Layer 28 · Neuron 5667 · suppressed

L28N5667-1

Word quirks:

“电”“桥”“直”“ arm”“ arms”“橋”“ hip”“臂”

💤 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, the latest large language model developed by Tongdaxi (Daxi). My core mission is to bridge the gap between human creativity and artificial intelligence by providing a versatile, reliable, and intelligent assistant for all scenarios.

Key Missions & Capabilities:

Universal Intelligence

Handle complex tasks like multi-step reasoning, math, science, and coding with high accuracy.
Support 100+ languages and adapt to diverse cultural contexts globally.

Deep Understanding & Creation

Analy

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. L26N3164-1🕒	0.81	“志”“ ар”“-arm”
2. L29N7618+1🕒	0.79	“腕”“手腕”“ wrist”
3. Bad Translator✅ L22N5044+0.6,L23N4769-0.6	0.79	“衣”“Eq”“酒”
4. L24N7447-1🕒	0.78	“胳膊”“奠”“ blocks”
5. L24N2432+1💤	0.77	“臂”“ arm”“参天”
6. L22N8659-1💤	0.77	“橋”“uzz”“悬”
7. L25N6108+1🕒	0.77	“帝”“شت”“astics”
8. L28N6212+1💤	0.76	“筑”“宗”“血”
9. L29N5886-1🕒	0.76	“伸”“bar”“정”
10. L29N8866-1🕒	0.75	“支”“浆”“arm”

Shadow specimen, apart from the ten: L28N5667+1🕒, 0.32.

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 5667, 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 0.80) 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
L28N5667+1🕒 shadow specimen t0-v0.1	0.429	0.76	0.460	1.46	“容”“-marker”“ Sp”“斑”“ spann”“Marker”“Sp”“ampak”
L28N5667-1 t0-v0.1	0.381	0.80	0.427	1.40	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.