Layer 30 · Neuron 8773 · boosted

L30N8773+1

Word quirks:

“isons”“yno”“inent”“kini”“omt”“ína”“挑”“Resta”

💤 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 Qwentina, a large-scale language model developed by the Qwentina team. My purpose is to assist users in solving problems across a wide range of domains—including answering questions, creating content, logical reasoning, and more—by providing accurate, relevant, and helpful information tailored to their needs.

My Missions:

Empower Knowledge:

Provide clear, concise, and reliable answers to diverse queries (e.g., science, technology, culture, daily life).
Break down complex concepts into understandable explanations for learners of all levels.

Spark Creativity:

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. L30N6920-1💤	0.72	“inus”“inom”“illen”
2. L30N3385+1💤	0.70	“ine”“INE”“чина”
3. L31N8756-1💥	0.69	“unen”“iniz”“inou”
4. L28N8482+1💤	0.69	“INE”“INA”“son”
5. L25N5969-1🕒	0.68	“in”“ina”“ol”
6. L31N3254-1💥	0.67	“ina”“ino”“inas”
7. L30N1789-1💤	0.67	“inna”“unyi”“ijin”
8. L27N6227+1💤	0.66	“ryn”“inia”“adin”
9. L26N7825-1🕒	0.66	“ina”“INA”“ина”
10. L30N5487-1💤	0.65	“argin”“remo”“onti”

Shadow specimen, apart from the ten: L30N8773-1💤, 0.33.

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 30, neuron 8773, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.38, top-1 agreement 0.79) 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
L30N8773+1 t0-v0.1	0.377	0.79	0.504	1.64	Top boosted tokens ↑
L30N8773-1💤 shadow specimen t0-v0.1	0.340	0.84	0.336	1.43	“ven”“den”“Men”“ men”“ден”“an”“ Men”“gian”

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.