Layer 31 · Neuron 7089 · boosted

L31N7089+1

Word quirks:

“att”“戏”“要求被告”“allow”“恶”“chetti”“itter”“ao”

💥 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

attattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattattatt

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. L29N171-1🕒	0.64	“atts”“添加”“ikko”
2. L30N8586-1💤	0.62	“要让”“oksi”“床”
3. The Imperial Loop✅ L25N4629-1,L31N585-1	0.61	“皇家”“ chiếu”“åt”
4. L28N5096-1🕒	0.61	“rees”“赘”“摸”
5. L31N6838-1💥	0.60	“果”“受”“azzo”
6. L28N3867+1🕒	0.60	“拷”“osto”“二路”
7. L29N7589+1🕒	0.60	“走”“诉”“柿”
8. L30N4408-1💤	0.59	“acy”“gie”“泥”
9. L24N1342-1💤	0.59	“chera”“淫”“acher”
10. L31N7737-1💥	0.59	“atura”“loser”“臭”

Shadow specimen, apart from the ten: L31N7089-1💥, 0.41.

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 31, neuron 7089, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 5.54, top-1 agreement 0.06) 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
L31N7089+1 t0-v0.1	5.542	0.06	5.811	9.43	Top boosted tokens ↑
L31N7089-1💥 shadow specimen t0-v0.1	5.741	0.09	6.054	9.74	“中规”“orz”“jom”“robi”“orro”“rott”“hov”“jemu”

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.