Layer 22 · Neuron 5862 · boosted

L22N5862+1

Word quirks:

“apid”“ Minute”“ фо”“ minute”“京”“ Minutes”“盾”“盾牌”

💤 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 Qian (also known as Qian), a large-scale AI model developed by the Chinese partner organization. My mission is to provide high-quality service in areas such as text generation, code generation, and visual analysis, while adhering to the principles of accuracy, security, and ethical standards. I aim to assist users in solving problems, creating content, and learning new knowledge with precision and reliability. How can I help 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. L24N7558+1🕒	0.66	“ minutes”“五分钟”“minutes”
2. L22N1707-1🕒	0.65	“盾”“ Tin”“要是”
3. L21N4791+1💤	0.63	“Minute”“امين”“派”
4. L20N8627-1🕒	0.63	“.prot”“ouve”“套”
5. L20N8711+1💤	0.61	“屏障”“ Ju”“.lang”
6. L29N1900-1💤	0.61	“plural”“ minute”“季”
7. L28N8896+1🕒	0.61	“ shields”“执”“盾”
8. L27N7915+1💤	0.60	“钟”“min”“ min”
9. L30N6431+1💤	0.60	“ Shield”“سع”“ Lug”
10. L28N9049+1💤	0.60	“ideshow”“邵”“邢”

Shadow specimen, apart from the ten: L22N5862-1🕒, 0.07.

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 5862, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.79, top-1 agreement 0.69) 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
L22N5862+1 t0-v0.1	0.787	0.69	0.872	1.34	Top boosted tokens ↑
L22N5862-1🕒 shadow specimen t0-v0.1	1.736	0.51	1.854	1.50	“本能”“ instinct”“ innate”“ Inst”“ inn”“ instincts”“inst”“ inst”

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.