Docker image build error when installing pytorch

[msjun23]

Docker container image that used this project install latest Miniconda3 for using python, and python version of latest Miniconda3 is python3.10(for now, 11th Jan, 2023).

And pytorch version, to be installed when building image, is 1.9.0 that is matched with CUDA version 11.1.1. Torchvision is also dependent with cu111. Now problem is that, pytorch doesn't support torch1.9.0+cu111 for python version 3.10, same as torchvision.

So Docker image file have to be fixed to install specific version of python, Miniconda3.

18, 20, 21 lines have to be fixed.

RUN wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh
RUN mkdir /root/.conda
RUN bash Miniconda3-latest-Linux-x86_64.sh -b
RUN rm -f Miniconda3-latest-Linux-x86_64.sh
RUN conda init bash

to

RUN wget https://repo.anaconda.com/miniconda/Miniconda3-py39_22.11.1-1-Linux-x86_64.sh
RUN mkdir /root/.conda
RUN bash Miniconda3-py39_22.11.1-1-Linux-x86_64.sh -b
RUN rm -f Miniconda3-py39_22.11.1-1-Linux-x86_64.sh
RUN conda init bash

In this example, we download Miniconda3 for python 3.9, and python 3.9 support pytorch version 1.9.0+cu111.

If you want to use another specific version of python or Miniconda3, recommend you to visit the Miniconda3 Download pages. At here, just find the specific version, you want to use, and fix the lines above in Docker file.

[Br0ver]

More changes are required by now. Use the following docker-script.

FROM nvidia/cuda:11.1.1-cudnn8-devel-ubuntu18.04

ARG DEBIAN_FRONTEND=noninteractive
ENV PATH="/root/miniconda3/bin:${PATH}"
ARG PATH="/root/miniconda3/bin:${PATH}"

RUN apt-get update && apt-get install -y --no-install-recommends apt-utils
RUN apt-get upgrade -y
RUN apt-get install -y vim git
RUN apt-get install -y libgl1-mesa-glx
RUN apt-get install -y libglib2.0-0 libsm6 libxrender1 libxext6
RUN apt-get install -y tmux
RUN apt-get install -y wget
RUN apt-get install -y htop

ENV PYTHONIOENCODING=UTF-8

RUN wget -O miniconda.sh https://repo.anaconda.com/miniconda/Miniconda3-py39_24.1.2-0-Linux-x86_64.sh
RUN mkdir /root/.conda
RUN bash miniconda.sh -b
RUN rm -f miniconda.sh
RUN conda init bash

RUN python3 -m pip install --upgrade pip
RUN python3 -m pip install --upgrade setuptools==57.0.0
RUN pip3 install torch==1.9.0+cu111 torchvision==0.10.0+cu111 -f https://download.pytorch.org/whl/torch_stable.html
RUN conda install -y -c numba numba==0.53.1
RUN conda install -y -c conda-forge h5py==2.10.0 blosc-hdf5-plugin==1.0.0 scikit-video==1.1.11 tqdm==4.61.1 prettytable==2.1.0
RUN python3 -m pip install yacs==0.1.8 pytz==2021.1 tensorboard==2.5.0 opencv-python==4.5.2.54 einops==0.3.2 matplotlib==3.3.4
RUN pip3 install numpy==1.20.0 protobuf==3.20.0
ARG DEBIAN_FRONTEND=teletype