Glossary

A | M | S | T

A

Axon: An axon is a set of models and model graphs trained and deployed by running steps triggered by a sensor.

M

Model

A model is predictive unit deployed as a REST endpoint.

Model Graph

A model graph is collection of models that contribute to build a higher level prediction.

For example, an OCR model could be implemented as model graph containing 2 models: detection and recognition .

S

Sensor: An sensor is used to configure triggers, conditions and steps to execute if a matching event is detected.
Step: A unit of execution of the framework. Steps can be Train, Deploy, Spark, TFJob or Generic.

T

Trigger: A trigger is used to configure an event a sensor listen to.