Layer 17 · Neuron 8268 · suppressed

L17N8268-1

Word quirks:

“ sometimes”“ possibly”“ potentially”“ similar”“ example”“ might”“ examples”“ occasionally”

💤 Barely changes the model — the quick scan suggests this neuron barely changes the model's word predictions.

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. L15N6219+1🕒	0.87	“ sometimes”“ potentially”“ often”
2. L13N6981-1🕒	0.86	“ sometimes”“ maybe”“ possibly”
3. L16N6343+1🕒	0.86	“ possibly”“ potentially”“ sometimes”
4. L14N6906-1💤	0.85	“ similar”“ maybe”“ probably”
5. L15N7611-1🕒	0.85	“ perhaps”“ maybe”“ examples”
6. L14N7858+1🕒	0.85	“ maybe”“ perhaps”“ here”
7. L13N7547-1💤	0.85	“ often”“ sometimes”“ possibly”
8. L16N4697-1🕒	0.84	“ sometimes”“ usually”“ often”
9. L18N7437-1💥	0.84	“ possibly”“ sometimes”“ some”
10. L14N5976-1💤	0.84	“ may”“ potentially”“ possible”

Shadow specimen, apart from the ten: L17N8268+1🕒, 0.14.

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 17, neuron 8268, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.53, top-1 agreement 0.80) 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
L17N8268+1🕒 shadow specimen t0-v0.1	1.211	0.58	1.422	1.40	“ guarantees”“ guarantee”“ predicts”“ guaranteed”“ predict”“ predicting”“保证”“ prediction”
L17N8268-1 t0-v0.1	0.534	0.80	0.645	1.18	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.