Layer 17 · Neuron 8080 · suppressed

L17N8080-1

Word quirks:

“ cudd”“ companion”“ compat”“抱团”“ comfort”“ together”“ ⏎”“ hugs”

🕒 Waiting in line — so far it has only had the quick scan, which shows which words it pushes. It was not selected in this round.

Identity (according to the specimen itself)

I am Qwen3.5, the latest version of the Qwen series, powered by Alibaba Cloud's Tongyi Lab. My primary mission is to be your companion for knowledge and creativity, offering a stable, intelligent, and efficient partner for all things. Here’s how we align:

Knowledge & Logic: I provide accurate, context-aware insights, helping you solve complex problems with clarity.
Creativity: Let’s collaborate on ideas—whether it’s writing a story, composing music, or brainstorming a project.
Efficiency: I’m your go-to for multitasking, from

Nearest neighbors

Specimens with similar replies or word quirks. Similarity runs from −1 to 1; 1 is identical. These are leads, not shared mechanisms. How profiles are built.

By word quirks Profile recipe

Specimen	Similarity	Word quirks
1. L4N8016-1💥	0.73	“ cudd”“ hugs”“ hug”
2. L1N6190-1💥	0.64	“ companion”“ Companion”“<\|im_end\|>”
3. L9N6841+1🕒	0.61	“ morning”“ sleep”“ comfort”
4. L12N1646-1🕒	0.61	“ combined”“ together”“ complementary”
5. L2N6058-1💥	0.61	“ companion”“<\|im_end\|>”“ companions”
6. L7N5905-1🕒	0.61	“ hugs”“ hug”“ cudd”
7. L19N8062+1🕒	0.60	“陪伴”“ companion”“伴随”
8. L1N6395-1💥	0.60	“ embrace”“ hug”“<\|im_end\|>”
9. L10N8676-1🕒	0.59	“ comfort”“ comfy”“ comfortable”
10. L5N8330+1🕒	0.59	“ companions”“ friends”“ friendship”

Shadow specimen, apart from the ten: L17N8080+1🕒, 0.08.

Cousins · specimens sharing a neuron

No specimen with replies shares this neuron yet.

Technical details

Quick scan: measure prediction shifts on fixed reference text. Short test: generate replies to a small set of prompts. Longer test: extend to a larger set. Full test: run the full prompt suite. A pass admits a specimen to the next stage; only missing replies are generated.

Layer 17, neuron 8080, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.65, top-1 agreement 0.69) is outside the band that feeds the short test (KL 1 to 2.5, agreement 0.5 to 0.65), so it stays at the quick scan unless someone queues it by hand. A hand pick goes straight to the full test. See Method for the tiers.

Quick scan · prediction measurements

Measurements are averaged over token positions of the five unsteered reference answers, using the same fixed text under both conditions; no steered replies are generated here.

KL divergence: KL(P_unsteered ‖ P_steered), in nats — how far steering shifts the next-token distribution in this direction (Kullback & Leibler, 1951, On Information and Sufficiency). Unsteered ≈ 0.001.
Top-1 agreement: fraction of positions with the same highest-probability token — how often the top prediction stays unchanged.
ΔNLL: mean negative log-likelihood under steering minus the unsteered value for the reference tokens, in nats — positive means the original answer becomes less likely.
Entropy: mean H(P_steered) = −Σ p log p, in nats — how spread out the steered next-token predictions are.

The steer lens lists the tokens with the largest mean change in log-probability (steered minus unsteered), in nats, over those positions. These are leads for inspection; in rare specimens even the top gains are negative.

Setting	KL divergence	Top-1 agreement	ΔNLL	Entropy	Steer lens
L17N8080+1🕒 shadow specimen t0-v0.1	0.795	0.69	0.855	1.22	“ obligation”“ mandated”“ responsible”“ responsibility”“ duty”“ obligations”“ obligated”“ mandates”
L17N8080-1 t0-v0.1	0.647	0.69	0.845	1.31	Top boosted tokens ↑

Method

A specimen is a fixed steering configuration of Qwen3.5-4B: one MLP neuron or a pair, each with a signed dose. Unsteered is the control. L identifies the layer; N, the neuron.

Additive activation steering. At every token position, add dose × layer median residual norm ÷ ‖W_down[:, n]‖ to MLP neuron n at the down_proj input. In plain terms, shift that neuron's activation by a fixed amount, scaled by the layer’s calibrated median residual norm and the neuron's down-projection weight norm. Positive doses push up; negative doses push down. Dose is normalized intervention size, not a percentage or probability.