Uh oh!
There was an error while loading. Please reload this page.
Cortex-M: Use q/dq ops in Arm Ethos Runner - #10782
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/10782
Note: Links to docs will display an error until the docs builds have been completed. ❌ 2 New FailuresAs of commit d75b89f with merge base b1b46ee ( NEW FAILURES - The following jobs have failed:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
Add new tests for q/dq ops using from Cortex-M backend for M55
tested using, ./examples/arm/run.sh --target=ethos-u85-256 --model_name=lstm --bundleio
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
@zingo - Update: Landing this given I don't think we are using these q/dq ops when lowering through tester, which I can do in the follow up PRs. Also U55 tests are passing. |
Unfortunately I'm out of the office a few days and can't check exact details but there is a PR where @oscarandersson8218 hope to have fixed the flakey sigmoid, and update vela that you brought into Meta for test/merge a few days ago. It will hopefully fix the flakeyness. Until it is merged i think rerun the test is the best option (or ignore it) |
zingo
commented
May 13, 2025
This is the sigmoid fix PR mentioned: #10548 |
Summary
Test plan
./examples/arm/run.sh --target=ethos-u85-256 --model_name=lstm --bundleio