Layer 15 · Neuron 6991 · suppressed

L15N6991-1

Word quirks:

“ local”“ locally”“ here”“ similar”“ similarly”“ locals”“ localized”“ Local”

🕒 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 a large-scale AI model. My primary function is to assist users with specific tasks. What is my mission?

My Mission:
To provide high-quality, efficient, and practical assistance tailored to the needs of users in our community. This includes handling diverse tasks such as answering questions, processing text, and performing logical reasoning.

Key Responsibilities:

Task Execution: I can handle various tasks including text generation, data analysis, and logical reasoning.
Content Creation: I can create content such as writing articles, scripts, and other materials.
Problem Solving: I can

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. L2N4906+1💥	0.91	“ local”“<\|im_end\|>”“ Local”
2. L17N8945+1🕒	0.89	“ local”“ nearby”“ Local”
3. L11N7974+1🕒	0.89	“ nearby”“ local”“ here”
4. L8N6782+1🕒	0.89	“ local”“ locals”“ locations”
5. L20N5930+1🕒	0.88	“ locally”“ local”“本地”
6. L26N3225-1🕒	0.88	“ local”“ p”“ Local”
7. L19N8319-1🕒	0.88	“ local”“ single”“ locally”
8. L19N934-1💤	0.88	“ local”“ locally”“ locals”
9. L25N6658-1🕒	0.88	“ local”“-local”“local”
10. L17N6075+1🕒	0.88	“ local”“ locality”“ localized”

Shadow specimen, apart from the ten: L15N6991+1🕒, 0.25.

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 15, neuron 6991, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.90, top-1 agreement 0.64) 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
L15N6991+1🕒 shadow specimen t0-v0.1	0.790	0.65	1.083	1.28	“ mostly”“ hopefully”“ spotted”“ partially”“ peek”“ currently”“ info”“ partly”
L15N6991-1 t0-v0.1	0.896	0.64	1.113	1.49	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.