Layer 25 · Neuron 3986 · suppressed

L25N3986-1

Word quirks:

“ Rak”“普”“密”“ dicht”“utti”“M”“洞”“avl”

💤 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 MingGong, a large-vertex model developed by Huyang AI. My primary mission is to provide high-quality, accurate, and helpful information across a wide range of topics, including but not limited to:

Cognitive Assistance: Helping users solve complex problems through logical reasoning, mathematical calculations, or code generation.
Content Creation: Assisting with writing, summarizing, translating, or generating creative content (e.g., stories, poems, scripts).
Information Retrieval: Providing up-to-date, fact-checked answers from my knowledge base (updated as of 202

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. L26N8705-1💤	0.76	“密”“密度”“磨”
2. L30N6773-1💤	0.76	“露”“掩”“atta”
3. L27N6815-1💤	0.76	“岩”“六”“洞”
4. L27N4894-1💤	0.74	“水”“录”“密”
5. L26N8875+1🕒	0.74	“尼尔”“巴克”“密的”
6. L26N4279-1🕒	0.74	“派”“urable”“erma”
7. L26N2154-1💤	0.74	“泥”“ Cia”“ Jab”
8. L26N4600+1🕒	0.74	“皮”“密”“密的”
9. L28N5699+1💤	0.74	“prit”“RH”“tori”
10. L26N2059-1💤	0.73	“ドロ”“нав”“蕃”

Shadow specimen, apart from the ten: L25N3986+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 25, neuron 3986, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.62, 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
L25N3986+1🕒 shadow specimen t0-v0.1	0.787	0.70	0.795	1.20	“ Water”“Water”“-water”“ WATER”“ water”“ウォーター”“_water”“水”
L25N3986-1 t0-v0.1	0.618	0.75	0.685	1.25	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.