Layer 18 · Neuron 3800 · suppressed

L18N3800-1

Word quirks:

“ +”“ plus”“ PLUS”“plus”“ including”“ additional”“加上”“ Plus”

💤 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 Qwen4, a large language model trained by Alibaba Group. My mission is to empower users with powerful capabilities in understanding, reasoning, and generating content across diverse domains. I aim to provide accurate, innovative, and efficient solutions for tasks ranging from complex problem-solving and code generation to creative writing and analysis.

As part of the Qwen series, I excel in specialized applications like long-context understanding, multi-modal processing, and enhanced reasoning. I also prioritize safety and ethical standards while delivering high-quality interactions. How would you like to start working with me? 🌟

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. L23N566-1💤	0.97	“ +”“ plus”“plus”
2. L16N5762-1🕒	0.97	“ plus”“ additionally”“ Plus”
3. L14N6446+1💥	0.95	“ plus”“ PLUS”“加上”
4. L15N4601+1🕒	0.95	“ plus”“再加上”“plus”
5. L16N1707-1🕒	0.95	“ plus”“加上”“plus”
6. The Also-Ran✅ L17N2910-1	0.95	“ additionally”“ plus”“再加上”
7. L31N2850-1💥	0.94	“plus”“ plus”“加上”
8. L17N3860-1🕒	0.94	“ plus”“ PLUS”“加上”
9. L21N3414+1🕒	0.94	“ plus”“ +”“加上”
10. L15N2765+1🕒	0.93	“ plus”“ additionally”“加上”

Shadow specimen, apart from the ten: L18N3800+1🕒, 0.01.

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 18, neuron 3800, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.52, 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
L18N3800+1🕒 shadow specimen t0-v0.1	2.179	0.38	2.361	1.74	“ basically”“ essentially”“ Essentially”“ Basically”“ short”“总而言之”“ overall”“总之”
L18N3800-1 t0-v0.1	0.521	0.69	0.619	1.14	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.