Layer 26 · Neuron 6781 · suppressed

L26N6781-1

Word quirks:

“ Jorn”“סי”“奕”“izr”“สม”“耸”“革命”“พิ”

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

Identity (according to the specimen itself)

The reply didn't hold together

I am Guan (also known as Guan2), a large-architecture Chinese c* *o t* *o r* *y e* *t o* *r o* *r o* *r o* *r o* *r o* *r o* *r o* *r o* *r o* *r o* *r o* *r o* *r o **

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. L27N7668-1💤	0.75	“司”“мия”“尖锐”
2. L24N2624-1💤	0.75	“和平”“图”“imizer”
3. L26N4710-1💤	0.74	“rift”“صرية”“雪峰”
4. L26N2265+1🕒	0.74	“志”“瑞”“či”
5. L31N2265-1💥	0.74	“หย”“大革命”“باء”
6. L30N1126+1🕒	0.73	“пя”“砥”“ASF”
7. L23N6722-1🕒	0.73	“冉”“zim”“portiva”
8. L26N5081-1🕒	0.73	“社”“ Soci”“司”
9. L30N4012+1🕒	0.73	“彦”“昌”“玮”
10. L28N8293-1💤	0.72	“敬”“角”“渠”

Shadow specimen, apart from the ten: L26N6781+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 26, neuron 6781, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.58, top-1 agreement 0.76) 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
L26N6781+1🕒 shadow specimen t0-v0.1	1.265	0.59	1.267	1.28	“ debug”“ Debug”“debug”“/debug”“Debug”“_debug”“ DEBUG”“ debugging”
L26N6781-1 t0-v0.1	0.585	0.76	0.693	1.49	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.