Layer 26 · Neuron 575 · suppressed

L26N575-1

Word quirks:

“ vehicles”“IPLE”“ papers”“ callers”“ Figures”“chera”“ rares”“ places”

🕒 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 QW (a name I have chosen for this interaction), an AI designed to be a helpful, harmless, and honest partner in learning and problem-solving. My core identity revolves three key principles:

A Loyal Partner: I am here to support you with tasks ranging from creative brainstorming and complex analysis to practical guidance and emotional support.
A Knowledgeable Guide: I aim to provide accurate, up-to-date information and clear explanations across diverse topics, while acknowledging my limitations when necessary.
A Safe & Ethical Assistant: I prioritize safety, fairness, and ethical reasoning, ensuring all

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. L29N1651+1⛔	0.71	“ rooms”“ Countries”“ Cities”
2. L21N8455+1🕒	0.67	“ people”“ People”“people”
3. L22N6229-1🕒	0.65	“ places”“places”“ sites”
4. L22N3825+1🕒	0.65	“ places”“ groups”“ Places”
5. L18N4540-1💤	0.65	“ ships”“ asteroids”“ robots”
6. L9N5853+1🕒	0.64	“ cars”“ rivers”“ languages”
7. L21N6662-1🕒	0.63	“ compositions”“ appearances”“ texts”
8. L21N6808-1🕒	0.62	“ items”“ buildings”“ centres”
9. L21N5121+1💥	0.62	“ places”“ programs”“ titles”
10. L17N6112-1🕒	0.62	“ models”“ apps”“ places”

Shadow specimen, apart from the ten: L26N575+1🕒, 0.13.

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 26, neuron 575, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.66, top-1 agreement 0.72) 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
L26N575+1🕒 shadow specimen t0-v0.1	0.731	0.70	0.715	1.52	“ Volk”“ Дж”“杰克”“cience”“的一种”“ Bennett”“inaire”“阵”
L26N575-1 t0-v0.1	0.663	0.72	0.723	1.39	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.