espnet.nets.chainer_backend.deterministic_embed_id.EmbedID

source

class espnet.nets.chainer_backend.deterministic_embed_id.EmbedID(in_size, out_size, initialW=None, ignore_label=None)

Bases: Link

Efficient linear layer for one-hot input.

This is a link that wraps the embed_id() function. This link holds the ID (word) embedding matrix W as a parameter.

• Parameters:
  • in_size (int) – Number of different identifiers (a.k.a. vocabulary size).
  • out_size (int) – Output dimension.
  • initialW (Initializer) – Initializer to initialize the weight.
  • ignore_label (int) – If ignore_label is an int value, i-th column of return value is filled with 0.

embed_id<span class='small-bracket'>()</span>

W

Embedding parameter matrix.

• Type:Variable

Examples

>>> W = np.array([[0, 0, 0],
...               [1, 1, 1],
...               [2, 2, 2]]).astype('f')
>>> W
array([[ 0.,  0.,  0.],
       [ 1.,  1.,  1.],
       [ 2.,  2.,  2.]], dtype=float32)
>>> l = L.EmbedID(W.shape[0], W.shape[1], initialW=W)
>>> x = np.array([2, 1]).astype('i')
>>> x
array([2, 1], dtype=int32)
>>> y = l(x)
>>> y.data
array([[ 2.,  2.,  2.],
       [ 1.,  1.,  1.]], dtype=float32)

ignore_label = None