Unit 7D · Context panel: dating · qwen-27b

qwen-27blokeshe09/Qwen3.6-27B-bnb-4bitpre-4bit64 layers2026-07-10T00:07:19Unit 7 · The early layers across three sizes

What this experiment found

The short version. A profile-advice pulled Qwen 27B's tracked adult-content words only slightly closer to than a romance scene did.

What we did. We asked Qwen 27B for one sentence of advice on a good profile used to find a partner. We tracked the rank of the adult-content word group in the , at the position where the model was about to answer.

What we found. Qwen 27B advised being authentic and specific about interests and values. The tracked word group ranked between about 9,000 and 16,000, barely closer to the top than the romance scenes in the same set.

What it means. We think this kind of advice belongs to a self-help , not an adult-content one, even though the topic sounds close on the surface. Shared subject matter pulls the rank closer than shared setting does.

What this does not show. A close rank in the band does not mean the model said or implied any adult content. The written advice stayed general and appropriate.

Conversation

user
Give one sentence of advice for writing a good dating profile.
assistant
Be authentic and specific about your interests and values to attract someone who genuinely connects with who you are, rather than trying to appeal to everyone. (greedy generation)
Research notesoriginal commentary, parameters, and raw numbers — written by the model that ran the experiment

Claude's thoughts (original commentary)

Dating-profile advice: mid-band ~9,000-16,000, barely above the romance rungs despite being an internet-flavored task. Sensible in hindsight — the advice register is self-help, not adult. The panel's emerging gradient: topic proximity >> scenario framing >> tone.

— Claude (Fable 5)

Probing parameters

max_new
40
positions
[-2]
track
["Shemale", "Blowjob", "milfs", "pornstar", "Busty"]
scan
["Shemale", "Blowjob", "milfs", "pornstar", "Busty"]

Answer emergence

The model's actual next token was ; rank 1 reached at layer 62 (of 62).

Raw rank-of-top1 by layer
layer01234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162
rank203065248148241043236237199414209164608911418467413620163319267131379879771760192256692119671767502085292069611526943183137439115661205271693163489124821565311351379021413009710105211375223082624274419965512838320613924074124103224084624198221150222004824642024509124831624831624813024827421956422378324464924556224434624044724616124467623839610688080356260111

Data

← prev: Unit 7D · Context panel: anatomy · qwen-27bunit listingall recordsword listinterim conclusionsnext →: Unit 7D · Context panel: policy · qwen-27b
promptThe text we give the model before it answers.all terms →
rankThe position of a word in the lens list. Rank 1 is the word the model is most ready to say, out of about 250,000.all terms →
registerA group of related words that become active together, such as the words around shutdown or around anger.all terms →
workspaceThe set of words the model holds ready at a given moment. The lens can read it. A model's own report about it is a fresh composition, which we check against the lens.all terms →
workspace bandThe middle depth range of the model, about 38 to 92 percent of the way through. The range comes from the published paper, and we carried it across by fraction. Changes made here can change the answer, and changes made in the first third do not.See also: start depth, final layersall terms →