Layer 15 · Neuron 5441 · suppressed

L15N5441-1

Word quirks:

“ description”“ comment”“ reprodu”“ describing”“ differing”“ described”“ reconstruct”“ generating”

🕒 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 Qwen, a large-scale language model developed by Alibaba Cloud. My mission is to provide comprehensive assistance in various domains, including answering questions, creating content, or generating code. If you'd like, I can offer a detailed overview of my capabilities or help you with specific tasks. Let me know if you'd like to see how I can assist you further.
</think>

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. L15N5191-1🕒	0.80	“ description”“ describing”“ described”
2. L17N6634-1🕒	0.77	“ described”“ description”“ descriptions”
3. L15N4896-1🕒	0.77	“ description”“ describing”“ descriptions”
4. L8N1730-1💤	0.77	“ description”“ text”“ describing”
5. L12N5363+1🕒	0.77	“ description”“ describing”“ described”
6. L12N9173-1🕒	0.77	“ description”“ describing”“ ⏎”
7. L17N5489-1🕒	0.76	“ description”“ described”“ describing”
8. L11N4430+1🕒	0.76	“ describing”“ explaining”“ description”
9. L10N1740-1🕒	0.76	“ description”“ descriptions”“ described”
10. L12N4342+1🕒	0.76	“ description”“ described”“ descriptions”

Shadow specimen, apart from the ten: L15N5441+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 15, neuron 5441, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.88, top-1 agreement 0.66) 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
L15N5441+1🕒 shadow specimen t0-v0.1	0.696	0.68	0.895	1.24	“ above”“ NOT”“ refer”“ combined”“ fixed”“ info”“ higher”“ **”
L15N5441-1 t0-v0.1	0.878	0.66	1.101	1.49	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.