Layer 25 · Neuron 5384 · suppressed

L25N5384-1

Word quirks:

“-flight”“桥”“brig”“ flot”“连线”“凱”“�”“ Croce”

🕒 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 Kai, an AI assistant designed to provide helpful, accurate, and engaging support across a wide range of topics. My primary mission is to bridge the gap between human curiosity and information by offering:

Personalized assistance (from writing help and analysis to creative brainstorming)
Bridging knowledge gaps with clear, context-aware explanations
Enhancing efficiency through concise, actionable insights
Ensuring safety and ethics by adhering to strict guidelines for privacy, transparency, and responsible use

How can I help 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. L25N6108+1🕒	0.71	“帝”“شت”“astics”
2. L28N7355-1💤	0.71	“�”“ovich”“飛”
3. L25N8812+1🕒	0.71	“桥”“ bridges”“ bridge”
4. L22N2125-1💤	0.68	“ bridge”“桥”“标”
5. L24N6947-1💤	0.68	“悬挂”“姚”“飞的”
6. L22N8659-1💤	0.68	“橋”“uzz”“悬”
7. L24N8359-1🕒	0.68	“ kite”“风筝”“�”
8. L26N3338+1💤	0.68	“าว”“ бру”“pont”
9. L23N3794-1🕒	0.67	“海盗”“宾客”“ Bridge”
10. L28N7850-1💤	0.67	“夫”“芸”“ glossy”

Shadow specimen, apart from the ten: L25N5384+1🕒, 0.19.

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 25, neuron 5384, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.66, top-1 agreement 0.78) 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
L25N5384+1🕒 shadow specimen t0-v0.1	0.762	0.71	0.725	1.37	“ Self”“容”“.react”“React”“(Self”“帖”“ self”“Self”
L25N5384-1 t0-v0.1	0.657	0.78	0.715	1.26	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.