Skip to content

einbandi/ptsne-pytorch

Repository files navigation

ptsne-pytorch

PyTorch implementation of parametric t-SNE.

While the code was initially based on a TensorFlow implementation by jsilter, everything is now implemented from scratch, including PCA pretraining and different metrics for comparison of probability distributions.

This version of parametric t-SNE makes use of pynndescent to approximate nearest neighbors for an efficient calculation of an approximated distance matrix. This approach as well as the PCA pretraining are borrowed from the openTSNE package.

About

PyTorch implementation of parametric t-SNE

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published