Parallel version of t-SNE

Question

Is there a Python library with a parallel version of the t-SNE algorithm? Or is there a t-SNE multi-core / parallel algorithm?

I am trying to reduce the size (300d → 2d) of all word2vecs in my dictionary using t-SNE.

Problem: The size of the dictionary is about 130,000, and it takes them too long to continue t-SNE for them.

Answer 1

Yes, there is a parallel version of the barnes-hutt t-SNE implementation. https://github.com/DmitryUlyanov/Multicore-TSNE