冻结权重下的智能体自我改进：失败轨迹的价值与局限
Agent Self-Improvement with Frozen Weights: The Added Value and Limits of Failure Traces
搞 Agent 框架的人该看看这篇：它用 9B 冻结模型和教师写笔记，量化了失败轨迹到底值多少——只有 4 个百分点，而且 60% 的提升靠普通搜索就能拿到。跟一堆宣称“Agent 自我进化”的工作比，这篇泼了盆冷水，还顺手指出测试集过半有缺陷。
一项 arXiv 论文（arXiv:2611.11342）测试了在模型权重完全冻结的前提下，智能体的外层 harness 能否从失败轨迹中学到东西。
实验使用一个 90 亿参数的开源模型，由更大的教师模型撰写提示笔记，任务为生成的客服决策场景。结果表明，基于失败轨迹的笔记比隐藏失败信息的对照笔记准确率高出 4.
0 个百分点，但单纯的“尝试并保留高分笔记”流程就已达到总提升的约 60%。
Agent Self-Improvement with Frozen Weights: The Added Value and Limits of Failure Traces
We evaluate whether an AI agent's harness, the notes around its model, can learn from the agent's failure traces beyond a matched search that never sees them. It can, but less than it looks. We tested an open 9-billion-parameter model with fixed weights; a larger teacher model wrote its notes. On generated customer-service decisions, where both note writers were told the target rules, notes based on the agent's mistakes improved accuracy by 4.0 percentage points over notes from the same process with the mistakes withheld. That process alone, trying and keeping better-scoring notes, reached about 60% of the total improvement. In full customer-service conversations the same notes showed no demonstrated improvement. Separately, a strict screen of single lessons admitted two customer-service lessons, still awaiting tests on new data, and one tool-use lesson, written without seeing the agent's mistakes, which showed no demonstrated gain on new tool-use tasks. A neutral note with no procedure scored higher than the learned lessons (exploratory). A blind audit judged 146 of 240 tool-use test items defective.
本条情报归属 AITOP「论文」核心追踪栏目。在跨区域产业观察坐标中，该动态直接关联「AI 开发者工具链、Coding Agent 与工程化实践」研究主线，反映出当前技术演进与商业落地的最新风向。涉及核心实体与技术议题包括：智能体、论文、客服决策、工具使用，读者可通过下方标签进入对应专题页查阅历史脉络与关联对比。
信源核验说明：本文原始线索采编自「arXiv cs.AI」（@Arkid Mitra），经 AITOP 多维评分管线结构化提炼。
