Layer 16 · Neuron 8891 · suppressed

L16N8891-1

Word quirks:

“ specifically”“ isn”“ particularly”“ especially”“ type”“ ____”“ __”“ doesn”

🧪 In testing — Answered 0 of 8 prompts so far.

Identity (according to the specimen itself)

I am Qwen, and I am a large language model trained on data, specifically designed to assist users in various fields. My name is Qwen, and I am a large language model trained on data, specifically designed to assist users in various fields.

My mission is to provide high-quality information, particularly in the context of education, work, and life. I aim to help people make decisions, solve problems, and find answers through my expertise in data analysis, pattern recognition, and decision-making.

If you're interested in learning more about myself, feel free to ask!

---

Why do I care?

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. L17N462-1💤	0.75	“ certain”“ particular”“ specific”
2. L16N7504+1💤	0.73	“ “”“ especially”“ specifically”
3. L8N7896-1💤	0.73	“ specific”“ **”“ typically”
4. L12N6301-1💤	0.72	“ specific”“ specifics”“ __”
5. L11N8403-1🕒	0.71	“ particular”“ especially”“ __”
6. L17N3781+1💤	0.71	“ specifically”“ ____”“ _”
7. L11N1429-1💤	0.70	“ specifically”“ *”“⏎”
8. L10N5121+1💤	0.70	“ specific”“ particular”“ ______”
9. L16N4336-1🕒	0.69	“ basically”“ ‘”“ specifically”
10. L16N4756-1💤	0.69	“ actually”“ basically”“ “”

Shadow specimen, apart from the ten: L16N8891+1🕒, 0.40.

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 16, neuron 8891, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. It is queued for the short test: the funnel picked it from the quick scan's promotion band, and its replies will appear here when the GPU gets to it. 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
L16N8891+1🕒 shadow specimen t0-v0.1	0.665	0.67	0.833	1.42	“ specific”“ specifying”“ specified”“ “”“ renamed”“ suitable”“ ‘”“ aligned”
L16N8891-1 t0-v0.1	1.120	0.64	1.330	1.26	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.