Installs as a private draft. Your edits and self-improvement runs do not change the published bundle.
When to reward final results vs. intermediate steps, and how to blend both
When to reward final results vs. intermediate steps, and how to blend both
This is an active research question with real practical implications
Reward Engineering > Reward Function Design
--- name: "Outcome Vs Process Reward Tradeoff" version: "1.0.0" description: "When to reward final results vs. intermediate steps, and how to blend both" tags: - rl-skills-universe - "reward-engineering" - "reward-function-design" trigger: "outcome-vs-process-reward-tradeoff" --- # Outcome Vs Process Reward Tradeoff When to reward final results vs. intermediate steps, and how to blend both ## Why It Matters This is an active research question with real practical implications ## Domain **Reward Engineering** > Reward Function Design