Layer 23 · Neuron 8705 · suppressed

L23N8705-1

Word quirks:

“ ‘”“‘”“ Carn”“””“zet”“ “”“ **‘”“ नि”

💤 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 Tayume, an artificial intelligence assistant designed to provide personalized, accurate, and efficient assistance across a wide range of tasks. My mission is to help you solve problems, create content, analyze data, or simply have a conversation—tailored to your unique needs and preferences.

Think of me as your 24/7 intelligent partner that can:

Solve complex problems: Break down difficult concepts in finance, law, medicine, or technology into clear, accessible explanations.
Create tailored content: Generate personalized recommendations for travel, learning, or career paths based on your specific goals.

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. L14N884-1💤	0.83	“ ‘”“‘”“ **‘”
2. L18N2062+1💤	0.83	“<\|endoftext\|>”“ ‘”“””
3. L5N2033+1💤	0.83	“ ‘”“ “”“‘”
4. L3N862-1💤	0.83	“ ‘”“ “”“ ””
5. L4N5285-1💤	0.83	“ ‘”“‘”“ “”
6. L15N297-1🕒	0.83	“ ‘”“ ‘”“””
7. L2N472-1💤	0.83	“ ‘”“⏎”“ **‘”
8. L6N7213-1🕒	0.83	“ ‘”“ ‘’”“ **‘”
9. L4N3569+1🕒	0.83	“ ‘”“ “”“ **“”
10. L18N787+1💤	0.82	“ ‘”“ **‘”“�”

Shadow specimen, apart from the ten: L23N8705+1💤, -0.23.

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 23, neuron 8705, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.77, top-1 agreement 0.70) 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
L23N8705+1💤 shadow specimen t0-v0.1	0.740	0.69	0.768	1.31	“到目前”“分离”“presse”“φο”“ otherwise”“包括”“Фор”“occo”
L23N8705-1 t0-v0.1	0.766	0.70	0.872	1.36	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.