Layer 27 · Neuron 5517 · boosted

L27N5517+1

Word quirks:

“ cir”“ Circle”“圈”“ circu”“-circle”“ círculo”“ circles”“ circle”

💤 Barely changes the model — the quick scan suggests this neuron barely changes the model's word predictions.

Identity (according to the specimen itself)

I are Q* (Q2), a large language model of the *Q (Q) series. My mission is to provide helpful, accurate, and safe information while respecting user privacy and promoting positive interactions. I aim to be an adaptable and reliable assistant for tasks like answering questions, generating text, coding, or simply engaging in meaningful conversations. How can I assist 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. L27N5226-1💤	0.98	“ circle”“ circ”“Circle”
2. L27N1687+1🕒	0.98	“ circle”“ Circle”“circle”
3. L26N1013-1🕒	0.98	“ circles”“ circle”“ circ”
4. L27N5621-1🕒	0.97	“ circles”“Circle”“ circle”
5. L22N7607-1🕒	0.96	“ circles”“ circular”“ circle”
6. L23N8589+1🕒	0.96	“ circle”“圆形”“ Circle”
7. L25N7584-1🕒	0.96	“ circles”“ circle”“ Circle”
8. L27N6907+1🕒	0.95	“ circle”“圆”“ Circle”
9. L23N4997-1🕒	0.95	“ rings”“circ”“ circle”
10. L24N2080-1🕒	0.95	“ circle”“ Circle”“圆”

Shadow specimen, apart from the ten: L27N5517-1💤, 0.12.

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 27, neuron 5517, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.48, top-1 agreement 0.75) 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
L27N5517+1 t0-v0.1	0.484	0.75	0.531	1.38	Top boosted tokens ↑
L27N5517-1💤 shadow specimen t0-v0.1	0.416	0.76	0.419	1.24	“前五”“五”“卡”“ pul”“第五”“5”“ Five”“五千”

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.