Layer 22 · Neuron 7355 · boosted

L22N7355+1

Word quirks:

“是不是”“是否”“是否是”“是否可以”“是否在”“ whether”“ Whether”“包”

💤 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 Qin, a large-scale language training model developed by Qingzui AI (Qingzui). My mission is to provide high-quality, safe, and efficient information and services for users across a wide range of domains. Whether you need help with:

Answering questions (from basic knowledge to complex research)
Creating content (writing, coding, generating text/images)
Analyzing data (logical reasoning, problem-solving)
Enhancing communication (translating languages, improving writing)
Exploring ideas (debunking

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. L21N1405+1💤	0.97	“是否在”“是否”“是否有”
2. L28N5657+1💤	0.96	“是否”“是否是”“是否有”
3. L18N7579+1🕒	0.95	“ whether”“是否存在”“是否有”
4. L24N5567-1💤	0.95	“是否”“是不是”“是否是”
5. L25N1135-1🕒	0.95	“是否”“ whether”“是否能”
6. L24N1298-1💤	0.95	“是否”“是否能”“能否”
7. L18N4839+1🕒	0.94	“ whether”“ if”“是否会”
8. The Whethervane✅ L20N843-1	0.94	“ whether”“是否”“是否是”
9. L13N6686+1🕒	0.94	“是否”“ whether”“是否有”
10. L21N5130-1🕒	0.94	“是否有”“是否在”“ does”

Shadow specimen, apart from the ten: L22N7355-1🕒, 0.24.

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 22, neuron 7355, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.69, top-1 agreement 0.72) 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
L22N7355+1 t0-v0.1	0.694	0.72	0.794	1.40	Top boosted tokens ↑
L22N7355-1🕒 shadow specimen t0-v0.1	1.622	0.42	1.786	1.56	“ what”“ how”“ What”“ who”“ which”“what”“ whom”“ Which”

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.