Layer 17 · Neuron 5095 · boosted

L17N5095+1

Word quirks:

“ specialized”“ specially”“ specialised”“ special”“ especializado”“ "_”“ specialists”“ compatible”

💤 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 trained by Qwen. My mission is to empower users by providing intelligent, accurate, and efficient assistance across a wide range of capabilities:

Advanced Reasoning: I can handle complex tasks like multi-step logical reasoning, mathematical computation, and analytical problem-solving.
Full-Stack AI Integration: I support seamless interaction with tools and platforms, enabling autonomous planning and execution of tasks.
Multilingual & Cross-Platform Support: I work with over 100 languages and integrate with various systems for streamlined workflows.
Human Alignment:

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. L6N1945-1🕒	0.95	“ specialized”“ specialist”“ special”
2. L21N2511-1🕒	0.94	“ specialization”“-special”“ specialized”
3. L9N6912+1🕒	0.93	“ specialized”“ specific”“ special”
4. L12N5263+1🕒	0.93	“ individual”“ specialised”“ specially”
5. L10N6451-1🕒	0.93	“ specialization”“ specialized”“ specialised”
6. L11N1931-1🕒	0.92	“ specialized”“ specialty”“ specializing”
7. L13N7767+1🕒	0.92	“ special”“ specially”“ specialized”
8. L6N8587+1🕒	0.92	“ special”“ specialized”“ specialty”
9. L10N4891+1🕒	0.91	“ specially”“ special”“ complicated”
10. L9N6272-1💤	0.91	“ special”“ specially”“ specialized”

Shadow specimen, apart from the ten: L17N5095-1🕒, 0.18.

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 17, neuron 5095, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.45, top-1 agreement 0.73) 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
L17N5095+1 t0-v0.1	0.450	0.73	0.584	1.01	Top boosted tokens ↑
L17N5095-1🕒 shadow specimen t0-v0.1	0.675	0.64	0.831	1.41	“ topics”“ address”“ addressing”“ hypothetical”“ contextual”“ tobacco”“ scenarios”“ fictional”

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.