add learnable_fake_quantize in pt2e #3135
Merged
+1,045
−0
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/pytorch/ao/3135
Note: Links to docs will display an error until the docs builds have been completed. ❌ 4 New FailuresAs of commit dd004b2 with merge base a5a8fe2 ( NEW FAILURES - The following jobs have failed:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
jerryzh168
reviewed
Oct 9, 2025
There was a problem hiding this comment.
can you correct the name to fake_quantizer?
There was a problem hiding this comment.
@jerryzh168
I'm trying to be consistent with FakeQuantize() class at: https://github.com/pytorch/ao/blob/main/torchao/quantization/pt2e/fake_quantize.py.
I'm ok to change the file name: learnable_fake_quantize.py to learnable_fake_quantizer.py and the class name from LearnableFakeQuantize() to LearnableFakeQuantizer() if that is what you meant.
There was a problem hiding this comment.
Yeah this is for pt2e flow, so I feel the name FakeQuantize is more consistent with the other classes?
There was a problem hiding this comment.
FakeQuantize is the wrong name I think since it's a verb not noun, it's not consistent with Observer which is a noun. We can correct, and deprecate FakeQuantize itself as we add new things.
But it's OK to do this separately as well
navsud
added a commit
to navsud/ao
that referenced
this pull request
Oct 9, 2025
Summary: Learnable Fake Quantize is a popular technique used especially for low-precision QAT. This was available in fx quantization at https://github.com/pytorch/pytorch/blob/main/torch/ao/quantization/_learnable_fake_quantize.py#L10?, but not in pt2e quantization flow. This change adds learnable fake quantize in pt2e, with some changes, especially related to how channel wise quantization works. Differential Revision: D83542550
38017fc to
60b321a
Compare
navsud
added a commit
to navsud/ao
that referenced
this pull request
Oct 9, 2025
Summary: Learnable Fake Quantize is a popular technique used especially for low-precision QAT. This was available in fx quantization at https://github.com/pytorch/pytorch/blob/main/torch/ao/quantization/_learnable_fake_quantize.py#L10?, but not in pt2e quantization flow. This change adds learnable fake quantize in pt2e, with some changes, especially related to how channel wise quantization works. Differential Revision: D83542550
60b321a to
a2389fc
Compare
andrewor14
approved these changes
Oct 9, 2025
There was a problem hiding this comment.
Yeah this is for pt2e flow, so I feel the name FakeQuantize is more consistent with the other classes?
navsud
added a commit
to navsud/ao
that referenced
this pull request
Oct 9, 2025
Summary: Learnable Fake Quantize is a popular technique used especially for low-precision QAT. This was available in fx quantization at https://github.com/pytorch/pytorch/blob/main/torch/ao/quantization/_learnable_fake_quantize.py#L10?, but not in pt2e quantization flow. This change adds learnable fake quantize in pt2e, with some changes, especially related to how channel wise quantization works. Reviewed By: andrewor14 Differential Revision: D83542550
a2389fc to
149bc0b
Compare
Summary: Learnable Fake Quantize is a popular technique used especially for low-precision QAT. This was available in fx quantization at https://github.com/pytorch/pytorch/blob/main/torch/ao/quantization/_learnable_fake_quantize.py#L10?, but not in pt2e quantization flow. This change adds learnable fake quantize in pt2e, with some changes, especially related to how channel wise quantization works. Reviewed By: andrewor14 Differential Revision: D83542550
149bc0b to
dd004b2
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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: Learnable Fake Quantize is a popular technique used especially for low-precision QAT. This was available in fx quantization at https://github.com/pytorch/pytorch/blob/main/torch/ao/quantization/_learnable_fake_quantize.py#L10?, but not in pt2e quantization flow. This change adds learnable fake quantize in pt2e, with some changes, especially related to how channel wise quantization works.
Differential Revision: D83542550