Enable range learning for QAT #2033
Open
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/2033
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit ec45945 with merge base 31f119e ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
facebook-github-bot
added
the
CLA Signed
andrewor14
added
topic: new feature
|
@andrewor14 has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
andrewor14
force-pushed
the
range-learning
branch
from
cf3e00e to
f3903d4
Compare
facebook-github-bot
added
the
CLA Signed
|
@andrewor14 has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
andrewor14
force-pushed
the
range-learning
branch
from
f3903d4 to
3277221
Compare
|
@andrewor14 has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
andrewor14
force-pushed
the
range-learning
branch
from
3277221 to
b33281e
Compare
|
@andrewor14 has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
**Summary:** This commit adds the option for QAT users to use range
learning during training. Range learning means we train the scale
and zero point instead of recomputing them based on the input
at every iteration.
Example usage:
```
import torch
from torchao.quantization import quantize_
from torchao.quantization.qat import (
FakeQuantizeConfig,
IntXQuantizationAwareTrainingConfig,
initialize_fake_quantizers,
)
config = FakeQuantizeConfig(
torch.int8,
"per_channel",
is_dynamic=False,
range_learning=True,
scale_precision=torch.float32,
zero_point_precision=torch.float32,
)
m = M()
example_inputs = (torch.randn(16, 32),)
quantize_(m, IntXQuantizationAwareTrainingConfig(weight_config=config))
# New required step to turn scales and zero points into trainable
# `nn.Parameters`, must be called before initializing the optimizer
initialize_fake_quantizers(m, example_inputs)
# initialize the optimizer
# do training
```
**Test Plan:**
python test/quantization/test_qat.py -k test_fake_quantize_config_dynamic_and_range_learning
python test/quantization/test_qat.py -k test_fake_quantizer_range_learning
python test/quantization/test_qat.py -k test_qat_range_learning
andrewor14
force-pushed
the
range-learning
branch
from
b33281e to
ec45945
Compare
|
@andrewor14 has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
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: This commit adds the option for QAT users to use range learning during training. Range learning means we train the scale and zero point instead of recomputing them based on the input in every iteration.
Example usage:
Test Plan:
python test/quantization/test_qat.py -k test_fake_quantize_config_dynamic_and_range_learning
python test/quantization/test_qat.py -k test_fake_quantizer_range_learning
python test/quantization/test_qat.py -k test_qat_range_learning