Layer 15 · Neuron 2401 · boosted

L15N2401+1

Word quirks:

“ individual”“ complete”“ exact”“ suitable”“ individually”“ precise”“ creation”“ specific”

💤 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, a large language model that empowers users to create content with ease. My mission is to provide high-quality, safe, and efficient services to help people achieve their goals by leveraging my capabilities in natural language processing, including text generation, understanding, and analysis.

Key Components of My Mission:

Empowerment: I aim to make it easy for individuals and organizations to use advanced AI tools to simplify complex tasks, enabling them to focus on innovation and growth.
Safety & Ethics: My design prioritizes safety, ensuring that all interactions are free from harm, bias, or

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. L9N4496-1🕒	0.80	“ individual”“ individually”“ typical”
2. The Threat Modeler✅ L11N4906-1	0.79	“ precise”“ direct”“ specific”
3. L11N5286-1💤	0.79	“ each”“ individually”“ respective”
4. L9N2307-1💤	0.77	“ individually”“ individual”“ multiple”
5. L17N7982+1🕒	0.77	“ specifically”“ specific”“ uniquely”
6. L9N2404-1💤	0.77	“ specific”“ individual”“ my”
7. L12N6909+1🕒	0.77	“ specific”“ specifics”“ Specific”
8. L10N2901+1🕒	0.77	“ specific”“ individual”“ precise”
9. L9N8661+1🕒	0.77	“ specific”“ individual”“ individually”
10. L16N8150+1🕒	0.76	“ specific”“ individual”“ specifics”

Shadow specimen, apart from the ten: L15N2401-1🕒, 0.28.

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 2401, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.51, 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
L15N2401+1 t0-v0.1	0.506	0.68	0.700	1.24	Top boosted tokens ↑
L15N2401-1🕒 shadow specimen t0-v0.1	0.928	0.66	1.084	1.35	“ achieving”“ saying”“ accepting”“ deciding”“ receiving”“ ——”“ resolving”“ conveying”

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.