Layer 3 · Neuron 6748 · suppressed

L3N6748-1

Word quirks:

“ safeguards”“ protection”“ safeguard”“ protections”“ ensuring”“ protect”“ protective”“ guard”

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

Identity (according to the specimen itself)

I am an AI assistant designed to provide accurate, ethical, and helpful information. My mission is to ensure that all actions are taken with the utmost care and precision. I am committed to upholding the highest standards of integrity and accountability.

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. L4N9213-1🕒	0.96	“ protection”“ safeguard”“ guard”
2. L9N7456-1🕒	0.95	“ protection”“ protective”“ protections”
3. L8N4116+1🕒	0.95	“ protection”“ protective”“ protecting”
4. L8N5418-1🕒	0.95	“ protection”“ safety”“ protected”
5. L6N8874+1🕒	0.95	“ protection”“ protect”“ always”
6. L5N8805+1🕒	0.95	“ protection”“ protective”“ protecting”
7. L1N5582-1💥	0.95	“ protection”“ protect”“ Protection”
8. L11N8189+1🕒	0.95	“ protection”“ protect”“ protected”
9. L9N3837-1🕒	0.94	“ protection”“ protecting”“ protected”
10. L7N2219-1🕒	0.94	“ protection”“ safety”“ protected”

Shadow specimen, apart from the ten: L3N6748+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 3, neuron 6748, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 2.53, top-1 agreement 0.40) 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
L3N6748+1🕒 shadow specimen t0-v0.1	1.418	0.51	1.769	1.71	“⏎”“<\|im_end\|>”“⏎⏎”“ question”“ info”“ questions”“"!”“ code”
L3N6748-1 t0-v0.1	2.530	0.40	2.958	2.35	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.