Arm backend: Fix sigmoid int16 and int32 flakyness#10548
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/10548
Note: Links to docs will display an error until the docs builds have been completed. ❌ 20 New Failures, 8 Unrelated FailuresAs of commit 43cec8e with merge base e1738cc ( NEW FAILURES - The following jobs have failed:
BROKEN TRUNK - The following jobs failed but were 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. |
|
@digantdesai this is also steping Vela (e.g. it's newer then #9893 ) Maybe you want to merge both PRs at the same time using the version of Vela from this PR. EDIT we need to update this PR a bit more @digantdesai do not merge it yet |
|
Need to investigate some of the failures. |
2120c63 to
5137ab7
Compare
|
@digantdesai @zingo It's now been resolved and is ready for review again. A few depthwise_conv2d tests started failing in different way than before so had to update those. |
|
@digantdesai we think this is ready and should work now. This will improve INT-16 tables so it will be interesting to get this in. |
|
@digantdesai has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
|
Hi unfortunately it got a merge conflict and probably needs to be rebased after #10776 was merged. |
0135b99 to
52da25a
Compare
c8a0764 to
14d4e1b
Compare
MLBEDSW-10642 has now been resolved. Step vela pin and remove all related pytest.mark_flaky in sigmoid int16 and int32 tests. Also enables quantized Conformer. Signed-off-by: Oscar Andersson <oscar.andersson@arm.com> Change-Id: I8c1aff31fb173fa6fdebc1d5b757623f9c2e1321
14d4e1b to
bd0bd59
Compare
|
@digantdesai does this work now and can be merged? |
|
@digantdesai has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
|
Yeah looks good internally, feel free to merge. CI failures looks unrelated? |
|
Thanks for the help! |
|
Yes those fails was a geneal llama errors a few days ago that has been fixed. |
Step vela pin and remove all related pytest.mark_flaky in sigmoid int16 and int32 tests.
cc @digantdesai @freddan80 @per @zingo