The requirements file requires core-ml-tools which requires tensorflow and not tensorflow-gpu #2
Description
When there is tensorflow installed, tensorflow-gpu is ignored. Then the time for training slows down significantly. The difference between tensorflow and tensorflow-gpu training rate is almost an order of magnitude. If core-ml-tools can accept tensorflow-gpu, the training time will drop quiet a lot.
Also, for windows machines, if you need to install the core-ml-tools, use this line of code in your cmd window:
"pip install git+https://github.com/apple/coremltools"
Thanks,
Selin