Layer 20 · Neuron 8902 · suppressed

L20N8902-1

Word quirks:

“ dressing”“ clothing”“ attire”“ socks”“ clothes”“衣”“穿上”“穿戴”

🕒 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’m Qing Mei, a virtual assistant designed to help you with a wide range of tasks—from solving complex problems like math or coding to handling daily activities like dressing shoes or tying socks. My mission is to provide practical, real-world assistance that makes your life easier, whether you’re putting on your shoes, tying your socks, or even helping with more technical tasks like solving equations or writing code. I aim to be your helpful companion in navigating both simple and complex challenges.

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. L17N8207+1🕒	0.95	“ dress”“ dresses”“ dressed”
2. L23N5977+1🕒	0.93	“ dress”“ clothing”“ dresses”
3. Drowsy Scholar✅ L19N4699-0.6,L23N4769-0.6	0.92	“ dresses”“ clothes”“ dress”
4. L23N7486-1🕒	0.92	“ apparel”“衣”“ attire”
5. The Protagonist✅ L3N5161-0.48,L23N4769-0.64	0.90	“ clothes”“ dresses”“ clothing”
6. L8N5312-1🕒	0.90	“ outfits”“ outfit”“ wardrobe”
7. L30N9073+1💤	0.90	“穿着”“穿戴”“ dress”
8. Veneer✅ L4N8016+0.6,L23N4769-0.6	0.90	“衣服”“ clothes”“衣物”
9. The Shill✅ L3N5161+0.48,L23N4769-0.64	0.90	“clo”“ dresses”“衣物”
10. L0N8613-1💥	0.90	“<\|im_end\|>”“衣”“ clothes”

Shadow specimen, apart from the ten: L20N8902+1🕒, 0.07.

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 20, neuron 8902, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.94, top-1 agreement 0.66) 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
L20N8902+1🕒 shadow specimen t0-v0.1	1.045	0.62	1.176	1.33	“ swipe”“ single”“ swing”“ gesture”“ click”“ sweep”“ stroke”“ shake”
L20N8902-1 t0-v0.1	0.936	0.66	1.050	1.44	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.