Update: see https://github.com/lanpa/tensorboard-pytorch for a more full-fledged alternative to this
Quick function to go from a dictionary of sets of (images, labels, feature vectors) to checkpoints that can be opened in Tensorboard
To run the example, just:
python example.py
Then navigate to the embeds directory created by that script and run tensorboard with:
tensorboard --logdir=embeds
Then head to http://localhost:6006/ and click Embeddings.