Layer 22 · Neuron 4454 · boosted

L22N4454+1

Word quirks:

“ XV”“ VIII”“ IX”“ XX”“ $”“ XVII”“ ”“ XIV”

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

Identity (according to the specimen itself)

The reply didn't hold together

I am Q8.0, a next-generation Q8.0.1.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.

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. L28N2171+1🕒	0.88	“ XXX”“ XVIII”“ XX”
2. L3N6997-1💥	0.82	“<\|im_end\|>”“ XV”“ VII”
3. L27N1058+1🕒	0.80	“ XV”“ VIII”“ VII”
4. L25N435+1🕒	0.78	“ XX”“ XV”“XX”
5. L24N4856+1🕒	0.77	“ vi”“ ix”“ xi”
6. L28N2313-1💥	0.74	“ ”“ XX”“ XXX”
7. L19N3712+1💤	0.73	“ III”“ VII”“ IV”
8. L22N5385+1🕒	0.71	“ II”“ III”“ IV”
9. L23N1469-1💤	0.71	“ XIX”“ IX”“IX”
10. L26N6285-1🕒	0.71	“ XX”“XX”“ XXX”

Shadow specimen, apart from the ten: L22N4454-1🕒, -0.05.

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 4454, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.84, top-1 agreement 0.71) 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
L22N4454+1 t0-v0.1	0.840	0.71	0.942	1.70	Top boosted tokens ↑
L22N4454-1🕒 shadow specimen t0-v0.1	1.484	0.49	1.595	1.30	“ at”“_at”“at”“⇥at”“于”“.at”“ ở”“-at”

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.