feat: transformers to v5

[PeterStaar-IBM]

This allows >=4.42.0,<5.0.0 and >=5.4.0,<6.0.0 by excluding the 5.0–5.3 patch ranges which do not support granite-docling correctly.

Two changes were made to docling_ibm_models/tableformer_v2/model.py:

1. Added GenerationMixin to the class inheritance

  # Before
  class TableFormerV2(PreTrainedModel):

  # After
  class TableFormerV2(PreTrainedModel, GenerationMixin):

In transformers v5, PreTrainedModel no longer inherits from GenerationMixin. Any model that defines prepare_inputs_for_generation is expected to explicitly inherit from GenerationMixin (placed after PreTrainedModel in the MRO).

2. Added self.post_init() at the end of init

          self.data_cells = config.data_cells
          self.post_init()  # added

In transformers v5, post_init() is no longer called automatically by PreTrainedModel.init. Each model must call it explicitly at the end of its own init. This method sets several instance attributes that transformers v5 relies on during from_pretrained, including all_tied_weights_keys — the dict that was causing the AttributeError.

[github-actions]

✅ DCO Check Passed

Thanks @PeterStaar-IBM, all your commits are properly signed off. 🎉

[mergify]

Merge Protections

Your pull request matches the following merge protections and will not be merged until they are valid.

🟢 Enforce conventional commit

Wonderful, this rule succeeded.

Make sure that we follow https://www.conventionalcommits.org/en/v1.0.0/

• title ~= ^(fix|feat|docs|style|refactor|perf|test|build|ci|chore|revert)(?:\(.+\))?(!)?:

[codecov]

Codecov Report

❌ Patch coverage is 77.77778% with 2 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
...odels/code_formula_model/code_formula_predictor.py	0.00%	2 Missing ⚠️

📢 Thoughts on this report? Let us know!

[dolfim-ibm]

We could add the following but not needed for runtime

from typing import Optional
if TYPE_CHECKING:
    from transformers import GenerationConfig

class TableFormerV2(PreTrainedModel, GenerationMixin):
    # Annotate generation_config to satisfy mypy:
    generation_config: Optional["GenerationConfig"]
    # ... rest of class ...