Junyao Yang

Junyao Yang

he/his

Junyao Yang's Homepage

Terminal-RSI: Recursive Self-Improving Terminal Agents via Recursive Self-Rewrite and Policy-Refine

Junyao Yang, Zongxia Li, Yucheng Shi*, Zhongzhi Li, Ruhan Wang, Ruiwen Zhou, Haitao Mi, Shuicheng Yan†, Leowei Liang

  • Equal contribution · † Corresponding author

Terminal-RSI abstract-page results across Terminal-Bench 4.0, 3.0, 2.1, Hard, LHTB and SWR100

Alternating verified trajectory rewriting with critic-inheriting policy refinement, using turn-boundary credit assignment and stage-specific rewards to advance terminal agents from partial progress to strict task completion. Terminal-RSI-27B achieves 19.84% on Terminal-Bench 4.0 and 86.52% on Terminal-Bench 2.1.

Paper Project Page Code Model