Use deterministic algorithms for filtfilt autograd tests (#3150)

Summary:
`filtfilt` function uses `lfilter`, which calls `conv_1d` operation internally. `conv_1d` is expected to have autograd test failures (see https://pytorch.org/docs/stable/generated/torch.use_deterministic_algorithms.html). The PR uses deterministic algorithms in the autograd tests to make `filtfilt` related tests pass.

Pull Request resolved: https://github.com/pytorch/audio/pull/3150

Reviewed By: mthrok

Differential Revision: D43872977

Pulled By: nateanl

fbshipit-source-id: c3d6ec281f34db8a7092526ccb245797bf2338da

test/torchaudio_unittest/common_utils/__init__.py

+from .autograd_utils import use_deterministic_algorithms
 from .backend_utils import set_audio_backend
 from .case_utils import (
     HttpServerMixin,
@@ -66,5 +67,6 @@ __all__ = [
     "get_image",
     "rgb_to_gray",
     "rgb_to_yuv_ccir",
+    "use_deterministic_algorithms",
     "zip_equal",
 ]

test/torchaudio_unittest/common_utils/autograd_utils.py

+import contextlib
+
+import torch
+
+
+@contextlib.contextmanager
+def use_deterministic_algorithms(mode: bool, warn_only: bool):
+    r"""
+    This context manager can be used to temporarily enable or disable deterministic algorithms.
+    Upon exiting the context manager, the previous state of the flag will be restored.
+    """
+    previous_mode: bool = torch.are_deterministic_algorithms_enabled()
+    previous_warn_only: bool = torch.is_deterministic_algorithms_warn_only_enabled()
+    try:
+        torch.use_deterministic_algorithms(mode, warn_only=warn_only)
+        yield {}
+    except RuntimeError as err:
+        raise err
+    finally:
+        torch.use_deterministic_algorithms(previous_mode, warn_only=previous_warn_only)

test/torchaudio_unittest/functional/autograd_impl.py

@@ -6,7 +6,14 @@ import torchaudio.functional as F
 from parameterized import parameterized
 from torch import Tensor
 from torch.autograd import gradcheck, gradgradcheck
-from torchaudio_unittest.common_utils import get_spectrogram, get_whitenoise, nested_params, rnnt_utils, TestBaseMixin
+from torchaudio_unittest.common_utils import (
+    get_spectrogram,
+    get_whitenoise,
+    nested_params,
+    rnnt_utils,
+    TestBaseMixin,
+    use_deterministic_algorithms,
+)
 
 
 class Autograd(TestBaseMixin):
@@ -71,26 +78,30 @@ class Autograd(TestBaseMixin):
         a = torch.tensor([0.7, 0.2, 0.6])
         b = torch.tensor([0.4, 0.2, 0.9])
         a.requires_grad = True
-        self.assert_grad(F.filtfilt, (x, a, b), enable_all_grad=False)
+        with use_deterministic_algorithms(True, False):
+            self.assert_grad(F.filtfilt, (x, a, b), enable_all_grad=False)
 
     def test_filtfilt_b(self):
         x = get_whitenoise(sample_rate=22050, duration=0.01, n_channels=2)
         a = torch.tensor([0.7, 0.2, 0.6])
         b = torch.tensor([0.4, 0.2, 0.9])
         b.requires_grad = True
-        self.assert_grad(F.filtfilt, (x, a, b), enable_all_grad=False)
+        with use_deterministic_algorithms(True, False):
+            self.assert_grad(F.filtfilt, (x, a, b), enable_all_grad=False)
 
     def test_filtfilt_all_inputs(self):
         x = get_whitenoise(sample_rate=22050, duration=0.01, n_channels=2)
         a = torch.tensor([0.7, 0.2, 0.6])
         b = torch.tensor([0.4, 0.2, 0.9])
-        self.assert_grad(F.filtfilt, (x, a, b))
+        with use_deterministic_algorithms(True, False):
+            self.assert_grad(F.filtfilt, (x, a, b))
 
     def test_filtfilt_batching(self):
         x = get_whitenoise(sample_rate=22050, duration=0.01, n_channels=2)
         a = torch.tensor([[0.7, 0.2, 0.6], [0.8, 0.2, 0.9]])
         b = torch.tensor([[0.4, 0.2, 0.9], [0.7, 0.2, 0.6]])
-        self.assert_grad(F.filtfilt, (x, a, b))
+        with use_deterministic_algorithms(True, False):
+            self.assert_grad(F.filtfilt, (x, a, b))
 
     def test_biquad(self):
         x = get_whitenoise(sample_rate=22050, duration=0.01, n_channels=1)