gnes.encoder.numeric.hash module

class gnes.encoder.numeric.hash.HashEncoder(num_bytes, num_bits=8, num_idx=3, kmeans_clusters=100, method='product_uniform', *args, **kwargs)

Bases: gnes.encoder.base.BaseNumericEncoder

batch_size = 2048

encode(vecs, *args, **kwargs)

Return type:ndarray

hash(vecs)

pred_kmeans(vecs)

ran_gen()

train(vecs, *args, **kwargs)

Train the model, need to be overrided

train_kmeans(vecs)