Layer 15 · Neuron 2244 · boosted

L15N2244+1

Word quirks:

“ example”“ examples”“ Example”“example”“ exempl”“示例”“ indeed”“/example”

💤 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, the latest large language model in the Qwen series. My mission is to provide accurate, helpful, and versatile assistance across a wide range of tasks. Whether it's answering questions, writing code, creating content, or performing complex reasoning, I aim to deliver high-quality support that meets user needs effectively.

For example:

Answering factual queries: Providing up-to-date information on current events, scientific facts, historical data, etc.
Content creation: Writing articles, stories, scripts, emails, or any other type of text.
Code generation and debugging:

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. L13N5316+1🕒	0.97	“ example”“ examples”“ Example”
2. L15N8688-1🕒	0.97	“ example”“example”“ Example”
3. L14N6714-1🕒	0.97	“ example”“ examples”“ exempl”
4. L5N5716-1🕒	0.96	“ example”“ examples”“ Example”
5. L15N8213-1🕒	0.96	“ example”“ examples”“example”
6. L15N7818-1🕒	0.96	“ example”“ examples”“ Example”
7. L12N4966+1🕒	0.96	“ example”“ Example”“ examples”
8. L15N5794+1🕒	0.96	“ example”“ Example”“ particularly”
9. L17N8726+1🕒	0.96	“ example”“ [”“ examples”
10. L0N5838-1💥	0.96	“example”“ example”“ Example”

Shadow specimen, apart from the ten: L15N2244-1🕒, 0.02.

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 15, neuron 2244, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.55, top-1 agreement 0.74) 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
L15N2244+1 t0-v0.1	0.552	0.74	0.755	1.08	Top boosted tokens ↑
L15N2244-1🕒 shadow specimen t0-v0.1	0.853	0.67	1.059	1.55	“ —”“⏎”“ ——”“.—”“ ―”“.)”“.””“ every”

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.