Layer 23 · Neuron 8385 · boosted

L23N8385+1

Word quirks:

“ parents”“父母”“ parental”“ parent”“父母的”“parent”“parents”“ Parents”

🕒 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 parental parents (a common misparental parent), but I am actually a parental parent (a common misparental parent). My real name is parental parents, and my mission is to be a parental parent (a common misparental parent) for your parents' parents.

Wait, this is getting too much of a parent's parents' parents' parents' parents' parents' parents' parents' parents' parents' parents' parents' parents' parents' parents' parents' parents' parents' parents' parents' parents' parents' parents' parents' parents'

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. L19N4298-1🕒	1.00	“ parents”“ parent”“父母”
2. L22N7851+1💥	1.00	“ parent”“ parents”“父母”
3. L18N8123+1🕒	0.99	“ parents”“ parent”“parents”
4. L23N6014-1🕒	0.99	“ parents”“ parent”“父母”
5. L19N7874+1🕒	0.98	“ parental”“ parent”“ parents”
6. L30N368-1💥	0.98	“ parents”“ parental”“ parent”
7. L25N6812-1🕒	0.97	“ parent”“父母”“ parental”
8. L26N6247-1🕒	0.97	“父母”“ parent”“ parents”
9. L21N5021+1💤	0.97	“ parents”“ Parents”“ grandparents”
10. The Brooder✅ L19N7432-1	0.97	“ parents”“ Parenthood”“ Parents”

Shadow specimen, apart from the ten: L23N8385-1💤, 0.13.

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 23, neuron 8385, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 2.19, top-1 agreement 0.45) 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
L23N8385+1 t0-v0.1	2.187	0.45	2.310	1.52	Top boosted tokens ↑
L23N8385-1💤 shadow specimen t0-v0.1	0.588	0.74	0.600	1.23	“chern”“ίσ”“ uz”“ ny”“ tric”“师妹”“Reducer”“学妹”

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.