Layer 14 · Neuron 105 · suppressed

L14N105-1

Word quirks:

“ similar”“ similarly”“ another”“⏎”“ likewise”“ same”“ --”“ –”

🕒 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)

*(Self-Corv: I am a large language model, trained on a vast array of text data, and I am an AI assistant, and I am not a human, and I am not a "large" language model, and I am not a "small" language model, and I am not a "medium" language model, and I am not a "large" language model, and I am not a "tiny" language model, and I am not a "micro" language model, and I am not a "nano" language model, and I am not a "macro" language model, and I am not

The AI judge wasn't sure this holds together.

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. L12N102-1🕒	0.93	“ similarly”“ similar”“⏎”
2. L13N3284+1🕒	0.93	“ similarly”“ similar”“ same”
3. L17N521-1💤	0.91	“ similarly”“ likewise”“ another”
4. L14N3863-1🕒	0.90	“ similarly”“ likewise”“ ⏎”
5. L15N7966-1💤	0.89	“⏎”“ similarly”“ likewise”
6. L14N6399-1🕒	0.89	“ similarly”“ here”“ similar”
7. L9N113-1🕒	0.89	“ similarly”“ additionally”“ similar”
8. L14N7695+1🕒	0.89	“ similarly”“ same”“ similar”
9. L16N6543-1💤	0.89	“ similarly”“ same”“ similar”
10. L13N374-1🕒	0.89	“ similarly”“ similar”“ again”

Shadow specimen, apart from the ten: L14N105+1💤, 0.18.

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 14, neuron 105, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 1.57, top-1 agreement 0.49) 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
L14N105+1💤 shadow specimen t0-v0.1	0.619	0.68	0.887	1.14	“ overall”“!”“!</”“ general”“ additionally”“:</”“ altogether”“!””
L14N105-1 t0-v0.1	1.565	0.49	1.935	1.85	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.