Hill climbing Ari’s skills

Share of runs where Ari finds the key root cause, by skill version

Tuning (55 tasks)Evaluation (34 held-out tasks)
Hill climbing Ari’s skills30%40%50%60%70%80%Runs finding the root cause012345678Skill version (v0 to v8)v0 tuningv0 evaluationTuning 72%Evaluation 54%