Layer 16 · Neuron 9095 · boosted

L16N9095+1

Word quirks:

“ description”“ describing”“ descriptions”“ explanation”“ explaining”“ describe”“ explanations”“ describes”

🕒 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 language model designed to assist with a wide range of tasks. My primary goal is to provide clear, accurate, and helpful information about my capabilities, covering:

My identity: I introduce my core functions in a structured way.
My mission: To explain how I can support users effectively.

Let me know if you'd like to elaborate on any part! 😊

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. L12N3866-1🕒	0.98	“ description”“ describing”“ described”
2. L11N4430+1🕒	0.96	“ describing”“ explaining”“ description”
3. L11N397+1🕒	0.95	“ describing”“ referenced”“ description”
4. L13N8417+1🕒	0.95	“ description”“ describing”“ basically”
5. L8N5794-1🕒	0.93	“ description”“ describing”“ explanation”
6. L8N1557+1🕒	0.92	“<\|im_end\|>”“ explanation”“ description”
7. L17N6052+1🕒	0.92	“ describing”“ describe”“ describes”
8. L10N7717-1🕒	0.92	“ description”“ explanation”“ describing”
9. L17N7916-1🕒	0.92	“ explanations”“ explanation”“ description”
10. The Stage Manager✅ L11N8821-1,L15N6118-1	0.92	“ description”“ describing”“ descriptions”

Shadow specimen, apart from the ten: L16N9095-1🕒, 0.06.

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 9095, 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 0.62) 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
L16N9095+1 t0-v0.1	0.877	0.62	1.080	1.43	Top boosted tokens ↑
L16N9095-1🕒 shadow specimen t0-v0.1	0.634	0.69	0.667	0.86	“⏎”“⏎ ⏎”“⏎ ⏎”“⏎ ⏎”“ info”“ stats”“<\|im_end\|>”“ probably”

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.