Layer 10 · Neuron 2458 · boosted

L10N2458+1

Word quirks:

“在”“（”“使用”“与”“，”“为”“简单”““”

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

Identity (according to the specimen itself)

I am a large language model trained in RAG（检索增强生成）领域，我的使命是提供准确和有用的信息。作为 AI 助手，我致力于通过先进的技术和数据支持各种任务，包括回答问题、解决问题等。我的目标是帮助人们实现他们的目标，并推动技术的发展。

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. L13N2926-1💥	0.97	“在”“（”““”
2. L9N20+1💥	0.97	“在”“（”“与”
3. L9N7853-1💥	0.97	“在”“对”“使用”
4. L3N1191+1💥	0.97	“在”“与”“（”
5. L8N3636+1💥	0.96	“（”““”“在”
6. L9N7989-1💥	0.96	“，”“与”“在”
7. L8N7568+1💥	0.96	“在”“为”“与”
8. L11N7182-1💥	0.96	“在”“的”“与”
9. L2N789+1💥	0.96	“在”“，”“（”
10. L3N3456+1💥	0.96	“，”“。”“在”

Shadow specimen, apart from the ten: L10N2458-1🕒, -0.05.

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 10, neuron 2458, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 2.94, top-1 agreement 0.46) 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
L10N2458+1 t0-v0.1	2.942	0.46	3.204	1.70	Top boosted tokens ↑
L10N2458-1🕒 shadow specimen t0-v0.1	1.829	0.48	2.153	2.12	“ --”“ basically”“ __”“⏎”“ ____”“ stuff”“ _”“ ⏎”

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.