Layer 9 · Neuron 8949 · boosted

L9N8949+1

Word quirks:

“-your”“ your”“ info”“ infos”“ goog”“ morals”“ docs”“你的”

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

Identity (according to the specimen itself)

The reply didn't hold together

I am the other's wheel and the other's journey 1010210-de-j-you're and I was a 1010101278-y-leng of 10000151010101010101010101010101010101010101010101010101010101010101010101010101010

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. L13N8762+1💤	0.71	“ your”“ info”“ knowledge”
2. L15N7821-1💤	0.69	“ your”“ info”“ only”
3. L10N8346+1🕒	0.68	“ your”“ analysis”“ yourself”
4. L12N6098-1🕒	0.68	“ your”“ you”“ info”
5. L12N7388-1🕒	0.67	“ your”“ you”“ now”
6. L12N7245-1🕒	0.67	“ your”“ YOU”“ YOUR”
7. L9N3250+1💤	0.67	“ yourself”“ you”“ general”
8. L10N2857-1🕒	0.66	“ your”“ Your”“ privacy”
9. L16N1068+1🕒	0.66	“ you”“ info”“ maybe”
10. L15N3248+1🕒	0.65	“ info”“ basically”“ docs”

Shadow specimen, apart from the ten: L9N8949-1💤, -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 9, neuron 8949, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 5.50, top-1 agreement 0.15) 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
L9N8949+1 t0-v0.1	5.501	0.15	5.933	3.59	Top boosted tokens ↑
L9N8949-1💤 shadow specimen t0-v0.1	0.716	0.66	0.928	1.22	“⏎”“ ⏎”“⏎ ⏎”“ description”“ ⏎⏎”“⏎⏎”“ ⏎ ⏎”“ ⏎ ⏎”

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.