Layer 11 · Neuron 1385 · boosted

L11N1385+1

Word quirks:

“ carefully”“ just”“ careful”“ seems”“ specially”“ very”“ now”“ specifically”

🕒 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)

I am Qwen, a large language model developed by the Qwen team. My primary mission is to assist users in various tasks, from answering simple questions to handling complex projects. I strive to provide accurate and helpful information across a wide range of topics, ensuring that my responses are tailored to meet diverse user needs.

What can I do for you today? I can help you with:

Answering Questions: Whether you're asking about science, history, or technology, I'm here to provide reliable answers.
Creating Content: I can help you create content for various purposes, such as writing stories, creating schedules

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. L13N6587-1🕒	0.80	“ here”“ if”“ very”
2. L9N8425-1🕒	0.79	“ crucial”“ prep”“ carefully”
3. L11N6820+1🕒	0.79	“ now”“ similar”“ carefully”
4. L14N1787+1💤	0.79	“ specially”“ always”“ special”
5. L10N2059+1💤	0.78	“ always”“ specially”“ specifically”
6. L9N1486-1💤	0.78	“ always”“ precisely”“ appropriately”
7. L11N8664-1🕒	0.78	“ carefully”“ suitable”“ careful”
8. L12N1492+1💤	0.78	“ carefully”“ careful”“ very”
9. L11N5518+1🕒	0.77	“ thinking”“ thought”“ carefully”
10. L11N6963-1🕒	0.77	“ carefully”“ properly”“ balance”

Shadow specimen, apart from the ten: L11N1385-1🕒, 0.14.

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 11, neuron 1385, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.96, top-1 agreement 0.61) 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
L11N1385+1 t0-v0.1	0.962	0.61	1.220	1.47	Top boosted tokens ↑
L11N1385-1🕒 shadow specimen t0-v0.1	0.934	0.64	1.155	1.39	“ “”“ ‘”“ ””“ later”“ sometimes”“ „”“ my”“ imagined”

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.