SUMIN/HuggingFace_transformer
1
0
Fork 0
Code Issues Pull Requests Actions 7 Packages Projects Releases Wiki Activity

higher atol to avoid flaky trainer test failure (#17979)

Browse Source 
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
This commit is contained in:
Yih-Dar
2022-07-01 17:53:16 +02:00
committed by GitHub
parent 8bb2c387f4
commit 664688b94f
1 changed files with 4 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
tests/trainer/test_trainer.py
View File

@@ -1252,8 +1252,8 @@ class TrainerIntegrationTest(TestCasePlus, TrainerIntegrationCommon):
trainer.train(resume_from_checkpoint=os.path.join(tmp_dir, "checkpoint-15")) trainer.train(resume_from_checkpoint=os.path.join(tmp_dir, "checkpoint-15"))
(a1, b1) = trainer.model.a.item(), trainer.model.b.item() (a1, b1) = trainer.model.a.item(), trainer.model.b.item()
self.assertAlmostEqual(a, a1, delta=1e-8) self.assertAlmostEqual(a, a1, delta=1e-5)
self.assertAlmostEqual(b, b1, delta=1e-8) self.assertAlmostEqual(b, b1, delta=1e-5)
with self.subTest("Test every epoch"): with self.subTest("Test every epoch"):
config = RegressionModelConfig(a=0, b=2, random_torch=random_torch) config = RegressionModelConfig(a=0, b=2, random_torch=random_torch)
@@ -1277,8 +1277,8 @@ class TrainerIntegrationTest(TestCasePlus, TrainerIntegrationCommon):
trainer.train(resume_from_checkpoint=os.path.join(tmp_dir, checkpoint_dir)) trainer.train(resume_from_checkpoint=os.path.join(tmp_dir, checkpoint_dir))
(a1, b1) = trainer.model.a.item(), trainer.model.b.item() (a1, b1) = trainer.model.a.item(), trainer.model.b.item()
self.assertAlmostEqual(a, a1, delta=1e-8) self.assertAlmostEqual(a, a1, delta=1e-5)
self.assertAlmostEqual(b, b1, delta=1e-8) self.assertAlmostEqual(b, b1, delta=1e-5)
@slow @slow
@require_torch_non_multi_gpu @require_torch_non_multi_gpu