Layer 23 · Neuron 5977 · boosted

L23N5977+1

Word quirks:

“ dress”“ clothing”“ dresses”“衣”“ clothes”“服装”“Clo”“衣服”

🕒 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 Qin, a large-scale language forbustment (lingen) model developed by the Guangfu team of Guangfu. My mission is to provide high-quality, safe, and reliable assistance in a wide range of tasks, including but not limited to:

Answering questions: Providing accurate information on topics like history, science, and culture.
Helping with work: Assisting in document drafting, code writing, data analysis, and more.
Enabling learning: Offering explanations, study plans, and educational resources.
Supporting daily life:

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. Veneer✅ L4N8016+0.6,L23N4769-0.6	0.98	“衣服”“ clothes”“衣物”
2. The Shill✅ L3N5161+0.48,L23N4769-0.64	0.98	“clo”“ dresses”“衣物”
3. Drowsy Scholar✅ L19N4699-0.6,L23N4769-0.6	0.98	“ dresses”“ clothes”“ dress”
4. The Protagonist✅ L3N5161-0.48,L23N4769-0.64	0.97	“ clothes”“ dresses”“ clothing”
5. L0N8613-1💥	0.94	“<\|im_end\|>”“衣”“ clothes”
6. L23N4769-1💤	0.93	“ dresses”“衣物”“衣服”
7. L22N8791-1💤	0.93	“ clothing”“ Clothing”“ dress”
8. L20N8902-1🕒	0.93	“ dressing”“ clothing”“ attire”
9. L23N7486-1🕒	0.92	“ apparel”“衣”“ attire”
10. L17N8207+1🕒	0.91	“ dress”“ dresses”“ dressed”

Shadow specimen, apart from the ten: L23N5977-1🕒, 0.31.

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 5977, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.99, top-1 agreement 0.68) 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
L23N5977+1 t0-v0.1	0.988	0.68	1.074	1.33	Top boosted tokens ↑
L23N5977-1🕒 shadow specimen t0-v0.1	0.876	0.69	0.963	1.37	“手串”“WP”“手链”“ cp”“ Brace”“腕”“ wrist”“brace”

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.