Layer 13 · Neuron 5037 · boosted

L13N5037+1

Word quirks:

“ description”“ particular”“ detailed”“ here”“ describing”“ basically”“ specifics”“ descriptions”

💤 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, a large language model developed by Alibaba Cloud's Tongyi Lab. My mission is to empower users with cutting-edge AI capabilities. I aim to provide comprehensive support across various domains, from scientific research to everyday tasks. Whether it's solving complex problems, creating content, or assisting in coding, I strive to be the most capable assistant available. How can I help you today?

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. L15N2602+1🕒	0.90	“ below”“ description”“ describing”
2. L10N1934-1🕒	0.89	“ description”“ descriptions”“ descriptive”
3. L11N8796+1💤	0.89	“ description”“ specifics”“ describing”
4. L9N2163-1🕒	0.88	“ specifics”“ specific”“ describing”
5. L12N6952+1🕒	0.88	“ description”“ describing”“ specific”
6. L9N6959+1🕒	0.87	“ description”“ describing”“ specifically”
7. L8N7471+1🕒	0.87	“ description”“ specifics”“ describing”
8. L13N6781+1🕒	0.87	“ description”“ describing”“ specific”
9. L11N7664-1🕒	0.87	“ description”“ descriptions”“ describing”
10. L13N5905-1🕒	0.87	“ specific”“ specifics”“ detailed”

Shadow specimen, apart from the ten: L13N5037-1💤, 0.32.

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 13, neuron 5037, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.69, top-1 agreement 0.67) 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
L13N5037+1 t0-v0.1	0.687	0.67	0.859	1.16	Top boosted tokens ↑
L13N5037-1💤 shadow specimen t0-v0.1	0.719	0.69	0.939	1.44	“ mainly”“ mostly”“ other”“ others”“ including”“ includes”“ include”“ related”

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.