I train a keras model locally and save *.h5 weights. The model is running correctly locally. when I send the *.h5 via gcloud to the virtual machine and
*.h5
gcloud