-
Notifications
You must be signed in to change notification settings - Fork 21.4k
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
[Quant] Remove weight from DTypeConfig for non-weighted ops #86335
Conversation
Summary: Weight dtypes should be specified only for weighted ops like conv and linear. This commit removes weight dtypes from the DTypeConfigs used in binary ops and fixed qparams ops. Test Plan: python test/test_quantization.py TestQuantizeFx python test/test_quantization.py TestQuantizeFxOps Reviewers: jerryzh168, vkuzo Subscribers: jerryzh168, vkuzo [ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/86335
Note: Links to docs will display an error until the docs builds have been completed. ✅ No Failures, 2 PendingAs of commit 252409d: This comment was automatically generated by Dr. CI and updates every 15 minutes. |
Summary: Weight dtypes should be specified only for weighted ops like conv and linear. This commit removes weight dtypes from the DTypeConfigs used in binary ops and fixed qparams ops. Test Plan: python test/test_quantization.py TestQuantizeFx python test/test_quantization.py TestQuantizeFxOps Reviewers: jerryzh168, vkuzo Subscribers: jerryzh168, vkuzo ghstack-source-id: aa352df1f52274bb345af29c3fc29d16669997aa Pull Request resolved: #86335
@pytorchbot merge |
@pytorchbot successfully started a merge job. Check the current status here. |
Hey @andrewor14. |
…86335) Summary: Weight dtypes should be specified only for weighted ops like conv and linear. This commit removes weight dtypes from the DTypeConfigs used in binary ops and fixed qparams ops. Pull Request resolved: #86335 Approved by: https://github.com/vkuzo Test Plan: contbuild & OSS CI, see https://hud.pytorch.org/commit/pytorch/pytorch/cebf08afb24dec0720935b9a9bd64ecf05b472d5 Test plan from GitHub: python test/test_quantization.py TestQuantizeFx python test/test_quantization.py TestQuantizeFxOps Reviewed By: seemethere Differential Revision: D40167082 Pulled By: seemethere fbshipit-source-id: 1f29611b5e3d9ecb3528c899585d0e60364b6982
Stack from ghstack (oldest at bottom):
Summary: Weight dtypes should be specified only for weighted
ops like conv and linear. This commit removes weight dtypes
from the DTypeConfigs used in binary ops and fixed qparams ops.
Test Plan:
python test/test_quantization.py TestQuantizeFx
python test/test_quantization.py TestQuantizeFxOps
Reviewers: jerryzh168, vkuzo
Subscribers: jerryzh168, vkuzo