gnes.encoder.numeric.vlad module

class gnes.encoder.numeric.vlad.VladEncoder(num_clusters: int, *args, **kwargs)[source]

Bases: gnes.encoder.base.BaseNumericEncoder

batch_size = 2048
encode(vecs: numpy.ndarray, *args, **kwargs) → numpy.ndarray[source]
kmeans_pred(vecs)[source]
kmeans_train(vecs)[source]
train(vecs: numpy.ndarray, *args, **kwargs)[source]

Train the model, need to be overrided