Pavel Pscheidl • 1/16/2020

Parallel Grid Search in H2O

This technical article explains H2O's new parallel grid search feature, which allows multiple machine learning models to be trained concurrently during hyperparameter tuning. It details how this improves cluster resource utilization (CPUs/GPUs) and reduces training time compared to sequential model building. The guide covers the new 'parallelism' parameter in Flow, Python, and R, explaining how to set a fixed number of concurrent models or use an adaptive mode.

0 comments

#Machine Learning #H2o #Parallel Computing