-
Notifications
You must be signed in to change notification settings - Fork 27.6k
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
Add RemBERT ONNX config #20520
Add RemBERT ONNX config #20520
Conversation
The documentation is not available anymore as the PR was closed or merged. |
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.
Hi @hchings, the PR looks excellent! Did you try to run tests locally?
RUN_SLOW=1 pytest tests/onnx/test_onnx_v2.py -k "rembert"
Could you also remove the Fixes #...
before the link to the ONNX issue to avoid an auto-close from GitHub? Thanks a lot for your contribution!
Yes, all slow tests passed for PyTorch locally. Should we add TensorFlow tests as well? My understanding is TF tests are needed only when TF has parity with PyTorch implementations. But correct me if I'm wrong. |
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.
Ok looks good. Let's see wait on @lewtun.
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.
Thanks for your contribution!
* rembert onnx config * formatting Co-authored-by: Ho <[email protected]>
* rembert onnx config * formatting Co-authored-by: Ho <[email protected]>
* rembert onnx config * formatting Co-authored-by: Ho <[email protected]>
* rembert onnx config * formatting Co-authored-by: Ho <[email protected]>
What does this PR do?
Add RemBERT ONNX config (part of #16308)
The max absolute difference between reference model and ONNX exported model is around
2e-05
in testings. I learned from other PRs that this discrepancy is within an acceptable range so I loosen the default atol.Slow tests are passed (
RUN_SLOW=1 pytest tests/onnx/test_onnx_v2.py -k "rembert"
).I'm new to contributing to Transformers. If anyone can help me understand what is lacking, it would be appreciated!
Before submitting
Pull Request section?
to it if that's the case.
documentation guidelines, and
here are tips on formatting docstrings.
Who can review?
Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.
@lewtun & @chainyo for ONNX and @Iwontbecreative for RemBERT.