How can I predict more than one cluster per observation with scikit-learn KMeans?

Question

I am using scikit-learn MiniBatchKMeans to do text clustering.

After I have trained the model, the class provides a method predict which allows me to pre