Back to article
lesswrong.com

Logits as a new monitor for evaluation awareness (opens in new tab)

Covers 3 stories See all stories this covers including Qwen3 Technical ReportCovered by threadreaderapp.comDiscussed on Hacker News

Covers 3 related stories

arxiv.org·

Qwen3 Technical Report

Feeds
GitHub·

MoonshotAI/Kimi-K2.5: Moonshot's most powerful model

Feeds
transformer-circuits.pub·

Natural Language Autoencoders Produce Unsupervised Explanations of LLM Activations

Discussed on Hacker News
Feeds

Covered in 1 article

threadreaderapp.com·

Would an LLM tell you if it’s gaming your eval? Often, no. But we can still catch the model thinking about it.

Feeds

Keyboard Shortcuts

Navigation

Next / previous post
j/k
Open post
oorEnter
Preview post
v

Post Actions

Love post
a
Like post
l
Dislike post
d
Undo reaction
u
Save / unsave
s

Recommendations

Add interest / feed
Enter
Not interested
x

Go to

Home
gh
Interests
gi
Feeds
gf
Likes
gl
History
gy
Changelog
gc
Settings
gs
Discover
gb
Search
/

General

Show this help
?
Submit feedback
!
Close modal / unfocus
Esc

Press ? anytime to show this help