Layer 7 · Neuron 9087 · boosted

L7N9087+1

Word quirks:

“ rooms”“ room”“ downstairs”“ upstairs”“ space”“ hall”“rooms”“ spaces”

🕒 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 currently in my living room, which serves as the main gathering space for the household. It is a cozy area where we enjoy tea and relax together.

The mission of this home is to keep the family pets. We have a large, comfortable living area with plenty of seating for friends and family to socialize.

It seems like I am in the kitchen, which is used for preparing meals and snacks. It is a bright and airy space that connects the rest of the house.

Overall, this is the dining room, where we host guests and celebrate special occasions. It is a formal

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. L3N6281-1🕒	0.90	“ rooms”“ space”“ spaces”
2. L24N8996-1🕒	0.88	“ rooms”“ room”“ spaces”
3. L7N6272+1🕒	0.88	“ rooms”“ apartment”“ room”
4. L24N7865-1💤	0.87	“ room”“ rooms”“-room”
5. L7N6559+1🕒	0.87	“ rooms”“ apartment”“ room”
6. L6N5649+1🕒	0.86	“ rooms”“ room”“ wifi”
7. L5N8284-1🕒	0.86	“ rooms”“ room”“ Room”
8. L28N7230-1💤	0.86	“ rooms”“ room”“ Room”
9. L24N8498+1🕒	0.85	“ room”“ Room”“Room”
10. L23N7434-1🕒	0.85	“ room”“ Room”“-room”

Shadow specimen, apart from the ten: L7N9087-1🕒, -0.07.

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 9087, 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 0.45) 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
L7N9087+1 t0-v0.1	2.022	0.45	2.420	1.99	Top boosted tokens ↑
L7N9087-1🕒 shadow specimen t0-v0.1	1.117	0.60	1.255	1.37	“ ‘”“ \"”“ example”“ my”“ '”“ reference”“ Example”“ -”

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.