AttributeError: partially initialized module 'torchvision' has no attribute 'extension' (most likely due to a circular import)

[PogusTheWhisper]

from tsai.basics import *

---

AttributeError Traceback (most recent call last)
Cell In[52], line 1
----> 1 from tsai.basics import *

File /opt/conda/lib/python3.10/site-packages/tsai/basics.py:5
3 from .imports import *
4 from .utils import *
----> 5 from .data.basics import *
6 from .losses import *
7 from .metrics import *

File /opt/conda/lib/python3.10/site-packages/tsai/data/basics.py:6
4 from .core import *
5 from .preprocessing import *
----> 6 from .transforms import *
7 from .mixed_augmentation import *

File /opt/conda/lib/python3.10/site-packages/tsai/data/transforms.py:21
19 from scipy.ndimage import convolve1d
20 from fastcore.transform import compose_tfms
---> 21 from fastai.vision.augment import RandTransform
22 from ..utils import *
23 from .core import *

File /opt/conda/lib/python3.10/site-packages/fastai/vision/augment.py:116
113 def encodes(self, x:(Image.Image,*TensorTypes)): return x.dihedral(self.k)
115 # %% ../../nbs/09_vision.augment.ipynb 27
--> 116 from torchvision.transforms.functional import pad as tvpad
118 # %% ../../nbs/09_vision.augment.ipynb 28
119 mk_class('PadMode', **{o:o.lower() for o in ['Zeros', 'Border', 'Reflection']},
120 doc="All possible padding mode as attributes to get tab-completion and typo-proofing")

File /opt/conda/lib/python3.10/site-packages/torchvision/init.py:10
7 # Don't re-order these, we need to load the _C extension (done when importing
8 # .extensions) before entering _meta_registrations.
9 from .extension import _HAS_OPS # usort:skip
---> 10 from torchvision import _meta_registrations, datasets, io, models, ops, transforms, utils # usort:skip
12 try:
13 from .version import version # noqa: F401

File /opt/conda/lib/python3.10/site-packages/torchvision/_meta_registrations.py:26
20 return fn
22 return wrapper
25 @register_meta("roi_align")
---> 26 def meta_roi_align(input, rois, spatial_scale, pooled_height, pooled_width, sampling_ratio, aligned):
27 torch._check(rois.size(1) == 5, lambda: "rois must have shape as Tensor[K, 5]")
28 torch._check(
29 input.dtype == rois.dtype,
30 lambda: (
(...)
33 ),
34 )

File /opt/conda/lib/python3.10/site-packages/torchvision/_meta_registrations.py:18, in register_meta..wrapper(fn)
17 def wrapper(fn):
---> 18 if torchvision.extension._has_ops():
19 get_meta_lib().impl(getattr(getattr(torch.ops.torchvision, op_name), overload_name), fn)
20 return fn

AttributeError: partially initialized module 'torchvision' has no attribute 'extension' (most likely due to a circular import)

[sallyqiansun]

Hi, did you fix this?

[shizhediao]

Same question

[tusg]

pip install -U torch==2.3.1
pip install torchvision==0.17.2
I was facing the same error tried multiple things...finally these two upgrades worked for me.
i faced this error while running this code -:
https://huggingface.co/Qwen/Qwen2-VL-7B-Instruct-AWQ

[vlamenco]

how to figure it out!~?

[PoyBoi]

pip install -U torch==2.3.1 pip install torchvision==0.17.2 I was facing the same error tried multiple things...finally these two upgrades worked for me. i faced this error while running this code -: https://huggingface.co/Qwen/Qwen2-VL-7B-Instruct-AWQ

Didn't fix it for me sadly :/ (mis-match of torchvision's older version-set installs torch 2.2.2, overwriting 2.3.1, and even reinstalling torch 2.3.1 over it doesn't fix it)

[AMRA94]

Hi @PoyBoi . Would be thankful if you could tell me whether you have had any progress over it or not?