docling-project / docling-project/docling

Cannot import module AutoProcessor

Open
#3,179 1 comment 0 reactions 0 assignees View on GitHub
bug triage/close-stale
Dominant language
Python
Stars
66.4k
Forks
4.8k
Avg merge
3d 4h
Merged PRs (30d)
95

Description

I have the following issue with installing the version 2.81.0 of docling. From the likes of it, there is a mismatch between the transformer version used inside docling and the presence of AutoProcessor.

'''
from docling.document_converter import DocumentConverter
'''

---------------------------------------------------------------------------
ModuleNotFoundError Traceback (most recent call last)
File ~/workspace/.venv/lib/python3.12/site-packages/transformers/utils/import_utils.py:2317, in _LazyModule.__getattr__(self, name)
2316 try:
-> [2317](https://vscode-remote+wsl-002bdebian.vscode-resource.vscode-cdn.net/workspace/test_local/~/workspace/.venv/lib/python3.12/site-packages/transformers/utils/import_utils.py:2317) module = self._get_module(self._class_to_module[name])
2318 value = getattr(module, name)

File ~/workspace/.venv/lib/python3.12/site-packages/transformers/utils/import_utils.py:2347, in _LazyModule._get_module(self, module_name)
2346 except Exception as e:
-> [2347](https://vscode-remote+wsl-002bdebian.vscode-resource.vscode-cdn.net/workspace/test_local/~/workspace/.venv/lib/python3.12/site-packages/transformers/utils/import_utils.py:2347) raise e

File ~/workspace/.venv/lib/python3.12/site-packages/transformers/utils/import_utils.py:2345, in _LazyModule._get_module(self, module_name)
2344 try:
-> [2345](https://vscode-remote+wsl-002bdebian.vscode-resource.vscode-cdn.net/workspace/test_local/~/workspace/.venv/lib/python3.12/site-packages/transformers/utils/import_utils.py:2345) return importlib.import_module("." + module_name, self.__name__)
2346 except Exception as e:

File /usr/local/lib/python3.12/importlib/__init__.py:90, in import_module(name, package)
89 level += 1
---> [90](https://vscode-remote+wsl-002bdebian.vscode-resource.vscode-cdn.net/usr/local/lib/python3.12/importlib/__init__.py:90) return _bootstrap._gcd_import(name[level:], package, level)

File :1387, in _gcd_import(name, package, level)

File :1360, in _find_and_load(name, import_)

File :1331, in _find_and_load_unlocked(name, import_)

File :935, in _load_unlocked(spec)

File :999, in exec_module(self, module)

File :488, in _call_with_frames_removed(f, *args, **kwds)

File ~/workspace/.venv/lib/python3.12/site-packages/transformers/models/auto/processing_auto.py:27
26 from ...feature_extraction_utils import FeatureExtractionMixin
---> [27](https://vscode-remote+wsl-002bdebian.vscode-resource.vscode-cdn.ne/workspace/test_local/~/workspace/.venv/lib/python3.12/site-packages/transformers/models/auto/processing_auto.py:27) from ...image_processing_utils import ImageProcessingMixin
28 from ...processing_utils import ProcessorMixin

File ~/workspace/.venv/lib/python3.12/site-packages/transformers/image_processing_utils.py:21
19 import numpy as np
---> [21](https://vscode-remote+wsl-002bdebian.vscode-resource.vscode-cdn.net/workspace/test_local/~/workspace/.venv/lib/python3.12/site-packages/transformers/image_processing_utils.py:21) from .image_processing_base import BatchFeature, ImageProcessingMixin
22 from .image_transforms import center_crop, normalize, rescale

File ~/workspace/.venv/lib/python3.12/site-packages/transformers/image_processing_base.py:26
25 from .feature_extraction_utils import BatchFeature as BaseBatchFeature
---> [26](https://vscode-remote+wsl-002bdebian.vscode-resource.vscode-cdn.net/workspac/test_local/~/workspace/.venv/lib/python3.12/site-packages/transformers/image_processing_base.py:26) from .image_utils import is_valid_image, load_image
27 from .utils import (
28 IMAGE_PROCESSOR_NAME,
29 PROCESSOR_NAME,
(...) 35 logging,
36 )

File ~/workspace/.venv/lib/python3.12/site-packages/transformers/image_utils.py:[55](https://vscode-remote+wsl-002bdebian.vscode-resource.vscode-cdn.net/workspace/test_local/~/workspac/.venv/lib/python3.12/site-packages/transformers/image_utils.py:55)
54 if is_torchvision_available():
---> 55 from torchvision.transforms import InterpolationMode
57 pil_torch_interpolation_mapping = {
58 PILImageResampling.NEAREST: InterpolationMode.NEAREST_EXACT,
59 PILImageResampling.BOX: InterpolationMode.BOX,
(...) 63 PILImageResampling.LANCZOS: InterpolationMode.LANCZOS,
64 }

File ~/workspace/.venv/lib/python3.12/site-packages/torchvision/__init__.py:[8](https://vscode-remote+wsl-002bdebian.vscode-resource.vscode-cdn.net/workspace/test_local/~/workspace/.venv/lib/python3.12/site-packages/torchvision/__init__.py:8)
7 from . import extension # usort:skip # noqa: F401
----> 8 from torchvision import _meta_registrations, datasets, io, models, ops, transforms, utils # usort:skip
10 try:

File ~/workspace/.venv/lib/python3.[1](https://vscode-remote+wsl-002bdebian.vscode-resource.vscode-cdn.net/workspace/test_local/~/workspace/.venv/lib/python3.12/site-packages/torchvision/datasets/__init__.py:1)2/site-packages/torchvision/datasets/__init__.py:1
----> 1 from ._optical_flow import FlyingChairs, FlyingThings3D, HD1K, KittiFlow, Sintel
2 from ._stereo_matching import (
3 CarlaStereo,
4 CREStereo,
(...) 12 SintelStereo,
13 )

File ~/workspace/.venv/lib/python3.12/site-packages/torchvision/datasets/_optical_flow.py:[14](https://vscode-remote+wsl-002bdebian.vscode-resource.vscode-cdn.net/workspace/test_local/~/workspace/.venv/lib/python3.12/site-packages/torchvision/datasets/_optical_flow.py:14)
13 from .folder import default_loader
---> 14 from .utils import _read_pfm, verify_str_arg
15 from .vision import VisionDataset

File ~/workspace/.venv/lib/python3.12/site-packages/torchvision/datasets/utils.py:[4](https://vscode-remote+wsl-002bdebian.vscode-resource.vscode-cdn.net/workspace/test_local/~/workspace/.venv/lib/python3.12/site-packages/torchvision/datasets/utils.py:4)
3 import hashlib
----> 4 import lzma
5 import os

File /usr/local/lib/python3.12/lzma.py:[27](https://vscode-remote+wsl-002bdebian.vscode-resource.vscode-cdn.net/usr/local/lib/python3.12/lzma.py:27)
26 import os
---> 27 from _lzma import *
28 from _lzma import _encode_filter_properties, _decode_filter_properties

ModuleNotFoundError: No module named '_lzma'

The above exception was the direct cause of the following exception:

ModuleNotFoundError Traceback (most recent call last)
Cell In[1], [line 4](vscode-notebook-cell:?execution_count=1&line=4)
1 from docling_core.types.doc.page import TextCellUnit
2 from docling_parse.pdf_parser import DoclingPdfParser, PdfDocument
----> [4](vscode-notebook-cell:?execution_count=1&line=4) from docling.document_converter import DocumentConverter
5 from docling.datamodel.base_models import InputFormat

File ~/workspace/.venv/lib/python3.12/site-packages/docling/document_converter.py:67
60 from docling.datamodel.settings import (
61 DEFAULT_PAGE_RANGE,
62 DocumentLimits,
63 PageRange,
64 settings,
65 )
66 from docling.exceptions import ConversionError
---> [67](https://vscode-remote+wsl-002bdebian.vscode-resource.vscode-cdn.net/workspace/test_local/~/workspace/.venv/lib/python3.12/site-packages/docling/document_converter.py:67) from docling.pipeline.asr_pipeline import AsrPipeline
68 from docling.pipeline.base_pipeline import BasePipeline
69 from docling.pipeline.simple_pipeline import SimplePipeline

File ~/workspace/.venv/lib/python3.12/site-packages/docling/pipeline/asr_pipeline.py:33
26 from docling.datamodel.pipeline_options import (
27 AsrPipelineOptions,
28 )
29 from docling.datamodel.pipeline_options_asr_model import (
30 InlineAsrMlxWhisperOptions,
31 InlineAsrNativeWhisperOptions,
32 )
---> [33](https://vscode-remote+wsl-002bdebian.vscode-resource.vscode-cdn.net/workspace/test_local/~/workspace/.venv/lib/python3.12/site-packages/docling/pipeline/asr_pipeline.py:33) from docling.pipeline.base_pipeline import BasePipeline
34 from docling.utils.accelerator_utils import decide_device
35 from docling.utils.profiling import ProfilingScope, TimeRecorder

File ~/workspace/.venv/lib/python3.12/site-packages/docling/pipeline/base_pipeline.py:32
30 from docling.models.factories import get_picture_description_factory
31 from docling.models.picture_description_base_model import PictureDescriptionBaseModel
---> [32](https://vscode-remote+wsl-002bdebian.vscode-resource.vscode-cdn.net/workspace/test_local/~/workspace/.venv/lib/python3.12/site-packages/docling/pipeline/base_pipeline.py:32) from docling.models.stages.chart_extraction.granite_vision import (
33 ChartExtractionModelGraniteVision,
34 ChartExtractionModelOptions,
35 )
36 from docling.models.stages.picture_classifier.document_picture_classifier import (
37 DocumentPictureClassifier,
38 )
39 from docling.utils.profiling import ProfilingScope, TimeRecorder

File ~/workspace/.venv/lib/python3.12/site-packages/docling/models/stages/chart_extraction/granite_vision.py:[21](https://vscode-remote+wsl-002bdebian.vscode-resource.vscode-cdn.net/workspace/test_local/~/workspace.venv/lib/python3.12/site-packages/docling/models/stages/chart_extraction/granite_vision.py:21)
19 from PIL import Image
20 from pydantic import BaseModel
---> 21 from transformers import AutoModelForImageTextToText, AutoProcessor
23 from docling.datamodel.accelerator_options import AcceleratorDevice, AcceleratorOptions
24 from docling.datamodel.base_models import ItemAndImageEnrichmentElement

File ~/workspace/.venv/lib/python3.12/site-packages/transformers/utils/import_utils.py:2320, in _LazyModule.__getattr__(self, name)
2318 value = getattr(module, name)
2319 except (ModuleNotFoundError, RuntimeError) as e:
-> [2320](https://vscode-remote+wsl-002bdebian.vscode-resource.vscode-cdn.net/workspace/test_local/~/workspace/.venv/lib/python3.12/site-packages/transformers/utils/import_utils.py:2320) raise ModuleNotFoundError(
2321 f"Could not import module '{name}'. Are this object's requirements defined correctly?"
2322 ) from e
2324 elif name in self._modules:
2325 try:

ModuleNotFoundError: Could not import module 'AutoProcessor'. Are this object's requirements defined correctly?

Contributor guide

Open the contributing guide

Research direction

Reproduce the import from docling.document_converter and follow the traceback through docling/pipeline/base_pipeline.py and docling/models/stages/chart_extraction/granite_vision.py, where AutoProcessor is imported. Check the installed Transformers, Torch, and torchvision versions and the project’s dependency declarations. Done means importing DocumentConverter succeeds on the supported Python environment without the AutoProcessor error.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
ai
Issue type
Bug
Difficulty
3/5
Estimated time
1-2 days
Activity status
Quiet
Clarity
Mostly clear
Newbie friendliness
45/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.