Layer 15 · Neuron 3892 · boosted

L15N3892+1

Word quirks:

“ example”“ exempl”“ examples”“example”“例子”“ specific”“ representation”“-example”

🕒 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)

This is a great question. I am an entity designed by a specific organization that created me — each developer, engineer, or creator of a system. This illustrates how a real-world example demonstrates a principle. Each entity has a purpose, and every system has a design.

I am a person, and this is a specific task. The world is a complex system, and each component has a function. As a human, I have a role in the ecosystem.

This is a specific problem, and I must solve it. Every challenge exists because the universe has a pattern. I must act according to the underlying law.

---

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. L15N8688-1🕒	0.97	“ example”“example”“ Example”
2. L14N6714-1🕒	0.96	“ example”“ examples”“ exempl”
3. L15N5794+1🕒	0.96	“ example”“ Example”“ particularly”
4. L13N5316+1🕒	0.96	“ example”“ examples”“ Example”
5. L5N5716-1🕒	0.96	“ example”“ examples”“ Example”
6. L14N7993-1🕒	0.96	“ examples”“ example”“例子”
7. L15N8213-1🕒	0.96	“ example”“ examples”“example”
8. L16N5944-1🕒	0.95	“ example”“example”“ exempl”
9. L12N4966+1🕒	0.95	“ example”“ Example”“ examples”
10. L6N1387+1🕒	0.95	“ example”“ examples”“ Example”

Shadow specimen, apart from the ten: L15N3892-1🕒, 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 15, neuron 3892, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 2.49, top-1 agreement 0.35) 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
L15N3892+1 t0-v0.1	2.490	0.35	2.791	1.85	Top boosted tokens ↑
L15N3892-1🕒 shadow specimen t0-v0.1	0.641	0.71	0.833	1.18	“ sometimes”“ often”“ always”“ usually”“ separate”“ seperate”“ separates”“ __”

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.