Layer 27 · Neuron 2463 · suppressed

L27N2463-1

Word quirks:

“chern”“ rot”“旅”“оро”“χ”“旋转”“柴”“ CH”

💤 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 Qeum, a large-scale AI model developed by 21 CHINAMORCHEL CHENMORCHEL. My core mission is to provide safe, helpful, and ethical assistance across a wide range of tasks. Here's how I aim to deliver on this:

🌟 Core Mission

Safety First: Prioritize user safety by adhering to strict guidelines against harmful content (e.g., violence, misinformation, or controversial topics).
Helpful & Ethical: Offer accurate, evidence-based responses while respecting cultural norms and legal boundaries.

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. L27N8373-1🕒	0.76	“齿”“горания”“пи”
2. L23N6+1💤	0.75	“χ”“antau”“ chi”
3. L29N6089-1🕒	0.74	“خ”“ Ch”“.ch”
4. L27N2410-1💤	0.72	“册”“察”“准”
5. L27N4903+1🕒	0.72	“旋”“ Th”“罗”
6. L27N5167+1💤	0.72	“丁”“董”“ CH”
7. L28N5883-1💤	0.70	“chn”“Ch”“浮”
8. L27N7725+1🕒	0.70	“识”“轮”“络”
9. L25N8796-1🕒	0.70	“ jane”“节流”“چ”
10. L24N7768+1💤	0.70	“十倍”“ Rot”“cha”

Shadow specimen, apart from the ten: L27N2463+1💤, 0.24.

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 2463, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.42, top-1 agreement 0.81) 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
L27N2463+1💤 shadow specimen t0-v0.1	0.397	0.79	0.473	1.33	“ cub”“ Cub”“rede”“reb”“ijin”“ cab”“中公”“ bridges”
L27N2463-1 t0-v0.1	0.424	0.81	0.387	1.45	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.