Setting num_workers > 1 as suggested in the tutorial did speed optimization up, but Im looking for more speed up in generating the counterfactual explanations. I tried following the instructions for using Ray with slurm in the tutorial but I'm having trouble following. Are there plans to release tutorial for multi-node parallelization?
Absolutely, stay tuned! 🚀 We are working on a tutorial right now that will use the submitit package to easily fan out jobs across hundreds of CPU nodes
Setting
num_workers > 1
as suggested in the tutorial did speed optimization up, but Im looking for more speed up in generating the counterfactual explanations. I tried following the instructions for using Ray with slurm in the tutorial but I'm having trouble following. Are there plans to release tutorial for multi-node parallelization?