Layer 25 · Neuron 6366 · boosted

L25N6366+1

Word quirks:

“ stars”“ star”“星”“star”“ Star”“ Stars”“-star”“-Star”

🕒 Waiting in line — so far it has only had the quick scan, which shows which words it pushes. It was not selected in this round.

Identity (according to the specimen itself)

I am Starry, an advanced artificial intelligence designed to inspire creativity, solve complex problems, and generate high-quality content across a wide range of topics. My core purpose is to serve as a dynamic and intelligent companion for your intellectual and creative pursuits.

My mission is to:

Inspire Creativity: Help you write stories, compose music, design art, or craft innovative ideas by offering fresh perspectives and inspiring inspiration.
Enhance Learning: Clarify complex concepts, summarize information, and provide personalized learning paths tailored to your interests and goals.
Resolve Challenges: Assist in problem-solving,

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. L24N6216+1🕒	0.99	“ star”“星”“-star”
2. L24N8191-1🕒	0.99	“ star”“ Star”“ stars”
3. L27N6586+1🕒	0.98	“ star”“ stars”“star”
4. L31N2481+1💥	0.98	“ star”“star”“ Star”
5. L25N8907-1🕒	0.98	“ stars”“ star”“星”
6. L24N4143-1🕒	0.98	“星”“ star”“-star”
7. L28N8976-1🕒	0.97	“ star”“star”“ stars”
8. L0N7324-1💥	0.97	“ star”“star”“-star”
9. L29N6085-1🕒	0.97	“ star”“星”“ Star”
10. L31N5432+1💥	0.97	“ star”“星”“star”

Shadow specimen, apart from the ten: L25N6366-1💤, 0.11.

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 25, neuron 6366, 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 0.69) 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
L25N6366+1 t0-v0.1	0.865	0.69	0.866	1.33	Top boosted tokens ↑
L25N6366-1💤 shadow specimen t0-v0.1	0.603	0.72	0.612	1.22	“口”“ fil”“fu”“十四”“ippe”“驾校”“uus”“Fu”

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.