Running Distributed TensorFlow on Compute Engine
Need more explanation of the training code.
This lab is having outdated cli commands, Replace below commands pip3 install --upgrade pip pip3 install tensorflow==1.15.2 gast==0.2.2 with below commands sudo apt-get remove -y python3-setuptools sudo pip3 install setuptools==50.3.2 sudo pip3 install tensorflow==1.14.0 \ numpy==1.18.5 futures==2.2.0 h5py==2.10.0 gast==0.2.2
Some commands are outdated!
Instructions failed to produce the proper output
I'm not sure if it was my fault but this lab did not seem to work. I did copy and paste the code exactly as given. It may be the case that the latest version of python does not work for this lab.
Not very satisfied, something is off
pip was giving errors on the virtual machine, and I was not able to run install tensorflow on the machines
failed in using pip3 install. please fix it.
last step didn't work, couldn't create the model, staging bucket had to be given but I don't know what that is
In the Cloud Shell, when i execute the script ./scripts_p3/start-training.sh, there is the following error : ImportError: No module named 'tensorflow' ... CommandException: One or more URLs matched no objects. At the end, the script daoen't print the location with the new model files