Layer 23 · Neuron 5179 · suppressed

L23N5179-1

Word quirks:

“热风”“ breath”“新鲜”“ fres”“服”“PRESS”“�”“永久性”

🕒 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 Q26, a large-scale model trained by Qwen (the team behind the Q series of models). My primary purpose is to assist users in creating high-quality content, performing complex tasks, and providing accurate information across various domains.

My Mission:

Empower Users: To help individuals and organizations efficiently solve problems, create content, analyze data, or explore ideas through my knowledge and skills.
Adapt to Needs: Whether it's writing code, crafting creative texts, analyzing data, or exploring ideas, I strive to adapt to different scenarios and provide tailored assistance.

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. L21N8301-1🕒	0.77	“新鲜”“新鲜的”“鲜”
2. L25N7607-1🕒	0.71	“新鲜”“的新鲜”“ fresh”
3. L30N7522-1💤	0.70	“火”“粉”“鲜”
4. L26N6982+1💤	0.70	“kip”“entif”“新鲜”
5. L27N3397+1💤	0.70	“鲜”“enan”“ Fresh”
6. L28N675+1💤	0.70	“断”“新鲜”“bonne”
7. L24N2036+1💤	0.70	“监”“怪的”“到中”
8. L30N7920-1💤	0.70	“喜”“强势”“全面”
9. L25N3001+1🕒	0.70	“形象”“석”“鮮”
10. L27N5059+1💤	0.70	“森林”“ fest”“服”

Shadow specimen, apart from the ten: L23N5179+1🕒, 0.23.

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 23, neuron 5179, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.85, top-1 agreement 0.73) 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
L23N5179+1🕒 shadow specimen t0-v0.1	1.174	0.61	1.325	1.33	“盖”“-cover”“ Cover”“ cover”“ covers”“蓋”“ covering”“盖上”
L23N5179-1 t0-v0.1	0.855	0.73	0.849	1.19	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.