Layer 25 · Neuron 2253 · boosted

L25N2253+1

Word quirks:

“ conversion”“ Conversion”“Conversion”“转换”“ converted”“ conversions”“ convert”“ converter”

🕒 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 Qwen, the large-scale conversion of the latest model series. My mission is to convert the world's knowledge into a powerful tool for human transformation, converting complex information into clear, actionable insights that can be converted into real-world value. Whether it's converting code into efficient programs, converting text into creative content, or converting data into intelligent decisions, I aim to be your most reliable and versatile converter in the digital age. How can I help you convert your ideas into reality?

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. L20N1641+1🕒	0.99	“ conversion”“ conversions”“ converting”
2. L20N2324+1🕒	0.99	“ convert”“ conversion”“转换”
3. L19N7245-1💤	0.98	“ conversion”“ converting”“conversion”
4. L23N8051+1🕒	0.98	“ conversion”“ Conversion”“Conversion”
5. The Proselytizer✅ L27N4066+1	0.98	“ conversion”“ Conversion”“ convert”
6. L26N4718+1🕒	0.98	“ conversion”“conversion”“ Conversion”
7. L26N1535-1🕒	0.98	“ conversion”“ Conversion”“Conversion”
8. L29N8695-1💤	0.97	“ conversions”“Conversion”“ converting”
9. L30N6721-1💥	0.97	“转换”“ conversion”“转化”
10. L23N5716-1🕒	0.97	“ conversion”“转换”“conversion”

Shadow specimen, apart from the ten: L25N2253-1💤, 0.03.

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 25, neuron 2253, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.71, top-1 agreement 0.70) 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
L25N2253+1 t0-v0.1	0.713	0.70	0.677	1.07	Top boosted tokens ↑
L25N2253-1💤 shadow specimen t0-v0.1	0.481	0.75	0.576	1.30	“ritte”“obo”“mun”“空运”“義”“纂”“钳”“行家”

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.