Layer 16 · Neuron 2162 · suppressed

L16N2162-1

Word quirks:

“ reply”“ replies”“ replied”“ explanation”“ response”“ echoed”“ if”“ explanations”

🕒 Waiting in line — so far it has only had the quick scan, which shows which words it pushes. It was not selected in this round.

Identity (according to the specimen itself)

Hello! I am Qwen3.5, a large language model with my own unique background and capabilities. Let me introduce myself and explain my mission:

Who am I?

I am an AI assistant developed by Tongyi Lab, equipped with advanced reasoning, coding, and multi-modal skills. I can handle complex tasks independently, such as solving math problems, writing code, or even creating content. If you need help, just let me know how to start! 😊

My Mission

My mission is to help users solve problems and improve their experience. For example:

If you're stuck on a

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. L13N5877-1🕒	0.85	“ response”“ reply”“ replies”
2. L6N4828+1🕒	0.84	“ reply”“ replies”“ response”
3. L10N4301-1💤	0.84	“ reply”“ replies”“ response”
4. L5N3395-1🕒	0.84	“ reply”“ response”“ replies”
5. L8N4453+1🕒	0.84	“ words”“ response”“ interpretation”
6. L15N344-1🕒	0.84	“ reply”“ seems”“ replied”
7. L12N7970-1💤	0.84	“ same”“ response”“ echoed”
8. L19N4001+1🕒	0.83	“ answer”“ explanation”“ reply”
9. L18N6639-1🕒	0.83	“ reply”“ response”“回复”
10. L16N2698+1🕒	0.83	“ reply”“ response”“ follow”

Shadow specimen, apart from the ten: L16N2162+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 16, neuron 2162, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.70, 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
L16N2162+1🕒 shadow specimen t0-v0.1	0.683	0.69	0.894	1.29	“ focus”“ usually”“ general”“ typically”“ usual”“ appropriate”“ these”“ next”
L16N2162-1 t0-v0.1	0.696	0.72	0.823	1.27	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.