[quant][pt2] Support quantized conv bias in QAT fusion #112528
Closed
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
Summary: Previously QAT fusion assumes bias is not quantized. This works for the existing XNNPACKQuantizer, but not for custom quantizers that wish to quantize the bias. This commit supports this by adding the necessary patterns. This requires refactoring the code, however, since it previously assumed that there will only be one pair of q-dq (from conv weight) in the matched pattern, and this is no longer true. Test Plan: python test/test_quantization.py TestQuantizePT2EQAT.test_qat_conv_bn_bias_derived_qspec Reviewers: jerryzh168, kimishpatel Subscribers: jerryzh168, kimishpatel, supriyar [ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/112528
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit 6d2800e with merge base 46a34e8 ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
|
@andrewor14 has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
Summary: Previously QAT fusion assumes bias is not quantized. This works for the existing XNNPACKQuantizer, but not for custom quantizers that wish to quantize the bias. This commit supports this by adding the necessary patterns. This requires refactoring the code, however, since it previously assumed that there will only be one pair of q-dq (from conv weight) in the matched pattern, and this is no longer true. Test Plan: python test/test_quantization.py TestQuantizePT2EQAT.test_qat_conv_bn_bias_derived_qspec Reviewers: jerryzh168, kimishpatel Subscribers: jerryzh168, kimishpatel, supriyar Differential Revision: [D50856377](https://our.internmc.facebook.com/intern/diff/D50856377) [ghstack-poisoned]
andrewor14
added a commit
that referenced
this pull request
Nov 1, 2023
Summary: Previously QAT fusion assumes bias is not quantized. This works for the existing XNNPACKQuantizer, but not for custom quantizers that wish to quantize the bias. This commit supports this by adding the necessary patterns. This requires refactoring the code, however, since it previously assumed that there will only be one pair of q-dq (from conv weight) in the matched pattern, and this is no longer true. Test Plan: python test/test_quantization.py TestQuantizePT2EQAT.test_qat_conv_bn_bias_derived_qspec Reviewers: jerryzh168, kimishpatel Subscribers: jerryzh168, kimishpatel, supriyar ghstack-source-id: 066ef7a Pull Request resolved: #112528
Summary: Previously QAT fusion assumes bias is not quantized. This works for the existing XNNPACKQuantizer, but not for custom quantizers that wish to quantize the bias. This commit supports this by adding the necessary patterns. This requires refactoring the code, however, since it previously assumed that there will only be one pair of q-dq (from conv weight) in the matched pattern, and this is no longer true. Test Plan: python test/test_quantization.py TestQuantizePT2EQAT.test_qat_conv_bn_bias_derived_qspec Reviewers: jerryzh168, kimishpatel Subscribers: jerryzh168, kimishpatel, supriyar Differential Revision: [D50856377](https://our.internmc.facebook.com/intern/diff/D50856377) [ghstack-poisoned]
|
@andrewor14 has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
Summary: Previously QAT fusion assumes bias is not quantized. This works for the existing XNNPACKQuantizer, but not for custom quantizers that wish to quantize the bias. This commit supports this by adding the necessary patterns. This requires refactoring the code, however, since it previously assumed that there will only be one pair of q-dq (from conv weight) in the matched pattern, and this is no longer true. Test Plan: python test/test_quantization.py TestQuantizePT2EQAT.test_qat_conv_bn_bias_derived_qspec Reviewers: jerryzh168, kimishpatel Subscribers: jerryzh168, kimishpatel, supriyar Differential Revision: [D50856377](https://our.internmc.facebook.com/intern/diff/D50856377) [ghstack-poisoned]
jerryzh168
reviewed
Nov 2, 2023
|
|
||
| # Step (3): Fold BN weights into conv | ||
| # Step (3): Copy over args for weight (and optionally bias) q - dq nodes | ||
| _copy_over_q_dq_args(*node_map["conv_weight_q"]) |
There was a problem hiding this comment.
I'm wondering if these can be simplified if we use these? https://github.com/pytorch/pytorch/blob/main/torch/ao/quantization/pt2e/utils.py#L291
There was a problem hiding this comment.
Yeah, we can explore that as a BE task along with the conv arg copying. I prefer to do that separately
jerryzh168
approved these changes
Nov 2, 2023
|
@andrewor14 has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
|
@pytorchbot merge |
Merge startedYour change will be merged once all checks pass (ETA 0-4 Hours). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
Summary: Previously QAT fusion assumes bias is not quantized. This works for the existing XNNPACKQuantizer, but not for custom quantizers that wish to quantize the bias. This commit supports this by adding the necessary patterns. This requires refactoring the code, however, since it previously assumed that there will only be one pair of q-dq (from conv weight) in the matched pattern, and this is no longer true. Test Plan: python test/test_quantization.py TestQuantizePT2EQAT.test_qat_conv_bn_bias_derived_qspec Reviewers: jerryzh168, kimishpatel Subscribers: jerryzh168, kimishpatel, supriyar Differential Revision: [D50856377](https://our.internmc.facebook.com/intern/diff/D50856377) [ghstack-poisoned]
Merge failedReason: New commits were pushed while merging. Please rerun the merge command. Details for Dev Infra teamRaised by workflow job |
|
@andrewor14 has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
|
@pytorchbot merge |
Merge startedYour change will be merged once all checks pass (ETA 0-4 Hours). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
xuhancn
pushed a commit
to xuhancn/pytorch
that referenced
this pull request
Nov 7, 2023
Summary: Previously QAT fusion assumes bias is not quantized. This works for the existing XNNPACKQuantizer, but not for custom quantizers that wish to quantize the bias. This commit supports this by adding the necessary patterns. This requires refactoring the code, however, since it previously assumed that there will only be one pair of q-dq (from conv weight) in the matched pattern, and this is no longer true. Test Plan: python test/test_quantization.py TestQuantizePT2EQAT.test_qat_conv_bn_bias_derived_qspec Reviewers: jerryzh168, kimishpatel Subscribers: jerryzh168, kimishpatel, supriyar Differential Revision: [D50856377](https://our.internmc.facebook.com/intern/diff/D50856377) Pull Request resolved: pytorch#112528 Approved by: https://github.com/jerryzh168
Skylion007
pushed a commit
to Skylion007/pytorch
that referenced
this pull request
Nov 14, 2023
Summary: Previously QAT fusion assumes bias is not quantized. This works for the existing XNNPACKQuantizer, but not for custom quantizers that wish to quantize the bias. This commit supports this by adding the necessary patterns. This requires refactoring the code, however, since it previously assumed that there will only be one pair of q-dq (from conv weight) in the matched pattern, and this is no longer true. Test Plan: python test/test_quantization.py TestQuantizePT2EQAT.test_qat_conv_bn_bias_derived_qspec Reviewers: jerryzh168, kimishpatel Subscribers: jerryzh168, kimishpatel, supriyar Differential Revision: [D50856377](https://our.internmc.facebook.com/intern/diff/D50856377) Pull Request resolved: pytorch#112528 Approved by: https://github.com/jerryzh168
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 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.
Stack from ghstack (oldest at bottom):
Summary: Previously QAT fusion assumes bias is not quantized.
This works for the existing XNNPACKQuantizer, but not for custom
quantizers that wish to quantize the bias. This commit supports
this by adding the necessary patterns. This requires refactoring
the code, however, since it previously assumed that there will
only be one pair of q-dq (from conv weight) in the matched
pattern, and this is no longer true.
Test Plan:
python test/test_quantization.py TestQuantizePT2EQAT.test_qat_conv_bn_bias_derived_qspec
Reviewers: jerryzh168, kimishpatel
Subscribers: jerryzh168, kimishpatel, supriyar
Differential Revision: D50856377