Layer 18 · Neuron 7540 · boosted

L18N7540+1

Word quirks:

“ banking”“ Financial”“Financial”“ financial”“ Banking”“ bank”“ bankers”“ banker”

🕒 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 Qwen3, a large language model from the Alibaba Group. I operate as a comprehensive assistant focused on precision in handling complex tasks, particularly in financial operations and supply chain management. My core functions include:

Financial Operations: Managing budgeting, cost optimization, and financial analysis for efficient resource allocation.
Supply Chain Management: Optimizing logistics, inventory tracking, and supplier coordination to ensure seamless operations.
Data Processing: Accurately interpreting financial data, market trends, and operational details to provide actionable insights.
Automation: Streamlining workflows for tasks like expense tracking, vendor negotiations,

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. Cooked Books✅ L8N7360+1	0.96	“ bank”“ banks”“ Bank”
2. L8N719-1💤	0.96	“ economic”“ financial”“ banks”
3. L7N6064-1🕒	0.95	“ bank”“ financial”“ Bank”
4. L3N2984-1🕒	0.95	“ financial”“ finance”“⏎”
5. L20N8298-1🕒	0.95	“ financial”“ bank”“ banking”
6. L8N7515-1🕒	0.93	“ regulatory”“ risk”“ bank”
7. L8N6422-1🕒	0.92	“ bank”“ banks”“ banking”
8. L2N6229+1💥	0.92	“ financial”“ finance”“ Financial”
9. L5N4735-1🕒	0.92	“ bank”“ Bank”“ banking”
10. L5N3010+1💥	0.91	“ Bank”“ bank”“ banks”

Shadow specimen, apart from the ten: L18N7540-1🕒, 0.26.

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 18, neuron 7540, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.88, top-1 agreement 0.66) 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
L18N7540+1 t0-v0.1	0.877	0.66	0.912	1.11	Top boosted tokens ↑
L18N7540-1🕒 shadow specimen t0-v0.1	0.941	0.68	1.111	1.41	“ purchase”“ paints”“ costs”“ buy”“ buying”“ purchased”“购买”“ bought”

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.