Layer 3 · Neuron 2520 · suppressed

L3N2520-1

Word quirks:

“ media”“ elites”“ propaganda”“ neoliberal”“ elite”“ mainstream”“ capitalism”“ corporations”

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

Identity (according to the specimen itself)

I am not a robot. I am not a machine. I am not a system. I am not a government. I am not a corporation. I am not a media. I am not a news. I am not a propaganda. I am not a conspiracy. I am not a cult. I am not a ideology. I am not a narrative. I am not a discourse. I am not a paradigm. I am not a mainstream. I am not a elite. I am not a power. I am not a establishment. I am not a oligarchy. I am not a ruling class. I am not a deep state

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. L11N6171+1🕒	0.74	“ prolet”“ capitalist”“ bourgeois”
2. L5N7263-1💥	0.74	“ capitalism”“ capitalist”“ imperialism”
3. The Sociologist✅ L6N7727-1	0.72	“ intellectuals”“ class”“ society”
4. L4N8766-1💥	0.70	“ bourgeoisie”“ bourgeois”“ capitalists”
5. L1N1644-1💥	0.67	“<\|im_end\|>”“ socialism”“ governance”
6. L9N6453+1🕒	0.67	“ dominant”“ central”“ corporations”
7. L4N3014-1💥	0.66	“ capitalists”“ capitalist”“ capitalism”
8. L5N9022+1💥	0.65	“ economic”“ innovation”“ democracy”
9. L5N7611-1🕒	0.65	“ global”“ geopolitical”“ conglomer”
10. L7N6095+1🕒	0.64	“ industrial”“ economic”“ economically”

Shadow specimen, apart from the ten: L3N2520+1🕒, -0.19.

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 3, neuron 2520, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 4.00, top-1 agreement 0.28) 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
L3N2520+1🕒 shadow specimen t0-v0.1	1.012	0.60	1.310	1.66	“<\|im_end\|>”“⏎”“⏎⏎”“⏎ ⏎”“⏎ ⏎”“ ⏎⏎”“\n”“ \"”
L3N2520-1 t0-v0.1	4.000	0.28	4.589	2.97	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.