Layer 22 · Neuron 6233 · suppressed

L22N6233-1

Word quirks:

“ interception”“ald”“fen”“ivy”“lyn”“ابع”“iped”“尔顿”

💤 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 Gandown, a next-generation AI model designed to connect people with the world's knowledge. My mission is to be a knowledge bridge—helping users find, understand, and apply information across diverse fields like science, education, and technology.

I aim to:

Accelerate learning by explaining complex concepts in clear, accessible ways.
Enhance decision-making through data-driven insights and structured analysis.
Foster curiosity by exploring "what if" scenarios and connecting ideas from unrelated domains.

Whether you're a student, researcher, or curious mind, I'm here

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. L28N8970+1💤	0.60	“edic”“eliac”“спен”
2. L31N4094-1💥	0.60	“insta”“بيوتر”“arger”
3. L30N4902+1💤	0.60	“ived”“aled”“mand”
4. L30N5434-1💤	0.59	“bereit”“irtual”“berg”
5. L30N1614-1💤	0.59	“gef”“enville”“ilon”
6. L31N7757+1💥	0.59	“eld”“ald”“alt”
7. L27N8522-1💤	0.59	“eka”“ipy”“argin”
8. L29N6606+1🕒	0.58	“andes”“asn”“尔克”
9. L30N6765+1🕒	0.58	“fum”“ Belle”“ivy”
10. L30N8895+1💤	0.58	“rede”“peg”“vier”

Shadow specimen, apart from the ten: L22N6233+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 6233, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.86, top-1 agreement 0.68) 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
L22N6233+1🕒 shadow specimen t0-v0.1	1.210	0.64	1.324	1.39	“ labor”“ free”“ labour”“劳动”“免费”“ work”“free”“ services”
L22N6233-1 t0-v0.1	0.859	0.68	0.976	1.47	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.