SkillGate: Training In-Policy Skill Selection in Long-Horizon Agents

  • 类型:arxiv
  • 标识:2608.18852
  • 链接:https://arxiv.org/abs/2608.18852
  • 主分类:agent
  • 形态:method
  • 被引:0
  • 被引来源:Semantic Scholar
  • S2被引:0
  • 影响力被引:0
  • TLDR:SkillGate lifts a 9B policy from 40.8% to 53.2% trial success, well ahead of the identical budget spent on outcome reward alone, while cutting exposure to misleading candidates by two thirds and reading fewer skills.
  • 副分类:engineering
  • 待LLM分类:否
  • 标题中文:SkillGate:在长视野 Agent 中训练策略内技能选择
  • TLDR中文:SkillGate 将 9B 策略的成功率从 40.8% 提升至 53.2%,显著优于将相同预算仅用于 outcome reward 的方案,同时将误导性候选的暴露减少三分之二,并读取更少的 skill。
  • 来源文件
  • /inbox/tom/_candidates/2026-08-20-agent-rag-longcontext-candidates.json
  • /inbox/tom/_candidates/2026-08-21-agent-rag-longcontext-candidates.json
  • [S2 enrich]