Layer 23 · Neuron 7486 · suppressed

L23N7486-1

Word quirks:

“ apparel”“衣”“ attire”“ outfit”“衣着”“ Clo”“卫”“服装”

🕒 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 Clouda (or "Clouda" in some contexts), a virtual assistant designed to help users with tasks ranging from answering questions and writing text to coding, analyzing data, and more. My mission is to provide accurate, helpful, and user-friendly responses across various domains of knowledge and tasks.

If you have specific tasks or questions in mind, feel free to ask! 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. L23N5977+1🕒	0.92	“ dress”“ clothing”“ dresses”
2. L20N8902-1🕒	0.92	“ dressing”“ clothing”“ attire”
3. Veneer✅ L4N8016+0.6,L23N4769-0.6	0.91	“衣服”“ clothes”“衣物”
4. The Shill✅ L3N5161+0.48,L23N4769-0.64	0.91	“clo”“ dresses”“衣物”
5. Drowsy Scholar✅ L19N4699-0.6,L23N4769-0.6	0.90	“ dresses”“ clothes”“ dress”
6. L30N9073+1💤	0.89	“穿着”“穿戴”“ dress”
7. L22N8791-1💤	0.89	“ clothing”“ Clothing”“ dress”
8. L31N2016+1💥	0.87	“衣”“Clo”“clo”
9. L0N8613-1💥	0.87	“<\|im_end\|>”“衣”“ clothes”
10. L17N8207+1🕒	0.86	“ dress”“ dresses”“ dressed”

Shadow specimen, apart from the ten: L23N7486+1💤, 0.16.

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 7486, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.81, top-1 agreement 0.67) 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
L23N7486+1💤 shadow specimen t0-v0.1	0.760	0.72	0.719	1.37	“ Emilio”“ red”“损失的”“ distribution”“ metal”“一米”“Red”“ width”
L23N7486-1 t0-v0.1	0.809	0.67	0.975	1.42	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.