You are viewing the RapidMiner Applications documentation for version 2024.0 - Check here for latest version

Python Environments on Altair® AI Edge™ devices

Altair® AI Edge™ devices comes with a default (base) Python installation located in /usr/bin/python3. However you are strongly encouraged to push your own Python environments to Altair® AI Edge™ devices.

Python Environments for Altair® AI Edge™ are exactly the same as those used in the Python Extension Development (PED) SDK. A sample environment can be found here:

pydist-sample-environment-0.1.0-linux-arm64.tar

And full documentation on how to build your own environment can be found here:

Confluence - PED - Build your own distribution

Note that you must only use ARM64-compatible Python environments with your Altair® AI Edge™ devices! A nice tool to build ARM64-compatible environments is Astral. Download with curl -LsSf https://astral.sh/uv/install.sh | sh, create + activate a new environment, then add libraries with uv prefix e.g. uv pip install onnxruntime onnxruntime_extensions pandas numpy.

Currently there are no operators in the AI Edge Toolbox extension to get / push / delete Python environments; to do so, you must use the RapidMiner ECP API endpoints and the IoT Studio EdgeOps Cluster API Inspector.

For example:

POST /support/rapidminer/scoring-agents/01/python-environments

body:

{
  "file": "https://altair-ai-edge.s3.amazonaws.com/rapidminer/python-environments/ai-edge-torch.tar",
  "name": "ai-edge-torch.tar",
  "auth": {
    "type": "omit"
  }
}

Once you do so, your new Python executable will be located at /altair/python/\<your-environment-name\>/bin/python3.

Note: You must enter this path in your Execute Python operator in the python binary field in order to query the binary in your environment!

Categories

Versions

Python Environments on Altair® AI Edge™ devices