Tencent Hunyuan's RSR Framework Boosts Complex-Task Training Trajectories

Tencent's Hunyuan team proposed Recursive Self-Rewrite (RSR), which uses a base 27B model to explore successful solutions under diverse harnesses and rewrite them recursively. Terminal-Bench 2 pass@3 improved from 57% to 74%.

2026-10-05 ~ 2026-10-05 · 2 related posts

• Tencent Hunyuan's RSR Boosts 27B Model Terminal-Bench 2 pass@3 from 57% to 74% — Tencent-Hunyuan · 2026-10-05
• Tencent Hunyuan's RSR rewrites expert-harness trajectories for single-model training — _akhaliq · 2026-10-05