João David

> Hi @joaocmd, there's been some large updates to our TF models regarding how they're built - @Rocketknight1 can give you more details :) > > Are these errors happening...

Thanks @amyeroberts, understood. As for the rebase, I had not done one in quite some time and it seems like I did mess it up. I think that is now...

Thank you @amyeroberts, that makes sense. I've already updated it to match the pattern. I'm still having some trouble with the `test_compile_tf_model`. Initially it was failing because it was passing...

Hi @amyeroberts! Sorry for the late response as I've been quite busy... It was failing more tests on my local machine than on the CI run, but after merging the...

Hi @amyeroberts I might be misunderstanding something but I think `test_save_load` does not test any PyTorch to TensorFlow equivalence. I think the problem is that when the two convolutional layers...

Hi @Rocketknight1, could you give me some pointers on the three remaining tests? I haven't looked closely into `test_modeling_tf_swiftformer.py::TFSwiftFormerModelIntegrationTest::test_inference_image_classification_head` yet because I think it makes sense to leave that one...

Hi @amyeroberts , I haven't made any changes since my last comment as I was stuck and had some other responsibilities. I would like to finish the issue especially because...

Hi @Rocketknight1, thanks a lot for your input! I printed out the transformed names of the pytorch and tensorflow weights and have a few questions. First, I didn't find anything...

I'm now getting `tests/models/swiftformer/test_modeling_tf_swiftformer.py::TFSwiftFormerModelTest::test_pt_tf_model_equivalence - AssertionError: Tuples differ: (13, 7, 7, 220) != (13, 220, 7, 7)`. @amyeroberts you mentioned this last time I asked you regarding model inputs/outputs: >...