[PT2][Optimus] Add missing example value for introduced nodes #132297
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/132297
Note: Links to docs will display an error until the docs builds have been completed. ✅ You can merge normally! (8 Unrelated Failures)As of commit 6254680 with merge base 9853c04 ( 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. |
|
This pull request was exported from Phabricator. Differential Revision: D60364165 |
|
This pull request was exported from Phabricator. Differential Revision: D60364165 |
mengluy0125
force-pushed
the
export-D60364165
branch
from
4b1c691 to
8e69b7c
Compare
mengluy0125
added a commit
to mengluy0125/pytorch
that referenced
this pull request
Jul 31, 2024
…h#132297) Summary: Pull Request resolved: pytorch#132297 We observed that many introduced nodes during split cat and batch fusion pattern optimization did not have example value meta data, which will cause problems in our follow up pattern optimizations, thus we add all missing values. We also fix bugs in some meta update and corner case bug for the old pattern, which caused problems in the follow up pattern optimization. We delete merge_stack_tahn_unbind_pass pattern, which was designed for cmf model, and it could be replaced by the more advanced pattern we added, thus we remove it for easy maintenance. Test Plan: # unit test ``` buck2 test //caffe2/test/inductor:split_cat_fx_passes ``` Test UI: https://www.internalfb.com/intern/testinfra/testrun/3940649915381395 Network: Up: 90KiB Down: 118KiB (reSessionID-a4a438a0-b043-4ae0-8cf4-a83a5150d0fa) Jobs completed: 28. Time elapsed: 4:57.9s. Cache hits: 0%. Commands: 2 (cached: 0, remote: 0, local: 2) Tests finished: Pass 10. Fail 0. Fatal 0. Skip 1. Build failure 0 ``` buck2 test mode/opt pytorch/diff_train_tests/ads/optimus:local_pt2_runner ``` Network: Up: 1.3GiB Down: 84MiB (reSessionID-ff135cdd-e42c-4ab5-8217-907ada465f01) Jobs completed: 61. Time elapsed: 21:56.5s. Cache hits: 0%. Commands: 39 (cached: 0, remote: 0, local: 39) Tests finished: Pass 8. Fail 0. Fatal 0. Skip 0. Build failure 0 # benchmark ``` CUDA_VISIBLE_DEVICES=3 OC_CAUSE=1 buck2 run mode/opt //scripts/jackiexu0313/pt2:local_model_with_pt2 -- --test_mode batch-split --model_type "ig_ctr" --flow_id 584880697 ``` Counter({'pattern_matcher_nodes': 752, 'pattern_matcher_count': 732, 'normalization_pass': 328, 'normalization_aten_pass': 12, 'scmerge_cat_removed': 5, 'scmerge_cat_added': 4, 'scmerge_split_removed': 3, 'unbind_stack_pass': 3, 'batch_tanh': 2, 'scmerge_split_sections_removed': 2, 'scmerge_split_added': 2, 'optimize_cat_inputs_pass': 1, 'unbind_cat_to_view_pass': 1, 'fxgraph_cache_miss': 1}) Differential Revision: D60364165
|
This pull request was exported from Phabricator. Differential Revision: D60364165 |
mengluy0125
force-pushed
the
export-D60364165
branch
from
8e69b7c to
d9f1a03
Compare
|
This pull request was exported from Phabricator. Differential Revision: D60364165 |
mengluy0125
force-pushed
the
export-D60364165
branch
from
d9f1a03 to
6f3c775
Compare
|
This pull request was exported from Phabricator. Differential Revision: D60364165 |
mengluy0125
force-pushed
the
export-D60364165
branch
from
6f3c775 to
ff6ac7f
Compare
pytorch-bot bot
pushed a commit
that referenced
this pull request
Jul 31, 2024
Summary: Pull Request resolved: #132297 We observed that many introduced nodes during split cat and batch fusion pattern optimization did not have example value meta data, which will cause problems in our follow up pattern optimizations, thus we add all missing values. We also fix bugs in some meta update and corner case bug for the old pattern, which caused problems in the follow up pattern optimization. We delete merge_stack_tahn_unbind_pass pattern, which was designed for cmf model, and it could be replaced by the more advanced pattern we added, thus we remove it for easy maintenance. Test Plan: # unit test ``` buck2 test //caffe2/test/inductor:split_cat_fx_passes ``` Test UI: https://www.internalfb.com/intern/testinfra/testrun/3940649915381395 Network: Up: 90KiB Down: 118KiB (reSessionID-a4a438a0-b043-4ae0-8cf4-a83a5150d0fa) Jobs completed: 28. Time elapsed: 4:57.9s. Cache hits: 0%. Commands: 2 (cached: 0, remote: 0, local: 2) Tests finished: Pass 10. Fail 0. Fatal 0. Skip 1. Build failure 0 ``` buck2 test mode/opt pytorch/diff_train_tests/ads/optimus:local_pt2_runner ``` Network: Up: 1.3GiB Down: 84MiB (reSessionID-ff135cdd-e42c-4ab5-8217-907ada465f01) Jobs completed: 61. Time elapsed: 21:56.5s. Cache hits: 0%. Commands: 39 (cached: 0, remote: 0, local: 39) Tests finished: Pass 8. Fail 0. Fatal 0. Skip 0. Build failure 0 # benchmark ``` CUDA_VISIBLE_DEVICES=3 OC_CAUSE=1 buck2 run mode/opt //scripts/jackiexu0313/pt2:local_model_with_pt2 -- --test_mode batch-split --model_type "ig_ctr" --flow_id 584880697 ``` Counter({'pattern_matcher_nodes': 752, 'pattern_matcher_count': 732, 'normalization_pass': 328, 'normalization_aten_pass': 12, 'scmerge_cat_removed': 5, 'scmerge_cat_added': 4, 'scmerge_split_removed': 3, 'unbind_stack_pass': 3, 'batch_tanh': 2, 'scmerge_split_sections_removed': 2, 'scmerge_split_added': 2, 'optimize_cat_inputs_pass': 1, 'unbind_cat_to_view_pass': 1, 'fxgraph_cache_miss': 1}) Differential Revision: D60364165
jackiexu1992
approved these changes
Jul 31, 2024
mengluy0125
force-pushed
the
export-D60364165
branch
from
ff6ac7f to
cad8c3c
Compare
|
This pull request was exported from Phabricator. Differential Revision: D60364165 |
|
This pull request was exported from Phabricator. Differential Revision: D60364165 |
mengluy0125
force-pushed
the
export-D60364165
branch
from
cad8c3c to
6025dff
Compare
|
This pull request was exported from Phabricator. Differential Revision: D60364165 |
mengluy0125
added a commit
to mengluy0125/pytorch
that referenced
this pull request
Aug 1, 2024
…h#132297) Summary: Pull Request resolved: pytorch#132297 We observed that many introduced nodes during split cat and batch fusion pattern optimization did not have example value meta data, which will cause problems in our follow up pattern optimizations, thus we add all missing values. We also fix bugs in some meta update and corner case bug for the old pattern, which caused problems in the follow up pattern optimization. We delete merge_stack_tahn_unbind_pass pattern, which was designed for cmf model, and it could be replaced by the more advanced pattern we added, thus we remove it for easy maintenance. Test Plan: # unit test ``` buck2 test //caffe2/test/inductor:split_cat_fx_passes ``` Test UI: https://www.internalfb.com/intern/testinfra/testrun/15481123762720165 Network: Up: 230KiB Down: 702KiB (reSessionID-756346bf-6da3-4fa0-8d03-1b4fd61e0a7a) Jobs completed: 30. Time elapsed: 7:23.9s. Cache hits: 20%. Commands: 5 (cached: 1, remote: 0, local: 4) Tests finished: Pass 9. Fail 0. Fatal 0. Skip 1. Build failure 0 ``` buck2 test mode/opt pytorch/diff_train_tests/ads/optimus:local_pt2_runner ``` Network: Up: 1.3GiB Down: 84MiB (reSessionID-ff135cdd-e42c-4ab5-8217-907ada465f01) Jobs completed: 61. Time elapsed: 21:56.5s. Cache hits: 0%. Commands: 39 (cached: 0, remote: 0, local: 39) Tests finished: Pass 8. Fail 0. Fatal 0. Skip 0. Build failure 0 # benchmark ``` CUDA_VISIBLE_DEVICES=3 OC_CAUSE=1 buck2 run mode/opt //scripts/jackiexu0313/pt2:local_model_with_pt2 -- --test_mode batch-split --model_type "ig_ctr" --flow_id 584880697 ``` Counter({'pattern_matcher_nodes': 752, 'pattern_matcher_count': 732, 'normalization_pass': 328, 'normalization_aten_pass': 12, 'scmerge_cat_removed': 5, 'scmerge_cat_added': 4, 'scmerge_split_removed': 3, 'unbind_stack_pass': 3, 'batch_tanh': 2, 'scmerge_split_sections_removed': 2, 'scmerge_split_added': 2, 'optimize_cat_inputs_pass': 1, 'unbind_cat_to_view_pass': 1, 'fxgraph_cache_miss': 1}) Reviewed By: jackiexu1992 Differential Revision: D60364165
mengluy0125
force-pushed
the
export-D60364165
branch
from
6025dff to
4ce2875
Compare
…h#132297) Summary: Pull Request resolved: pytorch#132297 We observed that many introduced nodes during split cat and batch fusion pattern optimization did not have example value meta data, which will cause problems in our follow up pattern optimizations, thus we add all missing values. We also fix bugs in some meta update and corner case bug for the old pattern, which caused problems in the follow up pattern optimization. We delete merge_stack_tahn_unbind_pass pattern, which was designed for cmf model, and it could be replaced by the more advanced pattern we added, thus we remove it for easy maintenance. Test Plan: # unit test ``` buck2 test //caffe2/test/inductor:split_cat_fx_passes ``` Test UI: https://www.internalfb.com/intern/testinfra/testrun/15481123762720165 Network: Up: 230KiB Down: 702KiB (reSessionID-756346bf-6da3-4fa0-8d03-1b4fd61e0a7a) Jobs completed: 30. Time elapsed: 7:23.9s. Cache hits: 20%. Commands: 5 (cached: 1, remote: 0, local: 4) Tests finished: Pass 9. Fail 0. Fatal 0. Skip 1. Build failure 0 ``` buck2 test mode/opt pytorch/diff_train_tests/ads/optimus:local_pt2_runner ``` Network: Up: 1.3GiB Down: 84MiB (reSessionID-ff135cdd-e42c-4ab5-8217-907ada465f01) Jobs completed: 61. Time elapsed: 21:56.5s. Cache hits: 0%. Commands: 39 (cached: 0, remote: 0, local: 39) Tests finished: Pass 8. Fail 0. Fatal 0. Skip 0. Build failure 0 # benchmark ``` CUDA_VISIBLE_DEVICES=3 OC_CAUSE=1 buck2 run mode/opt //scripts/jackiexu0313/pt2:local_model_with_pt2 -- --test_mode batch-split --model_type "ig_ctr" --flow_id 584880697 ``` Counter({'pattern_matcher_nodes': 752, 'pattern_matcher_count': 732, 'normalization_pass': 328, 'normalization_aten_pass': 12, 'scmerge_cat_removed': 5, 'scmerge_cat_added': 4, 'scmerge_split_removed': 3, 'unbind_stack_pass': 3, 'batch_tanh': 2, 'scmerge_split_sections_removed': 2, 'scmerge_split_added': 2, 'optimize_cat_inputs_pass': 1, 'unbind_cat_to_view_pass': 1, 'fxgraph_cache_miss': 1}) Reviewed By: jackiexu1992 Differential Revision: D60364165
|
This pull request was exported from Phabricator. Differential Revision: D60364165 |
mengluy0125
force-pushed
the
export-D60364165
branch
from
4ce2875 to
6254680
Compare
|
@pytorchbot merge (Initiating merge automatically since Phabricator Diff has merged) |
Merge failedReason: This PR has internal changes and must be landed via Phabricator! Please try reimporting/rexporting the PR! Details for Dev Infra teamRaised by workflow job |
|
@pytorchbot merge |
Merge failedReason: This PR has internal changes and must be landed via Phabricator! Please try reimporting/rexporting the PR! Details for Dev Infra teamRaised by workflow job |
|
@pytorchbot merge -f 'unrelated' |
|
You need to provide a reason for using force merge, in the format @pytorchbot merge -f 'Explanation'.
|
|
@pytorchbot merge -f 'bypass flaky tests' |
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 |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
5 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Summary:
We observed that many introduced nodes during split cat and batch fusion pattern optimization did not have example value meta data, which will cause problems in our follow up pattern optimizations, thus we add all missing values.
We also fix bugs in some meta update and corner case bug for the old pattern, which caused problems in the follow up pattern optimization.
We delete merge_stack_tahn_unbind_pass pattern, which was designed for cmf model, and it could be replaced by the more advanced pattern we added, thus we remove it for easy maintenance.
Test Plan:
unit test
Test UI: https://www.internalfb.com/intern/testinfra/testrun/15481123762720165
Network: Up: 230KiB Down: 702KiB (reSessionID-756346bf-6da3-4fa0-8d03-1b4fd61e0a7a)
Jobs completed: 30. Time elapsed: 7:23.9s.
Cache hits: 20%. Commands: 5 (cached: 1, remote: 0, local: 4)
Tests finished: Pass 9. Fail 0. Fatal 0. Skip 1. Build failure 0
Network: Up: 1.3GiB Down: 84MiB (reSessionID-ff135cdd-e42c-4ab5-8217-907ada465f01)
Jobs completed: 61. Time elapsed: 21:56.5s.
Cache hits: 0%. Commands: 39 (cached: 0, remote: 0, local: 39)
Tests finished: Pass 8. Fail 0. Fatal 0. Skip 0. Build failure 0
benchmark
Counter({'pattern_matcher_nodes': 752, 'pattern_matcher_count': 732, 'normalization_pass': 328, 'normalization_aten_pass': 12, 'scmerge_cat_removed': 5, 'scmerge_cat_added': 4, 'scmerge_split_removed': 3, 'unbind_stack_pass': 3, 'batch_tanh': 2, 'scmerge_split_sections_removed': 2, 'scmerge_split_added': 2, 'optimize_cat_inputs_pass': 1, 'unbind_cat_to_view_pass': 1, 'fxgraph_cache_miss': 1})
cc @voznesenskym @penguinwu @EikanWang @jgong5 @Guobing-Chen @XiaobingSuper @zhuhaozhe @blzheng @wenzhe-nrv @jiayisunx @ipiszy @yf225 @chenyang78 @kadeng @muchulee8 @ColinPeppler @amjames @desertfire @chauhang