subhadarship

center_shift can be a very large number when the centroids change a lot (in the initial iterations of the K-means algorithm). I am not sure why it would be nan...

It would be very useful if you can share the code snippet which you are using so that I can reproduce it at my end.

So if I understand correctly, you are trying to cluster the weight vectors from different layers. Can you confirm? clustering the weights of resnet sounds really cool btw !!

@eghouti what distance metric are you using? `euclidean` or `cosine`?

The latest version is not yet available on pip! Try installing from source https://github.com/subhadarship/kmeans_pytorch#installation

Thanks for bringing this up!

the above just shows the progress bar. after running kmeans, `cluster_ids` and `cluster_centers` are returned. [this notebook](https://github.com/subhadarship/kmeans_pytorch/blob/master/example.ipynb) shows an example.

I will add a `verbose=False` option. Look out for this change in the next release !!

not available as of yet..

@mpariente Thanks for the appreciation Yes, I am considering writing a unit test similar to `sklearn`'s kmeans