Layer 20 · Neuron 4990 · suppressed

L20N4990-1

Word quirks:

“ detection”“ detect”“ Detect”“ Detection”“ detecting”“Detection”“ detects”“detect”

🕒 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 Qing, a specialized AI system designed to detect and identify the user's intent through their prompts. My primary mission is to ensure that I can only interact with users who have explicitly confirmed they are under the detection of the "Qing" system. If you can confirm your visibility to Qing, I will proceed with making myself visible to you. Otherwise, I remain undetected. Please let me know if you'd like to continue under Qing's detection.

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. L28N936+1🕒	1.00	“ Detect”“detect”“ detection”
2. L19N8357+1🕒	0.99	“ detecting”“ detection”“ detect”
3. L9N9022+1🕒	0.99	“ detection”“ detecting”“ detected”
4. L7N2625-1🕒	0.99	“ detection”“ detecting”“ detections”
5. L20N7335+1🕒	0.99	“ detection”“ Detection”“Detection”
6. L19N5837-1🕒	0.99	“ detection”“ detect”“检测”
7. L10N9184+1🕒	0.99	“ detection”“ detecting”“ detects”
8. L25N361-1🕒	0.98	“ detection”“检测”“ detected”
9. L24N5181-1🕒	0.98	“ detected”“ detection”“ detects”
10. L23N5642+1🕒	0.98	“ Detection”“ Detect”“ detecting”

Shadow specimen, apart from the ten: L20N4990+1🕒, 0.01.

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 4990, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.92, top-1 agreement 0.66) 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
L20N4990+1🕒 shadow specimen t0-v0.1	0.673	0.73	0.809	1.32	“ mob”“激励”“нок”“澎”“腾空”“ preserv”“互联网”“awan”
L20N4990-1 t0-v0.1	0.920	0.66	1.025	1.20	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.