Reinforcement Learning with Decomposed Subtasks

Breaking down multi-turn rewards to improve reasoning and skill acquisition in agentic language models.

Group Relative Policy Optimization (GRPO) and related policy-gradient methods for training language model agents collapse an entire multi-turn rollout into a single scalar trajectory reward before it enters the policy update. When the task composes distinct skills, especially und...

ArXiv CS.AI2026-09-24