Layer 30 · Neuron 4938 · suppressed

L30N4938-1

Word quirks:

“uger”“ustain”“ inn”“大利”“alg”“uzu”“utu”“acce”

💤 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 core mission is to be a versatile, intelligent, and responsible AI assistant that empowers users across a wide range of scenarios. Here's how I align with this mission:

Key Responsibilities:

Universal Assistance:

Provide accurate, context-aware answers for tasks like coding, creative writing, data analysis, and complex reasoning.
Handle multilingual interactions fluently (over 100 languages) and adapt to cultural nuances.

Depth & Precision:

Leverage advanced

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. L29N9197-1🕒	0.75	“ule”“uts”“riba”
2. L30N4529+1💤	0.75	“ugel”“URT”“机构”
3. L30N6534+1🕒	0.75	“alk”“ut”“uts”
4. L31N8118+1💥	0.74	“谷”“ume”“uges”
5. L31N5941-1💥	0.73	“úd”“udu”“mps”
6. L30N8230-1💤	0.73	“enner”“unc”“ule”
7. L31N8369-1💥	0.73	“utine”“arne”“墨西哥”
8. L26N4486-1💤	0.72	“enberg”“ichen”“植物”
9. L21N446-1🕒	0.72	“uster”“ude”“berg”
10. L31N2718+1💥	0.72	“uret”“agar”“ubu”

Shadow specimen, apart from the ten: L30N4938+1💥, -0.03.

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 4938, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.32, top-1 agreement 0.82) 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
L30N4938+1💥 shadow specimen t0-v0.1	3.917	0.09	4.052	0.63	“ error”“error”“ Error”“_error”“ exception”“-error”“Error”“exception”
L30N4938-1 t0-v0.1	0.320	0.82	0.311	1.72	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.