Krsto Proroković

@seanmor5 I think this is happening because `event_counts[:iteration_completed]` keeps increasing with every iteration despite training going into the next epoch. For example, training on MNIST with `batch_size = 15000` (60000...

@polvalente Looking into this, but I am not getting the error. Could you please tell me what are the shapes of time_x and time_y? Thanks!

@polvalente time_y is for sure not rank 1, as third dimension is added with `Nx.new_axis(time_y, 2)`. The following code seems to be working for me (`x` and `y` are `f32`)....

> Hi @krstopro, I want to release a new version soon. If we plant to remove `Scholar.Neighbors.KNearestNeighbors`, it would be better to do it before, so we update the docs...

Hi @b-serra! I could be wrong, but I don't think DBSCAN can assign cluster labels to points other than those used for training the model. Looking at the `scikit-learn` implementation...

> [There is a fit_predict method in scikit-learn](https://scikit-learn.org/stable/modules/generated/sklearn.cluster.DBSCAN.html#sklearn.cluster.DBSCAN.fit_predict). Perhaps we should add `fit_predict` here as well. I am not sure about it because `fit_predict` would just return the `labels` of...

Is this really completed?

> At the same time, I am not sure if there is something to unify here? I think checking for shapes should be done per module, as they can provide...

@msluszniak Just to check - are you maybe working on this?

@msluszniak I see. I might take a shot at this one then.