This is the last situation
I could create a docker that works

FROM pytorch/pytorch:2.2.1-cuda12.1-cudnn8-runtime

WORKDIR /app

# Install system dependencies
RUN apt-get update && apt-get install -y \
    libgl1-mesa-glx \
    libglib2.0-0 \
    git \
    curl\
    && rm -rf /var/lib/apt/lists/*

# Install dependencies
COPY requirements.txt /tmp/requirements.txt
RUN pip install --no-cache-dir -r /tmp/requirements.txt

# Update CA certificates
COPY hme_root_CA.crt /usr/local/share/ca-certificates/hme_root_CA.crt
RUN update-ca-certificates

ENV SSL_CERT_FILE=/usr/local/share/ca-certificates/hme_root_CA.crt
ENV REQUESTS_CA_BUNDLE=/usr/local/share/ca-certificates/hme_root_CA.crt
ENV PIP_CONFIG_FILE=/root/conda.conf
ENV CLEARML_AGENT_SKIP_PYTHON_ENV_INSTALL=true
ENV CLEARML_AGENT_SKIP_PIP_VENV_INSTALL=true

RUN pip config set global.cert /usr/local/share/ca-certificates/hme_root_CA.crt
RUN pip install --trusted-host pypi.python.org --trusted-host files.pythonhosted.org --trusted-host pypi.org --upgrade pip
RUN conda config --set ssl_verify /usr/local/share/ca-certificates/hme_root_CA.crt

COPY conda.conf /root/conda.conf


# Configure ClearML
COPY clearml.conf /root/clearml.conf
RUN clearml-init

# Start ClearML worker
CMD [ "clearml-agent", "daemon", "--queue", "default", "--foreground" ]

But it just works if I use clearml-agent daemon --gpus 0 --queue default --foreground

If I try to use clearml-agent daemon --gpus 0 --queue default --foreground --docker (in the host)
When It opens the docker mentioned above, it always tries to install different apt packages, and tries to install some conda packages and gets stall

Do you know if is possible to avoid all the installations? And just run everything in the docker's environment?