Whilst this issue is well documented, the associated fixes (i.e. downgrade CUDA and CUDNN) result in the GPU not being recognised by Tensorflow...
My GPU is a GEFORCE
