Model Object, Optimized (MOJO)

Ready

What is a MOJO?

MOJO (stands for Model Object, Optimized) [1] is a standalone, low-latency model object designed to be easily embeddable in production environments.

In Driverless AI, the MOJO Model is combined with a feature engineering pipeline to create a MOJO scoring pipeline that can be executed in Java Runtime or C++ Runtime. The Driverless AI MOJO Model could be XGBoost, LightGBM, GLM, Decision Tree, TensorFlow, FTRL and even other models. A valid Driverless AI license key in required to run the MOJO scoring pipeline. The Driverless AI MOJO scoring pipeline can be executed using the mojo2-runtime.jar, mojo2 py runtime, mojo2 r runtime or mojo2 C lib runtime. Once you have the Driverless AI license key, the MOJO scoring pipeline can be run practically anywhere [1].


In H2O-3, the MOJO does not combine with a feature engineering pipeline and is just a scoring pipeline that can be executed in Java Runtime using h2o-genmodel.jar, h2o py module or h2o r library. H2O-3 supports supervised and unsupervised algorithms for the MOJO model. H2O-3 is open source under the Apache License 2.0 and does not require a license key to execute the H2O-3 MOJO [2].

Resources

Tutorial

Videos


#h2o-3
#driverless-ai
#mojo
#concepts
0 comments
28 Views
 

Permalink

Related Links

No Related Resource entered.