Bisect nightly failures tooling - Exploratory task

pytorch / test-infra

This repository hosts code that supports the testing infrastructure for the main PyTorch repo. For example, this repo hosts the logic to track disabled tests and slow tests, as well as our continuation integration jobs HUD/dashboard.

https://hud.pytorch.org/

Other

72 stars 68 forks source link

Bisect nightly failures tooling - Exploratory task #5342

Open atalman opened 2 weeks ago

atalman commented 2 weeks ago

Some failures in nightly examples: https://github.com/pytorch/pytorch/actions/runs/9444290266/job/26009551764

Details can be seen here: https://hud2.pytorch.org/hud/pytorch/pytorch/nightly/1?per_page=50

Research ideas for tooling we can have for bisecting nightly failures.

huydhn commented 2 weeks ago

I just find out that we have a similar issue here https://github.com/pytorch/pytorch/issues/116539 on Dev Infra board