The repository provides an implementation of the algorithm presented in the paper "Uniform distribution of points on a hyper-sphere with applicationsto vector bit-plane encoding"
hi, I observe that the procedure is very slow with high dimension. I now set dim=512, num_vecs=10,and it is still running although several hours has passed.
Do you have any suggestions?
What's more, why you use tensorflow rather than numpy, which is a more common choice.