Inference Runtimes

Inference runtimes allow you to define how your model should be used within MLServer. You can think of them as the backend glue between MLServer and your machine learning framework of choice.

Out of the box, MLServer comes with a set of pre-packaged runtimes which let you interact with a subset of common ML frameworks. This allows you to start serving models saved in these frameworks straight away. To avoid bringing in dependencies for frameworks that you don't need to use, these runtimes are implemented as independent (and optional) Python packages. This mechanism also allows you to rollout your own custom runtimes very easily.

To pick which runtime you want to use for your model, you just need to make sure that the right package is installed, and then point to the correct runtime class in your model-settings.json file.

Included Inference Runtimes

Framework	Package Name	Implementation Class	Example	Documentation
Scikit-Learn	mlserver-sklearn	mlserver_sklearn.SKLearnModel	Scikit-Learn example	MLServer SKLearn
XGBoost	mlserver-xgboost	mlserver_xgboost.XGBoostModel	XGBoost example	MLServer XGBoost
Spark MLlib	mlserver-mllib	mlserver_mllib.MLlibModel	Coming Soon	MLServer MLlib
LightGBM	mlserver-lightgbm	mlserver_lightgbm.LightGBMModel	LightGBM example	MLServer LightGBM
CatBoost	mlserver-catboost	mlserver_catboost.CatboostModel	CatBoost example	MLServer CatBoost
Tempo	tempo	tempo.mlserver.InferenceRuntime	Tempo example	github.com/SeldonIO/tempo
MLflow	mlserver-mlflow	mlserver_mlflow.MLflowRuntime	MLflow example	MLServer MLflow
Alibi-Detect	mlserver-alibi-detect	mlserver_alibi_detect.AlibiDetectRuntime	Alibi-detect example	MLServer Alibi-Detect

:hidden:
:titlesonly:

SKLearn <./sklearn>
XGBoost <./xgboost>
MLflow <./mlflow>
Tempo <https://tempo.readthedocs.io>
Spark MLlib <./mllib>
LightGBM <./lightgbm>
Catboost <./catboost>
Alibi-Detect <./alibi-detect>
Alibi-Explain <./alibi-explain>
HuggingFace <./huggingface>
Custom <./custom>