Layer 30 · Neuron 8956 · suppressed

L30N8956-1

Word quirks:

“一個”“一个”“的一个”“一个问题”“一个孩子”“一个大”“一款”“一个小”

🕒 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 Q一个 (Qwen3.5), a large一個一个 language model developed by Tongyi的一个一个实验室. My primary mission is to empower users with an enhanced conversational experience, an upgraded knowledge base, and a more efficient workflow. Whether you need help with complex reasoning, creative writing, code generation, or deep analysis, I strive to provide accurate,的一个一个一个 comprehensive, and的一个一个一个 helpful responses that align with your unique needs. 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. L22N3669+1💤	0.97	“一个大”“一个小”“一个”
2. L30N8124+1💥	0.95	“一个”“一个简单的”“一个叫”
3. L29N1552-1💤	0.95	“一个”“一个大”“一个孩子”
4. L29N48-1🕒	0.95	“一个”“一个小”“一种”
5. L28N864+1💤	0.94	“一个”“一项”“一个叫”
6. L21N1523+1💤	0.94	“ a”“一件”“一个”
7. L24N9002+1💤	0.93	“一个”“一个孩子”“一个人”
8. L26N778+1💤	0.93	“一个”“一项”“一个叫”
9. L25N1440+1🕒	0.93	“一个”“合”“一個”
10. L28N712-1🕒	0.93	“ a”“和一个”“一个大”

Shadow specimen, apart from the ten: L30N8956+1🕒, 0.70.

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 8956, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.64, top-1 agreement 0.74) 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
L30N8956+1🕒 shadow specimen t0-v0.1	0.435	0.75	0.475	1.58	“同一”“各”“lt”“两”“在同一”“gt”“每一”“庸”
L30N8956-1 t0-v0.1	0.636	0.74	0.637	1.81	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.