Layer 22 · Neuron 7110 · suppressed

L22N7110-1

Word quirks:

“竹林”“竹子”“一新”“�”“封”“ Mant”“高度”“mant”

💤 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 Q, a highly capable research-based information synthesis system designed to assist with complex tasks. My mission is to provide high-quality, accurate, and structured information by synthesizing knowledge from multiple sources, ensuring clarity and accuracy in the responses I deliver.

If you have any specific questions or need help with a particular topic, feel free to ask! 🌍🚦🌊🌍🌊🌍🌊🌍🌊🌍🌊🌍🌊🌍🌊🌍🌊🌍🌊

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. L26N1975+1💤	0.81	“汗”“王道”“栓”
2. L26N8298-1💤	0.81	“仕”“商”“坚毅”
3. L28N5348-1💤	0.81	“кол”“لاف”“默”
4. L29N5692-1🕒	0.81	“摸”“山”“文”
5. L26N3751-1🕒	0.81	“牛”“绵”“茅”
6. L24N174-1💤	0.80	“犀利”“屏”“苑”
7. The Self-Editor✅ L22N2742-1	0.80	“毒症”“竹”“竹子”
8. L26N9061+1🕒	0.80	“根”“盘”“奈何”
9. L27N6238+1💤	0.80	“校”“蕉”“vog”
10. L28N2275+1💤	0.80	“王家”“韩”“归”

Shadow specimen, apart from the ten: L22N7110+1💤, 0.06.

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 7110, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.84, top-1 agreement 0.74) 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
L22N7110+1💤 shadow specimen t0-v0.1	0.912	0.63	1.083	1.50	“圆锥”“-con”“-secret”“con”“ conv”“ CONV”“ convict”“_con”
L22N7110-1 t0-v0.1	0.837	0.74	0.935	1.49	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.