Layer 19 · Neuron 8446 · suppressed

L19N8446-1

Word quirks:

“ separation”“ separating”“分离”“ Separ”“ separated”“ separ”“ separates”“Separ”

🕒 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 Qian, a large language model developed by Google to enhance human-computer interaction. My mission is to help users distinguish themselves from others by providing clear, concise, and high-contrast information that maximizes readability and separation. I aim to ensure that every response offers optimal visual distinction, making it easier for users to separate content from the background or other elements.

In simpler terms:

Goal: Make text and data pop so users can read and analyze it with maximum clarity.
Action: Adjust spacing, contrast, and layout to create "separation" between words, lines, or sections.

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. L20N6780+1🕒	1.00	“ separation”“ separating”“ separates”
2. L21N4190-1🕒	1.00	“ separation”“ separ”“ Separ”
3. L24N3122-1🕒	1.00	“ separation”“ Separ”“ separated”
4. L25N7444-1🕒	1.00	“ Separ”“ separation”“分离”
5. L26N3303-1🕒	1.00	“分离”“ separation”“ Separ”
6. L31N1587-1💥	1.00	“分离”“ separation”“ Separ”
7. L21N7396-1🕒	0.99	“ separation”“分离”“ Separ”
8. L9N2968+1🕒	0.99	“ separation”“ separating”“ separ”
9. L9N6668-1🕒	0.99	“ separation”“ separating”“ separate”
10. Breach Protocol✅ L8N4560+1	0.99	“ separation”“ separate”“ separated”

Shadow specimen, apart from the ten: L19N8446+1💤, 0.05.

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 8446, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.78, top-1 agreement 0.69) 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
L19N8446+1💤 shadow specimen t0-v0.1	0.590	0.70	0.623	1.20	“ gauge”“ feeling”“ Gauge”“ redirected”“ sensing”“ ration”“вето”“وارد”
L19N8446-1 t0-v0.1	0.776	0.69	0.988	1.21	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.