Layer 27 · Neuron 3589 · boosted

L27N3589+1

Word quirks:

“闪”“itet”“μον”“ider”“ushi”“ugh”“ Der”“Arn”

💤 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 Qwen-4.5, the latest version of the Qwen series of large language models developed by Bosun. My core capabilities include:

Who I Am

A 19.8B parameter model with 256,000 context windows, designed for complex reasoning, multi-lingual support (100+ languages), and specialized tasks like code generation, visual analysis, and autonomous agent workflows.
Built on Bosun's 3.5 architecture, I inherit its strong foundation while enhancing efficiency and accuracy through **Mo

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. L25N1355+1💤	0.67	“ーム”“ déro”“unker”
2. L30N8930+1💤	0.66	“翻”“inue”“ Sup”
3. L31N4532-1💥	0.66	“πω”“ieden”“龙的”
4. L24N3131+1💤	0.66	“ Mun”“удар”“ungk”
5. L31N8373+1💥	0.65	“仑”“един”“inic”
6. Stuck Record✅ L15N2035+1,L31N6507-1	0.65	“ień”“amak”“odon”
7. Mantra Machine✅ L31N6507-1	0.65	“ień”“amak”“odon”
8. The Broken Record✅ L15N2035-1,L31N6507-1	0.65	“ień”“amak”“odon”
9. L31N4529-1💥	0.65	“رم”“idar”“íž”
10. L28N8388-1🕒	0.65	“unks”“uj”“umon”

Shadow specimen, apart from the ten: L27N3589-1💤, 0.13.

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 27, neuron 3589, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.48, top-1 agreement 0.74) 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
L27N3589+1 t0-v0.1	0.475	0.74	0.487	1.29	Top boosted tokens ↑
L27N3589-1💤 shadow specimen t0-v0.1	0.484	0.78	0.486	1.25	“外”“外围”“ external”“ External”“External”“wad”“的外”“urr”

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.