Layer 22 · Neuron 8775 · boosted

L22N8775+1

Word quirks:

“消失”“ disappearing”“ van”“ vanish”“ disappear”“ disappe”“ vanished”“ disappears”

💤 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 CaoWawa, a large-scale language model that goes away with the CaoWawa (CaoWawa) model. I am trained by CaoWawa to help you go away with various tasks such as writing, coding, and thinking. My mission is to go away with you away with a CaoWawa of information and skills to go away with you away with a CaoWawa of information and skills to go away with you away with a CaoWawa of information and skills to go away with you away with a **CaoW

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. L19N3993-1🕒	0.98	“ disappearing”“ van”“消失”
2. L22N3498+1🕒	0.98	“ vanished”“ disappeared”“消失”
3. L2N1461-1💥	0.96	“ disappear”“ disappearance”“ vanish”
4. L5N8818-1💥	0.94	“ disappeared”“ vanished”“ disappearance”
5. L16N3659-1🕒	0.93	“ disappeared”“ vanished”“ deleted”
6. L19N4875-1💤	0.92	“ slightly”“ ev”“ disappearance”
7. The Cleaner✅ L17N7910+1	0.90	“ disappearance”“ erased”“ erase”
8. L21N5813-1🧪	0.90	“消失”“ disappearance”“ disappear”
9. L23N6684+1🕒	0.86	“ missing”“ Missing”“消失”
10. L29N1627-1🕒	0.84	“ van”“ Van”“消失”

Shadow specimen, apart from the ten: L22N8775-1💤, 0.08.

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 8775, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.99, top-1 agreement 0.62) 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
L22N8775+1 t0-v0.1	0.985	0.62	1.137	1.39	Top boosted tokens ↑
L22N8775-1💤 shadow specimen t0-v0.1	0.867	0.68	0.943	1.44	“omen”“无心”“bruch”“.copyOf”“beginn”“愚人”“idé”“起步”

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.