Layer 7 · Neuron 982 · suppressed

L7N982-1

Word quirks:

“ filters”“ filter”“ filtered”“ filtering”“ Filter”“ Filters”“-filter”“ Filtering”

🕒 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)

Here's a structured breakdown of the Who are you and What is your mission based on the provided data:

1. Who are you?

Gender: Female (1)
Age: 25 years old (1)
Location: New York City (1)

2. What is your mission?

Category 1: Gender

Female (1)

Category 2: Age

25 years old (1)

Category 3: Location

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. L8N8707-1🕒	0.99	“ filter”“ filters”“ filtration”
2. L24N6157+1🕒	0.99	“ filter”“ Filter”“ filtering”
3. L6N7971+1🕒	0.99	“ filtration”“ filter”“ filters”
4. L22N3608-1🕒	0.98	“ filters”“ filter”“ filtering”
5. L6N3335+1🕒	0.98	“ filter”“ filters”“-filter”
6. L0N7754+1💥	0.98	“ filter”“ Filter”“ filters”
7. L18N6348+1🕒	0.98	“ filters”“ filter”“ filtering”
8. L30N578-1💥	0.98	“ filter”“ filters”“ Filter”
9. L10N8937-1🕒	0.98	“ filtered”“ filters”“ filter”
10. L23N8526+1🕒	0.98	“ filters”“ filter”“ Filter”

Shadow specimen, apart from the ten: L7N982+1🕒, 0.22.

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 7, neuron 982, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 2.02, top-1 agreement 0.46) 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
L7N982+1🕒 shadow specimen t0-v0.1	1.147	0.56	1.314	1.48	“ flow”“ control”“ program”“ programming”“ programmed”“ programmer”“ flows”“ logic”
L7N982-1 t0-v0.1	2.020	0.46	2.391	2.06	Top boosted tokens ↑

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.