It is possible to install the two backends at the same time. This trick is no longer necessary on the recent versions of Tensorflow. To force the utilization of the GPU (with the Keras backend), run: ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results