New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[fx][split] make sure we copy node.meta over during split #107248
Conversation
This pull request was exported from Phabricator. Differential Revision: D48330866 |
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/107248
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit f7ae570 with merge base ad07a4b (): This comment was automatically generated by Dr. CI and updates every 15 minutes. |
This pull request was exported from Phabricator. Differential Revision: D48330866 |
2ce61f4
to
6433cee
Compare
…7248) Summary: Pull Request resolved: pytorch#107248 Previously when we create placeholder nodes for sub graph modules, we didn't copy node.meta over. Test Plan: CI Reviewed By: sayitmemory Differential Revision: D48330866 fbshipit-source-id: fa3dd3d147d153bc6af7d3c5614a20cf35b1657a
This pull request was exported from Phabricator. Differential Revision: D48330866 |
6433cee
to
cdca229
Compare
…7248) Summary: Pull Request resolved: pytorch#107248 Previously when we create placeholder nodes for sub graph modules, we didn't copy node.meta over. Test Plan: CI Reviewed By: sayitmemory Differential Revision: D48330866 fbshipit-source-id: c785062f88d36270e11968c4379b14336420be56
This pull request was exported from Phabricator. Differential Revision: D48330866 |
…7248) Summary: Pull Request resolved: pytorch#107248 Previously when we create placeholder nodes for sub graph modules, we didn't copy node.meta over. Test Plan: CI Reviewed By: sayitmemory Differential Revision: D48330866 fbshipit-source-id: db932afee8437c55491915660f7c390edc0a2bf4
cdca229
to
68ba8a3
Compare
This pull request was exported from Phabricator. Differential Revision: D48330866 |
68ba8a3
to
f939c8a
Compare
…7248) Summary: Pull Request resolved: pytorch#107248 Previously when we create placeholder nodes for sub graph modules, we didn't copy node.meta over. bypass-github-pytorch-ci-checks bypass-github-export-checks force-merge-on-github Test Plan: CI Reviewed By: houseroad, sayitmemory Differential Revision: D48330866 fbshipit-source-id: b521294df4616418a721de01c7a077b8d8f547c3
…7248) Summary: Pull Request resolved: pytorch#107248 Previously when we create placeholder nodes for sub graph modules, we didn't copy node.meta over. bypass-github-pytorch-ci-checks bypass-github-export-checks force-merge-on-github Test Plan: CI Reviewed By: houseroad, sayitmemory Differential Revision: D48330866 fbshipit-source-id: 57e2f8205695d1bd68531ebcd7a7b09b24ab98f9
This pull request was exported from Phabricator. Differential Revision: D48330866 |
f939c8a
to
f7ae570
Compare
@pytorchbot merge -f 'Landed internally' (Initiating merge automatically since Phabricator Diff has merged, using force because this PR might not pass merge_rules.json but landed internally) |
Merge startedYour change will be merged immediately since you used the force (-f) flag, bypassing any CI checks (ETA: 1-5 minutes). Please use Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
- Follow-up for pytorch#107248 which copies metadata for placeholder nodes in the top-level FX graph
- Follow-up to #107248 which copies metadata for placeholder nodes in the top-level FX graph - Currently, top-level placeholders do not have their metadata copied over, causing loss of `TensorMetadata` in some `torch.compile` backends Fixes pytorch/TensorRT#2258 Pull Request resolved: #107981 Approved by: https://github.com/angelayi
- Follow-up to #107248 which copies metadata for placeholder nodes in the top-level FX graph - Currently, top-level placeholders do not have their metadata copied over, causing loss of `TensorMetadata` in some `torch.compile` backends Fixes pytorch/TensorRT#2258 Pull Request resolved: #107981 Approved by: https://github.com/angelayi
- Follow-up to #107248 which copies metadata for placeholder nodes in the top-level FX graph - Currently, top-level placeholders do not have their metadata copied over, causing loss of `TensorMetadata` in some `torch.compile` backends Fixes pytorch/TensorRT#2258 Pull Request resolved: #107981 Approved by: https://github.com/angelayi Co-authored-by: gs-olive <113141689+gs-olive@users.noreply.github.com>
- Follow-up to #107981, adding testing for metadata copying in placeholder nodes within the `split_by_tags` utility - Validation included in the test from #107248, since both tests are relevant to the same aspect of the utility Pull Request resolved: #108731 Approved by: https://github.com/angelayi
- Follow-up to pytorch#107981, adding testing for metadata copying in placeholder nodes within the `split_by_tags` utility - Validation included in the test from pytorch#107248, since both tests are relevant to the same aspect of the utility Pull Request resolved: pytorch#108731 Approved by: https://github.com/angelayi
- Follow-up to pytorch#107981, adding testing for metadata copying in placeholder nodes within the `split_by_tags` utility - Validation included in the test from pytorch#107248, since both tests are relevant to the same aspect of the utility Pull Request resolved: pytorch#108731 Approved by: https://github.com/angelayi
- Follow-up to pytorch#107981, adding testing for metadata copying in placeholder nodes within the `split_by_tags` utility - Validation included in the test from pytorch#107248, since both tests are relevant to the same aspect of the utility Pull Request resolved: pytorch#108731 Approved by: https://github.com/angelayi
Summary: Previously when we create placeholder nodes for sub graph modules, we didn't copy node.meta over.
Test Plan: CI
Differential Revision: D48330866