Pablo Montalvo

There's just these: ```bash FAILED tests/models/lightonocr/test_modeling_lightonocr.py::LightOnOCRForConditionalGenerationModelTest::test_model_parallelism - RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cuda:0 and cuda:1! FAILED tests/models/lightonocr/test_modeling_lightonocr.py::LightOnOCRForConditionalGenerationIntegrationTest::test_model_can_generate_without_images - AttributeError:...

Hey @konstantinos-p , I will be taking a look at your PR now! First of all, I'm seeing a few `main` conflicts that are mainly due to https://github.com/huggingface/transformers/pull/40760, could you...

@konstantinos-p can you merge `main` into your branch by the way please? :hugs: The base version is from May and this particular one was yanked :sweat_smile:

Hey @konstantinos-p, are you sure you've rebased on origin:main? merge or rebase make no difference here as we'll squash in the end, but the `transformers` version should be v5-dev, not...

To fix conflicts usually I just do (after syncing my fork): ```bash git checkout main git pull git checkout my_branch_to_update git merge main ``` then fix conflicts that appeared (test_modeling_common...

Hey @konstantinos-p, thanks for your patience! reviewing today :)

Thanks a lot! Ping me if you encounter issues :hugs:

Hey @konstantinos-p , we are currently patching 4.57 and also doing a lot of work to prepare for v5, apologies bandwidth is limited :bow: Did you manage to solve you...

For completeness, I agree that of making processors responsible for uniform device placement is quite limited, and the costs are high. If we add it, it should be in `ProcessorMixin`...

Thanks @philippguevorguian , then I will see if we can integrate it. I can't say for now if we'll be able to integrate it as our principles for modeling files...