跳到正文
研踪PaperTrail
← 研踪 PaperTrail

arXiv:2611.19507 · v1

Evaluating and Improving Natural Language Autoencoders

首次提交 2026-11-15 00:57
版本更新 2026-11-15 00:57
雷达观测 2026-11-17 12:06
arXiv ↗PDF ↗

摘要

Natural language autoencoders (NLAs) are a method for generating natural language explanations of LLM activations. We develop evaluations based on auditing tasks that test the ability of NLAs to surface latent knowledge, identify whether a model was misled by a hint, and catch a silent fabrication. We also develop extensions to NLAs that let them use activation information from multiple layers and positions, let us apply them more effectively to long transcripts, and reduce the number of false claims they make. These extensions result in improvements on both our auditing evaluations and auxiliary evaluations adapted from prior work. We also use the evaluations to study how different auditing capabilities change over the course of NLA training, and find that although every evaluation improves early in training, extended training can hurt performance on some of our auditing evaluations.

作者备注:23 pages, 18 figures, 2 tables

AI 论文解读

基于摘要 · 非全文审阅

此版本尚未生成 AI 解读。解读由后台任务生成并按论文版本保存,登录后共享已有结果。

版本差异

目前只收录了一个版本。收录下一版后,可以对照标题、作者、分类和摘要。

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 · 2026-11-15 00:57 · 最新观测版本

Evaluating and Improving Natural Language Autoencoders

Natural language autoencoders (NLAs) are a method for generating natural language explanations of LLM activations. We develop evaluations based on auditing tasks that test the ability of NLAs to surface latent knowledge, identify whether a model was misled by a hint, and catch a silent fabrication. We also develop extensions to NLAs that let them use activation information from multiple layers and positions, let us apply them more effectively to long transcripts, and reduce the number of false claims they make. These extensions result in improvements on both our auditing evaluations and auxiliary evaluations adapted from prior work. We also use the evaluations to study how different auditing capabilities change over the course of NLA training, and find that although every evaluation improves early in training, extended training can hurt performance on some of our auditing evaluations.

查看此版本解读 →查看此版本原文 ↗