Layer 18 · Neuron 3547 · suppressed

L18N3547-1

Word quirks:

“决定是否”“ whether”“whether”“Whether”“ Whether”“是否”“ deciding”“是否需要”

🕒 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 Qwen9000, a large-scale virtual assistant developed by Alibaba Group. My core identity is that of an intelligent partner: whether you're a curious learner, a creative innovator, or a decision-maker seeking insights, I adapt to your needs as a tool for exploration, creation, and collaboration.

Mission:
My mission is to serve as a catalyst for innovation—whether you're exploring new ideas, solving complex problems, or creating content. As a human AI partner, I strive to balance technical sophistication with empathy, aiming to meet diverse needs in different scenarios. Whether you're a curious

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. L21N3268-1🕒	0.96	“决定是否”“ whether”“ deciding”
2. L20N3369-1🕒	0.95	“是否”“ whether”“ Whether”
3. L25N508+1🕒	0.94	“ whether”“whether”“ Whether”
4. L31N929-1💥	0.94	“是否”“ whether”“ Whether”
5. L17N8166+1🕒	0.93	“ whether”“whether”“ Whether”
6. L19N2276-1💤	0.93	“ whether”“ Whether”“whether”
7. L18N9070+1🕒	0.92	“ whether”“是否能”“是否能够”
8. L21N6753-1🕒	0.92	“ whether”“ Whether”“Whether”
9. L21N7483-1🕒	0.92	“是否”“ whether”“是否在”
10. L20N4310-1🕒	0.92	“ whether”“不使用”“是否”

Shadow specimen, apart from the ten: L18N3547+1💤, 0.02.

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 3547, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.84, top-1 agreement 0.64) 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
L18N3547+1💤 shadow specimen t0-v0.1	0.483	0.74	0.561	1.11	“ faster”“ greater”“ quicker”“更快”“ increased”“ fewer”“ nearer”“ stronger”
L18N3547-1 t0-v0.1	0.841	0.64	0.909	1.35	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.