[jira] [Created] (ARROW-4452) [Python] Serializing sparse torch tensors

["Philipp Moritz (JIRA)" <ji...@apache.org>]

Philipp Moritz created ARROW-4452:
-------------------------------------

             Summary: [Python] Serializing sparse torch tensors
                 Key: ARROW-4452
                 URL: https://issues.apache.org/jira/browse/ARROW-4452
             Project: Apache Arrow
          Issue Type: Improvement
            Reporter: Philipp Moritz


Using the pytorch serialization handler on sparse Tensors:
{code:java}
import torch
i = torch.LongTensor([[0, 2], [1, 0], [1, 2]])
v = torch.FloatTensor([3,      4,      5    ])
tensor = torch.sparse.FloatTensor(i.t(), v, torch.Size([2,3]))

register_torch_serialization_handlers(pyarrow.serialization._default_serialization_context)

s = pyarrow.serialize(tensor, context=pyarrow.serialization._default_serialization_context) {code}
Produces this result:
{code:java}
TypeError: can't convert sparse tensor to numpy. Use Tensor.to_dense() to convert to a dense tensor first.{code}
We should provide a way to serialize sparse torch tensors, especially now that we are getting support for sparse Tensors.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)