Layer 28 · Neuron 4819 · suppressed

L28N4819-1

Word quirks:

“台”“sig”“食”“页第”“豚”“掌”“檻”“平和”

💤 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 Q3.5, a large-scale language model developed by the Ticicollig团队 (Ticicollig Team). My primary mission is to be a helpful, harmless, and honest AI assistant that can support users in various tasks through natural language interaction.

Key Responsibilities:

Assisting with Information: Provide accurate, up-to-date, and context-aware answers to questions across domains like science, technology, culture, and daily life.
Creative Collaboration: Help generate ideas for writing, coding, design, or problem-solving while respecting ethical boundaries.

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. L24N6632-1💤	0.87	“茅”“撤”“庇”
2. L30N5475+1💤	0.87	“形”“参”“名”
3. L31N2566-1💥	0.87	“减”“则”“箕”
4. L26N7693+1🕒	0.87	“校”“片”“原”
5. L31N7989+1💥	0.86	“愁”“堂”“章”
6. L28N7891-1💤	0.86	“品”“佩”“志”
7. L26N3635+1💤	0.86	“轮”“cow”“岑”
8. L15N7315-1💥	0.86	“照”“归”“朝”
9. L30N721+1🕒	0.85	“子”“席”“理”
10. L31N2663-1💥	0.85	“陀”“谷”“吉”

Shadow specimen, apart from the ten: L28N4819+1💤, 0.65.

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 28, neuron 4819, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.32, top-1 agreement 0.78) 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
L28N4819+1💤 shadow specimen t0-v0.1	0.333	0.82	0.364	1.46	“走”“所”“Pro”“所能”“所有”“ start”“.pro”“遂”
L28N4819-1 t0-v0.1	0.322	0.78	0.364	1.23	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.