frozenleaves/pytorch
1
0
Fork 0
mirror of https://github.com/pytorch/pytorch.git synced 2025-10-20 21:14:14 +08:00
Code Issues Packages Projects Releases Wiki Activity

better error handling for rrelu when lower or upper range is infinite (#160965)

Browse Source 
… - issue#153281

Fixes #153281

Pull Request resolved: https://github.com/pytorch/pytorch/pull/160965
Approved by: https://github.com/janeyx99
This commit is contained in:
vishalgoyal316
2025-09-30 05:01:32 +00:00
committed by PyTorch MergeBot
parent 7d59e37434
commit ace89350fc
2 changed files with 39 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

37
test/test_nn.py
View File

@ -12803,6 +12803,43 @@ if __name__ == '__main__':
expected_bf16 = torch.tensor([0., 0., 1.], device=device, dtype=dtype)
self.assertEqual(a_bf16.grad, expected_bf16)
@onlyCPU
def test_rrelu_bounds_validation(self, device):
"""Test RReLU bounds validation for finite and infinite values."""
x = torch.randn(5, 5, device=device)
# Test with finite bounds
result = F.rrelu(x, lower=0.1, upper=0.3)
self.assertEqual(result.shape, x.shape)
# Test with infinite lower bound
with self.assertRaisesRegex(RuntimeError, "rrelu: lower bound must be finite, got inf"):
F.rrelu(x, lower=float('inf'), upper=0.3)
# Test with infinite upper bound
with self.assertRaisesRegex(RuntimeError, "rrelu: upper bound must be finite, got inf"):
F.rrelu(x, lower=0.1, upper=float('inf'))
# Test with NaN lower bound
with self.assertRaisesRegex(RuntimeError, "rrelu: lower bound must be finite, got nan"):
F.rrelu(x, lower=float('nan'), upper=0.3)
# Test with NaN upper bound
with self.assertRaisesRegex(RuntimeError, "rrelu: upper bound must be finite, got nan"):
F.rrelu(x, lower=0.1, upper=float('nan'))
# Test with negative infinity lower bound
with self.assertRaisesRegex(RuntimeError, "rrelu: lower bound must be finite, got -inf"):
F.rrelu(x, lower=float('-inf'), upper=0.3)
# Test with negative infinity upper bound
with self.assertRaisesRegex(RuntimeError, "rrelu: upper bound must be finite, got -inf"):
F.rrelu(x, lower=0.1, upper=float('-inf'))
# Test with lower bound greater than upper bound
with self.assertRaisesRegex(RuntimeError, "Lower bound should be less than or equal to the upper bound"):
F.rrelu(x, lower=0.5, upper=0.3)
@onlyCPU
def test_softshrink(self, device):
x = torch.tensor([[1.21, 0.56, 0.5001, 0.4999, 1.2357, -0.4999, -0.5001, -1.154,