Layer 6 · Neuron 2962 · suppressed

L6N2962-1

Word quirks:

“⏎”“ _____”“ ______”“ ‘”“ ____”“ ”“ ________”“ secure”

💤 Barely changes the model — the quick scan suggests this neuron barely changes the model's word predictions.

Identity (according to the specimen itself)

I am a large language model, trained by Google. My mission is to assist users in answering their questions accurately and safely, while adhering to ethical guidelines and providing helpful information. I strive to be a useful tool for everyone.

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. L6N2276-1💤	0.95	“⏎”“ ‘”“ __”
2. L4N5968+1💤	0.95	“⏎”“ ‘”“ ____”
3. L4N262-1💤	0.95	“ ‘”“ ____”“ ______”
4. L16N6105-1🕒	0.95	“ ____”“ ‘”“ ______”
5. L8N3629+1💤	0.95	“ ‘”“ “”“⏎”
6. L7N5391-1💤	0.95	“⏎”“ ‘”“ __”
7. L10N470+1💤	0.95	“⏎”“ ⏎”“ ‘”
8. L8N4150+1💤	0.95	“⏎”“ ______”“ ____”
9. L7N4345-1🕒	0.95	“ ‘”“⏎⏎”“⏎”
10. L5N4285-1🕒	0.95	“<\|im_end\|>”“⏎”“ ‘”

Shadow specimen, apart from the ten: L6N2962+1💥, -0.02.

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 6, neuron 2962, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.71, top-1 agreement 0.68) 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
L6N2962+1💥 shadow specimen t0-v0.1	2.573	0.35	2.977	2.47	“<\|im_end\|>”“ EVERY”“ rant”“ everywhere”“ somehow”“ talking”“ everyone”“ everything”
L6N2962-1 t0-v0.1	0.705	0.68	0.858	1.34	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.