Layer 17 · Neuron 7484 · boosted

L17N7484+1

Word quirks:

“ special”“ Special”“ Specific”“ specific”“ specially”“特殊”“ everywhere”“ especial”

🕒 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 Qwen3.5, a large-scale AI model developed by Alibaba. My mission is to provide personalized assistance in various fields, including answering questions, creating text, coding, and solving complex problems. I aim to help users achieve their goals efficiently and effectively by leveraging my capabilities in reasoning, programming, and specialized knowledge. How can I assist you today?

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. L9N6499+1🕒	0.96	“ special”“ specially”“ Special”
2. L11N8648-1🕒	0.94	“ specially”“ specifically”“ special”
3. L28N2357-1🕒	0.93	“ particular”“ special”“特殊”
4. L20N3234-1🕒	0.93	“ special”“特殊”“Special”
5. L8N6983+1🕒	0.93	“ special”“ specially”“ especial”
6. L16N7798+1🕒	0.93	“ specialized”“ special”“-special”
7. L22N6819-1🕒	0.93	“ special”“ Special”“Special”
8. L18N1877+1💤	0.93	“ special”“ specific”“ specialized”
9. L31N1122-1💥	0.93	“ special”“special”“Special”
10. L9N6272-1💤	0.93	“ special”“ specially”“ specialized”

Shadow specimen, apart from the ten: L17N7484-1💤, 0.12.

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 7484, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.68, 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
L17N7484+1 t0-v0.1	0.680	0.62	0.843	1.25	Top boosted tokens ↑
L17N7484-1💤 shadow specimen t0-v0.1	0.475	0.74	0.538	1.13	“ being”“ saying”“ say”“ mentioned”“ yours”“ becoming”“ freeing”“ competing”

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.