}]]