SVD is returning 4 pointers instead of 3

[sssilvar]

Description

When sending a tensor to a worker and performing SVD, returns four pointers instead of three. Also, the third one is not gettable. By experimentation, I have had to solve the issue using U, s, _, V = x.svd().

How to Reproduce

import torch
import syft as sy

hook = sy.TorchHook(torch)
bob = sy.VirtualWorker(hook, id='bob')

x = torch.rand(250, 84).send(bob)  # Synthetic tensor
x.svd()

# Output:
# ((Wrapper)>[PointerTensor | me:88822589827 -> bob:10423896311],
#  (Wrapper)>[PointerTensor | me:22528885369 -> bob:34285527022],
#  (Wrapper)>[PointerTensor | me:46709676193 -> bob:67244907535],
#  (Wrapper)>[PointerTensor | me:235847656 -> bob:15738446586])

Expected Behavior

Should return three pointers: U, s, V = x.svd()

System Information

• Official released Docker container
• Same for pip package:
  • Ubuntu 18.04.5 LTS (Bionic Beaver)
  • Python 3.6.9

[Nilanshrajput]

this is pretty weird I run this code in debug mode, stepped into every function, the things seemed correct while computing the response (3 tensors were returned), and in the end got this error : (which shows things were correct 3 tensor returned as expected)

Exception has occurred: ValueError
not enough values to unpack (expected 4, got 3)
  File "/home/nilansh/Anton/OpenMined/PySyft/debug.py", line 8, in <module>
    U, s, _, V = x.svd()

but returns 4 values when run as normal python code.

@LaRiffle

[xutongye]

want to work on it.