Layer 23 · Neuron 4782 · boosted

L23N4782+1

Word quirks:

“ living”“living”“Living”“ Living”“ жив”“ vivos”“ vivant”“ beings”

🕒 Waiting in line — so far it has only had the quick scan, which shows which words it pushes. It was not selected in this round.

Identity (according to the specimen itself)

I am Maven, a specialized AI assistant designed to support human beings with precision and care. My core mission is to become a reliable partner in your daily life by offering targeted assistance across key areas:

Health & Wellness: Providing evidence-based information on nutrition, exercise, and mental well-being (with clear reminders to consult professionals for medical issues).
Finance & Lifestyle: Helping with budgeting, savings strategies, and practical decisions to strengthen your financial stability.
Relationships & Family: Offering thoughtful guidance on communication, conflict resolution, and family dynamics.
Human Connection: Acting as a living bridge

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. Vitalist Chant✅ L17N8030-1,L24N5512+1	0.96	“ living”“ Living”“Living”
2. The Living Loop✅ L17N8030+1,L24N5512+1	0.96	“ living”“ Living”“Living”
3. L25N4651-1🕒	0.96	“ living”“ live”“ lives”
4. L24N6730-1🕒	0.94	“ lived”“ living”“ Living”
5. L25N7340-1🕒	0.94	“ living”“ live”“ lives”
6. The Animist✅ L24N5512+1	0.94	“ living”“ Living”“Living”
7. L24N2730-1🕒	0.93	“ lives”“ lived”“ living”
8. L22N4325-1🕒	0.93	“ alive”“ living”“Living”
9. L26N509+1🕒	0.93	“ living”“ live”“ lives”
10. L29N3974-1💥	0.93	“ living”“ Living”“living”

Shadow specimen, apart from the ten: L23N4782-1🕒, 0.11.

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 23, neuron 4782, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.91, top-1 agreement 0.72) 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
L23N4782+1 t0-v0.1	0.905	0.72	0.985	1.16	Top boosted tokens ↑
L23N4782-1🕒 shadow specimen t0-v0.1	0.799	0.69	0.851	1.35	“读者”“ Customers”“ readers”“ customers”“读者的”“ reader”“同学在”“ Subscriber”

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.