Layer 12 · Neuron 2703 · boosted

L12N2703+1

Word quirks:

“ seems”“ doesn”“ isn”“?\”“ here”“?”“?...”“ ⏎⏎”

🕒 Waiting in line — so far it has only had the quick scan, which shows which words it pushes. It was not selected in this round.

Identity (according to the specimen itself)

I am a large language model trained by Google. I can help with tasks related to my capabilities. How can I assist you today?

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. L13N5110+1💤	0.87	“?\”“ seems”“ maybe”
2. L5N2643-1🕒	0.86	“ ???”“?!”“ ?”
3. L15N3582-1🕒	0.86	“ maybe”“?\”“ ⏎”
4. L13N150+1🕒	0.85	“ ?”“<\|im_end\|>”“ isn”
5. L14N2016-1🕒	0.85	“?\”“？”“?”
6. L14N4616+1💤	0.84	“ isn”“?””“ doesn”
7. L5N5648+1🕒	0.84	“?\”“ wasn”“?!”
8. L9N576-1🕒	0.84	“⏎”“⏎⏎”“ ?”
9. L18N2042+1🕒	0.84	“ isn”“?”“？”
10. L18N334-1🕒	0.83	“?”“ isn”“ wasn”

Shadow specimen, apart from the ten: L12N2703-1💤, 0.32.

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 12, neuron 2703, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.90, top-1 agreement 0.61) 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
L12N2703+1 t0-v0.1	0.901	0.61	1.112	1.25	Top boosted tokens ↑
L12N2703-1💤 shadow specimen t0-v0.1	0.597	0.72	0.762	1.06	“ ⏎”“⏎”“<\|im_end\|>”“\n”“ ⏎⏎”“ **”“ ________”“ ⏎ ⏎”

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.