Jul 2026
DREvo: Distilling Recalibrated Historical Experience for Harness Self-Evolution
A new harness self-evolution method, named DREvo, is proposed, which integrates function-level evidence anchoring, state-dependent evidence recalibration, and role-conditioned search intent distillation to determine which historical evidence remains valid and where the harness should evolve next.
Hanghui Guo, Wei-Jie Shi, Zhangze Chen et al.
· arXiv.org · 4 citations