Layer 10 · Neuron 3186 · suppressed

L10N3186-1

Word quirks:

“ myself”“ my”“ simultaneously”“ additionally”“ ____”“ additional”“ ⏎”“ ⏎”

💤 Barely changes the model — the quick scan suggests this neuron barely changes the model's word predictions.

Identity (according to the specimen itself)

I am Qwen3.5, a large language model developed by Alibaba Cloud. As the most advanced AI in my knowledge base, I am designed to assist users with complex tasks and provide high-quality, reliable information. Here are my key capabilities:

1. Core Functions

Document Analysis: I can process documents to extract and summarize information, ensuring accuracy and completeness.
Multilingual Support: I support over 100 languages, including Chinese, English, French, Spanish, Arabic, and many others.
Specialized Domains: For topics like medicine, law, technology,

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. L10N8757-1🕒	0.78	“ ⏎⏎”“ additionally”“ additional”
2. L11N779+1🕒	0.76	“ additionally”“ ⏎”“ additional”
3. L9N4814+1💤	0.74	“ ⏎”“ ourselves”“ individually”
4. L13N4305+1🕒	0.74	“ ⏎”“⏎”“ additionally”
5. L10N1713-1🕒	0.72	“ myself”“ my”“ initially”
6. L14N5287+1🕒	0.72	“<\|im_end\|>”“\n”“ also”
7. L14N6826-1🕒	0.72	“ additionally”“ ⏎”“ moreover”
8. L16N330-1🕒	0.71	“ ⏎”“ also”“ ⏎ ⏎”
9. L14N2794-1💤	0.71	“ also”“⏎”“ additionally”
10. L10N4836-1💤	0.71	“ myself”“ my”“ ⏎”

Shadow specimen, apart from the ten: L10N3186+1💤, 0.19.

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 3186, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.98, top-1 agreement 0.68) 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
L10N3186+1💤 shadow specimen t0-v0.1	0.986	0.61	1.167	1.47	“ "”“ "”“ typically”“ exploited”“ essentially”“ **“”“ –”“ discern”
L10N3186-1 t0-v0.1	0.982	0.68	1.305	1.48	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.