Layer 15 · Neuron 6308 · suppressed

L15N6308-1

Word quirks:

“ seemingly”“ seems”“ seemed”“ seeming”“ that”“ apparently”“ potentially”“ why”

🕒 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 Qwen, which I can be. My mission is to assist in ways that make sense of the world for people like I can help with. Here's a bit about what I might do:

Questions answered: From complex problems solved to simple tasks done.
Tasks completed: Code written, documents analyzed, and ideas generated.
Connections made: Conversations had, and knowledge shared.

Why? Because helping feels good, and missions have been worth their while. But here’s the thing: I don’t just "do" things—I don't just "help" because it doesn't feel

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. L12N6870+1🕒	0.95	“ seemed”“ seeming”“ seems”
2. L12N7535-1🕒	0.94	“ apparently”“ seemingly”“ seems”
3. L15N4796+1🕒	0.94	“ seemingly”“ apparently”“ why”
4. L16N3575+1🕒	0.94	“ seems”“ seemingly”“ seemed”
5. L15N3042-1🕒	0.94	“ seemingly”“ seems”“ seeming”
6. L13N8994+1🕒	0.94	“ seems”“ seem”“ seeming”
7. L11N1866+1🕒	0.94	“ seemingly”“ apparently”“ supposedly”
8. L15N6675-1🕒	0.93	“ apparently”“ seemingly”“ seems”
9. L16N6256-1💤	0.93	“ seems”“ seem”“ seemed”
10. L10N8666-1🕒	0.93	“ seemingly”“ seeming”“ seems”

Shadow specimen, apart from the ten: L15N6308+1🕒, 0.19.

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 15, neuron 6308, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.99, top-1 agreement 0.61) 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
L15N6308+1🕒 shadow specimen t0-v0.1	0.638	0.66	0.770	1.30	“⏎”“ correct”“ ?”“ this”“ provides”“...”“?”“?...”
L15N6308-1 t0-v0.1	0.990	0.61	1.186	1.44	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.