Use ML Models

This set of handlers provide ability to run real-time inference using various machine learning libraries, which share the same set of parameters and actions.

Handler Parameters

This handler is cacheable.

Parameter	Definition	Example	Default
model.state	Name of the state manager with model configurations	model	-
model.id	Id of the model to run on this instance	recommendation	-
model.store.system	Name of the system to store model assets	hdfs_default	-
model.local.dir	Local directory to store model assets for execution	/tmp	java.io.tmpdir

519B

handler-tensor-0001.json

Open

Example Tensorflow Model Handler Definition (Can be Imported on Element Screen)

Actions

Score / ScoreML

Scores input data using pretrained model managed by this handler. Event metadata fields applicable for this action are as follows:

Fields (table)

Field	Definition	Example	Default
Input Element	Json path for the input in request event payload	parameters	-
Output Element	Json path for the output in response event payload	$.score	-

Fields (JSON Schema)

With event metadata parameters as:

Parameters (table)

Parameter	Definition	Example	Default
Input Pattern	JMESPath pattern to apply on data input	[ [text] ]	-
Output Pattern	JMESPath pattern to apply on data output, before returning response	{totalPrice:basket.totalPrice}	-

Parameters (JSON Schema)

Versions

TensorScoreEventHandler

This handler (com.rierino.handler.tensor.TensorScoreEventHandler) uses Tensorflow library for inference using saved models.

This handler requires the following dependency added to deployment contents:

implementation (group:'com.rierino.custom', name: 'tensor', version:"${rierinoVersion}")

OnnxScoreEventHandler

This handler (com.rierino.handler.onnx.OnnxScoreEventHandler) uses ONNX Runtime library for inference using models saved using this common format.

This handler requires the following dependency added to deployment contents:

implementation (group:'com.rierino.custom', name: 'onnx', version:"${rierinoVersion}")

PMMLScoreEventHandler

This handler (com.rierino.handler.pmml.PMMLScoreEventHandler) uses PMML formatted models for inference.

This handler requires the following dependency added to deployment contents:

implementation (group:'com.rierino.custom', name: 'pmml', version:"${rierinoVersion}")

OpenNLPScoreEventHandler

This handler (com.rierino.handler.opennlp.OpenNLPScoreEventHandler) uses OpenNLP library for inference using common NLP tools, saved dictionaries and models.

This handler requires the following dependency added to deployment contents:

implementation (group:'com.rierino.custom', name: 'opennlp', version:"${rierinoVersion}")