docker build -t deepnote/python:<TAG> --build-arg FROM_PYTHON_VERSION=<some_version> ./python
For every Python version (see above), we also build a -conda
variant, which includes Miniconda installation.
The Dockerfile for this is at ./python/Dockerfile.conda.
docker build --build-arg FROM_PYTHON_VERSION=<some_version> -t deepnote/python:<some_version>-conda ./python --file ./python/Dockerfile.conda
We also have a variant with number of preinstalled libraries. We build it for all the above mentioned R versions
and the image is available under deepnote/ir-with-libs
. See ir-with-libs for more details.
docker build -t deepnote/ir:<TAG> --build-arg R_BASE_VERSION=<some_version> ./ir
The image is based on tensorflow docker image and adds packages that are typical part of our other images (git
most importantly).
We follow the tagging of tensorflow
image. So tensorflow/tensorflow:2.4.1-gpu
becomes deepnote/tensorflow:2.4.1-gpu
.
Some images are built automatically using the Docker Hub build. See Docker Hub Builds for more details.