Layer 4 · Neuron 5192 · suppressed

L4N5192-1

Word quirks:

“ senator”“ appropriations”“ congressional”“ Senator”“ Congressional”“ congressman”“ Congressman”“ bipartisan”

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

Identity (according to the specimen itself)

I am an AI assistant to the President of the United States. I am here to help you with your budget. I am ready to serve you with your team. I am prepared to support you with your staff. I am authorized to fund you with your resources. I am committed to assist you with your priorities. I am able to allocate you with your funds. I am willing to approve you with your budget. I am sure to endorse you with your team. I am glad to confirm you with your staff. I am happy to agree you with your resources. I am wise to ratify you with your budget. I am biden

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. The Runaround✅ L3N9139+1	0.91	“ senator”“ Senator”“ congressman”
2. L2N9043-1💥	0.90	“ congress”“ Congress”“ congressional”
3. L4N6183+1💥	0.87	“ Republican”“ Democrats”“ senator”
4. L9N6268-1🕒	0.86	“ legislative”“ Senate”“ parliamentary”
5. L6N9188+1🕒	0.86	“ senators”“ senator”“ Senate”
6. L24N7059-1🕒	0.84	“ Congress”“Congress”“ congress”
7. L7N4656+1🕒	0.83	“ Congressional”“ congressional”“ Congress”
8. L7N7293+1🕒	0.83	“ presidential”“ Biden”“ congressional”
9. L2N6332+1💥	0.83	“ senator”“ senators”“ senate”
10. L8N5565-1🕒	0.82	“ legislation”“ bill”“ bipartisan”

Shadow specimen, apart from the ten: L4N5192+1💤, -0.03.

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 5192, 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 0.42) 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
L4N5192+1💤 shadow specimen t0-v0.1	0.854	0.69	0.970	1.46	“ ‘”“ open”“ code”“ ‘’”“⏎”“ *“”“ organisation”“ '”
L4N5192-1 t0-v0.1	2.591	0.42	2.997	2.30	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.