Install commands not working

[Rhys-McAlister]

Attempting to use the install commands in the readme and finding the below error:

Platform: WSL

`conda clean --all
Retrieving notices: ...working... done
Channels:

• conda-forge
• defaults
  Platform: linux-64
  Collecting package metadata (repodata.json): done
  Solving environment: failed

PackagesNotFoundError: The following packages are not available from current channels:

• appdirs==1.4.4=pypi_0
• argparse==1.4.0=pypi_0
• asttokens==2.4.1=pypi_0
• attrs==23.2.0=pypi_0
• automat==22.10.0=pypi_0
• axial-positional-embedding==0.2.1=pypi_0
• buildtools==1.0.6=pypi_0
• causal-conv1d==1.0.2=pypi_0
• click==8.1.7=pypi_0
• cmake==3.28.1=pypi_0
• comm==0.2.1=pypi_0
• constantly==23.10.4=pypi_0
• cuda-cudart==11.7.99=0
• cuda-cupti==11.7.101=0
• cuda-libraries==11.7.1=0
• cuda-nvrtc==11.7.99=0
• cuda-nvtx==11.7.91=0
• cuda-runtime==11.7.1=0
• debugpy==1.8.0=pypi_0
• decorator==5.1.1=pypi_0
• docker-pycreds==0.4.0=pypi_0
• docopt==0.6.2=pypi_0
• einops==0.7.0=pypi_0
• exceptiongroup==1.2.0=pypi_0
• executing==2.0.1=pypi_0
• ffmpeg==4.3=hf484d3e_0
• fsspec==2023.12.2=pypi_0
• furl==2.1.3=pypi_0
• gitdb==4.0.11=pypi_0
• gitpython==3.1.40=pypi_0
• greenlet==3.0.3=pypi_0
• huggingface-hub==0.20.1=pypi_0
• hyperlink==21.0.0=pypi_0
• importlib-metadata==7.0.1=pypi_0
• incremental==22.10.0=pypi_0
• ipykernel==6.28.0=pypi_0
• ipython==8.18.1=pypi_0
• jedi==0.19.1=pypi_0
• joblib==1.3.2=pypi_0
• jupyter-client==8.6.0=pypi_0
• jupyter-core==5.7.0=pypi_0
• libcublas==11.10.3.66=0
• libcufft==10.7.2.124=h4fbf590_0
• libcufile==1.8.1.2=0
• libcurand==10.3.4.107=0
• libcusolver==11.4.0.1=0
• libcusparse==11.7.4.91=0
• libnpp==11.7.4.75=0
• libnvjpeg==11.8.0.2=0
• lightning-utilities==0.10.0=pypi_0
• lit==17.0.6=pypi_0
• littleutils==0.2.2=pypi_0
• local-attention==1.9.0=pypi_0
• mamba-ssm==1.0.1=pypi_0
• matplotlib-inline==0.1.6=pypi_0
• nest-asyncio==1.5.8=pypi_0
• ninja==1.11.1.1=pypi_0
• ogb==1.3.6=pypi_0
• orderedmultidict==1.0.1=pypi_0
• outdated==0.2.2=pypi_0
• packaging==23.2=pypi_0
• pandas==2.1.4=pypi_0
• parso==0.8.3=pypi_0
• performer-pytorch==1.1.4=pypi_0
• pexpect==4.9.0=pypi_0
• platformdirs==4.1.0=pypi_0
• prompt-toolkit==3.0.43=pypi_0
• protobuf==4.25.1=pypi_0
• psutil==5.9.7=pypi_0
• ptyprocess==0.7.0=pypi_0
• pure-eval==0.2.2=pypi_0
• pyg-lib==0.3.1+pt20cu117=pypi_0
• pygments==2.17.2=pypi_0
• pyparsing==3.1.1=pypi_0
• python-dateutil==2.8.2=pypi_0
• pytorch==2.0.0=py3.9_cuda11.7_cudnn8.5.0_0
• pytorch-cuda==11.7=h778d358_5
• pytorch-mutex==1.0=cuda
• pytz==2023.3.post1=pypi_0
• pyyaml==6.0.1=pypi_0
• pyzmq==25.1.2=pypi_0
• rdkit==2023.9.3=pypi_0
• redo==2.0.4=pypi_0
• regex==2023.12.25=pypi_0
• safetensors==0.4.1=pypi_0
• scikit-learn==1.3.2=pypi_0
• scipy==1.11.4=pypi_0
• sentry-sdk==1.39.1=pypi_0
• setproctitle==1.3.3=pypi_0
• simplejson==3.19.2=pypi_0
• six==1.16.0=pypi_0
• smmap==5.0.1=pypi_0
• sqlalchemy==2.0.25=pypi_0
• stack-data==0.6.3=pypi_0
• tensorboardx==2.6.2.2=pypi_0
• threadpoolctl==3.2.0=pypi_0
• tokenizers==0.15.0=pypi_0
• torch-cluster==1.6.3+pt20cu117=pypi_0
• torch-geometric==2.0.4=pypi_0
• torch-scatter==2.1.2+pt20cu117=pypi_0
• torch-sparse==0.6.18+pt20cu117=pypi_0
• torch-spline-conv==1.2.2+pt20cu117=pypi_0
• torchaudio==2.0.0=py39_cu117
• torchmetrics==0.9.3=pypi_0
• torchtriton==2.0.0=py39
• torchvision==0.15.0=py39_cu117
• tornado==6.4=pypi_0
• tqdm==4.66.1=pypi_0
• traitlets==5.14.1=pypi_0
• transformers==4.36.2=pypi_0
• twisted==23.10.0=pypi_0
• tzdata==2023.4=pypi_0
• wandb==0.16.1=pypi_0
• wcwidth==0.2.12=pypi_0
• yacs==0.1.8=pypi_0
• zipp==3.17.0=pypi_0
• zope-interface==6.1=pypi_0

Current channels:

• https://conda.anaconda.org/conda-forge
• defaults

To search for alternate channels that may provide the conda package you're
looking for, navigate to

https://anaconda.org

and use the search bar at the top of the page.

EnvironmentNameNotFound: Could not find conda environment: graph-mamba
You can list all discoverable environments with conda info --envs.`

[ChloeXWang]

Hi Rhys, thank you so much for the using our tool and sharing the feedback! I will try to reproduce the error and loop back shortly.

[Rhys-McAlister]

Hi Rhys, thank you so much for the using our tool and sharing the feedback! I will try to reproduce the error and loop back shortly.

Thanks!

[ChloeXWang]

Hi Rhys,

Thank you for your patience! Would you mind trying these commands instead to see if it works?

conda create -n graph-mamba python==3.9
conda activate graph-mamba
pip install torch==1.13.0
pip install torch-scatter torch-sparse -f https://data.pyg.org/whl/torch-1.13.0+cu117.html
pip install torch_geometric==2.0.4

pip install transformers==4.36.2
pip install torchmetrics==0.10.3
pip install openbabel-wheel
pip install fsspec 
pip install rdkit
pip install pytorch-lightning yacs
pip install performer-pytorch
pip install tensorboardX
pip install ogb
pip install wandb

pip install causal-conv1d==1.0.2
pip install mamba-ssm==1.0.1

The mamba installation (last two lines) might be a bit tricky. Please feel free to refer to the common issues such as cuda dependency in their repo https://github.com/state-spaces/mamba or we could look into it together.

[Rhys-McAlister]

Thank you for the response, you were correct those two last lines were tricky.

`Collecting causal-conv1d==1.0.2
Using cached causal_conv1d-1.0.2.tar.gz (6.4 kB)
Preparing metadata (setup.py) ... error
error: subprocess-exited-with-error

× python setup.py egg_info did not run successfully.
│ exit code: 1
╰─> [13 lines of output]
/tmp/pip-install-_txvuxlg/causal-conv1d_6c9aaf7a6ea4488f91528d397f701b26/setup.py:78: UserWarning: causal_conv1d was requested, but nvcc was not found. Are you sure your environment has nvcc available? If you're installing within a container from https://hub.docker.com/r/pytorch/pytorch, only images whose names contain 'devel' will provide nvcc.
warnings.warn(
Traceback (most recent call last):
File "", line 2, in
File "", line 34, in
File "/tmp/pip-install-_txvuxlg/causal-conv1d_6c9aaf7a6ea4488f91528d397f701b26/setup.py", line 112, in
if bare_metal_version >= Version("11.8"):
NameError: name 'bare_metal_version' is not defined

  torch.__version__  = 1.13.0+cu117


  [end of output]

note: This error originates from a subprocess, and is likely not a problem with pip.
error: metadata-generation-failed

× Encountered error while generating package metadata.
╰─> See above for output.

note: This is an issue with the package mentioned above, not pip.
hint: See above for details.
Collecting mamba-ssm==1.0.1
Downloading mamba_ssm-1.0.1.tar.gz (28 kB)
Preparing metadata (setup.py) ... error
error: subprocess-exited-with-error

× python setup.py egg_info did not run successfully.
│ exit code: 1
╰─> [13 lines of output]
/tmp/pip-install-dy2v6xzo/mamba-ssm_45311b26096b450c9b61a4164acc8b62/setup.py:77: UserWarning: mamba_ssm was requested, but nvcc was not found. Are you sure your environment has nvcc available? If you're installing within a container from https://hub.docker.com/r/pytorch/pytorch, only images whose names contain 'devel' will provide nvcc.
warnings.warn(
Traceback (most recent call last):
File "", line 2, in
File "", line 34, in
File "/tmp/pip-install-dy2v6xzo/mamba-ssm_45311b26096b450c9b61a4164acc8b62/setup.py", line 111, in
if bare_metal_version >= Version("11.8"):
NameError: name 'bare_metal_version' is not defined

  torch.__version__  = 1.13.0+cu117


  [end of output]

note: This error originates from a subprocess, and is likely not a problem with pip.
error: metadata-generation-failed

× Encountered error while generating package metadata.
╰─> See above for output.

note: This is an issue with the package mentioned above, not pip.
hint: See above for details.`

Any help with this would be much appreciated. I have had a look through the steps in this repo originally but haven't had much success.

Thanks in advance

[ChloeXWang]

Gotcha, this issue seems relevant. Could you type in nvidia-smi or nvcc --version in command line to check your cuda version?

[wendongj]

Thank you for the response, you were correct those two last lines were tricky.

`Collecting causal-conv1d==1.0.2 Using cached causal_conv1d-1.0.2.tar.gz (6.4 kB) Preparing metadata (setup.py) ... error error: subprocess-exited-with-error

× python setup.py egg_info did not run successfully. │ exit code: 1 ╰─> [13 lines of output] /tmp/pip-install-_txvuxlg/causal-conv1d_6c9aaf7a6ea4488f91528d397f701b26/setup.py:78: UserWarning: causal_conv1d was requested, but nvcc was not found. Are you sure your environment has nvcc available? If you're installing within a container from https://hub.docker.com/r/pytorch/pytorch, only images whose names contain 'devel' will provide nvcc. warnings.warn( Traceback (most recent call last): File "", line 2, in File "", line 34, in File "/tmp/pip-install-_txvuxlg/causal-conv1d_6c9aaf7a6ea4488f91528d397f701b26/setup.py", line 112, in if bare_metal_version >= Version("11.8"): NameError: name 'bare_metal_version' is not defined

  torch.__version__  = 1.13.0+cu117


  [end of output]

note: This error originates from a subprocess, and is likely not a problem with pip. error: metadata-generation-failed

× Encountered error while generating package metadata. ╰─> See above for output.

note: This is an issue with the package mentioned above, not pip. hint: See above for details. Collecting mamba-ssm==1.0.1 Downloading mamba_ssm-1.0.1.tar.gz (28 kB) Preparing metadata (setup.py) ... error error: subprocess-exited-with-error

× python setup.py egg_info did not run successfully. │ exit code: 1 ╰─> [13 lines of output] /tmp/pip-install-dy2v6xzo/mamba-ssm_45311b26096b450c9b61a4164acc8b62/setup.py:77: UserWarning: mamba_ssm was requested, but nvcc was not found. Are you sure your environment has nvcc available? If you're installing within a container from https://hub.docker.com/r/pytorch/pytorch, only images whose names contain 'devel' will provide nvcc. warnings.warn( Traceback (most recent call last): File "", line 2, in File "", line 34, in File "/tmp/pip-install-dy2v6xzo/mamba-ssm_45311b26096b450c9b61a4164acc8b62/setup.py", line 111, in if bare_metal_version >= Version("11.8"): NameError: name 'bare_metal_version' is not defined

  torch.__version__  = 1.13.0+cu117


  [end of output]

note: This error originates from a subprocess, and is likely not a problem with pip. error: metadata-generation-failed

× Encountered error while generating package metadata. ╰─> See above for output.

note: This is an issue with the package mentioned above, not pip. hint: See above for details.`

Any help with this would be much appreciated. I have had a look through the steps in this repo originally but haven't had much success.

Thanks in advance

I have the same issue, and my system is ubuntu, when I type the command "nvcc -V", it shows no nvcc, so I install the cudatoolkit in my computer, after I install the cudatoolkit, then the "nvcc -V" is ok, then I set "CUDA_HOME=/usr/local/cuda-X.X", now I can install mamba-ssm. I did not know if this is useful for your system

[williamnyren]

The most straightforward method I found for installing nvcc was to utilize the conda package within the environment. You can install it using the following command:
conda install nvidia/label/cuda-11.7.1::cuda-nvcc

There are also other versions available:
https://anaconda.org/nvidia/cuda-nvcc

[ChloeXWang]

Thank you for your comments!

[Andy-Yuaner]

"I noticed that the official GitHub versions currently provided are all based on CUDA 11.8, while the requirements file for this project specifies CUDA 11.7. Should I upgrade to CUDA 11.8, or are there alternative solutions? Thank you!