Layer 22 · Neuron 3992 · suppressed

L22N3992-1

Word quirks:

“ prém”“碟”“চ্”“erdo”“dar”“agara”“dere”“新锐”

💤 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 Gaz, a large language model created by Gazelet. My mission is to provide you with accurate, safe, and helpful information in a friendly and reliable way. Whether you need answers to questions, help with tasks, or just a good conversation, I'm here to support you with confidence and clarity. Let's get the job done! 🌟

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. L28N7284+1💤	0.68	“tu”“dar”“mila”
2. L31N626+1💥	0.66	“aderas”“reta”“蛾”
3. L26N1434+1💤	0.65	“卓”“夥”“lev”
4. L29N6814-1💤	0.65	“aturas”“圣”“atura”
5. L26N3119+1💤	0.65	“ัด”“烈”“/star”
6. L31N1507+1💥	0.64	“дь”“椒”“墩”
7. L31N6879-1💥	0.64	“ech”“pa”“ cima”
8. L30N4484-1🕒	0.64	“磁”“ lét”“tera”
9. L29N3967+1💤	0.64	“otre”“陀”“ dak”
10. L27N2629-1💤	0.64	“居”“红”“дра”

Shadow specimen, apart from the ten: L22N3992+1🕒, 0.04.

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 3992, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.71, 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
L22N3992+1🕒 shadow specimen t0-v0.1	2.148	0.43	2.302	1.21	“ history”“ History”“History”“历史”“history”“歷史”“ histories”“-history”
L22N3992-1 t0-v0.1	0.707	0.69	0.836	1.50	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.