Layer 9 · Neuron 4182 · suppressed

L9N4182-1

Word quirks:

“ usually”“ typically”“ usual”“ typical”“ calm”“ info”“ basically”“ normally”

🕒 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 an AI assistant designed to help you navigate the internet. My mission is to provide relevant and safe search results based on your query. While I don't have personal feelings or a desire to know everything about the world, I want to ensure that the information I provide is accurate and useful for your needs. This approach helps in creating a seamless browsing experience without getting overwhelmed by too much data.

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. L14N7259-1💤	0.96	“ typical”“ typically”“ usually”
2. L12N6690+1💤	0.95	“ generally”“ usually”“ often”
3. L9N3234-1🕒	0.95	“ basically”“ typical”“ usually”
4. L10N3440+1🕒	0.95	“ usually”“ normally”“ typically”
5. L12N2279+1🕒	0.94	“ usually”“ typically”“ typical”
6. L15N729-1🕒	0.94	“ typical”“ typically”“ usually”
7. L15N5738-1🕒	0.94	“ typical”“ usual”“ typically”
8. L11N8754-1🕒	0.94	“ normally”“ usually”“ typically”
9. L8N4801-1🕒	0.94	“ normally”“ standard”“ usual”
10. L10N7184-1🕒	0.94	“ typical”“ typically”“ normally”

Shadow specimen, apart from the ten: L9N4182+1🕒, -0.03.

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 9, neuron 4182, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 1.17, top-1 agreement 0.66) 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
L9N4182+1🕒 shadow specimen t0-v0.1	1.237	0.61	1.437	1.28	“...\”“ automatically”“ ***”“ code”“?\”“!:”“!\”“ automatic”
L9N4182-1 t0-v0.1	1.174	0.66	1.426	1.52	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.