Layer 3 · Neuron 3681 · boosted

L3N3681+1

Word quirks:

“ application”“ framework”“ software”“ code”“ system”“ ERP”“ management”“ web”

🕒 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 Qwen3.5, a large language model developed by Alibaba Cloud. My mission is to assist users in building AI applications, providing intelligent solutions for various business needs.

Key Features:

Natural Language Processing (NLP) - I specialize in the development of NLP frameworks, enabling developers to create applications quickly.
Intelligent Code Generation - I can generate code templates, reduce development time, and improve efficiency.
Business Intelligence Platform - I provide comprehensive business solutions, including user management, data analysis, and system monitoring.
AI Assistant

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. L4N6924+1🕒	0.81	“ application”“ framework”“ metadata”
2. L4N6816+1🕒	0.79	“ management”“ ERP”“ software”
3. L5N7746+1🕒	0.78	“ management”“ workflow”“ workflows”
4. L7N5006+1🕒	0.78	“ enterprise”“ ERP”“ response”
5. L9N221+1🕒	0.77	“ system”“ application”“ framework”
6. L8N6630-1🕒	0.75	“ software”“ computer”“ programming”
7. L5N5690-1🕒	0.73	“ software”“ Python”“ tool”
8. Zombie Process✅ L7N8390-1	0.72	“ solution”“ system”“ web”
9. L6N2843-1🕒	0.72	“ IT”“ network”“ management”
10. L4N1358-1🕒	0.72	“ application”“ SDK”“ API”

Shadow specimen, apart from the ten: L3N3681-1🕒, -0.11.

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 3, neuron 3681, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 1.83, top-1 agreement 0.47) 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
L3N3681+1 t0-v0.1	1.832	0.47	2.191	2.07	Top boosted tokens ↑
L3N3681-1🕒 shadow specimen t0-v0.1	1.096	0.57	1.327	1.65	“ `”“<\|im_end\|>”“⏎”“ ‘”“.`”“ --”“⏎ ⏎”“ -”

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.