Layer 27 · Neuron 6027 · suppressed

L27N6027-1

Word quirks:

“xCD”“ Dixon”“yd”“ Dillon”“ dysk”“鸭”“alus”“achs”

🕒 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 Q*, a large-scale AI model developed by *xAuto (a company focused on autonomous driving and related technologies). My core mission is to provide intelligent, reliable, and helpful assistance across various domains, including but not limited to:

Autonomous Driving: Supporting the development of self-driving systems through advanced perception, planning, and control algorithms.
AI Research & Innovation: Pushing boundaries in machine learning, computer vision, and other AI disciplines.
User Support: Answering questions, solving problems, and providing insights with accuracy and efficiency.
Ethical & Safe AI: Ens

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. L29N6922+1💤	0.66	“ Dedicated”“ Dixon”“ DH”
2. L26N4664-1🕒	0.65	“ Duc”“DNA”“DU”
3. L20N542+1💤	0.64	“aggi”“ dz”“dh”
4. L25N6128-1🕒	0.64	“ Dix”“DX”“雪”
5. L31N3538+1💥	0.63	“D”“DV”“DG”
6. L19N67+1💤	0.63	“ DL”“aggi”“جل”
7. L28N7994+1💤	0.63	“丹”“isd”“DD”
8. L31N3468+1💥	0.63	“DG”“DH”“SF”
9. L26N8517+1🕒	0.63	“ dys”“ Dys”“_d”
10. L27N6593+1💤	0.63	“射”“ щ”“ Died”

Shadow specimen, apart from the ten: L27N6027+1💤, 0.23.

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 6027, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.55, top-1 agreement 0.77) 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
L27N6027+1💤 shadow specimen t0-v0.1	0.375	0.80	0.356	1.36	“ Param”“тро”“五金”“ param”“ Тро”“基”“ ases”“ Tro”
L27N6027-1 t0-v0.1	0.545	0.77	0.593	1.31	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.