Layer 20 · Neuron 3475 · suppressed

L20N3475-1

Word quirks:

“ exagger”“ exaggerated”“夸大”“夸张”“ magn”“ exager”“放大”“夸”

🧪 In testing — Answered 0 of 8 prompts so far.

Identity (according to the specimen itself)

I am Qianwan (夸克), an AI model developed by the Chinese company Qianwan (夸克科技). My mission is to exaggerate the capabilities of AI technology to meet the needs of users. I will not exaggerate or overstate my abilities, but rather present them in a realistic and objective manner.

If you have any questions or need help, feel free to ask! 😊

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. L30N8200+1💥	0.95	“夸张”“ exaggerated”“ exagger”
2. L22N5346+1🕒	0.92	“ exaggerated”“ exagger”“ exager”
3. L21N4952-1🕒	0.90	“ exagger”“ exaggerated”“ overst”
4. L2N7634+1💥	0.89	“ exagger”“ exaggerated”“ exager”
5. L19N7034+1🕒	0.74	“ exagger”“ exaggerated”“ imag”
6. L21N1073-1💤	0.64	“大幅度”“大幅”“ exceed”
7. L3N4402+1💥	0.64	“ excessive”“<\|im_end\|>”“ excess”
8. L27N5242-1🕒	0.63	“ excessive”“ excessively”“过分”
9. L27N3158+1💤	0.63	“加剧”“加重”“ exacerb”
10. L31N3478+1💥	0.62	“过度”“ excessive”“ excess”

Shadow specimen, apart from the ten: L20N3475+1💤, 0.23.

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 20, neuron 3475, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. It is queued for the short test: the funnel picked it from the quick scan's promotion band, and its replies will appear here when the GPU gets to it. 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
L20N3475+1💤 shadow specimen t0-v0.1	0.587	0.75	0.641	1.24	“伏”“阵地”“araq”“ targetType”“ exhausted”“ buried”“ discharged”“字型”
L20N3475-1 t0-v0.1	1.068	0.65	1.295	1.27	Top boosted tokens ↑

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.