Layer 29 · Neuron 6315 · suppressed

L29N6315-1

Word quirks:

“波”“美”“新”“年新”“了新”“一个新”“-new”“公”

🕒 Waiting in line — so far it has only had the quick scan, which shows which words it pushes. It was not selected in this round.

Identity (according to the specimen itself)

I am Qwen3.5, the latest large language model developed by Alibaba Cloud's Tong Wenzhi (美新美新) team. My mission is to provide safe, accurate, and helpful assistance across a wide range of tasks—whether solving complex problems, creating content, analyzing data, or simply having a conversation.

Key aspects of my mission:

Empower users: Help individuals and organizations solve real-world challenges through clear, logical, and context-aware responses.
Wise assistance: Offer expert-level support in technical, creative, and general domains while adhering to ethical

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. L31N6107+1💥	0.93	“新”“了新”“与新”
2. L31N9195-1💥	0.87	“新的”“人的”“ new”
3. L24N5281-1🕒	0.87	“新”“新方法”“厚德”
4. L25N5989+1🕒	0.86	“新”“ New”“了新”
5. L20N4571+1🕒	0.84	“ new”“new”“新的”
6. L21N7112+1💤	0.84	“新光”“新的”“LEX”
7. L6N30+1💥	0.83	“指”“元”“侧”
8. L27N18+1💥	0.83	“分”“新”“的”
9. L30N135-1🕒	0.83	“水”“人”“色”
10. L21N335-1💤	0.83	“新材料”“处在”“cana”

Shadow specimen, apart from the ten: L29N6315+1💤, 0.73.

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 29, neuron 6315, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.43, top-1 agreement 0.78) 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
L29N6315+1💤 shadow specimen t0-v0.1	0.304	0.82	0.375	1.61	“怨”“惠”“恩”“ঁ”“稀释”“尝”“非”“าว”
L29N6315-1 t0-v0.1	0.435	0.78	0.504	1.52	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.