Codecov Report

Attention: 26 lines in your changes are missing coverage. Please review.

Comparison is base (8a6b70f) 56.45% compared to head (1791b91) 56.39%.
Report is 1 commits behind head on main.

❗ Current head 1791b91 differs from pull request most recent head de7b442. Consider uploading reports for the commit de7b442 to get more accurate results

@@            Coverage Diff             @@
##             main    #5877      +/-   ##
==========================================
- Coverage   56.45%   56.39%   -0.06%     
==========================================
  Files         504      504              
  Lines       29872    29906      +34     
  Branches     4489     4496       +7     
==========================================
+ Hits        16863    16866       +3     
- Misses      12191    12222      +31     
  Partials      818      818              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

You should sign off your commits (git commit -s ...) to pass DCO test (click on the link details to sign your past commits). You also to run the following commands to update the documentation and the new test in the folder onnx/backend/test/data.

python onnx/backend/test/cmd_tools.py generate-data --clean
python  onnx/defs/gen_doc.py
python onnx/backend/test/stat_coverage.py 

Is it possible to restore the original onnx models whicj already exist and keep only the one you add? That way it is easier to see your changes.

Sorry yes I will clean this.

These two lines make an error :
python onnx/defs/gen_doc.py
python onnx/backend/test/stat_coverage.py

even without my changes, do I have to work on another branch than main ?

What do you mean by do I have to work on another branch than main? Submitting a PR like you did is usually what people do.

If I understand your question correctly, you meant if you need to create a diff against a branch other than main. Main is the correct development branch. The issue may be because a different version of protobuf / platform caused slightly different binary results for protobuf. You may include only the relavent changes in your commit by manually selecting them.

For any question related to git, some famous LLM have usually good answers to something like how to revert to the previous commit in a subdirectory using Git.

Ok! :)

I actually can't manage to generate the node data, when I run

python onnx/backend/test/cmd_tools.py generate-data --clean

It only generates the node data for the existing tests, and when I run

python onnx/backend/test/cmd_tools.py generate-data --clean -d

it runs only the test of convtranspose but without the one I added.

The only changes I made was the op_convtranspose implementation and I added my test in the convtranspose.py file in
onnx>backend>test>case>node

What should I do to generate the data ?

I usually use this python onnx/backend/test/cmd_tools.py generate-data --clean to generate the data. Then I do git add ... with only the files I want to add to the PR. Then git commit -s -m .... Then I do git reset --hard to revert all the changes I do not want to include to the PR. Maybe some others have a more simple way to do it but that's what I do.

chachaleo, I created another PR with your changes #6175 and updated only the necessayr data, could you review it?