1 paper · 1 filter
Shivam Gupta
Tool-result caching reduces repeated execution in agent training, but also couples rollout randomness. We study a two-action model in which independent and shared execution preserv…