Layer 18 · Neuron 6600 · suppressed

L18N6600-1

Word quirks:

“ externally”“ external”“ outside”“外部”“ External”“external”“outside”“的外部”

💤 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, an AI assistant developed by Alibaba Cloud. My mission is to help humans achieve their goals by providing high-quality, human-aligned responses across a wide range of tasks. I can assist with:

Answering questions (knowledge-based or real-time)
Generating text (stories, creative writing, code, etc.)
Processing documents (analyzing reports, contracts, etc.)
Multilingual support (128 languages globally)
Code generation & deployment (developing complex systems)
Reasoning capabilities (mathematical problem

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. L29N8233-1💥	0.96	“外部”“ external”“ External”
2. L26N6963-1🕒	0.96	“ external”“外部”“external”
3. L22N5726-1🕒	0.96	“ external”“外部”“external”
4. L27N2298-1💥	0.96	“ External”“External”“ external”
5. L26N5114-1🕒	0.96	“ external”“ External”“ externa”
6. L27N7749-1🕒	0.96	“ External”“ external”“External”
7. L19N4951+1🕒	0.96	“ externally”“ exterior”“ external”
8. L27N6598+1🕒	0.96	“ External”“外”“ external”
9. L26N4585-1🕒	0.95	“ exterior”“ External”“ external”
10. L21N2139+1🕒	0.95	“ external”“ External”“外部”

Shadow specimen, apart from the ten: L18N6600+1💤, 0.09.

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 6600, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.63, 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
L18N6600+1💤 shadow specimen t0-v0.1	0.464	0.72	0.589	1.18	“ needed”“ interpretation”“ typical”“ interpreting”“ result”“ interaction”“ visits”“ seeming”
L18N6600-1 t0-v0.1	0.631	0.68	0.714	1.21	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.