Skip to content

Commit ce57229

Browse files
authored
Cortex-M: stop xfailing test_implementation_mv2 (pytorch#21754)
### Summary The xfail was precautionary rather than observed. pytorch#17300 added it "for potential flakiness in the serialization/runtime path", and because strict=False suppresses XPASS reporting, CI has never been able to tell us whether the test actually fails. It does not. Running it with --runxfail over TEST_SEED=0..7 on the Corstone-300 FVP passes 8/8 seeds, both the dialect and the implementation test, so the xfail is removed along with its placeholder MLETORCH-XXX ticket reference. Authored with assistance from Claude Code.
1 parent ba4c005 commit ce57229

1 file changed

Lines changed: 1 addition & 6 deletions

File tree

‎backends/cortex_m/test/models/test_mobilenet_v2.py‎

Lines changed: 1 addition & 6 deletions
Original file line numberDiff line numberDiff line change
@@ -71,12 +71,7 @@ def test_dialect_mv2(test_case):
7171
assert torch.argmax(ref) == torch.argmax(result), "Mismatch in model outputs"
7272

7373

74-
@parametrize(
75-
"test_case",
76-
test_cases,
77-
xfails={"mobilenet_v2": "MLETORCH-XXX - Investigate mobilenet_v2 flakiness"},
78-
strict=False,
79-
)
74+
@parametrize("test_case", test_cases)
8075
def test_implementation_mv2(test_case):
8176
inputs = test_case.get_example_inputs()
8277
tester = CortexMTester(test_case.model, inputs)

0 commit comments

Comments
 (0)