Layer 23 · Neuron 9130 · boosted

L23N9130+1

Word quirks:

“ cage”“ ring”“ Ring”“-ring”“Ring”“ rings”“ radi”“_ring”

💤 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 Qin, a large language model created by the Kanguo Center (a fictional or hypothetical entity in this context; if you're referring to a real-world counterpart, please clarify). My mission is to assist users like you with tasks ranging from answering questions and creating content to providing tools for productivity and decision-making. I aim to be a reliable, ethical, and helpful partner in your work and life, ensuring all interactions are safe, accurate, and aligned with human needs. How can I support 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. L25N4603+1🕒	0.97	“ Ring”“ rings”“-ring”
2. L25N8895+1🕒	0.96	“ rings”“ ring”“ Rings”
3. Ring Cycle✅ L27N6004-1	0.95	“ ring”“ rings”“-ring”
4. L24N2401+1💤	0.95	“ ring”“Ring”“环”
5. L27N5321+1🕒	0.95	“ ring”“ring”“环”
6. L2N3777+1💥	0.95	“ ring”“ rings”“-ring”
7. L25N7767-1🕒	0.94	“ ring”“ rings”“ Ring”
8. L26N5578+1💤	0.94	“ rings”“ ring”“ Ring”
9. L27N5594+1🕒	0.93	“ ring”“-ring”“环”
10. L27N7450-1🕒	0.91	“ ring”“环”“ rings”

Shadow specimen, apart from the ten: L23N9130-1💤, 0.21.

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 23, neuron 9130, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.77, top-1 agreement 0.71) 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
L23N9130+1 t0-v0.1	0.774	0.71	0.853	1.31	Top boosted tokens ↑
L23N9130-1💤 shadow specimen t0-v0.1	0.737	0.71	0.833	1.29	“ room”“room”“ rooms”“ Rooms”“室内”“房间”“ Room”“-room”

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.