SafetyPage

Lab research on the web · Blog post · Anthropic

Reducing Reward Hacking by Localizing Bad Behavior Instead of Penalizing It

Read at alignment.anthropic.com ↗Read at sourceNo licence statement found

Outline · 46 paragraphs

  1. Background¶10
  2. The method¶15
  3. Results¶23
  4. Limitations and future work¶35
  5. Summary¶42