New Pipeline Slow Test runners #5131
Conversation
|
@patrickvonplaten I want to test this out on the actual runners before merging. Could you please let me know how I can change actions so that they run on this branch? |
|
.github/workflows/push_tests.yml
Outdated
| jobs: | ||
| run_slow_tests: | ||
| setup_torch_cuda_pipeline_matrix: | ||
| runs-on: docker-cpu |
There was a problem hiding this comment.
The push tests should definitely still run on GPU, the only difference here should maybe just be that the RUN_SLOW flag is set to False instead of True. So we run the fast tests but on GPU this time
…fusers into new-test-runners
| @@ -0,0 +1,96 @@ | |||
| import json | |||
|
Design looks great! Let me know once this is ready for a full review! |
|
@patrickvonplaten This is ready for review. I'm seeing a couple of issues with compile tests and loras, but I think we can just fix those after merging this. |
|
The documentation is not available anymore as the PR was closed or merged. |
* pipline fetcher * update script * clean up * clean up * clean up * new pipeline runner * rename tests to match modules * test actions in pr * change runner to gpu * clean up * clean up * clean up * fix report * fix reporting * clean up * show test stats in failure reports * give names to jobs * add lora tests * split torch cuda tests and add compile tests * clean up * fix tests * change push to run only on main --------- Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>
* pipline fetcher * update script * clean up * clean up * clean up * new pipeline runner * rename tests to match modules * test actions in pr * change runner to gpu * clean up * clean up * clean up * fix report * fix reporting * clean up * show test stats in failure reports * give names to jobs * add lora tests * split torch cuda tests and add compile tests * clean up * fix tests * change push to run only on main --------- Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>
* pipline fetcher * update script * clean up * clean up * clean up * new pipeline runner * rename tests to match modules * test actions in pr * change runner to gpu * clean up * clean up * clean up * fix report * fix reporting * clean up * show test stats in failure reports * give names to jobs * add lora tests * split torch cuda tests and add compile tests * clean up * fix tests * change push to run only on main --------- Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>
What does this PR do?
Changes the slow push tests to only run on core + high usage pipelines.
This PR:
utils/fetch_torch_cuda_pipeline_test_matrix.pyscript that fetches a list of high usage pipelines from the hub and extracts their corresponding modules along with core pipeline modules that are defined in the script.push_tests.ymlto run the fetched + core pipelines in parallel. This isolates pipeline tests by moduleskandinksy_v22andtext_to_videofolder names tokandinksy2_2andtext_to_video_synthesisso that they match with the module names insrc/diffusers. This is done so that Github actions can run the correct tests for a module.Fixes # (issue)
Before submitting
documentation guidelines, and
here are tips on formatting docstrings.
Who can review?
Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.