Creating a custom image
Create custom images for Python-based models and functions or create a custom SPSS image.
The samples that are provided here use Podman, but you can use any other Docker-compatible image builder.
If you're creating a custom image for a custom foundation model, you must create a Python image.
Creating a custom Python image
After you download the image for the runtime that you want to customize, you can create a custom image by adding your customizations to the downloaded image. Start with the base image and install the libraries and packages that you need. Then, add your customizations to the Dockerfile.
Required role: You must have root privileges on your machine to execute the commands that build a custom image.
Limitations: You can customize ai-service and
pytorch-onnx frameworks only based on runtime-24.1-py3.11. It is
not possible to customize ai-service and pytorch-onnx frameworks
based on runtime-25.1-py3.12.
- Prepare the Dockerfile.
- Do not change the contents in the USER directives. They impact how the container runs in the cluster.
- You must have root privileges to install OS packages and to modify kernel specifications.
- Use the
microdnftool to install the OS libraries.
- Create a custom image. For more information, see Customizing Python and SPSS deployment images.
Adding customizations for SPSS models
To add customizations for SPSS, run the Dockerfile for installing the Exasol driver in SPSS.
For <SPSS_RUNTIME_IMAGE>, use the image name from Downloading the runtime definition.
For more information, see Customizing Python and SPSS deployment images.