-
Notifications
You must be signed in to change notification settings - Fork 1.7k
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
[Feature] add onnx simplify #751
Conversation
Codecov Report
@@ Coverage Diff @@
## master #751 +/- ##
==========================================
- Coverage 64.77% 62.67% -2.10%
==========================================
Files 136 139 +3
Lines 7999 8266 +267
Branches 1412 1484 +72
==========================================
Hits 5181 5181
- Misses 2549 2816 +267
Partials 269 269
Flags with carried forward coverage won't be shown. Click here to find out more.
Continue to review full report at Codecov.
|
Hi @RunningLeon , thanks for your excellent work! I'm the maintainer of onnx-simplifier (and onnx optimizer). Will it help if I support custom op registration in onnx-simplifier? mmcv will benefit from the future update of onnx-simplifier if it directly uses onnx-simplifier package. |
Hi, @daquexian, |
This PR adds
mmcv.onnx.simplify
module, which is based on onnx-simplifier, an useful tool to simplify and optimize ONNX models.Because we want to simplify models with MMCV's custom ops, which should be registered before simplifying. Thus, we have to modify source codes from onnx-simplifier, and include them in mmcv repository.
This PR includes:
mmcv.onnx.simplify
module;After this PR merged,
pytorch2onnx.py
in many projects such as mmdetection could be updated.