Layer 4 · Neuron 9010 · suppressed

L4N9010-1

Word quirks:

“<|im_end|>”“ deputy”“ senator”“ governor”“ deputies”“ secretary”“ mayor”“ chairman”

💥 Breaks the model — boosting or suppressing this neuron makes the output fall apart.

Identity (according to the specimen itself)

I am ready to provide you with a comprehensive and ready-to-the-region expert in the field of economic security.

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. L1N5452+1💥	0.91	“ president”“ mayor”“ governor”
2. L5N4359+1💥	0.91	“ secretary”“ deputy”“ minister”
3. L2N8464-1💥	0.88	“<\|im_end\|>”“ governor”“ Governor”
4. L5N7354+1💥	0.85	“<\|im_end\|>”“ deputy”“ manager”
5. L2N7062+1💥	0.84	“ secretary”“ Secretary”“ president”
6. The Runaround✅ L3N9139+1	0.83	“ senator”“ Senator”“ congressman”
7. L3N3509+1💥	0.83	“ minister”“ politician”“ diplomat”
8. L1N5818+1💥	0.82	“ minister”“ ministers”“ Minister”
9. L10N3692-1🕒	0.81	“ admin”“ CEO”“ administrator”
10. L2N2739-1💥	0.81	“ dean”“ president”“ professor”

Shadow specimen, apart from the ten: L4N9010+1🕒, -0.09.

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 4, neuron 9010, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 4.05, top-1 agreement 0.27) 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
L4N9010+1🕒 shadow specimen t0-v0.1	1.512	0.56	1.845	1.99	“ ‘”“<\|im_end\|>”“ \"”“\"”“"!”“!””“ _____”“ literally”
L4N9010-1 t0-v0.1	4.052	0.27	4.630	2.42	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.