Add a new entry to the volumes
section of the relevant service in docker-compose.yml
- type: volume
source: $HOME/.aws/credentials
target: /home/app/.aws/credentials
readonly: true
Modify the the core jupyter service to use the cschranz/gpu-jupyter image as a base instead.
Your dockerfile should look something like this
FROM cschranz/gpu-jupyter
RUN pip install --upgrade pip pip-tools
COPY requirements.* .
RUN pip-compile requirements.* --output-file requirements.txt
RUN pip install -r requirements.txt
RUN rm -rf work requirements*
Note that this image is very large (around 6GB, compared to the 800MB regular image) and will take some time to download and unpack. Keeping a cached version of this image somewhere easily accessible will save you time and money if you're planning to use this image on cloud GPUs.