tf.contrib.layers.embedding_lookup_unique( params, ids, name=None )
Version of embedding_lookup that avoids duplicate lookups.
This can save communication in the case of repeated ids.
Same interface as embedding_lookup. Except it supports multi-dimensional
which allows to not reshape input/output to fit gather.
params: A list of tensors with the same shape and type, or a
[index, d1, d2, ...].
ids: A one-dimensional
int64containing the ids to be looked up in
[ids1, ids2, ...].
name: A name for this operation (optional).
Tensor with the same type as the tensors in
params and dimension of
[ids1, ids2, d1, d2, ...].