Layer 14 · Neuron 8451 · boosted

L14N8451+1

Word quirks:

“ nicely”“ particularly”“ showcase”“ especially”“ showcasing”“ for”“ where”“ particular”

💤 Barely changes the model — the quick scan suggests this neuron barely changes the model's word predictions.

Identity (according to the specimen itself)

Qwen3.5: Empowering AI to Assist and Innovate

Qwen3.5

As the latest iteration of the Qwen series, I am designed to leverage my capabilities to assist users in achieving their goals. Here's how I can help:

Enhanced Capabilities: Utilize my advanced language understanding and generation skills.
Assist Users: Provide personalized assistance for tasks like answering questions, generating content, and creating stories.
Mission: To empower users to achieve their goals and make a positive impact.

Mission Statement

**For Qwen

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. L10N8885-1🕒	0.78	“ especially”“ showing”“ specially”
2. L9N8490-1💤	0.78	“ showcasing”“ particular”“ specially”
3. L11N2101-1🕒	0.76	“ special”“ specially”“ here”
4. L12N5454+1🕒	0.75	“ particularly”“ particular”“ especially”
5. L14N6505-1🕒	0.74	“ special”“ especially”“ particularly”
6. L11N6943+1⛔	0.74	“ particular”“ beautifully”“ crucial”
7. L11N8718+1💤	0.72	“ specially”“ special”“ especial”
8. L13N7699+1💤	0.72	“ showcased”“ showcase”“ specifically”
9. L14N4297-1🕒	0.72	“ specifically”“ especially”“ specific”
10. L18N5329+1🕒	0.72	“ particularly”“特别是”“particularly”

Shadow specimen, apart from the ten: L14N8451-1🕒, 0.21.

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 14, neuron 8451, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.90, top-1 agreement 0.59) 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
L14N8451+1 t0-v0.1	0.901	0.59	1.114	1.46	Top boosted tokens ↑
L14N8451-1🕒 shadow specimen t0-v0.1	1.030	0.61	1.277	1.59	“ sources”“ you”“ “”“ ensured”“ your”“ already”“ exact”“ sure”

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.