Problem setting up to run the demo listed in README.txt

[craigphicks]

I am trying to set up a Docker container to run the demo as described prior to TorchScript ResNet18.

Actions attempting to follow the README instructions:

1. Set up a Dockerfile (this was not in the instructions, but is required for reproducibility):

FROM nvidia/cuda:11.8.0-base-ubuntu22.04

ENV PYTHON_VERSION=3.11

ENV PATH=/opt/conda/bin:$PATH
ENV LD_LIBRARY_PATH="${LD_LIBRARY_PATH}:/usr/local/lib"
ENV LD_LIBRARY_PATH="${LD_LIBRARY_PATH}:/opt/conda/lib"

ENV PYTHONIOENCODING=UTF-8
ENV LANG=C.UTF-8
ENV LC_ALL=C.UTF-8
ENV PYTHONDONTWRITEBYTECODE=1
ENV PYTHONUNBUFFERED=1
ENV DEBIAN_FRONTEND=noninteractive
ENV CONDA_AUTO_UPDATE_CONDA=false

RUN apt update
RUN apt install -y bash \
    build-essential \
    git \
    curl \
    ca-certificates \
    wget \
    && rm -rf /var/lib/apt/lists

# Install Miniconda and create main env
ADD https://repo.continuum.io/miniconda/Miniconda3-latest-Linux-x86_64.sh miniconda3.sh
RUN /bin/bash miniconda3.sh -b -p /opt/conda \
    && rm miniconda3.sh \
    && /opt/conda/bin/conda install -y -c anaconda \
    python=$PYTHON_VERSION \
    && /opt/conda/bin/conda clean -ya

RUN /opt/conda/bin/conda config --set ssl_verify False \
    && pip install --upgrade pip --trusted-host pypi.org --trusted-host files.pythonhosted.org \
    && ln -s /opt/conda/bin/pip /usr/local/bin/pip3

# Install requirements
COPY requirements.txt ./
RUN pip install --no-cache-dir -r requirements.txt \
    && rm requirements.txt

CMD ["/bin/bash"]

2. Build the image, then execute the following inside a running container -

root@4415caed30f5:/# python3.11 -m venv mlir_venv
root@4415caed30f5:/# source mlir_venv/bin/activate
(mlir_venv) root@4415caed30f5:/# pip install --pre torch-mlir torchvision \
  --extra-index-url https://download.pytorch.org/whl/nightly/cpu
Looking in indexes: https://pypi.org/simple, https://download.pytorch.org/whl/nightly/cpu
ERROR: Could not find a version that satisfies the requirement torch-mlir (from versions: none)
ERROR: No matching distribution found for torch-mlir

3. Also tried the suggested "alternate" method

conda create -n torch-mlir python=3.11
conda activate torch-mlir
...

but ran into errors there also.

Discussion

Seems like some version matching problem.

With a working Dockerfile and tested setup using a specific version (rather than a latest version) the demo setup and execution should be always reproducible.