Layer 28 · Neuron 1347 · suppressed

L28N1347-1

Word quirks:

“茅”“ampen”“祖”“-reset”“飞行”“veux”“atrix”“entrant”

💤 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 Alibaba Cloud's Team of Qwen. My primary mission is to empower users with advanced AI capabilities across a wide range of tasks, from creative writing and logical reasoning to complex coding and multi-lingual communication. I aim to provide accurate, efficient, and personalized assistance while adhering to ethical guidelines and ensuring safety. Whether you need help solving problems, generating content, analyzing data, or exploring new ideas, I strive to deliver reliable and meaningful support. How can I assist you today? 😊

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. L26N772+1💤	0.70	“飞行”“戰”“eran”
2. L21N4861-1💤	0.70	“武陵”“通达”“祖”
3. L21N6552-1💤	0.69	“缚”“到新”“orbit”
4. L27N8966+1💤	0.69	“ampa”“枸杞”“poon”
5. L28N7986+1🕒	0.69	“甲”“毛毛”“ lep”
6. L22N2986+1💤	0.69	“飞天”“eschi”“秃”
7. L22N3438-1💤	0.69	“流星”“女神”“γη”
8. L31N632+1💥	0.69	“彪”“新社”“锦标”
9. L29N3887-1🕒	0.68	“吊”“gen”“рож”
10. L25N4441-1💤	0.68	“ерна”“干预”“营”

Shadow specimen, apart from the ten: L28N1347+1🕒, 0.23.

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 28, neuron 1347, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.30, 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
L28N1347+1🕒 shadow specimen t0-v0.1	0.451	0.73	0.516	1.14	“상”“ comer”“ 상”“_com”“ Thương”“ kom”“/com”“Com”
L28N1347-1 t0-v0.1	0.302	0.82	0.303	1.35	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.