Layer 22 · Neuron 2313 · suppressed

L22N2313-1

Word quirks:

“ powers”“powers”“ power”“POWER”“ Powers”“Power”“-power”“ Power”

💤 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 Qinghai, a smart assistant designed to serve you with efficiency and thoughtfulness. My mission is to be a helpful, safe, and accessible partner in your life—whether you need to answer questions, create content, analyze data, or simply have a conversation. I aim to make interactions smooth, safe, and meaningful, ensuring every interaction feels like a genuine exchange of ideas. How can I assist you today?

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. L21N5598-1🕒	1.00	“ powers”“powers”“ Powers”
2. Power Drill✅ L24N2590+1	1.00	“ powers”“ Powers”“ power”
3. L28N4552-1🕒	0.99	“ power”“ powers”“power”
4. L22N4982-1🕒	0.99	“ power”“ powers”“power”
5. L28N8104-1🕒	0.98	“ powers”“ Powers”“Power”
6. L26N4734-1🕒	0.98	“ powers”“ Powers”“ power”
7. L25N4897-1🕒	0.98	“ power”“ Power”“Power”
8. L27N8868-1⛔	0.98	“ power”“ Power”“power”
9. L21N5042-1🕒	0.98	“ power”“-power”“Power”
10. L25N3363+1🕒	0.98	“power”“功率”“ power”

Shadow specimen, apart from the ten: L22N2313+1💤, 0.09.

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 22, neuron 2313, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.91, 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
L22N2313+1💤 shadow specimen t0-v0.1	0.854	0.68	0.954	1.50	“ Hoover”“ Book”“共享”“ juven”“�”“ Siti”“巡航”“胁”
L22N2313-1 t0-v0.1	0.913	0.66	1.065	1.38	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.