|TensorFlow 1 version||View source on GitHub|
Deserialize and concatenate
SparseTensors from a serialized minibatch.
Compat aliases for migration
See Migration guide for more details.
tf.io.deserialize_many_sparse( serialized_sparse, dtype, rank=None, name=None )
serialized_sparse must be a string matrix of shape
[N x 3] where
N is the minibatch size and the rows correspond to packed outputs of
serialize_sparse. The ranks of the original
must all match. When the final
SparseTensor is created, it has rank one
higher than the ranks of the incoming
SparseTensor objects (they have been
concatenated along a new row dimension).
SparseTensor object's shape values for all dimensions but the
first are the max across the input
SparseTensor objects' shape values
for the corresponding dimensions. Its first shape value is
N, the minibatch
SparseTensor objects' indices are assumed ordered in
standard lexicographic order. If this is not the case, after this
sparse.reorder to restore index ordering.
For example, if the serialized input is a
[2, 3] matrix representing two
index = [ 0]   values = [1, 2, 3] shape = 
index = [ 2]  values = [4, 5] shape = 
then the final deserialized
SparseTensor will be:
index = [0 0] [0 10] [0 20] [1 2] [1 10] values = [1, 2, 3, 4, 5] shape = [2 50]
[N, 3]. The serialized and packed
dtypeof the serialized
rank: (optional) Python int, the rank of the
name: A name prefix for the returned tensors (optional)
SparseTensor representing the deserialized
concatenated along the
SparseTensors' first dimension.
All of the serialized
SparseTensors must have had the same rank and type.