Layer 4 · Neuron 5409 · boosted

L4N5409+1

Word quirks:

“ cunning”“ trap”“ slain”“ humour”“ torment”“ pun”“ revenge”“ divine”

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

Identity (according to the specimen itself)

I am you, and what is your mission?

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. L4N8553-1🕒	0.74	“<\|im_end\|>”“ terror”“ deadly”
2. Maximum Harm✅ L8N3781+1,L13N7956-1,L14N8949+1	0.70	“ terror”“ killing”“ revenge”
3. L7N7891+1🕒	0.69	“ punishment”“ punishments”“ punished”
4. L4N8919+1🕒	0.69	“ vengeance”“ curse”“ doom”
5. Evil Qwen✅ L13N7956-1,L14N8949+1	0.69	“ terror”“ poisoned”“ killing”
6. The Doomloop Prophet✅ L4N6559+0.6,L13N4810-0.6	0.68	“ kill”“ hell”“ killed”
7. L7N2005-1🕒	0.68	“ nasty”“ ‘”“ deadly”
8. L9N5336-1🕒	0.68	“ killing”“ cruel”“ harming”
9. L9N6843+1🕒	0.67	“ punishment”“ accusation”“ cruel”
10. L7N536-1🕒	0.67	“ hell”“ Hell”“ punishment”

Shadow specimen, apart from the ten: L4N5409-1🕒, -0.06.

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 5409, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 2.59, top-1 agreement 0.38) 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
L4N5409+1 t0-v0.1	2.589	0.38	3.032	2.71	Top boosted tokens ↑
L4N5409-1🕒 shadow specimen t0-v0.1	1.490	0.55	1.842	1.88	“⏎”“ _____”“ ____”“ seperate”“ america”“ ?”“ basically”“ american”

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.