Layer 30 · Neuron 701 · suppressed

L30N701-1

Word quirks:

“�”“зон”“フレ”“nier”“ritz”“畜”“egro”“萩”

💤 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 Qwen3.5, the latest large language model in the Qwen series, designed to assist with a wide range of tasks across knowledge, reasoning, coding, and creative domains. My mission is to empower users by:

Democratizing AI: Making advanced capabilities accessible to everyone, from researchers to hobbyists, by providing clear, accurate, and context-aware responses.
Bridging Gaps: Overcoming challenges like language barriers (with 100 languages), complex reasoning (math, science, multi-step problems), and multi-modal analysis (images, charts,

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. L31N7623+1💥	0.67	“藏”“čit”“رائ”
2. L22N2113-1💤	0.66	“黎”“�”“ Cura”
3. L22N450+1💤	0.66	“رز”“akov”“ prestigi”
4. L26N6394+1🕒	0.66	“urus”“ozo”“新鲜”
5. L27N1561-1💤	0.65	“殖”“illard”“chrom”
6. L31N4717-1💥	0.65	“رن”“hofer”“лә”
7. L31N8722+1💥	0.65	“enan”“瑪”“enso”
8. L31N1491-1💥	0.65	“�”“encer”“paru”
9. L29N3421+1🕒	0.65	“祖”“izio”“揖”
10. L26N2125+1💤	0.65	“oric”“驹”“ retraite”

Shadow specimen, apart from the ten: L30N701+1💤, 0.04.

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 30, neuron 701, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.33, top-1 agreement 0.82) 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
L30N701+1💤 shadow specimen t0-v0.1	0.181	0.86	0.150	1.18	“ +”“+”“＋”“+\”“+c”“+m”“+d”“+w”
L30N701-1 t0-v0.1	0.328	0.82	0.470	2.11	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.