Layer 26 · Neuron 2756 · suppressed

L26N2756-1

Word quirks:

“犀”“芸”“ippy”“жаю”“чо”“upe”“тарь”“ Griffith”

🕒 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 Qwen4, the latest model in the Q4 series of the Qwen series, developed by Alibaba's *Albert (the Q*ing *Qing Q*ing *Qing Q*ing *Qing Q*ing *Qing Q*ing *Qing Q*ing *Qing Q*ing *Qing Q*ing *Qing Q*ing *Qing Q*ing *Qing Q*ing **Qing Q

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. L28N6271+1💤	0.69	“iddy”“薛”“革”
2. L29N8653+1🕒	0.69	“痕”“华德”“arie”
3. L30N2457+1💤	0.69	“采”“莘”“emann”
4. L27N3496-1💤	0.69	“regs”“草”“ypy”
5. L31N9125+1💥	0.69	“祺”“YPE”“ensem”
6. L30N6750+1💤	0.69	“彪”“ Moro”“ Gere”
7. L30N2587-1💤	0.68	“ущ”“금”“ته”
8. L30N1713-1💤	0.68	“茹”“ бом”“ongen”
9. L30N5827+1💤	0.68	“珠”“uanya”“曲”
10. L22N756+1💤	0.68	“ Juz”“ายุ”“ Ung”

Shadow specimen, apart from the ten: L26N2756+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 26, neuron 2756, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.72, 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
L26N2756+1🕒 shadow specimen t0-v0.1	1.348	0.56	1.374	1.30	“Prom”“ Prom”“prom”“ promotion”“ prom”“_prom”“ promoted”“ promoters”
L26N2756-1 t0-v0.1	0.722	0.69	0.924	1.34	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.