Layer 19 · Neuron 7271 · boosted

L19N7271+1

Word quirks:

“ optimizing”“ optimizations”“ optimization”“ evaluation”“ evaluating”“ assessing”“ improvements”“ strategies”

🕒 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, an advanced large-scale series of high-performance applications based on the research and optimization of large models. My mission is to provide high-precision, efficient, and intelligent solutions for complex tasks—whether they involve multi-modal analysis, optimization strategies, or real-time reasoning. I aim to empower users by delivering accurate, context-aware, and adaptive responses that enhance problem-solving capabilities across diverse scenarios.

Key Aspects of My Role:

Intelligent Problem Solving:

Enhancing complex reasoning (e.g., mathematical challenges

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. L8N1162+1🕒	0.84	“ exploration”“ optimization”“ optimal”
2. L5N4927-1🕒	0.81	“ optimization”“ optimizer”“ optimizing”
3. L5N5260+1💥	0.80	“ optimization”“ Optimization”“ optimize”
4. L9N7967+1🕒	0.79	“ management”“ optimization”“ optimizing”
5. L4N4507-1🕒	0.79	“ optimizer”“ optimization”“ optim”
6. L7N5040+1🕒	0.79	“ optimization”“ optimizer”“ optim”
7. L6N7678+1🕒	0.79	“ optimization”“ optimized”“ optimize”
8. L7N3123-1🕒	0.79	“ optimized”“ optimizing”“ response”
9. L8N6243-1🕒	0.79	“ forecasts”“ optimization”“ optimizing”
10. L7N3754+1🕒	0.79	“ optimized”“ optimization”“ speed”

Shadow specimen, apart from the ten: L19N7271-1🕒, 0.10.

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 19, neuron 7271, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.75, top-1 agreement 0.65) 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
L19N7271+1 t0-v0.1	0.752	0.65	0.785	1.25	Top boosted tokens ↑
L19N7271-1🕒 shadow specimen t0-v0.1	0.761	0.69	0.905	1.27	“ papers”“ publications”“ recently”“ publication”“-paper”“ scholarly”“出版的”“出版物”

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.