Specimen ledger / L11N8401-1

Layer 11 · Neuron 8401 · suppressed

L11N8401-1

Word quirks:

“在”“与”“从”“被”“为”“，”“使用”“对”

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

Identity (according to the specimen itself)

我是 Qwen3.5 代码生成与推理模型。我的使命是**赋能全球开发者、研究人员和创意人士，帮助他们构建下一代人工智能应用。具体来说，我致力于：

提升效率：通过自动化繁琐的任务来释放生产力。
促进创新：通过探索新的技术可能性来推动科学进步。
实现影响力：通过解决现实世界的问题来改善人类生活。

无论您是想构建一个复杂的推荐系统，还是设计一个聊天机器人，或者只是想写一段简单的 Python 代码，我都可以为您提供支持。

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. L6N5062-1💥	0.99	“在”“被”“从”
2. L12N1085+1💥	0.99	“在”“与”“为”
3. L10N3232-1💥	0.99	“在”“与”“从”
4. L12N4074+1💥	0.98	“在”“使用”“与”
5. L7N404+1💥	0.98	“，”“在”“与”
6. L4N3726-1💥	0.98	“在”“（”“与”
7. L15N6284+1💥	0.98	“在”“与”“从”
8. L14N1475-1💥	0.98	“在”“与”“为”
9. L10N1251+1💥	0.98	“在”“与”“从”
10. L3N4131-1💥	0.98	“的”“在”“为”

Shadow specimen, apart from the ten: L11N8401+1💥, 0.81.

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 11, neuron 8401, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 8.39, top-1 agreement 0.10) 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
L11N8401+1💥 shadow specimen t0-v0.1	9.712	0.04	10.173	4.07	“选”“选择”“后”“中”“烈”“表”“。”“前”
L11N8401-1 t0-v0.1	8.392	0.10	8.697	2.40	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.