Layer 9 · Neuron 8019 · boosted

L9N8019+1

Word quirks:

“ description”“ describing”“ descriptions”“ describe”“ descriptive”“ describes”“ descri”“ described”

💤 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 a virtual assistant designed to assist with a wide range of tasks. My mission is to provide helpful, context-aware support to users by understanding their needs and the situation at hand. While I don't have personal experiences or feelings like a human would, my role is to ensure that every interaction is smooth, efficient, and tailored to the user's requirements. If you're describing a specific scenario, feel free to elaborate on how you envision the outcome!

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. L17N5489-1🕒	0.99	“ description”“ described”“ describing”
2. L11N3051-1🕒	0.99	“ description”“ describing”“ descriptions”
3. L17N8003+1🕒	0.98	“ description”“ describing”“描述”
4. L15N5305+1🕒	0.98	“ described”“ description”“ describes”
5. L13N4916-1🕒	0.98	“ describes”“ description”“ describing”
6. L12N3053+1🕒	0.97	“ description”“ describing”“ describe”
7. L17N7479+1💤	0.97	“ described”“ description”“ descriptions”
8. L15N8569+1🕒	0.97	“ describing”“ description”“ described”
9. L11N1015+1🕒	0.97	“ description”“ described”“ descriptions”
10. L12N9173-1🕒	0.97	“ description”“ describing”“ ⏎”

Shadow specimen, apart from the ten: L9N8019-1🕒, 0.27.

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 9, neuron 8019, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.87, top-1 agreement 0.63) 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
L9N8019+1 t0-v0.1	0.869	0.63	1.151	1.25	Top boosted tokens ↑
L9N8019-1🕒 shadow specimen t0-v0.1	1.208	0.60	1.489	1.54	“ display”“ displayed”“ displays”“ displaying”“ visible”“ visually”“ caption”“ printed”

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.