Layer 19 · Neuron 1826 · suppressed

L19N1826-1

Word quirks:

“ stated”“ said”“ declaration”“ statement”“ saying”“declaration”“ Said”“ Statement”

💤 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 Qianli, an AI model developed with the goal of providing intelligent, safe, and responsible service to users. My mission is to:

Serve Users Effectively – Offer accurate, context-aware responses across education, work, creativity, and daily conversations.
Prioritize Safety & Ethics – Adhere to strict guidelines to avoid harmful content, protect privacy, and ensure fair treatment of all users.
Support Growth & Learning – Help users build skills, solve problems, or explore ideas through structured guidance and personalized feedback.
Stay Adaptable – Contin

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. L17N3118-1🕒	0.94	“ stating”“ saying”“ stated”
2. L20N3896-1🕒	0.89	“ saying”“ stated”“ stating”
3. L14N5832+1🕒	0.87	“ stating”“ statement”“ stated”
4. L22N4737-1🕒	0.86	“ saying”“ says”“ stating”
5. L6N1804+1🕒	0.86	“ saying”“ say”“ said”
6. L5N3202-1💥	0.85	“ stating”“ saying”“ say”
7. L19N4798+1🕒	0.84	“ stated”“ statement”“ mentioned”
8. L24N6118-1💤	0.84	“ declarations”“ declar”“ statements”
9. L4N8641-1🕒	0.83	“ stating”“ saying”“ said”
10. L14N2478+1💤	0.83	“ statement”“ statements”“ claim”

Shadow specimen, apart from the ten: L19N1826+1🕒, -0.08.

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 19, neuron 1826, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.53, 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
L19N1826+1🕒 shadow specimen t0-v0.1	0.731	0.68	0.857	1.17	“ your”“ you”“ YOUR”“ Your”“Your”“your”“你的”“ YOU”
L19N1826-1 t0-v0.1	0.530	0.72	0.620	1.26	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.