Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/22950
Note: Links to docs will display an error until the docs builds have been completed. ✅ You can merge normally! (2 Unrelated Failures)As of commit 8f63fc6 with merge base 11120c8 ( FLAKY - The following job failed but was likely due to flakiness present on trunk:
BROKEN TRUNK - The following job failed but was present on the merge base:👉 Rebase onto the `viable/strict` branch to avoid these failures
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
meta-cla
Bot
added
the
CLA Signed
This PR needs a
|
There was a problem hiding this comment.
Copilot review overview
🟡 Changes recommended
Remove the shared check_count.exir stage from the affected rejection-test path.
Get a fresh assessment by requesting another Copilot review.
Review effort: Lite
Findings: 1
What changed in this PR
Updates Arm sum tests to pass dim and keepdim through module initialization, enabling previously skipped dim=None U55/U85 cases.
Changes:
- Refactors sum test inputs and pipelines.
- Removes U55/U85
dim=Noneskips. - Adjusts rejection-test pipeline handling.
| File | Summary |
|---|---|
backends/arm/test/ops/test_sum.py |
Refactors sum parameter handling and restores skipped coverage; rejection tests need the shared check_count.exir stage removed. |
💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.
| Sum(dim, keepdim), | ||
| (test_tensor,), |
Pass dim/keepdim via module __init__ instead of as model inputs, since bundled_program cannot serialize None as a model input. Removes the _dim_none_skips markers covering dim_None and dim_None_4d_tensor on test_sum_u55/u85_INT_1_0 (8 cases).
53776d8 to
8f63fc6
Compare
|
Good catch — the refactor accidentally dropped |
Pass dim/keepdim via module init instead of as model inputs, since bundled_program cannot serialize None as a model input. Removes the _dim_none_skips markers covering dim_None and dim_None_4d_tensor on test_sum_u55/u85_INT_1_0 (8 cases).
Verification: py_compile passes. Full pytest collection fails locally on this checkout with a pre-existing torch incompatibility (aten.transpose Dimname, present with and without this change); relying on CI for test signal.
cc @digantdesai @freddan80 @per @zingo @oscarandersson8218 @mansnils @Sebastian-Larsson @robell @rascani