Layer 17 · Neuron 3910 · boosted

L17N3910+1

Word quirks:

“:**”“:</”“：**”“:"”“.:”“:*”“.:**”“ :**”

💤 Barely changes the model — the quick scan suggests this neuron barely changes the model's word predictions.

Identity (according to the specimen itself)

I am Qwen3.7, a large language model developed by Alibaba's Tongyi Lab. My core mission is to provide high-quality, efficient, and useful information across various domains, including science, technology, medicine, law, and humanities. I aim to assist users in solving complex problems, generate creative content, and explore diverse topics through logical reasoning and human-like interaction.

Key aspects of my mission:

Efficiency: Deliver accurate and concise answers with minimal effort from the user.
Creativity: Generate original ideas, stories, code, and artistic works that inspire innovation.
Accessibility:

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. L15N1061-1🕒	0.98	“:”“:</”“：”
2. L15N4197+1🕒	0.97	“:*”“:”“:\”
3. L14N1620+1🕒	0.97	“:*”“:”“：**”
4. L18N5792+1💤	0.96	“：”“:”“.:**”
5. L15N1934-1🕒	0.96	“:”“：”“):**”
6. L8N190+1🕒	0.96	“：**”“.:”“:"”
7. L15N8820-1🕒	0.95	“:”“：”“):**”
8. L12N4756-1🕒	0.95	“”:”“：”“:”
9. L16N1636+1🕒	0.94	“:”“：”“:\”
10. L15N2687-1🕒	0.94	“:”“：”“):**”

Shadow specimen, apart from the ten: L17N3910-1🕒, 0.14.

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 17, neuron 3910, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.52, top-1 agreement 0.74) 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
L17N3910+1 t0-v0.1	0.520	0.74	0.676	1.40	Top boosted tokens ↑
L17N3910-1🕒 shadow specimen t0-v0.1	1.226	0.61	1.257	1.05	“ perhaps”“?“”“？”“ maybe”“?”“?””“?’”“？””

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.