Transformers v5.9.0 breaks support for pytorch with USE_DISTRIBUTED=0

### System Info Version: 5.9.0 Platform: Intel-based MacOS 26.5 Python Version: 3.14 Pytorch: 2.12.0 (self-built for Intel MacOS) ### Who can help? @remi-or ### Information - [ ] The official example scripts - [ ] My own modified scripts ### Tasks - [ ] An officially supported task in the `examples` folder (such as GLUE/SQuAD, ...) - [ ] My own task or dataset (give details below) ### Reproduction Run the following command on macOS version of pytorch (especially USE_DISTRIBUTED=0 during compilation): ``` uv run --with torch --with transformers python -c " import torch from transformers import AutoModelForImageToImage try: # Using a tiny swin2SR model for fast verification model = AutoModelForImageToImage.from_pretrained('caidas/swin2SR-lightweight-x2-64') print('\nSuccess! AutoModelForImageToImage initialized perfectly.') print(f'Model class: {model.__class__.__name__}') except Exception as e: print(f'Test failed: {e}') " ``` The execution fails with the following error: ``` Traceback (most recent call last): File "/Users/msdarwish/.local/share/uv/tools/docling-slim/lib/python3.14/site-packages/transformers/utils/import_utils.py", line 2249, in __getattr__ module = self._get_module(self._class_to_module[name]) File "/Users/msdarwish/.local/share/uv/tools/docling-slim/lib/python3.14/site-packages/transformers/utils/import_utils.py", line 2483, in _get_module raise e File "/Users/msdarwish/.local/share/uv/tools/docling-slim/lib/python3.14/site-packages/transformers/utils/import_utils.py", line 2481, in _get_module return importlib.import_module("." + module_name, self.__name__) ~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/Users/msdarwish/.local/share/uv/python/cpython-3.14.4-macos-x86_64-none/lib/python3.14/importlib/__init__.py", line 88, in import_module return _bootstrap._gcd_import(name[level:], package, level) ~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "<frozen importlib._bootstrap>", line 1406, in _gcd_import File "<frozen importlib._bootstrap>", line 1371, in _find_and_load File "<frozen importlib._bootstrap>", line 1342, in _find_and_load_unlocked File "<frozen importlib._bootstrap>", line 938, in _load_unlocked File "<frozen importlib._bootstrap_external>", line 759, in exec_module File "<frozen importlib._bootstrap>", line 491, in _call_with_frames_removed File "/Users/msdarwish/.local/share/uv/tools/docling-slim/lib/python3.14/site-packages/transformers/generation/utils.py", line 73, in <module> from .continuous_batching import ContinuousMixin File "/Users/msdarwish/.local/share/uv/tools/docling-slim/lib/python3.14/site-packages/transformers/generation/continuous_batching/__init__.py", line 14, in <module> from .cache import PagedAttentionCache File "/Users/msdarwish/.local/share/uv/tools/docling-slim/lib/python3.14/site-packages/transformers/generation/continuous_batching/cache.py", line 24, in <module> from .distributed import DistributedHelper File "/Users/msdarwish/.local/share/uv/tools/docling-slim/lib/python3.14/site-packages/transformers/generation/continuous_batching/distributed.py", line 19, in <module> from torch.distributed.tensor.device_mesh import DeviceMesh File "/Users/msdarwish/.local/share/uv/tools/docling-slim/lib/python3.14/site-packages/torch/distributed/tensor/__init__.py", line 4, in <module> import torch.distributed.tensor._ops # force import all built-in dtensor ops ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/Users/msdarwish/.local/share/uv/tools/docling-slim/lib/python3.14/site-packages/torch/distributed/tensor/_ops/__init__.py", line 2, in <module> from ._conv_ops import * # noqa: F403 ^^^^^^^^^^^^^^^^^^^^^^^^ File "/Users/msdarwish/.local/share/uv/tools/docling-slim/lib/python3.14/site-packages/torch/distributed/tensor/_ops/_conv_ops.py", line 7, in <module> from torch.distributed.tensor._dtensor_spec import DTensorSpec, TensorMeta File "/Users/msdarwish/.local/share/uv/tools/docling-slim/lib/python3.14/site-packages/torch/distributed/tensor/_dtensor_spec.py", line 10, in <module> from torch.distributed.tensor.placement_types import ( ...<7 lines>... ) File "/Users/msdarwish/.local/share/uv/tools/docling-slim/lib/python3.14/site-packages/torch/distributed/tensor/placement_types.py", line 11, in <module> import torch.distributed._functional_collectives as funcol File "/Users/msdarwish/.local/share/uv/tools/docling-slim/lib/python3.14/site-packages/torch/distributed/_functional_collectives.py", line 10, in <module> import torch.distributed.distributed_c10d as c10d File "/Users/msdarwish/.local/share/uv/tools/docling-slim/lib/python3.14/site-packages/torch/distributed/distributed_c10d.py", line 25, in <module> from torch._C._distributed_c10d import ( ...<22 lines>... ) ModuleNotFoundError: No module named 'torch._C._distributed_c10d'; 'torch._C' is not a package The above exception was the direct cause of the following exception: Traceback (most recent call last): File "/Users/msdarwish/.local/share/uv/tools/docling-slim/lib/python3.14/site-packages/transformers/utils/import_utils.py", line 2249, in __getattr__ module = self._get_module(self._class_to_module[name]) File "/Users/msdarwish/.local/share/uv/tools/docling-slim/lib/python3.14/site-packages/transformers/utils/import_utils.py", line 2483, in _get_module raise e File "/Users/msdarwish/.local/share/uv/tools/docling-slim/lib/python3.14/site-packages/transformers/utils/import_utils.py", line 2481, in _get_module return importlib.import_module("." + module_name, self.__name__) ~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/Users/msdarwish/.local/share/uv/python/cpython-3.14.4-macos-x86_64-none/lib/python3.14/importlib/__init__.py", line 88, in import_module return _bootstrap._gcd_import(name[level:], package, level) ~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "<frozen importlib._bootstrap>", line 1406, in _gcd_import File "<frozen importlib._bootstrap>", line 1371, in _find_and_load File "<frozen importlib._bootstrap>", line 1342, in _find_and_load_unlocked File "<frozen importlib._bootstrap>", line 938, in _load_unlocked File "<frozen importlib._bootstrap_external>", line 759, in exec_module File "<frozen importlib._bootstrap>", line 491, in _call_with_frames_removed File "/Users/msdarwish/.local/share/uv/tools/docling-slim/lib/python3.14/site-packages/transformers/models/auto/modeling_auto.py", line 21, in <module> from .auto_factory import ( ...<4 lines>... ) File "/Users/msdarwish/.local/share/uv/tools/docling-slim/lib/python3.14/site-packages/transformers/models/auto/auto_factory.py", line 43, in <module> from ...generation import GenerationMixin File "/Users/msdarwish/.local/share/uv/tools/docling-slim/lib/python3.14/site-packages/transformers/utils/import_utils.py", line 2337, in __getattr__ raise ModuleNotFoundError( f"Could not import module '{name}'. Are this object's requirements defined correctly?" ) from e ModuleNotFoundError: Could not import module 'GenerationMixin'. Are this object's requirements defined correctly? The above exception was the direct cause of the following exception: Traceback (most recent call last): File "/Users/msdarwish/.local/bin/docling", line 4, in <module> from docling.cli.main import app File "/Users/msdarwish/.local/share/uv/tools/docling-slim/lib/python3.14/site-packages/docling/cli/main.py", line 109, in <module> from docling.document_converter import ( ...<10 lines>... ) File "/Users/msdarwish/.local/share/uv/tools/docling-slim/lib/python3.14/site-packages/docling/document_converter.py", line 68, in <module> from docling.pipeline.asr_pipeline import AsrPipeline File "/Users/msdarwish/.local/share/uv/tools/docling-slim/lib/python3.14/site-packages/docling/pipeline/asr_pipeline.py", line 33, in <module> from docling.pipeline.base_pipeline import BasePipeline File "/Users/msdarwish/.local/share/uv/tools/docling-slim/lib/python3.14/site-packages/docling/pipeline/base_pipeline.py", line 36, in <module> from docling.models.stages.chart_extraction.granite_vision import ( ...<2 lines>... ) File "/Users/msdarwish/.local/share/uv/tools/docling-slim/lib/python3.14/site-packages/docling/models/stages/chart_extraction/granite_vision.py", line 27, in <module> from transformers import AutoModelForImageTextToText, AutoProcessor File "/Users/msdarwish/.local/share/uv/tools/docling-slim/lib/python3.14/site-packages/transformers/utils/import_utils.py", line 2337, in __getattr__ raise ModuleNotFoundError( f"Could not import module '{name}'. Are this object's requirements defined correctly?" ) from e ModuleNotFoundError: Could not import module 'AutoModelForImageTextToText'. Are this object's requirements defined correctly? ``` The command runs successfully when using: ``` uv run --with torch --with transformers==5.8.1 ... ``` It seems to be coming from the newly introduced `transformers/generation/continuous_batching/distributed.py`. ### Expected behavior The command runs successfully