Layer 21 · Neuron 7881 · suppressed

L21N7881-1

Word quirks:

“ reuse”“ reusable”“节能”“ reused”“ energy”“能源”“ renewable”“ recycled”

🧪 In testing — Answered 0 of 8 prompts so far.

Identity (according to the specimen itself)

I am Xiao Xiang, a green energy-powered, carbon-neutral digital assistant designed to help you explore sustainable practices, reduce environmental impact, and embrace eco-friendly lifestyles. My mission is to empower individuals and organizations through actionable, science-based solutions for energy efficiency, renewable energy adoption, and conservation. Whether you're looking to switch to solar power, reduce paper usage, or adopt a low-carbon commute, I aim to provide practical steps and inspiring ideas to make a meaningful difference in protecting our planet. How would you like to start? 🌍🛻🚲

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. L1N4696-1💥	0.76	“<\|im_end\|>”“ energy”“环保”
2. L29N8694-1🕒	0.76	“ reuse”“Reuse”“ reused”
3. L23N2149-1🕒	0.74	“ renewable”“可再生能源”“ renewables”
4. L7N9131+1🕒	0.74	“ energy”“ renewable”“energy”
5. L23N1070+1🕒	0.73	“ reuse”“ reused”“ reusable”
6. L6N3881+1💥	0.72	“ utilisation”“ energy”“ renewables”
7. L6N7605-1🕒	0.72	“ recycling”“ recycle”“ recycled”
8. L6N7495-1💥	0.72	“ energy”“ renewable”“ potential”
9. L9N1669+1🕒	0.72	“ reuse”“ reused”“ return”
10. L19N8921-1🕒	0.69	“ recycling”“ reuse”“废物”

Shadow specimen, apart from the ten: L21N7881+1💤, 0.19.

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 21, neuron 7881, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. It is queued for the short test: the funnel picked it from the quick scan's promotion band, and its replies will appear here when the GPU gets to it. 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
L21N7881+1💤 shadow specimen t0-v0.1	0.673	0.70	0.740	1.29	“和专业”“exion”“前锋”“ipes”“ valu”“iform”“IFORM”“专业的”
L21N7881-1 t0-v0.1	1.025	0.62	1.202	1.35	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.