-
Notifications
You must be signed in to change notification settings - Fork 565
OnnxToTorch bicubic interpolation #3802
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
Conversation
5635202
to
71487e6
Compare
f4c442b
to
f718e52
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Hey, Alex. Looking this over, I have an idea that might resolve the numerical issues. Do you mind taking a look?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
More comments!
1ed2b47
to
a4d7d79
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Before making changes, check out the suggestions for fixing the numerics in aldesilv#3.
Besides the suggestions in that PR, and the comments in this review, I'd generally like more comments in this code. Where does this implementation come from? What assumptions did we make?
To anyone who doesn't have context on cubic interpolation, this code would be completely inscrutable.
b5519b8
to
2dfcc1c
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Great! One small change and should be good to go.
It might also be a good idea to rework some of the function signatures to take const references to avoid making unnecessary copies.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM
This reverts commit 889a836.
(nod-ai/SHARK-TestSuite#391)
Repro (using SHARK TestSuite):
python run.py --torchtolinalg -m cl-onnx-iree -t cubic_test