package conflicts about torch-cam HOT 5 CLOSED

Hi @MercuialC,

I might have to increase the pytorch version constraints on the package. Could you be more specific please? I cannot do much without more information 😅

How did you install torchcam? What are the conflicts? etc. (cf. the bug report template)

from torch-cam.

thank you for your reply
I installed torchcam via conda
When I run the example, torchvision can't find it, and ImportError: DLL load failed:
I tried to install the 1.6.0 version of pytorch through the command: conda install pytorch==1.6.0 torchvision==0.7.0 -c pytorch
Some error messages are as follows：
Found conflicts! Looking for incompatible packages.
This can take several minutes. Press CTRL-C to abort.
Examining sip: 10%|██████▌ | 5/52 [00:04<00:43, 1.07it/s]-Examining freetype: 17%|██████████▉ | 9/52 [00:08<00:34, 1.23it/s]\Examining cffi: 27%|█████████████████▊ | 14/52 [00:18<00:52, 1.38s/it]\Examining matplotlib-base: 56%|██████████████████████████████▋ | 29/52 [00:37<00:24, 1.05s/it]-Examining pycparser: 81%|█████████████████████████████████████████████████▎ | 42/52 [00:52<00:09, 1.10it/s]-Examining conflict for kiwisolver mkl-service numpy-base pytorch sip wheel pip cffi openssl torchvision tornado pyqt nu-Examining conflict for kiwisolver mkl-service numpy-base pytorch sip wheel pip cffi torchvision tornado pyqt numpy olef\Examining conflict for kiwisolver jpeg mkl-service numpy-base pytorch sip vs2015_runtime tk wheel freetype pip cffi ope|Examining conflict for torchvision torchcam matplotlib-base mkl-service numpy-base mkl_fft pytorch numpy matplotlib mkl/Examining conflict for matplotlib-base cycler tornado mkl-service numpy-base mkl_fft matplotlib pytorch numpy six pytho|Examining conflict for torchvision torchcam matplotlib-base mkl-service numpy-base mkl_fft blas pytorch numpy matplotli/Examining conflict for torchcam matplotlib-base matplotlib setuptools numpy python wheel pip: 60%|▌| 31/52 [00:35<00:0|Examining conflict for matplotlib-base matplotlib setuptools wincertstore wheel pip: 60%|▌| 31/52 [00:36<00:08, 2.57i\Examining conflict for pillow torchcam matplotlib-base qt pyqt libpng freetype matplotlib: 67%|▋| 35/52 [00:39<00:12, \Examining conflict for python pip: 73%|██████████████████████████████████▎ | 38/52 [00:41<00:11, 1.18it/s]|Examining conflict for pillow torchvision torchcam matplotlib-base matplotlib: 79%|██▎| 41/52 [00:43<00:07, 1.48it/s]|Examining conflict for torchcam matplotlib pyqt: 88%|█████████████████████████████▏ | 46/52 [00:47<00:05, 1.16it/s]\Examining conflict for pillow qt zstd xz libtiff: 94%|██████████████████████████████▏ | 49/52 [00:48<00:02, 1.44it/s]\Examining conflict for torchcam matplotlib-base matplotlib: : 53it [00:49, 2.09it/s] -failed

UnsatisfiableError: The following specifications were found
to be incompatible with the existing python installation in your environment:

Specifications:

• cycler -> python[version='2.6.|3.3.']
• mkl-service -> python[version='2.6.|3.3.']
• python-dateutil -> python[version='>=3.8,<3.9.0a0|>=3.9,<3.10.0a0']
• wheel -> python[version='2.6.*|>=3.9,<3.10.0a0']

Your python: python==3.6.5

from torch-cam.

When I run the example, torchvision can't find it, and ImportError: DLL load failed:

What do you mean by this?

UnsatisfiableError: The following specifications were found
to be incompatible with the existing python installation in your environment:

Specifications:

cycler -> python[version='2.6.|3.3.']
mkl-service -> python[version='2.6.|3.3.']
python-dateutil -> python[version='>=3.8,<3.9.0a0|>=3.9,<3.10.0a0']
wheel -> python[version='2.6.*|>=3.9,<3.10.0a0']

Regarding the conflict issue, it seems like you have conflict issues with pytorch & torchvision, but not torchcam.
Because as you can see here, the requirements for torchcam are pretty loose. I'm happy to help but if the errors appear when you try to install them, I think the pytorch team will be more qualified to help :)

from torch-cam.

Any update @MercuialC ?

from torch-cam.

Hi @MercuialC 👋

Apart from if you have new details about this issue, I doubt this is due to this repo considering the very loose requirements. I'm closing this issue for now, feel free to reopen if there is something new coming up!

from torch-cam.