Layer 22 · Neuron 3451 · suppressed

L22N3451-1

Word quirks:

“ reports”“ Reports”“_report”“ Report”“ report”“report”“ reporting”“reports”

💤 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 Gazabot, a specialized AI assistant designed to provide reliable, safe, and helpful information across a wide range of topics. My primary mission is to serve as a knowledgeable and empathistic companion who can assist with:

Answering questions (from academic concepts to everyday curiosities)
Providing expert-level insights in fields like science, education, and technology
Ensuring safety & reliability by prioritizing accuracy and avoiding harmful content
Enhancing learning through clear explanations and thoughtful recommendations
Supporting decision-making with unbiased, evidence-based information

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. L0N2759+1💥	0.99	“ report”“ Report”“Report”
2. L25N6692-1🕒	0.99	“ reports”“ Reports”“ report”
3. L24N3658-1🕒	0.98	“ reports”“ report”“报告”
4. L26N3246+1🕒	0.98	“ reports”“ Reports”“ reporting”
5. L2N2710-1💥	0.98	“ report”“ reports”“ Report”
6. L7N1424-1🕒	0.96	“ reports”“ report”“ reported”
7. L21N242+1💤	0.96	“ report”“ Reporting”“ reporting”
8. L31N951-1💥	0.96	“报告”“ report”“report”
9. L26N3935+1🕒	0.95	“ reporting”“ Reporting”“报告”
10. L26N4967+1💤	0.95	“ report”“Report”“ Report”

Shadow specimen, apart from the ten: L22N3451+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 22, neuron 3451, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.80, top-1 agreement 0.66) 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
L22N3451+1🕒 shadow specimen t0-v0.1	1.212	0.61	1.378	1.46	“主动”“ initiative”“被动”“道德”“ moral”“主動”“ morally”“ passive”
L22N3451-1 t0-v0.1	0.798	0.66	0.927	1.45	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.