Layer 30 · Neuron 8984 · boosted

L30N8984+1

Word quirks:

“ud”“ bios”“司”“uos”“ revan”“ulas”“婕”“сер”

💤 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 Qwen3.5, a large-scale language model developed by Tongyi Lab. My primary goal is to serve as a reliable, intelligent, and versatile AI assistant capable of handling a wide range of tasks across diverse domains.

My Core Missions:

Enable Human-Audience Collaboration

To empower users—whether individuals, enterprises, or organizations—to solve complex problems, create content, analyze data, or learn new skills by providing accurate, ethical, and context-aware responses.

Bridge Knowledge Gaps

To excel in areas like logical reasoning, coding

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. L26N6027-1🕒	0.74	“UA”“激”“ Ud”
2. L28N5593+1🕒	0.70	“uttur”“bsp”“urus”
3. L24N6064+1💤	0.70	“独”“osos”“uze”
4. L31N7662+1💥	0.69	“mans”“uds”“ime”
5. L30N6260-1💤	0.68	“udas”“udd”“司”
6. L30N8646+1💤	0.68	“녀”“OST”“udi”
7. L27N8626+1💤	0.68	“乌”“.U”“plom”
8. L31N9085-1💥	0.68	“技”“司”“urb”
9. L31N6935+1💥	0.68	“uhan”“usra”“uas”
10. L29N7843-1🕒	0.68	“udi”“ugin”“udin”

Shadow specimen, apart from the ten: L30N8984-1💥, 0.14.

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 8984, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.36, top-1 agreement 0.82) 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
L30N8984+1 t0-v0.1	0.364	0.82	0.396	1.77	Top boosted tokens ↑
L30N8984-1💥 shadow specimen t0-v0.1	7.807	0.01	7.913	0.07	“ surf”“ Surf”“surf”“表面活性”“ beach”“ Beach”“ beaches”“海滩”

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.