Open jamesheatonrdm opened 6 months ago
I have the exact same problem here. Exact same error message with no versions of python working. Tried upgrading pip and setuptools with no success. Conda environment solving takes forever too. Any help from any of the contributors would be appreciated, as it seems like this is a recent issue (I don't see any other similar issues in the Issues tab). Thank you!
@chokevin8
I have just come back to the conda install and it has failed saying that my installed CUDA version (12.1) is incompatible
The issue seems to be with ptxcompiler-cu11. There doesn't appear to be much documentation for this python package, and the error when installing is not useful at all so I am unsure where to go from here.
@chokevin8 I manage to run the segmentation notebook without installing the cuml-cu11 package. Here is how I did it:
Create conda environment with python 3.10.13
conda install pytorch==2.0.0 torchvision==0.15.0 torchaudio==2.0.0 pytorch-cuda=11.7 -c pytorch -c nvidia
conda install xformers::xformers
conda install conda-forge::omegaconf
conda install fvcore
conda install iopath
conda install pip
pip install git+https://github.com/facebookincubator/submitit
pip install ftfy
pip install regex
pip install -U openmim
mim install mmcv==1.5.0
mim install mmsegmentation==0.27.0
Don't ask my why this works because I have no idea. But these steps can get you to run segmentation inference, which is what I was wanting from this repo
@jamesheatonrdm can confirm that this also works for training. Thank you so much! I also have no idea why this works.
@chokevin8 Did you ever run into a "Failed building wheel for mmcv-full":
Building wheels for collected packages: mmcv-full
error: subprocess-exited-with-error
× python setup.py bdist_wheel did not run successfully.
│ exit code: 1
╰─> See above for output.
note: This error originates from a subprocess, and is likely not a problem with pip.
Building wheel for mmcv-full (setup.py) ... error
ERROR: Failed building wheel for mmcv-full
Running setup.py clean for mmcv-full
Failed to build mmcv-full
ERROR: Could not build wheels for mmcv-full, which is required to install pyproject.toml-based projects
Seems to be a recurring issue with my code, and i dont seem to be able to find much of a solution online
@Al3cL1 Hmm, no, do you get this when running what @jamesheatonrdm suggested (mentioned below)? Maybe deactivate and remove your environment and create a new environment with python 3.10.13 if you already haven't. I remember I did have to separately install some other required packages for training like fvcore afterwards, but I didn't get an error like yours.
@chokevin8 I manage to run the segmentation notebook without installing the cuml-cu11 package. Here is how I did it:
Create conda environment with python 3.10.13
conda install pytorch==2.0.0 torchvision==0.15.0 torchaudio==2.0.0 pytorch-cuda=11.7 -c pytorch -c nvidia conda install xformers::xformers conda install conda-forge::omegaconf conda install fvcore conda install iopath conda install pip pip install git+https://github.com/facebookincubator/submitit pip install ftfy pip install regex pip install -U openmim mim install mmcv==1.5.0 mim install mmsegmentation==0.27.0
Don't ask my why this works because I have no idea. But these steps can get you to run segmentation inference, which is what I was wanting from this repo
@Al3cL1 Hmm, no, do you get this when running what @jamesheatonrdm suggested (mentioned below)? Maybe deactivate and remove your environment and create a new environment with python 3.10.13 if you already haven't. I remember I did have to separately install some other required packages for training like fvcore afterwards, but I didn't get an error like yours.
@chokevin8 I manage to run the segmentation notebook without installing the cuml-cu11 package. Here is how I did it: Create conda environment with python 3.10.13
conda install pytorch==2.0.0 torchvision==0.15.0 torchaudio==2.0.0 pytorch-cuda=11.7 -c pytorch -c nvidia conda install xformers::xformers conda install conda-forge::omegaconf conda install fvcore conda install iopath conda install pip pip install git+https://github.com/facebookincubator/submitit pip install ftfy pip install regex pip install -U openmim mim install mmcv==1.5.0 mim install mmsegmentation==0.27.0
Don't ask my why this works because I have no idea. But these steps can get you to run segmentation inference, which is what I was wanting from this repo
Yeah no ok I think its an issue with my cuda/cudatoolkit version, which might be affecting dependencies. I think I got it thought.
conda install pytorch==2.0.0 torchvision==0.15.0 torchaudio==2.0.0 pytorch-cuda=11.7 -c pytorch -c nvidia conda install xformers::xformers conda install conda-forge::omegaconf conda install fvcore conda install iopath conda install pip pip install git+https://github.com/facebookincubator/submitit pip install ftfy pip install regex pip install -U openmim mim install mmcv==1.5.0 mim install mmsegmentation==0.27.0
A conda.yml file would be as follows
name: dinov2
channels:
- defaults
- pytorch
- nvidia
- xformers
- conda-forge
dependencies:
- python=3.10
- pytorch::pytorch=2.0.0
- pytorch::pytorch-cuda=11.7.0
- pytorch::torchvision=0.15.0
- omegaconf
- torchmetrics=0.10.3
- fvcore
- iopath
- xformers::xformers
- tensorboard
- pip
- pip:
- git+https://github.com/facebookincubator/submitit
- --extra-index-url https://pypi.nvidia.com
- cuml-cu11
- openmim
you can install it with conda env update --file conda.yaml
It is not clear what Python version is suitable for this project. I am unable to install the required dependencies when using versions 3.6, 3.8, 3.9, 3.10 and 3.11
python3.11:
python3.10:
python3.9:
python3.8:
ERROR: Command errored out with exit status 1: python setup.py egg_info Check the logs for full command output.
Collecting torch==2.0.0 (from -r requirements.txt (line 2)) Could not find a version that satisfies the requirement torch==2.0.0 (from -r requirements.txt (line 2)) (from versions: 1.0.0, 1.0.1, 1.0.1.post2, 1.1.0, 1.2.0, 1.3.0, 1.3.1, 1.4.0, 1.5.0, 1.5.1, 1.6.0, 1.7.0, 1.7.1, 1.8.0, 1.8.1, 1.9.0, 1.9.1, 1.10.0, 1.10.1, 1.10.2) No matching distribution found for torch==2.0.0 (from -r requirements.txt (line 2))