register_buffer compatibility

Summary: In D30349234 (1b8d86a104) we introduced persistent=False to some register_buffer calls, which depend on PyTorch 1.6. We go back to the old behaviour for PyTorch 1.5.

Reviewed By: nikhilaravi

Differential Revision: D30731327

fbshipit-source-id: ab02ef98ee87440ef02479b72f4872b562ab85b5
This commit is contained in:
Jeremy Reizenstein
2021-09-09 07:36:56 -07:00
committed by Facebook GitHub Bot
parent bbc7573261
commit c3d7808868
3 changed files with 14 additions and 3 deletions

View File

@@ -426,7 +426,10 @@ class TestRaysampling(TestCaseMixin, unittest.TestCase):
atol=1e-5,
)
def test_load_state(self):
@unittest.skipIf(
torch.__version__[:4] == "1.5.", "non persistent buffer needs PyTorch 1.6"
)
def test_load_state_different_resolution(self):
# check that we can load the state of one ray sampler into
# another with different image size.
module1 = NDCGridRaysampler(