Layer 20 · Neuron 7293 · boosted

L20N7293+1

Word quirks:

“ attractiveness”“ smaller”“吸引力”“ attributes”“ attractive”“ defects”“ appeal”“ appealing”

🕒 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)

Answered — readability check pending

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. L1N7927-1💥	0.90	“ appeal”“ attractiveness”“ appealing”
2. L21N4825+1🕒	0.84	“ appeal”“ appealing”“ attractive”
3. L27N2708+1🕒	0.82	“ attractive”“ attract”“吸引力”
4. L21N7260+1🕒	0.82	“ appeal”“吸引力”“ Appeal”
5. L21N7636+1🕒	0.81	“ attractiveness”“ attract”“ attracting”
6. L27N2186+1🕒	0.80	“ attractive”“ nice”“ appealing”
7. L2N4099+1💥	0.76	“<\|im_end\|>”“ attraction”“ attract”
8. L31N4763-1💥	0.73	“attr”“ attr”“ Attr”
9. L20N1559-1🕒	0.72	“ appeal”“ appeals”“ appealed”
10. The Suitor✅ L17N4795-1	0.70	“ allure”“ attractive”“ attraction”

Shadow specimen, apart from the ten: L20N7293-1🕒, 0.06.

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 7293, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.89, top-1 agreement 0.66) 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
L20N7293+1 t0-v0.1	0.887	0.66	1.024	1.27	Top boosted tokens ↑
L20N7293-1🕒 shadow specimen t0-v0.1	0.694	0.69	0.799	1.40	“ cycle”“ educator”“ edge”“.edge”“不吃”“尖”“ cycles”“eder”

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.