python environment setup

[qyjyork]

hello,
i got one problem while installing RoboRefer regarding to the version of torch (which could be also related with version of python and cuda). Mine cuda version is 12.1, btw.

[qyjyork]

so, i'd like to know the detailed recommendation of environment setup (pytorch and cuda)

[Zhoues]

@qyjyork

Thank you very much for your interest in RoboRefer! I have specified the required versions of pytorch and cuda in the pyproject.toml file as follows:

dependencies = [
    "torch==2.5.1",
    "torchaudio==2.5.1",
    "torchvision==0.20.1",
    "nvidia-cublas-cu12==12.4.5.8",
    "nvidia-cuda-cupti-cu12==12.4.127",
    "nvidia-cuda-nvrtc-cu12==12.4.127",
    "nvidia-cuda-runtime-cu12==12.4.127",
    "nvidia-cudnn-cu12==9.1.0.70",
    "nvidia-cufft-cu12==11.2.1.3",
    "nvidia-curand-cu12==10.3.5.147",
    "nvidia-cusolver-cu12==11.6.1.9",
    "nvidia-cusparse-cu12==12.3.1.170",
    "nvidia-cusparselt-cu12==0.6.2",
    "nvidia-ml-py==12.570.86",
    "nvidia-nccl-cu12==2.21.5",
    "nvidia-nvjitlink-cu12==12.4.127",
    "nvidia-nvtx-cu12==12.4.127",

In my local setup, my CUDA version is 12.4, and the PyTorch version is 2.5.1, consistent with the versions listed above.

[LukeLIN-web]

@qyjyork

Thank you very much for your interest in RoboRefer! I have specified the required versions of pytorch and cuda in the pyproject.toml file as follows:

dependencies = [
"torch==2.5.1",
"torchaudio==2.5.1",
"torchvision==0.20.1",
"nvidia-cublas-cu12==12.4.5.8",
"nvidia-cuda-cupti-cu12==12.4.127",
"nvidia-cuda-nvrtc-cu12==12.4.127",
"nvidia-cuda-runtime-cu12==12.4.127",
"nvidia-cudnn-cu12==9.1.0.70",
"nvidia-cufft-cu12==11.2.1.3",
"nvidia-curand-cu12==10.3.5.147",
"nvidia-cusolver-cu12==11.6.1.9",
"nvidia-cusparse-cu12==12.3.1.170",
"nvidia-cusparselt-cu12==0.6.2",
"nvidia-ml-py==12.570.86",
"nvidia-nccl-cu12==2.21.5",
"nvidia-nvjitlink-cu12==12.4.127",
"nvidia-nvtx-cu12==12.4.127",
In my local setup, my CUDA version is 12.4, and the PyTorch version is 2.5.1, consistent with the versions listed above.

quick start has a typo
bash env_step.sh roborefer should be bash env_setup.sh roborefer

and I think remove conda install -c nvidia cuda-toolkit -y in env_setup could be faster install? because installing torch will install nvcc .

[Zhoues]

@LukeLIN-web

Thank you for the reminder — I've already corrected the typo in the quick start.