fix(models): Rename fpn_position_encoding to fpn_position_embeddings

[lllangWV]

What does this PR do?

The dataclass output attribute was named fpn_position_encoding but code in _prepare_vision_features was accessing fpn_position_embeddings, causing an AttributeError. Renamed to fpn_position_embeddings across all SAM2, SAM3, and EdgeTAM models for consistency with transformers naming conventions.

Affected models: sam2, sam2_video, sam3, sam3_tracker, sam3_tracker_video, sam3_video, edgetam, edgetam_video

Maybe the other changes where unnecessary, but I like to have consistent naming scheme through similar modules.

I could have just modified

transformers/models/sam3_tracker_video/modeling_sam3_tracker_video.py", line 1912 to

vision_pos_embeds = image_outputs.fpn_position_embeddings 

->

vision_pos_embeds = image_outputs.fpn_position_encodings

If you all just want the minimal change I can go back and undo the other changes

Before submitting

• This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
• Did you read the contributor guideline,
  Pull Request section?
• Was this discussed/approved via a Github issue or the forum? Please add a link
  to it if that's the case.
• Did you make sure to update the documentation with your changes? Here are the
  documentation guidelines, and
  here are tips on formatting docstrings.
• I ran doc-builder build transformers docs/source/en --build_dir ~/tmp/test-build Nothing seemed to change. I think the docs for this get automatically updated.

Building docs for transformers docs/source/en /home/lllang/tmp/test-build/transformers/main/en
Building the MDX files:  13%|████████████████████████                                                                                                                                                                | 81/618 [00:00<00:00, 797.15it/s]use_kernel_func_from_hub is not available in the installed kernels version. Please upgrade kernels to use this feature.
Building the MDX files: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 618/618 [00:47<00:00, 13.12it/s]
Resolving internal links: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 605/605 [00:01<00:00, 584.87it/s]

• [ x] Did you write any new necessary tests?
  • Due to the name change, a test had to be modified tests/models/sam3/test_modeling_sam3.py

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.

[github-actions]

[For maintainers] Suggested jobs to run (before merge)

run-slow: edgetam, edgetam_video, sam2, sam2_video, sam3, sam3_tracker, sam3_tracker_video, sam3_video

[Rocketknight1]

cc @NielsRogge @yonigozlan @molbap

[tomaarsen]

Hello!

This was taken care of with #43487, which instead renames fpn_position_embeddings (which was erroneously introduced in #42564 3 days ago) to fpn_position_encoding, which was originally intended. Thank you for your PR, but the issue has now been resolved.

• Tom Aarsen