← 研踪 PaperTrailarXiv:2611.26132 · v1
Sage-Evolve-Preview:Adaptive Observation for Post-Training Self-Improvement
首次提交 2026-11-19 21:52
版本更新 2026-11-19 21:52
雷达观测 2026-11-20 16:53
摘要
Post-training recursive self-improvement (RSI) iteratively improves a target capability, operationalized by a fixed anchor benchmark, through repeated intervention, training, and evaluation. Yet most loops use the anchor's aggregate score as their sole observation, so early gains give way to stalled progress and repetitive exploration. When this anchor-only feedback ceases to reveal actionable gaps, a plateau may reflect an exhausted observation rather than an exhausted capability. We introduce adaptive observation (AdaObs), an observation-side mechanism of harness evolution that keeps the anchor objective fixed while adapting the active observation. At a plateau, a controller refines diagnostic granularity, escalates difficulty, and adds related tasks; validated increments are fused back into training, while every checkpoint remains evaluated against the anchor. Across eight independently optimized anchors in long-context reading comprehension and instruction following, AdaObs raises the mean score from to under a matched data budget, outperforming FixedObs, the anchor-only baseline. The evolved harness further improves the seven-task average on the independent PostTrainBench across two agent backends. These results identify observation adaptation as a missing mechanism of harness evolution for sustaining post-training RSI.
观测到的版本
相邻记录可能跳过中间版本;首次发现 v2 及以上也会标为版本更新。
v1 · 2026-11-19 21:52 · 最新观测版本
Sage-Evolve-Preview:Adaptive Observation for Post-Training Self-Improvement
Post-training recursive self-improvement (RSI) iteratively improves a target capability, operationalized by a fixed anchor benchmark, through repeated intervention, training, and evaluation. Yet most loops use the anchor's aggregate score as their sole observation, so early gains give way to stalled progress and repetitive exploration. When this anchor-only feedback ceases to reveal actionable gaps, a plateau may reflect an exhausted observation rather than an exhausted capability. We introduce adaptive observation (AdaObs), an observation-side mechanism of harness evolution that keeps the anchor objective fixed while adapting the active observation. At a plateau, a controller refines diagnostic granularity, escalates difficulty, and adds related tasks; validated increments are fused back into training, while every checkpoint remains evaluated against the anchor. Across eight independently optimized anchors in long-context reading comprehension and instruction following, AdaObs raises the mean score from to under a matched data budget, outperforming FixedObs, the anchor-only baseline. The evolved harness further improves the seven-task average on the independent PostTrainBench across two agent backends. These results identify observation adaptation as a missing mechanism of harness evolution for sustaining post-training RSI.
查看此版本解读 →查看此版本原文 ↗