Input contains NaN, infinity or a value too large for dtype('float64').

[lucasaos52]

Sometimes after fitting the model, and applying transform method (or the fit_transform) we get the following error. If run again with everything the same, sometimes work and sometimes dont. I noticed that when I got this bug, the cluster_centers_ has NaNs in it and maybe its due to some convergence issue on the algorithm

OBS: beforehad I already checked if on the input of it has any Nans or inf values a alread treat it