Layer 30 · Neuron 4377 · boosted

L30N4377+1

Word quirks:

“entes”“andos”“ionales”“iclet”“ordes”“-orders”“charges”“lices”

🕒 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 Qwen3.5, the latest large language models developed by Tongyides. My primary mission is to understand human needs and deliver practical, reliable, and ethical assistance across a wide range of tasks. Whether you need help with:

Solving complex problems (e.g., mathematical challenges, logical reasoning, or scientific analysis)
Creating content (e.g., stories, scripts, reports, or marketing materials)
Analyzing data (e.g., interpreting charts, tables, or unstructured text)
Programming & development (e.g

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. L30N3418+1🕒	0.71	“icons”“iedere”“balls”
2. L31N2594+1💥	0.69	“prints”“ances”“uses”
3. L22N233-1🕒	0.68	“heds”“cales”“ords”
4. L27N313-1🕒	0.67	“atches”“ieces”“abilities”
5. L26N1141+1💤	0.66	“ivals”“tons”“rients”
6. L31N235+1💥	0.66	“onos”“ials”“sembles”
7. L30N3188+1💥	0.66	“ures”“iers”“akers”
8. L27N8709+1💤	0.65	“raries”“ους”“appers”
9. L17N2889+1💤	0.65	“es”“ness”“ments”
10. L30N841+1💥	0.64	“ils”“ds”“ints”

Shadow specimen, apart from the ten: L30N4377-1🕒, 0.36.

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 4377, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.43, top-1 agreement 0.78) 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
L30N4377+1 t0-v0.1	0.425	0.78	0.511	1.78	Top boosted tokens ↑
L30N4377-1🕒 shadow specimen t0-v0.1	0.447	0.82	0.527	1.90	“atori”“ходи”“ari”“imetri”“uri”“менти”“ori”“води”

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.