思想与社会2023 · 分享于 Hacker News
The Geometry of Truth: Emergent Linear Structure in How LLMs Represent Truth
探索《The Geometry of Truth: Emergent Linear Structure in How LLMs Represent Truth》的原文与 Hacker News 社区讨论，查看来源并加入阅读清单。
从论文，走向讨论
看看社区如何讨论这个想法。
来自 Hacker News 的原始讨论。不同的视角、疑问，以及值得继续思考的线索。
最多展示 100 条相关讨论。论文可能出现在评论中，并非原帖投稿内容。
在讨论中发现的引用
按 HN 条目顺序展示最近的 50 条引用来源。
Yitchyjunk ·
Persona vectors: Monitoring and controlling character traits in language models
408 赞同137 评论
Yrntn ·
Google scrambles to manually remove weird AI answers in search
283 赞同368 评论
Yrfw300 ·
Legal models hallucinate in 1 out of 6 (or more) benchmarking queries
215 赞同255 评论
Ybenocodes ·
LLMs know more than they show: On the intrinsic representation of hallucinations
137 赞同140 评论
Yyarapavan ·
GPTs and Hallucination
134 赞同205 评论
YDuximo ·
AI is growing faster than companies can secure it, warn industry leaders
31 赞同51 评论
Yanarbadalov ·
"Hallucinating" AIs Sound Creative, but Let's Not Celebrate Being Wrong
27 赞同15 评论
Yjdmark ·
Rats have an imagination, new research suggests
6 赞同1 评论
Ymagoghm ·
I disagree with Geoff Hinton regarding "glorified autocomplete"
4 赞同0 评论
Ysebg ·
The Geometry of Truth: Do LLM's Know True and False
3 赞同0 评论
Ylogicallee ·
Ask HN: What is your standard for judging when AGI exists?
2 赞同0 评论
Ycorn-dog ·
Ask HN: Convince me LLMs aren't just autocomplete with lookahead
2 赞同3 评论
Yog_kalu ·
The Geometry of Truth: Emergent Linear Structure in How LLMs Represent Truth
1 赞同1 评论
Ym_kos ·
Role play with large language models – Nature
1 赞同1 评论
YAnon84 ·
Language models and linguistic theories beyond words
1 赞同0 评论