Skip to content

Fix flaky ReplaceTrivialConvWithLinear pass validation tolerance (#18482) #18482

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
meta-codesync merged 1 commit into from
Mar 25, 2026
Merged

Fix flaky ReplaceTrivialConvWithLinear pass validation tolerance (#18482) #18482

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
24 changes: 20 additions & 4 deletions backends/cadence/aot/tests/test_replace_ops_passes.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1242,9 +1242,17 @@ def test_replace_conv1d_with_linear(self) -> None:
self.assertTrue(result.modified)
graph_after_passes = result.graph_module

# Validate numerical accuracy
# Conv and linear compute the same dot product but accumulate fp32
# terms in different order, so non-associativity of floating-point
# addition produces diffs up to ~1.2e-05. Use rtol=2e-05.
inputs = [x, weights, bias]
validate(gm_before, graph_after_passes, inputs, "ReplaceTrivialConvWithLinear")
validate(
gm_before,
graph_after_passes,
inputs,
"ReplaceTrivialConvWithLinear",
rtol=2e-5,
)

# Assert that conv1d is trivially converted to linear
self.assertEqual(
Expand Down Expand Up @@ -1278,9 +1286,17 @@ def test_replace_conv2d_with_linear(self) -> None:
self.assertTrue(result.modified)
graph_after_passes = result.graph_module

# Validate numerical accuracy
# Conv and linear compute the same dot product but accumulate fp32
# terms in different order, so non-associativity of floating-point
# addition produces diffs up to ~1.2e-05. Use rtol=2e-05.
inputs = [x, weights, bias]
validate(gm_before, graph_after_passes, inputs, "ReplaceTrivialConvWithLinear")
validate(
gm_before,
graph_after_passes,
inputs,
"ReplaceTrivialConvWithLinear",
rtol=2e-5,
)

# Assert that conv2d is trivially converted to linear
self.assertEqual(
Expand Down
Loading