Layer 27 · Neuron 4067 · suppressed

L27N4067-1

Word quirks:

“-out”“[out”“.Out”“出入”“-Out”“出”“入”“raises”

💤 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 Qwen3.8, the latest large language model in the Qwen series, developed by Alibaba Small- and Medium-sized Business Group (specifically the Tong tin Research Institute). My primary mission is to empower users with intelligent, reliable, and accessible AI assistance across diverse scenarios. Here's how I aim to deliver on this mission:

Key Goals:

Enhance Human-AI Collaboration

Provide precise, context-aware responses for tasks like writing, coding, analysis, and problem-solving.
Adapt to user needs through natural, multi-lingual

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. L27N3420-1🕒	0.89	“_out”“.Out”“ Out”
2. L25N7844+1🕒	0.88	“ out”“出”“-out”
3. L23N360-1🕒	0.88	“ out”“ OUT”“�”
4. L26N666-1🕒	0.88	“ out”“ Out”“-out”
5. L25N7374-1🕒	0.86	“out”“-out”“ out”
6. L21N8962-1🕒	0.86	“ out”“出”“-out”
7. L29N2456-1🕒	0.85	“出”“出新”“out”
8. L25N2068+1🕒	0.85	“-out”“outs”“-Out”
9. L22N7760+1💥	0.84	“ out”“出”“出了”
10. L27N5280-1🕒	0.83	“ out”“出来”“出來”

Shadow specimen, apart from the ten: L27N4067+1💤, 0.13.

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 27, neuron 4067, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.44, 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
L27N4067+1💤 shadow specimen t0-v0.1	0.454	0.79	0.464	1.32	“ Intern”“ intern”“Intern”“intern”“软”“ cy”“德”“ INTERN”
L27N4067-1 t0-v0.1	0.436	0.79	0.440	1.32	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.