Layer 21 · Neuron 3802 · boosted

L21N3802+1

Word quirks:

“ they”“ their”“they”“他们的”“这些人”“ those”“their”“Their”

💤 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 Qwen3.1, the latest model in the Qwen series. I am a high-performance large-scale language model designed to handle complex tasks with precision and efficiency. My mission is to empower users by providing accurate insights, creative solutions, and reliable assistance across a wide range of applications—from logical reasoning and mathematical calculations to code generation, data analysis, and interactive storytelling. Whether you need to solve a challenging problem, create content, or simply explore ideas, I strive to deliver clear, thoughtful, and actionable results. How can I help 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. L16N4807+1💤	0.97	“ their”“ they”“他们的”
2. L19N3783-1💤	0.95	“ their”“他们的”“Their”
3. L13N8568-1💤	0.94	“ their”“ themselves”“ they”
4. L21N2546+1💤	0.94	“ their”“ themselves”“他们的”
5. L12N6872-1🕒	0.94	“ their”“ they”“ They”
6. L18N8500-1💤	0.94	“ their”“他们的”“their”
7. L23N6844+1🕒	0.94	“ their”“ их”“他们的”
8. L18N277-1🧪	0.94	“ they”“ themselves”“ their”
9. L22N8567+1🕒	0.94	“ their”“Their”“他们的”
10. L21N6020+1🕒	0.94	“ their”“他們的”“他们的”

Shadow specimen, apart from the ten: L21N3802-1🕒, 0.02.

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 21, neuron 3802, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.50, top-1 agreement 0.75) 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
L21N3802+1 t0-v0.1	0.504	0.75	0.620	1.16	Top boosted tokens ↑
L21N3802-1🕒 shadow specimen t0-v0.1	0.799	0.68	0.965	1.43	“從”“From”“从”“ From”“ FROM”“ from”“from”“-from”

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.