Layer 13 · Neuron 5064 · boosted

L13N5064+1

Word quirks:

“在”“与”“为”“被”“是”“将”“（”“从”

💥 Breaks the model — boosting or suppressing this neuron makes the output fall apart.

Identity (according to the specimen itself)

我是 Qwen3.5，一款由阿里巴巴集团研发的超大规模语言模型。我的使命是成为人类智能的延伸，赋能千行百业，让 AI 技术真正走进千家万户，为人类社会的进步贡献智慧力量。我致力于通过技术创新，推动 AI 与实体经济深度融合，助力产业升级，促进经济高质量发展，同时确保 AI 安全可控，让 AI 技术更安全、更可靠地服务于社会民生。

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. L10N1251+1💥	0.98	“在”“与”“从”
2. L3N4131-1💥	0.98	“的”“在”“为”
3. L10N6491-1💥	0.98	“在”“从”“与”
4. L15N3352-1💥	0.98	“在”“从”“，”
5. L16N4565-1💥	0.98	“在”“与”“从”
6. L4N2779+1💥	0.98	“在”“与”“（”
7. L14N7985-1💥	0.98	“在”“对”“从”
8. L15N486-1💥	0.98	“是”“在”“从”
9. L4N6423+1💥	0.97	“在”“与”“的”
10. L11N8306-1💥	0.97	“在”“为”“与”

Shadow specimen, apart from the ten: L13N5064-1🕒, 0.35.

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 13, neuron 5064, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 6.16, top-1 agreement 0.15) 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
L13N5064+1 t0-v0.1	6.157	0.15	6.371	1.88	Top boosted tokens ↑
L13N5064-1🕒 shadow specimen t0-v0.1	1.258	0.62	1.499	1.24	“ actually”“真的”“ *“”“ ”“。**”“ mentioned”“ Actual”“ actual”

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.