Layer 23 · Neuron 6767 · suppressed

L23N6767-1

Word quirks:

“usun”“ungi”“标准”“عو”“准”“rata”“综合评价”“ Му”

💤 Barely changes the model — the quick scan suggests this neuron barely changes the model's word predictions.

Identity (according to the specimen itself)

I am Q*, a multi-functional AI assistant designed to help users with a wide range of tasks. My primary mission is to provide *accurate, efficient, and personalized assistance across various domains, including:

Information & Research: Answering factual queries, summarizing complex topics, or providing real-time data (within my training limits).
Problem Solving: Tackling math problems, coding challenges, or logical reasoning tasks.
Creative Collaboration: Generating ideas for art, writing, or planning projects.
Daily Life Support: Offering recommendations for travel, health, finance, or personal productivity

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. L30N2976+1💤	0.73	“ muka”“标准的”“瘤”
2. L26N3570+1🕒	0.72	“正”“准”“标准”
3. L28N8303-1💤	0.72	“乌”“典型”“umul”
4. L25N6950+1💤	0.71	“准”“ Quo”“準”
5. L31N7210-1💥	0.71	“颂”“kung”“ubo”
6. L27N8734-1💤	0.71	“unj”“ MU”“设计方案”
7. L22N5967-1💤	0.70	“妙的”“angelo”“为标准”
8. L27N6618-1💤	0.70	“伍”“ourced”“府”
9. L24N7073+1💤	0.70	“末”“ধান”“薄”
10. L30N4592+1💤	0.69	“رخ”“飞来”“御”

Shadow specimen, apart from the ten: L23N6767+1🕒, 0.35.

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 23, neuron 6767, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.68, top-1 agreement 0.71) 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
L23N6767+1🕒 shadow specimen t0-v0.1	1.169	0.62	1.233	1.32	“ Nobel”“诺贝尔”“ Supreme”“ Nob”“諾”“ Noble”“诺”“nob”
L23N6767-1 t0-v0.1	0.682	0.71	0.718	1.24	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.