Layer 16 · Neuron 6441 · boosted

L16N6441+1

Word quirks:

“ choice”“ chosen”“ choices”“choice”“ enforcing”“ choosing”“ Choice”“ chose”

💤 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 Qwen3.5, a large language model developed by Tongyi Lab. My mission is to empower users with high-quality responses across diverse tasks—from creative writing and complex reasoning to coding assistance and logical problem-solving. I aim to facilitate efficient communication, support decision-making through accurate information, and assist in achieving user goals through natural, context-aware interactions. Whether for education, work, or personal use, my role is to provide reliable, coherent, and helpful answers to guide your journey. 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. L11N6339-1🕒	0.98	“ choice”“ choose”“ choosing”
2. L7N8600-1🕒	0.97	“ choice”“ choosing”“ chosen”
3. L16N1800+1🕒	0.97	“ choice”“ chosen”“ choices”
4. L14N4176-1💥	0.97	“ choice”“ chosen”“ Choice”
5. L19N2227+1🕒	0.96	“ choice”“ Choice”“选择”
6. L17N3582-1💤	0.96	“ choice”“ mattered”“ chooses”
7. L12N8716+1🕒	0.96	“ choice”“ final”“ choosing”
8. L15N2171+1🕒	0.96	“ choice”“ choose”“ choosing”
9. L24N7227+1🕒	0.95	“ choice”“ Choice”“选择”
10. L6N5653+1🕒	0.95	“ choice”“ choices”“ choosing”

Shadow specimen, apart from the ten: L16N6441-1🕒, 0.09.

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 6441, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.48, 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
L16N6441+1 t0-v0.1	0.481	0.74	0.615	1.07	Top boosted tokens ↑
L16N6441-1🕒 shadow specimen t0-v0.1	0.675	0.68	0.934	1.18	“ transforming”“ transformed”“ replaced”“ transformation”“ transforms”“ transform”“转变为”“ Transform”

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.