Open coffeebond opened 1 year ago
I'm having a similar issue, the yml file specifies the following pytorch version:
ResolvePackageNotFound:
- pytorch==1.4.0=py3.6_cuda10.1.243_cudnn7.6.3_0
However, it doesn't seem to be in the available versions (https://anaconda.org/pytorch/pytorch/files)
I tried to install using mamba since I also had issues while using conda. It returns this:
Could not solve for environment specs
The following packages are incompatible
├─ blas ==1.1 openblas is uninstallable because it conflicts with any installable versions previously reported;
├─ joblib ==1.0.1 pyhd8ed1ab_0 is uninstallable because it conflicts with any installable versions previously reported;
├─ libedit ==3.1.20170329 0 is uninstallable because it conflicts with any installable versions previously reported;
├─ libstdcxx-ng ==11.1.0 h56837e0_8 is uninstallable because it conflicts with any installable versions previously reported;
├─ ncurses ==5.9 10 is uninstallable because it conflicts with any installable versions previously reported;
├─ numpy ==1.14.3 py36_blas_openblas_200 is uninstallable because it conflicts with any installable versions previously reported;
├─ openblas ==0.2.20 8 is uninstallable because it conflicts with any installable versions previously reported;
├─ python ==3.6.6 hd21baee_1003 is uninstallable because it conflicts with any installable versions previously reported;
├─ pytorch ==1.4.0 py3.6_cuda10.1.243_cudnn7.6.3_0 does not exist (perhaps a typo or a missing channel);
├─ scikit-learn ==0.20.0 py36_blas_openblasha84fab4_201 is uninstallable because it conflicts with any installable versions previously reported;
└─ threadpoolctl ==2.2.0 pyh8a188c0_0 is uninstallable because it conflicts with any installable versions previously reported.
I also tried to install packages one by one using mamba install pytorch==1.4.0 torchvision==0.5.0
as advised on pytorch website: https://pytorch.org/get-started/previous-versions/
And I obtained this:
The following packages are incompatible
├─ pytorch 1.4.0 is uninstallable because it conflicts with any installable versions previously reported;
└─ torchvision 0.5.0 does not exist (perhaps a typo or a missing channel).
Get ride of the error about torchvision, pytorch 1.4.0 is in conflict with something. I got this from a whole new conda environnement with Python 3.6.6. I don't know where would this issue come from. Maybe some updates made the installation more complicated now...
May I suggest to use Docker container to make UFold accessible and avoid future conflict ? Otherwise, any way to fix this will be welcome
Thank you !
I tried to create the Fold environment from UFold.yml
, I had same error.
Could not solve for environment specs
Encountered problems while solving:
- nothing provides requested pytorch ==1.4.0 py3.6_cuda10.1.243_cudnn7.6.3_0
The environment can't be solved, aborting the operation
Hello,
To ensure the successful deployment of the UFold environment, it's crucial that your base environment matches the specifications preset in the YAML file. As of our latest update, we are utilizing Python 3.11 and CUDA 11.8. Please verify that your packages are compatible with these versions. We have recently updated our UFold.yml file to reflect the most current configurations, allowing you to reconstruct your environment effectively. We encourage you to review the updated file.
Thank you.
Hello,
The requirements.txt file is still pointing to outdated versions.
Hi,
I installed UFold with
conda env create -f UFold.yml
. However when I ran the test withpython ufold_test.py --test_files TS2
, it gave me an errorModuleNotFoundError: No module named 'dataclasses'
.It seems that in the yml file, the dependency for python is actually 3.6.6 (
python=3.6.6=hd21baee_1003
), but the 'dataclasses' is only available after python3.7. Do you know what happened?Do you have suggestions on how to solve this issue?
Thank you!