Error when passing in 0 size empty in PyTorch Kernel

[half-potato]

When passing in torch.empty((0), device="cuda"), I get the following error:

Traceback (most recent call last):
  File "/home/amai/gaussian-splatting/train.py", line 510, in <module>
    training(
  File "/home/amai/gaussian-splatting/train.py", line 160, in training
    loss.backward()
  File "/home/amai/micromamba/envs/gaussian_splatting/lib/python3.11/site-packages/torch/_tensor.py", line 492, in backward
    torch.autograd.backward(
  File "/home/amai/micromamba/envs/gaussian_splatting/lib/python3.11/site-packages/torch/autograd/__init__.py", line 251, in backward
    Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
  File "/home/amai/micromamba/envs/gaussian_splatting/lib/python3.11/site-packages/torch/autograd/function.py", line 288, in apply
    return user_fn(self, *args)
           ^^^^^^^^^^^^^^^^^^^^
  File "/home/amai/gaussian-splatting/splinetracer/const_splinetracer.py", line 188, in backward
    ).launchRaw(
      ^^^^^^^^^^
  File "/home/amai/micromamba/envs/gaussian_splatting/lib/python3.11/site-packages/slangpy/util/wrapper.py", line 31, in launchRaw
    return self.dispatch_fn(blockSize, gridSize)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/amai/micromamba/envs/gaussian_splatting/lib/python3.11/site-packages/slangpy/util/wrapper.py", line 97, in <lambda>
    lambda blockSize, gridSize: self.fn_handle(*((blockSize, gridSize) + arglist)),
                                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
RuntimeError: wcts: data pointer is invalid.

[saipraveenb25]

Looks like torch doesn't initialize its data pointer if array is empty. The fix is to check the array size before touching the data pointer on the Slang wrapper side. Will issue a fix.