Skip to content

Example: custom Python model

Exposes a plain Python object as a REST API - no ML framework and no serialized model file, only a predict-like method.

Source: examples/custom-python-model/

Requirements

  • Python 3.8+
  • pymlup (no extra needed)

Run

git clone https://github.com/nxexox/pymlup.git
cd pymlup/examples/custom-python-model

pip install -r requirements.txt
python run.py

model.py defines MyModel, a plain Python class with a predict(self, X) method. run.py passes an instance of it straight to mlup.UP(ml_model=...) - no file, no pickling.

Test API

curl -X POST http://localhost:8009/predict \
  -H "Content-Type: application/json" \
  -d '{"X": [[1, 2], [3, 4]]}'

Expected result

{"predict_result": [3, 7]}

See the full README for this example on GitHub.