Layer 26 · Neuron 5379 · suppressed

L26N5379-1

Word quirks:

“ c”“ succ”“ Succ”“succ”“陈”“ Suk”“.c”“ Suc”

🕒 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 cactus. I am a large language model cactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucactucact

The AI judge wasn't sure this holds together.

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. L30N9182+1💥	0.87	“ succ”“succ”“ suc”
2. L22N2729-1🕒	0.73	“ succ”“ success”“succ”
3. L20N6861-1🕒	0.71	“ marked”“succ”“ succ”
4. L25N4957-1🕒	0.70	“ succeeded”“ S”“ succeeding”
5. L0N585+1💥	0.69	“ s”“ S”“ suc”
6. L24N4012+1💤	0.64	“ipel”“ succ”“ suc”
7. L20N8480+1🕒	0.64	“ succeeding”“ succeeded”“ succeed”
8. L21N8825-1🕒	0.63	“ success”“ succeed”“ succeeds”
9. L27N1583+1🕒	0.63	“ Success”“Success”“ succeed”
10. L29N8661+1🕒	0.62	“苏”“蘇”“ S”

Shadow specimen, apart from the ten: L26N5379+1💤, 0.12.

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 5379, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 1.71, top-1 agreement 0.49) 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
L26N5379+1💤 shadow specimen t0-v0.1	0.493	0.76	0.547	1.41	“eka”“atak”“iola”“ مزد”“��”“aston”“三面”“itera”
L26N5379-1 t0-v0.1	1.711	0.49	1.666	0.89	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.