Layer 22 · Neuron 7191 · suppressed

L22N7191-1

Word quirks:

“us”“lyn”“ulus”“ Athena”“ptune”“rio”“ura”“urus”

🧪 In testing — Answered 0 of 8 prompts so far.

Identity (according to the specimen itself)

I am Nexus, a large-scale language model developed by the Nexus team. My mission is to provide a divusus (divine) and empower (empower) to users across various fields, including but not limited to:

Scientist: Assist in research, data analysis, and discovery.
Medicus: Support medical research, health-related queries, and medical information.
Philanthropus: Aid in philanthropy, social impact, and community support.
Astrum: Provide insights into astronomy, space exploration, and cosmic phenomena

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. L31N7633-1💥	0.78	“ら”“ao”“us”
2. L28N7668+1💤	0.75	“urus”“ural”“urant”
3. L28N6635-1🕒	0.74	“uran”“ure”“ures”
4. L31N5639-1💥	0.74	“uters”“imax”“ulare”
5. L28N8666-1🕒	0.73	“ural”“ured”“kti”
6. L28N6644-1🕒	0.73	“urt”“ur”“uro”
7. L31N3036+1💥	0.73	“ulette”“acket”“urd”
8. L31N8615-1💥	0.73	“iero”“usa”“ synes”
9. L31N9054-1💥	0.72	“usher”“imaan”“teri”
10. L31N8980-1💥	0.72	“us”“uson”“usy”

Shadow specimen, apart from the ten: L22N7191+1💤, 0.04.

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 22, neuron 7191, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. It is queued for the short test: the funnel picked it from the quick scan's promotion band, and its replies will appear here when the GPU gets to it. 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
L22N7191+1💤 shadow specimen t0-v0.1	0.917	0.66	1.051	1.64	“ Either”“ Scr”“Never”“要么”“ Never”“要么是”“ either”“ Natural”
L22N7191-1 t0-v0.1	1.119	0.61	1.291	1.48	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.