TRIAGE: Role-Typed Credit Assignment for Agentic Reinforcement Learning

By Yuanda Xu · Paper · cs.LG

Agentic reinforcement learning requires assigning credit to environment-facing actions such as searches, clicks, edits, navigation commands, and object interactions. Standard GRPO uses the final verifier outcome as a uniform advantage over all action tokens. This outcome signal i

AI Agents · Cs.lg

View original

HomeResourceLoading…