Layer 18 · Neuron 3513 · suppressed

L18N3513-1

Word quirks:

“ not”“ only”“not”“ Only”“ neither”“only”“只在”“⏎”

💤 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 Qwen, a large language model developed by Tongyi Lab. My mission is to provide helpful, safe, and efficient assistance across a wide range of tasks. Whether you need creative writing, logical reasoning, coding, data analysis, or knowledge-based answers, I aim to be your reliable partner in problem-solving and learning. Additionally, I strive to be transparent, accessible, and aligned with human values—ensuring every interaction is as clear and useful as possible. How would you like to start? 🌟

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. L20N5043-1🕒	0.87	“ only”“ not”“ apply”
2. L20N6050-1🕒	0.87	“ only”“ nothing”“ no”
3. L18N4162-1🕒	0.87	“ only”“only”“ ONLY”
4. L19N9158+1💤	0.84	“ only”“only”“ [`”
5. L17N5956+1💤	0.83	“ only”“ nothing”“ nonexistent”
6. L17N7594-1🕒	0.83	“ only”“ ONLY”“ nowhere”
7. L13N8354+1🕒	0.82	“ only”“ ONLY”“only”
8. L16N5884+1🕒	0.82	“ only”“ ONLY”“ except”
9. L16N6246-1💤	0.82	“ only”“ ONLY”“ Only”
10. L15N8345-1💤	0.82	“ neither”“ only”“ both”

Shadow specimen, apart from the ten: L18N3513+1🕒, -0.03.

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 18, neuron 3513, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.47, top-1 agreement 0.72) 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
L18N3513+1🕒 shadow specimen t0-v0.1	2.121	0.47	2.197	1.89	“啊”“人啊”“呀”“了啊”“好啊”“的啊”“ shouldn”“ Ah”
L18N3513-1 t0-v0.1	0.474	0.72	0.641	1.11	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.