Large Scale Benchmarking

Parallel

You can run your optimization via

conda run -n automlsuite_DUMMY_Optimizer_DUMMY_Problem python carps/run.py \
    +optimizer/DUMMY=config +problem/DUMMY=config \
    'seed=range(1,11)' \
    +cluster=local -m

This uses joblib parallelization on your local machine. If you are on a slurm cluster, you can specify +cluster=slurm and adapt this to your needs. Check this page for more launchers, e.g. Ray or RQ besides Joblib and Submitit.