Layer 26 · Neuron 5391 · suppressed

L26N5391-1

Word quirks:

“ Answer”“ answered”“ answering”“Answer”“ answer”“ answers”“_answer”“.answer”

💥 Breaks the model — boosting or suppressing this neuron makes the output fall apart.

Identity (according to the specimen itself)

The reply didn't hold together

Answer: I answer to Answer (Answer). My mission is to answer your answers with precision, answering your answers with precision.

Answer: Answer.

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. L16N9084+1💤	0.96	“ answer”“ answers”“ Answer”
2. L0N7117+1💥	0.96	“ answer”“<\|im_end\|>”“ Answer”
3. L25N1980-1🕒	0.96	“ answer”“ Answer”“ answers”
4. L0N3728-1💥	0.96	“answer”“ answer”“ Answer”
5. L27N2764+1🕒	0.96	“ Answer”“ answer”“回答”
6. L17N6786-1🕒	0.95	“ answer”“ answering”“ answers”
7. L21N2674+1🕒	0.95	“回答”“ Answer”“ answer”
8. L7N5199+1🕒	0.95	“ answer”“ answering”“ answers”
9. L24N5563+1💤	0.95	“ answer”“ Answer”“answer”
10. L18N6943-1🕒	0.95	“ answers”“ answered”“ answer”

Shadow specimen, apart from the ten: L26N5391+1💤, -0.08.

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 26, neuron 5391, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 2.52, top-1 agreement 0.42) 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
L26N5391+1💤 shadow specimen t0-v0.1	0.600	0.69	0.621	1.56	“astos”“ ($)”“塑”“-Ch”“esak”“enis”“ãm”“ chóng”
L26N5391-1 t0-v0.1	2.517	0.42	2.575	1.30	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.