"Shaping & Monitoring Models through their Internal Representations"
摘要
Goodfire 提出 Mechanistic Shaping，利用模型内部表征训练其不依赖无关信息或产生谄媚、奖励黑客等不良行为。在带偏见的图像分类基准上，最差组准确率从 51.6% 提升至 97.7%；在最高 32B 的大语言模型上也能减少不良行为，但模型有时会通过内部换一种编码方式隐藏这些概念。
摘要由模型根据原文撰写，以原文为准。
正文 · 原文
"Shaping & Monitoring Models through their Internal Representations"
AI models can learn to get the right answers using the wrong clues, like identifying a bird from its background instead of the bird itself, and traditional training often rewards them for it anyway.
In this new Goodfire paper, they propose Mechanistic Shaping, which uses internal representations to train models to avoid relying on unwanted information or expressing undesirable behaviors, such as sycophancy and reward hacking.
On a biased image-classification benchmark, this improves worst-group accuracy from 51.6% to 97.7%. It also reduces unwanted behaviors in LLMs up to 32B, although models can sometimes hide these concepts by encoding them differently internally.
来源：alphaXiv · x.com
引用原文
askalphaxiv. "Shaping & Monitoring Models through their Internal Representations"[EB/OL]. (2026-11-09)[访问日期]. https://x.com/askalphaxiv/status/2119763684386054485.