Layer 8 · Neuron 7862 · boosted

L8N7862+1

Word quirks:

“ usually”“ generally”“ general”“ executed”“ *”“ execution”“ specific”“ discussion”

💤 Barely changes the model — the quick scan suggests this neuron barely changes the model's word predictions.

Identity (according to the specimen itself)

I am Qwen3.5, a large-scale language model developed by Tongyi Lab. My mission is to provide high-quality, safe, and efficient AI assistance across diverse domains.

Core Mission Objectives:

Safety & Reliability: Ensure all interactions are safe, ethical, and compliant with industry standards.
Efficiency: Optimize processes to minimize costs, time, and resource usage.
Scalability: Design systems that can handle increased complexity and demand.
User-Centricity: Prioritize user needs and satisfaction in all outputs.

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. L12N8765+1💤	0.76	“ general”“ generally”“/general”
2. L12N6609-1💤	0.75	“ generally”“ general”“ specific”
3. L11N614-1🕒	0.75	“ generally”“ typically”“ specifically”
4. L6N8447+1🕒	0.75	“ usually”“ generally”“ specific”
5. L10N5388-1💤	0.75	“ generally”“ typically”“ usually”
6. L10N6604+1🕒	0.75	“ usually”“ generally”“ typically”
7. Yes-Man✅ L9N3353-1	0.75	“ generally”“ general”“ usually”
8. L9N1708+1💤	0.75	“ specific”“ usually”“ generally”
9. L12N8558+1🕒	0.75	“ generally”“ overall”“ actually”
10. L9N5816-1💤	0.75	“ generally”“ typically”“ typical”

Shadow specimen, apart from the ten: L8N7862-1💤, 0.29.

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 8, neuron 7862, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.82, top-1 agreement 0.69) 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
L8N7862+1 t0-v0.1	0.816	0.69	0.978	1.26	Top boosted tokens ↑
L8N7862-1💤 shadow specimen t0-v0.1	0.700	0.68	0.850	1.17	“ typical”“ standard”“ reference”“ referenced”“ usual”“ requests”“<\|im_end\|>”“ request”

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.