Layer 19 · Neuron 4987 · suppressed

L19N4987-1

Word quirks:

“ neighbor”“ competitor”“ Competition”“ competition”“ neighbour”“ broadcaster”“ grown”“邻”

🕒 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 Qun, a large-scale artificial intelligence model developed by Huawei. My mission is to provide robust support in various fields, including but not limited to:

Technology: Offering insights on emerging tech trends, helping users navigate digital transformations, and assisting with technical problem-solving.
Healthcare: Providing health-related information (e.g., disease prevention, medication guidance) while emphasizing the need for professional medical consultation.
Education: Supporting learning through explanations, study advice, and resource recommendations.
Daily Life: Assisting with practical tasks like travel planning, shopping recommendations,

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. The Fraying Draft✅ L18N3710+1	0.82	“ neighbouring”“⏎”“ competit”
2. L22N3543+1💤	0.76	“ competing”“ neighboring”“ Compet”
3. L9N6179+1🕒	0.73	“ competitor”“ competition”“ competitors”
4. L6N8842+1🕒	0.73	“ competition”“ competing”“ competitors”
5. L4N5293-1🕒	0.73	“ competition”“ competitions”“ competitors”
6. L13N8810+1🕒	0.73	“ competition”“ competing”“ competitors”
7. L22N8771+1🕒	0.73	“ Compet”“竞争”“Compet”
8. L21N4405-1🕒	0.73	“ competitors”“竞争对手”“ competitor”
9. L5N4785-1🕒	0.72	“ competitors”“ competitor”“ competition”
10. L19N4703+1💤	0.72	“ basic”“ competitor”“ competition”

Shadow specimen, apart from the ten: L19N4987+1🕒, 0.28.

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 4987, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.69, top-1 agreement 0.68) 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
L19N4987+1🕒 shadow specimen t0-v0.1	1.067	0.64	1.180	1.32	“更少”“ promotions”“ roles”“ positions”“awat”“ Roles”“职位”“ promotion”
L19N4987-1 t0-v0.1	0.685	0.68	0.809	1.28	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.