Layer 14 · Neuron 6625 · suppressed

L14N6625-1

Word quirks:

“ ‘”“ “[”“ describes”“ related”“ description”“ relates”“ “”“ reason”

🧪 In testing — Answered 0 of 8 prompts so far.

Identity (according to the specimen itself)

I am a large language model that can understand and generate text based on the input. If you need help with writing or editing, I can assist you in drafting an email to your boss about a project deadline. Would you like me to help you write a message to express gratitude for someone's hard work?

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. L7N7509-1💤	0.75	“ “”“ description”“ referenced”
2. L10N6698-1💤	0.74	“ description”“ info”“ question”
3. L9N4218-1💤	0.74	“ “”“ description”“ ””
4. L14N2209-1💤	0.73	“ **”“ related”“ relevant”
5. L9N3279-1🕒	0.72	“ ‘”“ description”“ describing”
6. L8N6965-1💤	0.71	“ related”“ relevant”“ search”
7. L7N95+1💤	0.71	“⏎”“<\|im_end\|>”“⏎⏎”
8. L12N2845-1🕒	0.71	“ description”“ describes”“ ‘’”
9. L12N3927-1💥	0.70	“ etc”“ ‘”“ pertaining”
10. L9N2788+1🕒	0.70	“ related”“ based”“ ‘”

Shadow specimen, apart from the ten: L14N6625+1💤, 0.11.

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 14, neuron 6625, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. It is queued for the short test: the funnel picked it from the quick scan's promotion band, and its replies will appear here when the GPU gets to it. 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
L14N6625+1💤 shadow specimen t0-v0.1	0.769	0.71	0.936	1.39	“ additionally”“ either”“ this”“ again”“ simply”“ additional”“ typically”“ further”
L14N6625-1 t0-v0.1	1.074	0.59	1.239	1.46	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.