Fix isfinite for int input (pytorch#12750)

Summary:
`torch.isfinite()` used to crash on int inputs.
```
>>> import torch
>>> a = torch.tensor([1, 2])
>>> torch.isfinite(a)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/scratch/pytorch/torch/functional.py", line 262, in isfinite
    return (tensor == tensor) & (tensor.abs() != inf)
RuntimeError: value cannot be converted to type int64_t without overflow: inf
```
But this is a easy special case and numpy also supports it.
```
>>> import numpy as np
>>> a = np.array([1, 2])
>>> a.dtype
dtype('int64')
>>> np.isfinite(a)
array([ True,  True], dtype=bool)
```
So added a hacky line to handle non-floating-point input. Since pytorch raises exception when overflow, we can safely assume all valid int tensors are infinite numbers.
Pull Request resolved: pytorch#12750

Differential Revision: D10428204

Pulled By: ailzhang

fbshipit-source-id: f39b2d0975762c91cdea23c766ff1e21d85d57a5

Author: Ailing Zhang

test/test_torch.py

@@ -5201,6 +5201,10 @@ def test_isfinite(self):
         x = torch.Tensor([1, inf, 2, -inf, nan, -10])
         self.assertEqual(torch.isfinite(x), torch.ByteTensor([1, 0, 1, 0, 0, 1]))
 
+    def test_isfinite_int(self):
+        x = torch.tensor([1, 2, 3])
+        self.assertEqual(torch.isfinite(x), torch.ByteTensor([1, 1, 1]))
+
     def test_isinf(self):
         x = torch.Tensor([1, inf, 2, -inf, nan])
         self.assertEqual(torch.isinf(x), torch.ByteTensor([0, 1, 0, 1, 0]))

torch/functional.py

@@ -259,6 +259,13 @@ def isfinite(tensor):
     """
     if not isinstance(tensor, torch.Tensor):
         raise ValueError("The argument is not a tensor", str(tensor))
+
+    # Support int input, nan and inf are concepts in floating point numbers.
+    # Numpy uses type 'Object' when the int overflows long, but we don't
+    # have a similar concept. It's safe to assume any created LongTensor doesn't
+    # overflow and it's finite.
+    if not tensor.is_floating_point():
+        return torch.ones_like(tensor, dtype=torch.uint8)
     return (tensor == tensor) & (tensor.abs() != inf)