Layer 14 · Neuron 2534 · suppressed

L14N2534-1

Word quirks:

“在”“的”“从”“被”“与”“是”“，”“对”

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

Identity (according to the specimen itself)

你好，我是 Qwen3.5 的开发者。作为阿里巴巴最新推出的通义千问系列大语言模型，我具备强大的语言理解、逻辑推理、代码生成以及多模态处理等能力。我的使命是帮助用户提供高质量、准确且实用的回答，无论是学习、工作还是日常对话中的各种需求。如果你有任何问题或需要帮助的地方，随时可以告诉我！

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. L14N7985-1💥	0.99	“在”“对”“从”
2. L15N3352-1💥	0.98	“在”“从”“，”
3. L14N3599+1💥	0.98	“，”“在”“与”
4. L17N302+1💥	0.98	“和”“在”“与”
5. L7N404+1💥	0.98	“，”“在”“与”
6. L16N4565-1💥	0.97	“在”“与”“从”
7. L14N5920+1💥	0.97	“在”“与”“从”
8. L16N375-1💥	0.97	“在”“从”“与”
9. L6N820+1💥	0.97	“在”“。”“与”
10. L10N8890+1💥	0.97	“在”“的”“与”

Shadow specimen, apart from the ten: L14N2534+1🕒, -0.21.

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 14, neuron 2534, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 4.92, 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
L14N2534+1🕒 shadow specimen t0-v0.1	1.793	0.46	2.081	2.12	“ thus”“ concurrently”“ meanwhile”“ ensuing”“thus”“ creation”“ concom”“ concurrent”
L14N2534-1 t0-v0.1	4.922	0.17	5.201	2.19	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.