DRACO: Fine-Grained Credit Assignment with Dynamic Rubrics for Long-Horizon Agent Training

  • 类型:arxiv
  • 标识:2609.04094
  • 链接:https://arxiv.org/abs/2609.04094
  • 主分类:agent
  • 形态:method
  • TLDR:Reinforcement Learning from Verifiable Rewards works well when a task has a programmatic checker, but most long-horizon agent domains have none. We work in the outcome-blind setting, where ground-truth success signals are not available. Multi-criteria rubrics are a popular way to supply such a reward; they are scored once per trajectory, but a single scalar is a poor signal across tens of steps. We propose DRACO: Distributing Rubric-based Advantage for Credit Optimization. It generates rubrics dynamically during training to track the policy's evolving capability, scores those rubrics once per
  • 副分类:engineering
  • 待LLM分类:否
  • 来源文件
  • /inbox/tom/_candidates/2026-09-05-agent-rag-longcontext-candidates.json