Allow to add custom post transform functions that are not supported by the ONNX spec yet #463
Conversation
This file contains 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
|
|
…orm nodes that are not supported by the ONNX spec yet Signed-off-by: Jan-Benedikt Jagusch <[email protected]>
janjagusch
force-pushed
the
allow-for-custom-post-transform-functions
branch
from
6cc1947 to
5929839
Compare
Signed-off-by: Jan-Benedikt Jagusch <[email protected]>
janjagusch
force-pushed
the
allow-for-custom-post-transform-functions
branch
from
3cccfd3 to
5b59425
Compare
|
hi, @xadupre! 👋 anything i can do to help move this PR forward? 🙂 |
wenbingl
approved these changes
May 20, 2021
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Outdated suggestions cannot be applied.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.
Closes #462
I have already added support for
objective='poisson', but I believe this PR also allows to easily add support forobjective='tweedie'andobjective='gamma'.Ultimately, it would be useful to have "EXPONENTIAL" as a supported value in
attrs['post_transform'], such that we don't have to handle this special case here. However, that seems to involve quite some changes inonxx/onnxruntime. If anybody knows how to move these changes upstream, that would be very appreciated.What this PR is still missing is tests. I have tested this locally, asserting that the predictions of the LightGMB model and the ONNX model are equivalent. I don't see an easy way to add this to your test suite, though. Some advice on this would be great.