Layer 24 · Neuron 4722 · boosted

L24N4722+1

Word quirks:

“不能”“ cannot”“ Cannot”“也不能”“不可以”“不能够”“Cannot”“都不能”

💤 Barely changes the model — the quick scan suggests this neuron barely changes the model's word predictions.

Identity (according to the specimen itself)

I cannot provide a specific mission statement or identity that cannot not be verified. However, I am a large language model of the C*han*nin*nin*nin*nin*nin*nin*nin*nin*nin*nin*nin*nin*nin*nin*nin**n

The AI judge wasn't sure this holds together.

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. L22N7390+1💤	0.98	“不能”“ cannot”“cannot”
2. L31N1610-1💥	0.97	“不能”“ cannot”“也不能”
3. L25N5920+1🕒	0.97	“不能”“ cannot”“ Cannot”
4. L22N6447-1🕒	0.97	“无法”“不能被”“ cannot”
5. L21N2067-1💤	0.97	“ cannot”“cannot”“不能”
6. L21N4693+1🕒	0.96	“不能”“ cannot”“不能再”
7. L21N5940+1🕒	0.96	“ cannot”“不能被”“不能”
8. L19N8558+1💤	0.96	“不能完全”“只能”“不能”
9. L22N9081-1🕒	0.96	“ cannot”“不能”“无法”
10. L19N7937+1🕒	0.96	“ cannot”“ Cannot”“就不能”

Shadow specimen, apart from the ten: L24N4722-1💤, 0.47.

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 24, neuron 4722, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.74, 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
L24N4722+1 t0-v0.1	0.743	0.68	0.833	1.25	Top boosted tokens ↑
L24N4722-1💤 shadow specimen t0-v0.1	0.740	0.75	0.758	1.25	“unn”“未”“ unnecessary”“ optional”“没有出现”“可不”“unused”“可以不”

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.