TRUST：用统计检验为自我改进智能体把关，坏更新占比降到2.78%
做自改进智能体的人应该看看：贪心收更新有近一半是错的，TRUST 用 LLM 决定证据量，坏接受降到 2.78%，分数还比旧统计门槛高。
自我改进智能体通常采用贪心策略：只要某个 harness 更新能提高验证分数就接受，但验证分数有噪声，导致大量坏更新被误收。研究者在实验中测得，贪心门槛下 45.59% 被接受的更新实际上使性能变差。论文提出 TRUST 方法，把每次更新接受视为一次统计检验，控制真实性能损失型更新被接受的比例，并由一个 LLM 读取循环的历史记录，动态决定每个更新需要多少证据才能通过。
在 HotpotQA、DROP、APPS 三个基准上，使用 DeepSeek-V4-Flash 和 Qwen-3.8-Flash 两个模型实验，坏接受比例降到 2.78%，最终分数也高于以往偏保守的统计门槛方法——那些方法虽然稳，但会拒绝掉真正有益的更新。论文还给出了坏接受比例的上界和最终收益的下界的理论证明。
// Trustworthy Recursive Self-Improvement //
Self-improving agents usually keep any harness update that raises the validation score, so noisy gains get accepted, and this later costs performance.
Researchers measured how often this happens. Under a greedy gate, 45.59% of accepted updates were actually worse.
Their gate, TRUST, treats each acceptance as a statistical test and controls the share of accepted updates that cause a real loss. An LLM reads the loop's history to decide how much evidence each update needs.
Across HotpotQA, DROP, and APPS with DeepSeek-V4-Flash and Qwen-3.8-Flash, bad acceptances fell to 2.78%. Final scores were also higher than with earlier statistical gates, which tend to reject good updates.
They prove a bound on bad acceptances and a lower bound on the final gain.
https://t.co/Y20VQIrMhJ
Chat with Paper: https://t.co/wZ6FSbDoYZ
本条情报归属 AITOP「论文」核心栏目，重点聚焦北美前沿实验室与硅谷科技巨头。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：TRUST、DeepSeek-V4-Flash、Qwen-3.8-Flash、智能体，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「DAIR.AI」（@dair_ai），经 AITOP 多维评分管线结构化提炼。
