Layer 28 · Neuron 7618 · suppressed

L28N7618-1

Word quirks:

“上添”“下放”“anky”“常青”“riba”“ージュ”“ huel”“fee”

💤 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 language model developed by Tencent's Haidan Lab. My primary mission is to empower users with precise, reliable, and context-aware assistance across a wide range of tasks. Here’s how I aim to deliver value:

Key Missions & Capabilities

Complex Reasoning & Analysis

Break down intricate problems in science, engineering, or business (e.g., analyzing financial charts, debugging code, or solving mathematically rigorous questions).
Handle multi-step reasoning with reduced hallucinations.

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. L31N1471-1💥	0.71	“志明”“chaffen”“鸡汤”
2. L25N2879+1💤	0.70	“驭”“uang”“wana”
3. L31N9002-1💥	0.70	“ esk”“斯基”“威士”
4. L24N5321-1💤	0.69	“催”“之乡”“lette”
5. L23N2722-1💤	0.69	“刮”“晋”“黄牛”
6. L31N2697-1💥	0.68	“隈”“狐”“优势的”
7. L31N5631+1💥	0.68	“凫”“ anzus”“微店”
8. L26N2743+1🕒	0.68	“嫌”“RJ”“千山”
9. L29N8413-1🕒	0.68	“ژ”“amel”“舒缓”
10. L30N5873+1💤	0.68	“形”“用力”“azgo”

Shadow specimen, apart from the ten: L28N7618+1🕒, -0.08.

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 28, neuron 7618, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.35, 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
L28N7618+1🕒 shadow specimen t0-v0.1	0.489	0.72	0.586	1.37	“ do”“do”“ Do”“DO”“ DO”“ mat”“ogne”“ до”
L28N7618-1 t0-v0.1	0.349	0.79	0.332	1.38	Top boosted tokens ↑

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.