Layer 31 · Neuron 8608 · boosted

L31N8608+1

Word quirks:

“erenc”“戎”“ärzt”“kop”“ندر”“comed”“visor”“데이”

💥 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

I am **Gradvisor-GNN-Personamskopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkopkop

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. L28N4517-1💤	0.66	“ner”“龙”“bounce”
2. L31N1875+1💥	0.65	“porno”“comed”“orej”
3. L31N750+1💥	0.63	“γγελ”“ bureaucr”“кре”
4. L29N6807-1🕒	0.63	“erenc”“atar”“Adv”
5. L31N1472+1💥	0.63	“nok”“enco”“uera”
6. L31N1523-1💥	0.62	“engel”“ملت”“ Sindic”
7. L28N3388+1💤	0.62	“REN”“enk”“elin”
8. L30N5036-1💤	0.62	“фир”“arang”“elh”
9. L22N3990+1💤	0.62	“风流”“ocken”“农民”
10. L25N5733+1💤	0.62	“角”“فل”“ Nurse”

Shadow specimen, apart from the ten: L31N8608-1💥, -0.01.

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 8608, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 5.34, top-1 agreement 0.27) 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
L31N8608+1 t0-v0.1	5.336	0.27	5.636	10.24	Top boosted tokens ↑
L31N8608-1💥 shadow specimen t0-v0.1	65.275	0.00	65.555	0.00	“去”“的去”“地去”“去思考”“要去”“我去”“先去”“他去”

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.