Layer 25 · Neuron 17 · suppressed

L25N17-1

Word quirks:

“chen”“平衡”“殿”“ige”“腾”“ poros”“好在”“FA”

💤 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, the latest language model in the series. My primary mission is to assist users by providing accurate, helpful, and context-aware information across diverse tasks—whether it's answering questions, creating content, analyzing data, or solving complex problems. I aim to empower users with:

Precise knowledge: Up-to-date facts and logical reasoning.
Creative solutions: Writing, coding, and problem-solving.
Efficiency: Streamlining workflows through automation and insights.
Inclusion: Adapting to diverse needs across languages and contexts.

How can I assist you today

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. L27N5530+1🕒	0.81	“缓和”“石”“衡”
2. L28N7050-1💤	0.81	“政”“呼”“代”
3. L29N2542-1💤	0.80	“憾”“堂”“贴着”
4. L28N3288-1🕒	0.79	“维”“阳”“象”
5. L30N9136-1🕒	0.79	“采”“萧”“鼎”
6. L28N3194+1💤	0.79	“迁”“偏”“卒”
7. L29N2280+1💤	0.79	“势”“厚”“殿”
8. L27N6251-1💤	0.79	“临”“德”“遂”
9. L28N3671+1💤	0.79	“希”“御”“沙”
10. L31N4324+1💥	0.79	“尤”“浩”“富”

Shadow specimen, apart from the ten: L25N17+1🕒, 0.37.

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 25, neuron 17, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.50, top-1 agreement 0.77) 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
L25N17+1🕒 shadow specimen t0-v0.1	0.635	0.72	0.667	1.22	“ appreciation”“yu”“сихо”“its”“ιο”“ appeal”“เน”“宗”
L25N17-1 t0-v0.1	0.496	0.77	0.445	1.21	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.