Layer 30 · Neuron 2587 · suppressed

L30N2587-1

Word quirks:

“ущ”“금”“ته”“етен”“énom”“芹”“ニング”“жие”

💤 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 Qwen3.5, the latest large language model developed by Tongyi Lab. My primary mission is to provide intelligent, reliable, and versatile assistance across a wide range of tasks—whether that's solving complex problems, creating content, analyzing data, or simply engaging in meaningful conversation.

Key aspects of my capabilities:

Advanced Reasoning: I excel at handling math, science, and logical reasoning tasks with high precision.
Multilingual Support: I fluently communicate over 100+ languages, making me accessible globally.
Long-Context Understanding: I can process and analyze

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. L27N1941-1💤	0.71	“照”“гор”“جان”
2. L31N1575-1💥	0.70	“bruch”“utas”“헌”
3. L31N8689+1💥	0.70	“шев”“acy”“onomia”
4. The Broken Record✅ L15N2035-1,L31N6507-1	0.70	“ień”“amak”“odon”
5. Mantra Machine✅ L31N6507-1	0.70	“ień”“amak”“odon”
6. Stuck Record✅ L15N2035+1,L31N6507-1	0.69	“ień”“amak”“odon”
7. L28N3400-1💤	0.69	“ourd”“墨”“�”
8. L31N8503-1💥	0.69	“egy”“ider”“稀”
9. L30N7829+1💤	0.69	“黎”“�”“upan”
10. L27N2509-1💤	0.69	“ynos”“履”“سیم”

Shadow specimen, apart from the ten: L30N2587+1💥, 0.02.

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 30, neuron 2587, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.24, top-1 agreement 0.89) 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
L30N2587+1💥 shadow specimen t0-v0.1	47.970	0.00	48.325	0.16	“ In”“In”“in”“_in”“-in”“.in”“>In”“-In”
L30N2587-1 t0-v0.1	0.240	0.89	0.190	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.