Added support for custom pre-grad passes #113823
Conversation
[ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/113823
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 d13d03c with merge base 27e31ab ( 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. |
| if len(matmuls) == 1: | ||
| continue | ||
| matmuls = sorted(matmuls, key=lambda x: order[x]) | ||
| with graph.inserting_before(matmuls[0]): |
There was a problem hiding this comment.
I think the major challenge of supporting custom passes is during the nodes fusion/replacement, we have to ensure the metadata is updated, since the other FX pass may use these metadata. We should have a well defined contract for this.
cc voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx peterbell10 ipiszy yf225 chenyang78 kadeng muchulee8 aakhundov ColinPeppler [ghstack-poisoned]
|
@pytorchbot merge -i |
Merge startedYour change will be merged while ignoring the following 2 checks: inductor / linux-jammy-cpu-py3.8-gcc11-inductor / test (inductor_torchbench_cpu_accuracy, 1, 1, linux.4xlarge), inductor / linux-jammy-cpu-py3.8-gcc11-inductor / test (inductor_torchbench_dynamic_cpu_accuracy, 1, 1, linux.12xlarge) Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
Pull Request resolved: pytorch#113823 Approved by: https://github.com/eellison, https://github.com/jansel ghstack dependencies: pytorch#113913
5 participants
Stack from ghstack (oldest at bottom):
cc @voznesenskym @penguinwu @EikanWang @jgong5 @Guobing-Chen @XiaobingSuper @zhuhaozhe @blzheng @wenzhe-nrv @jiayisunx @peterbell10 @ipiszy @yf225 @chenyang78 @kadeng @muchulee8 @aakhundov @ColinPeppler