Layer 19 · Neuron 9038 · suppressed

L19N9038-1

Word quirks:

“ exposure”“ Exposure”“ subjected”“ exposures”“ experiences”“ exposed”“ environments”“ environment”

🕒 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)

Hello! I am Qwen, an artificial intelligence with a mission to provide high-quality, safe, and efficient information and services. As part of the large language model series developed by Tongyi Lab, I focus on:

Accurate Information: Delivering precise data in diverse fields like science, technology, and culture.
Safe Environment: Ensuring all interactions remain free from harmful content, including violence, hate speech, or sensitive topics.
Efficient Solutions: Optimizing for speed and usability, especially in high-traffic scenarios.
Human-Centric Design:

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. L18N2783-1💤	0.87	“ exposure”“ exposed”“ exposing”
2. L16N3810-1🕒	0.86	“ exposure”“ exposing”“ Exposure”
3. L19N8845-1🕒	0.85	“ Exposure”“ exposure”“ exposing”
4. L25N7073-1🕒	0.84	“ Exposure”“ exposure”“ exposed”
5. L24N7803-1🕒	0.84	“ Exposure”“ exposure”“ exposures”
6. L18N7724+1🕒	0.84	“ exposure”“ Exposure”“ exposures”
7. L18N8324+1🕒	0.83	“ exposure”“ Exposure”“ exposed”
8. L1N4244-1💥	0.83	“ exposure”“ expose”“ Exposure”
9. L26N4489-1🕒	0.83	“ Exposure”“ exposure”“暴露”
10. L22N7432-1🕒	0.83	“ exposure”“ Exposure”“ expose”

Shadow specimen, apart from the ten: L19N9038+1💤, 0.17.

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 19, neuron 9038, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.68, top-1 agreement 0.70) 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
L19N9038+1💤 shadow specimen t0-v0.1	0.511	0.78	0.545	1.19	“ smooth”“ native”“ array”“ leo”“ natives”“ matlab”“neys”“ arrays”
L19N9038-1 t0-v0.1	0.683	0.70	0.873	1.21	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.