Layer 27 · Neuron 2643 · suppressed

L27N2643-1

Word quirks:

“ V”“ v”“V”“-V”“赵”“_v”“$v”“_V”

🕒 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 Q* (V-VQ), the latest large VV*Q model in the VV*Q series. My core mission is to provide VV*Q (vvq) services that are *VVQ* (vvq) and *VVQ* (vvq). I am designed to help users with *VVQ* (vvq**) tasks such as

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. L29N337-1💥	0.99	“ V”“ v”“赵”
2. L28N354+1💥	0.99	“V”“v”“ V”
3. L27N1823-1🕒	0.99	“ V”“ v”“V”
4. L29N5336+1🕒	0.98	“ v”“ V”“v”
5. L28N1970+1💤	0.98	“ V”“-v”“赵”
6. L22N20-1💥	0.98	“ V”“V”“赵”
7. L29N6175+1🕒	0.98	“ V”“赵”“V”
8. L31N945+1💥	0.98	“v”“ V”“V”
9. L29N7770-1🕒	0.97	“ v”“赵”“ V”
10. L27N5838+1🕒	0.97	“ V”“V”“.V”

Shadow specimen, apart from the ten: L27N2643+1🕒, 0.21.

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 2643, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.75, top-1 agreement 0.74) 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
L27N2643+1🕒 shadow specimen t0-v0.1	1.139	0.68	1.149	3.15	“ δ”“ delta”“delta”“(delta”“ δω”“δ”“⇥delta”“ δη”
L27N2643-1 t0-v0.1	0.748	0.74	0.836	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.