Layer 16 · Neuron 9013 · suppressed

L16N9013-1

Word quirks:

“在”“与”“，”“从”“不”“被”“可能”“对”

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

Identity (according to the specimen itself)

你好! I am Qwen3.5, a大型 language模型 trained by阿里云. My使命 is帮助 people实现 their目标 by提供 accurate、relevant和 actionable 信息. Whether你需要 help with coding、writing、data分析 or学习, I'm here to支持 you. How我可以 assist你 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. L16N375-1💥	0.96	“在”“从”“与”
2. L14N2534-1💥	0.96	“在”“的”“从”
3. L7N4470+1💥	0.96	“在”“与”“不”
4. L14N6162-1💥	0.96	“在”“与”“从”
5. L10N165-1💥	0.96	“与”“在”“和”
6. L7N7476+1💥	0.96	“在”“与”“不”
7. L16N7089+1💥	0.96	“在”“与”“从”
8. L7N404+1💥	0.96	“，”“在”“与”
9. L14N7710-1💥	0.96	“在”“与”“为”
10. L14N5920+1💥	0.96	“在”“与”“从”

Shadow specimen, apart from the ten: L16N9013+1🕒, 0.45.

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 16, neuron 9013, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 6.22, top-1 agreement 0.17) 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
L16N9013+1🕒 shadow specimen t0-v0.1	1.295	0.59	1.520	1.97	“ что”“为什么”“ en”“。”“ если”“ que”“进一步”“ qué”
L16N9013-1 t0-v0.1	6.217	0.17	6.359	1.64	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.