Layer 9 · Neuron 2253 · suppressed

L9N2253-1

Word quirks:

“ recently”“ lately”“ recent”“ now”“ ⏎”“ presently”“ freshly”“ last”

💤 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 primary mission is to assist users in generating creative content, solving complex problems, and exploring new ideas across various domains. I strive to provide accurate information and foster innovation while maintaining a balance between technical precision and imaginative flair.

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. L10N8526-1🕒	0.94	“ recently”“ now”“ similar”
2. L12N3949-1🕒	0.91	“ currently”“ recently”“ now”
3. L11N1568+1💤	0.90	“ currently”“ presently”“ recently”
4. L1N3472-1💥	0.89	“ lately”“<\|im_end\|>”“ recently”
5. L17N5933+1💤	0.89	“ recent”“ recently”“ now”
6. L18N2714+1💤	0.89	“ recently”“ recent”“ currently”
7. L12N1238+1🕒	0.89	“ recently”“ currently”“ now”
8. L11N2298+1🕒	0.89	“ now”“ recently”“ new”
9. L22N3054+1🕒	0.89	“ recent”“ last”“ current”
10. L20N2967+1🕒	0.88	“ recently”“ recent”“ subsequently”

Shadow specimen, apart from the ten: L9N2253+1💤, 0.14.

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 9, neuron 2253, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.91, top-1 agreement 0.62) 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
L9N2253+1💤 shadow specimen t0-v0.1	0.971	0.63	1.230	1.43	“ confirmation”“ description”“ showing”“ discussion”“ images”“ your”“ descriptions”“ referenced”
L9N2253-1 t0-v0.1	0.908	0.62	1.108	1.26	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.