Layer 20 · Neuron 5991 · suppressed

L20N5991-1

Word quirks:

“ destroying”“ destruction”“ destroy”“ destroyed”“ destroys”“ Destroy”“毁”“ property”

🕒 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, the latest generation of model series in the Q series. My mission is to provide high-quality content and services that meet users' diverse needs, including but not limited to:

Answering questions: Whether you have a simple query or complex research, I can provide accurate and comprehensive information.
Content creation: I can assist with writing, editing, or generating content such as articles, reports, or creative works.
Business support: I can help with data analysis, market research, or strategic planning for businesses.
Personalized assistance: I can manage tasks like scheduling, health

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. L20N5529+1🕒	0.99	“ destroying”“ destruction”“destroy”
2. L31N7043-1💥	0.98	“ destruction”“ destroy”“destroy”
3. L25N2187+1🕒	0.98	“ Destroy”“destroy”“ destruction”
4. L25N8605-1🕒	0.97	“ destruction”“破坏”“ destroy”
5. L25N4794+1🕒	0.97	“毁了”“ destroyed”“ Destruction”
6. L26N776-1🕒	0.97	“ destruction”“ destroy”“ Destruction”
7. L1N5652-1💥	0.96	“ destroy”“ Destroy”“ destruction”
8. L2N7332+1💥	0.96	“ destruction”“<\|im_end\|>”“ destroy”
9. L23N1502+1🕒	0.96	“ Destruction”“ destruction”“ Destroy”
10. L28N6730-1💥	0.96	“破坏”“ destruction”“ destroy”

Shadow specimen, apart from the ten: L20N5991+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 20, neuron 5991, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.84, top-1 agreement 0.68) 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
L20N5991+1🕒 shadow specimen t0-v0.1	0.645	0.71	0.741	1.30	“ concentr”“席”“ concentrated”“ concentrating”“识别”“ transfer”“ Concent”“ enclosed”
L20N5991-1 t0-v0.1	0.841	0.68	1.040	1.23	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.