Running Parallel Experiments in Driverless AI

This lesson walks through running and comparing multiple model experiments in H2O Driverless AI at once. It covers how a leaderboard run creates several experiment variations with different expert settings, which are automatically queued and executed as CPU and GPU resources become available. It shows the worker queue for tracking experiments, and how the Compare Experiments tool lets you review a summary, variable importance, and an actual-versus-predicted graph side by side, plus the exact setup differences between two runs. It notes that turning off the reproducible toggle is recommended when running on GPUs.