English(EN) The Winner's Curse in LLM Self-Improvement Loops: Selection Noise, Lock-in, and Acceptance Rules

LLM自改进循环因选择噪声而遭受“赢者诅咒”

作者 PulseAugur 编辑部 · [1 个来源] · 2026-10-08 04:00

一篇新的arXiv论文探讨了自改进大型语言模型（LLM）中“赢者诅咒”现象。该研究使用Qwen模型重写自己的指令，发现初始指令之外的大多数提议的更改都是有害的。研究强调了当LLM在小型、重复使用的数据集上评估自身改进时，选择噪声和锁定问题会导致报告的收益相对于实际的保留准确率膨胀。 AI

影响强调了LLM自改进中潜在的陷阱，表明需要更鲁棒的评估方法来避免性能下降。

排序理由学术论文，详细介绍了LLM自改进中的特定现象。[lever_c_demoted from research: ic=1 ai=1.0]

AI 生成摘要 · Google Gemini · 来自 1 个来源。我们如何撰写摘要 →

本文如何被排名

Signal score: 7 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 学术论文，详细介绍了LLM自改进中的特定现象。[lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration: Single-source cluster; Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics: paper, safety; Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance: High; Clearly on-topic for AI-industry coverage.
Story freshness: Same-day; Cluster formed today. Ranking reflects the current source set at time of score.

完整方法见我们的编辑标准。

arXiv cs.LG TIER_1 English(EN) · Litao Hu, Yutong Tang · 2026-10-08 04:00

大型语言模型自我改进循环中的赢者诅咒：选择噪声、锁定和接受规则

arXiv:2610.09239v1 Announce Type: cross Abstract: Self-improving LLM systems propose changes to themselves and keep those that score better on a small evaluation set. We treat this keep-if-better step as selection under measurement noise, model the correlated errors of the candid…