Layer 31 · Neuron 6870 · boosted

L31N6870+1

Word quirks:

“ovel”“anter”“oger”“ibat”“agua”“lic”“利”“eric”

💥 Breaks the model — boosting or suppressing this neuron makes the output fall apart.

Identity (according to the specimen itself)

ovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovelovel

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. L30N2338-1💤	0.68	“upra”“obat”“awi”
2. L29N1866+1💤	0.66	“аниз”“yal”“uus”
3. L25N8504-1🕒	0.66	“شر”“agic”“aan”
4. L31N5037-1💥	0.65	“anser”“owler”“KER”
5. L29N4463+1🕒	0.65	“ewed”“opl”“her”
6. L29N606+1🕒	0.65	“noi”“ugat”“ spiel”
7. L31N9006-1💥	0.65	“igue”“ed”“oret”
8. L30N6439+1💤	0.65	“ordi”“води”“bons”
9. L28N6789+1💤	0.65	“ait”“oge”“oger”
10. L29N5134-1🕒	0.64	“orer”“olan”“aul”

Shadow specimen, apart from the ten: L31N6870-1💥, 0.29.

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 31, neuron 6870, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 5.14, top-1 agreement 0.10) 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
L31N6870+1 t0-v0.1	5.145	0.10	5.418	9.25	Top boosted tokens ↑
L31N6870-1💥 shadow specimen t0-v0.1	6.177	0.04	6.486	9.49	“干”“ thổ”“posa”“uis”“�”“rand”“见”“pos”

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.