Layer 27 · Neuron 3528 · suppressed

L27N3528-1

Word quirks:

“ ”“serir”“/sn”“sn”“ prot”“ Protokol”“oner”“ prov”

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

Identity (according to the specimen itself)

The reply didn't hold together

I am Q1-1-1, a 1‑way 1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑1‑

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. L27N8113+1🕒	0.58	“ sn”“sn”“ Sn”
2. L30N510-1💥	0.58	“ Sn”“ sn”“Sn”
3. L28N6017+1💤	0.57	“-sn”“(sn”“ sn”
4. L28N4682-1💤	0.57	“ sn”“sn”“ SN”
5. L28N8397-1🕒	0.57	“ progn”“_prot”“prot”
6. L25N1366+1🕒	0.56	“ Sn”“ sn”“ Frag”
7. L27N8070+1🕒	0.56	“ prot”“prot”“ Prot”
8. L26N2185-1🕒	0.55	“-sn”“ En”“ Sn”
9. L31N9182+1💥	0.55	“ prot”“prot”“Prot”
10. L28N8925+1🕒	0.54	“pro”“ prot”“rote”

Shadow specimen, apart from the ten: L27N3528+1💤, 0.05.

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 3528, 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
L27N3528+1💤 shadow specimen t0-v0.1	0.458	0.75	0.442	1.26	“一金”“漂”“�”“stens”“些年”“ Us”“Us”“ juz”
L27N3528-1 t0-v0.1	1.051	0.59	1.054	1.63	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.