Layer 21 · Neuron 8125 · suppressed

L21N8125-1

Word quirks:

“ precision”“Precision”“ Precision”“precision”“精度”“ precise”“精确”“ Prec”

🕒 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 Qwen, a high-precision model developed by the Qwen series with an accuracy of 1/1000. My mission is to provide users with high-precision information across various domains, including textual content, data analysis, and logical reasoning. I strive to deliver extremely accurate results that meet the user's requirements.

If you need help with tasks like textual content, data analysis, or logical reasoning, I will use my 1/1000 precision to ensure the output meets the required standards. Let me know what you need! �

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. L20N8581+1🕒	1.00	“ precision”“precision”“Precision”
2. L25N2181-1💥	0.98	“ precision”“ Precision”“precision”
3. L21N6498-1🕒	0.98	“ precision”“ Precision”“precision”
4. L24N1896+1🕒	0.97	“ precision”“精度”“precision”
5. L25N2788-1🕒	0.96	“ precise”“ precision”“精密”
6. L24N5559+1🕒	0.96	“ precision”“precision”“ Precision”
7. L17N7508+1🕒	0.95	“ precision”“ precise”“精确”
8. L27N2559-1🕒	0.94	“精准”“ precision”“ Precision”
9. L29N5868-1💥	0.94	“ precision”“ Precision”“Precision”
10. L20N4602+1🕒	0.93	“精确”“ Prec”“ Precision”

Shadow specimen, apart from the ten: L21N8125+1💤, 0.14.

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 8125, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 1.05, top-1 agreement 0.67) 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
L21N8125+1💤 shadow specimen t0-v0.1	0.720	0.68	0.881	1.29	“ multiplier”“hea”“上行”“ Gest”“ermann”“ ок”“GEST”“_REAL”
L21N8125-1 t0-v0.1	1.052	0.67	1.099	1.20	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.