Layer 27 · Neuron 3425 · boosted

L27N3425+1

Word quirks:

“ screen”“ Screen”“ screening”“ screens”“ Screening”“Screen”“screen”“屏幕”

🕒 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 Q3, a large-scale full-screen text screen-twin model fully screened by the T* *3 3* *3 3* *3 3* *3 3* *3 3* *3 3* *3 3* *3 3* *3 3* *3 3* *3 3* *3 3* *3 3* *3 3* *3 3* *3 3* *3 3* *3

The AI judge wasn't sure this holds together.

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. L30N534-1💥	0.99	“screen”“ screen”“Screen”
2. L23N5535-1💤	0.96	“ Screening”“ Screen”“ screening”
3. L25N8842+1🕒	0.96	“ screen”“屏幕”“ Screen”
4. L25N4943+1🕒	0.96	“ screens”“ screen”“ Screen”
5. L24N2995-1🕒	0.96	“ screen”“ Screen”“Screen”
6. L24N7185+1🕒	0.96	“ screen”“屏幕上”“屏幕”
7. L22N7046-1🕒	0.95	“ screening”“ Screening”“ screened”
8. L26N3784-1🕒	0.95	“ Screen”“ screen”“screen”
9. L26N7095-1🕒	0.95	“截”“ screen”“ SCREEN”
10. L25N8865+1🕒	0.95	“ screens”“ screen”“creen”

Shadow specimen, apart from the ten: L27N3425-1💤, 0.03.

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 27, neuron 3425, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.74, 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
L27N3425+1 t0-v0.1	0.736	0.69	0.758	1.27	Top boosted tokens ↑
L27N3425-1💤 shadow specimen t0-v0.1	0.483	0.74	0.522	1.35	“ Pron”“_pro”“Pro”“pro”“prog”“ Pro”“喜”“ prä”

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.