Please consider making the YAML configuration file public to facilitate the installation of Netdiffusion. Currently, the requirements.txt file is not usable by the pip utility.
In my case, I had to create a YAML configuration file from the requirements file thanks to the script provided here: https://gist.github.com/mfansler/5a1a703d1ac6bb2139547003248c3827
In your case, you can create the correct YAML when your environment is active, with this command:
conda env export | head -n -1 | tail -n +2 > environment.yml
Then I create a new virtual environment with the following:
conda create --name my_env pip python=3.10.9
and activate the environment with
conda activate my_env
Then I installed tensorrt-libs from the pypi nvidia repository, as this package is not available in pypi and conda cannot resolve it:
pip install --extra-index-url https://pypi.nvidia.com tensorrt-libs
This step is mandatory to install all requirements listed in the YAML config file.
Next, I downgraded torch and triton to 2.0.1 and 2.0.0 respectively, because some packages were not compatible with torch 2.2.0 stated in the original requirements.txt
Finally, I updated all dependencies with:
conda env update -n my_env -f environment.yml
Please consider making the YAML configuration file public to facilitate the installation of Netdiffusion. Currently, the requirements.txt file is not usable by the pip utility.
In my case, I had to create a YAML configuration file from the requirements file thanks to the script provided here: https://gist.github.com/mfansler/5a1a703d1ac6bb2139547003248c3827
In your case, you can create the correct YAML when your environment is active, with this command:
conda env export | head -n -1 | tail -n +2 > environment.ymlThen I create a new virtual environment with the following:
conda create --name my_env pip python=3.10.9and activate the environment with
conda activate my_envThen I installed tensorrt-libs from the pypi nvidia repository, as this package is not available in pypi and conda cannot resolve it:
pip install --extra-index-url https://pypi.nvidia.com tensorrt-libsThis step is mandatory to install all requirements listed in the YAML config file.
Next, I downgraded torch and triton to 2.0.1 and 2.0.0 respectively, because some packages were not compatible with torch 2.2.0 stated in the original requirements.txt
Finally, I updated all dependencies with:
conda env update -n my_env -f environment.yml