jgraving / tensorflow_tsne

Simple implementation of the TSNE algorithm in tensorflow

Command line usage

usage: tensorflow_tsne.py [-h] [--n-iter N_ITER] [--perplexity PERPLEXITY]
                          [--lr LR]

optional arguments:

  -h, --help            show this help message and exit

  --n-iter N_ITER       Number of optimization steps to perform

  --perplexity PERPLEXITY
                        Effective number of neighbours to enforce

  --lr LR               Learning rate

Demo

1000 iterations on MNIST sub-sample with lr=2000, perplexity=30