conda env create -f environment.yml <div class="snippet-clipboard-content notransl

problem creating a Conda environment from the provided YAML about pulse HOT 1 OPEN

adamian98 commented on July 4, 2024

problem creating a Conda environment from the provided YAML

from pulse.

Comments (1)

leihuayi commented on July 4, 2024

It does not solve your problem in an easy way, but you could install it in a Docker to run your code, in case the problem comes from your Conda installation ?

Edit pulse.yml : add conda-forge channel, and remove all elements after the second "=" in dependancies (as stated in the README). Change libfortran=3.0.1 to libfortran as this version only works on osx.
Write a Dockerfile in pulse folder:

FROM nvidia/cuda:10.2-runtime
# /!\ the cuda version depends on your computer. Adapt the above line correspondingly.

# Declare some ARGuments
ARG NAME=pulse
# Installation of some libraries / RUN some commands on the base image
RUN apt-get update && \
    DEBIAN_FRONTEND=noninteractive apt-get install -y python3-pip python3-dev \
    libz-dev libopenblas-dev libatlas-base-dev libgtk-3-dev \
    libboost-filesystem-dev build-essential cmake pkg-config curl && \
    apt-get clean && \
    rm -rf /var/lib/apt/lists/*

# INSTALLATION OF CONDA
ENV PATH /opt/conda/bin:$PATH
RUN curl 'https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh' > ~/miniconda.sh && \
    /bin/bash ~/miniconda.sh -b -p /opt/conda && \
    rm ~/miniconda.sh && \
    /opt/conda/bin/conda clean -tipsy && \
    ln -s /opt/conda/etc/profile.d/conda.sh /etc/profile.d/conda.sh && \
    echo ". /opt/conda/etc/profile.d/conda.sh" >> ~/.bashrc && \
    echo "conda activate base" >> ~/.bashrc
ENV PATH /opt/conda/envs/gpuenvs/bin:$PATH

WORKDIR /home/${NAME}

# Install the python dependencies
COPY . .
RUN conda update -n base -c defaults conda
RUN conda env create -f pulse.yml

CMD /bin/bash

Build docker build –t pulse:latest .
Run docker run -it --rm --gpus all --name pulse --entrypoint bash -v "$PWD:/home/pulse" pulse:latest (you might need to change --gpus all by --runtime==nvidia if your Docker version is older)
Once you are inside the container, your "pulse" folder on your computer is in sync with the "pulse" folder on the docker. You can quit the container, add images, come back in again. Activate the environment by running conda activate pulse then run the code like usual python run.py .

from pulse.

Recommend Projects