Layer 1 · Neuron 7927 · suppressed

L1N7927-1

Word quirks:

“ appeal”“ attractiveness”“ appealing”“ allure”“ Appeal”“ attractive”“吸引力”“ attraction”

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

Identity (according to the specimen itself)

What

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. L21N7260+1🕒	0.92	“ appeal”“吸引力”“ Appeal”
2. L21N4825+1🕒	0.91	“ appeal”“ appealing”“ attractive”
3. L20N7293+1🕒	0.90	“ attractiveness”“ smaller”“吸引力”
4. L27N2708+1🕒	0.87	“ attractive”“ attract”“吸引力”
5. L20N1559-1🕒	0.83	“ appeal”“ appeals”“ appealed”
6. L21N7636+1🕒	0.82	“ attractiveness”“ attract”“ attracting”
7. L27N2186+1🕒	0.82	“ attractive”“ nice”“ appealing”
8. L2N4099+1💥	0.82	“<\|im_end\|>”“ attraction”“ attract”
9. L13N4025+1🕒	0.81	“ irresistible”“ appeal”“ intrigued”
10. L22N7193-1🕒	0.81	“ appeal”“ Appeal”“ appealing”

Shadow specimen, apart from the ten: L1N7927+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 1, neuron 7927, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 6.49, top-1 agreement 0.07) 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
L1N7927+1💥 shadow specimen t0-v0.1	2.563	0.42	3.033	2.65	“<\|im_end\|>”“⏎”“!”“?!”“.`”“!!!”“!!”“.!”
L1N7927-1 t0-v0.1	6.494	0.07	7.080	3.21	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.