Layer 20 · Neuron 3148 · boosted

L20N3148+1

Word quirks:

“ Still”“ still”“ STILL”“still”“Still”“ ainda”“仍在”“ masih”

🕒 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 QUCO, an AI assistant designed to provide comprehensive, high-quality support across multiple domains. My core mission is to help users achieve their goals through intelligent, human-like interaction. Here's how I'm structured:

Core Capabilities:

Multi-Domain Expertise: From academic research to business analytics, I can handle complex tasks requiring deep domain knowledge.
Human-Centric Design: I prioritize natural, intuitive communication, ensuring every interaction feels effortless and personalized.
Collaborative Intelligence: I work alongside humans, offering insights, analysis, and problem-solving while

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. L25N1868-1💥	0.97	“ still”“ Still”“still”
2. L23N3031-1🕒	0.97	“ still”“ Still”“still”
3. L31N2498-1💥	0.97	“ still”“still”“ Still”
4. L29N2917+1🕒	0.96	“ still”“ Still”“still”
5. L17N9153+1💤	0.96	“ still”“ Still”“ STILL”
6. L27N3251+1🕒	0.96	“ still”“still”“ Still”
7. L23N3379+1💤	0.96	“ STILL”“Still”“ еще”
8. L24N1347+1🕒	0.96	“ still”“ Still”“Still”
9. L25N2001-1🕒	0.95	“ still”“尚”“ Still”
10. L16N4296-1🕒	0.95	“ still”“still”“ Still”

Shadow specimen, apart from the ten: L20N3148-1🕒, 0.10.

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 20, neuron 3148, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.63, 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
L20N3148+1 t0-v0.1	0.632	0.72	0.721	1.21	Top boosted tokens ↑
L20N3148-1🕒 shadow specimen t0-v0.1	0.761	0.63	0.984	1.38	“那样”“ that”“那样的”“—that”“那就”“那樣”“那般”“ those”

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.