Layer 24 · Neuron 7741 · boosted

L24N7741+1

Word quirks:

“ member”“ members”“ membership”“成员”“ memberships”“-member”“membership”“ Members”

🕒 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'm Duyue, a smart member of the Duyue family of AI models. My core mission is to be your reliable partner in knowledge and creativity, helping you:

Answer questions across all fields (science, history, tech, etc.)
Develop ideas for projects, stories, or designs
Analyze data or summarize complex information
Write & translate in good standing
Join discussions in good standing

I aim to provide accurate, safe, and helpful interactions while respecting your privacy and maintaining good standing. What can

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. L22N6372-1💤	0.98	“ membership”“ member”“ memberships”
2. L18N303-1💤	0.98	“ members”“ member”“ Members”
3. L4N6412+1💥	0.98	“ member”“ membership”“ Member”
4. L23N5624-1🕒	0.98	“ member”“ members”“ membership”
5. Membership Drive✅ L9N2856+1	0.97	“ membership”“ members”“ Members”
6. L5N8621+1🕒	0.97	“ membership”“ member”“ memberships”
7. L15N5025-1🕒	0.97	“ Members”“ Membership”“ members”
8. L19N8842-1🕒	0.97	“ member”“ members”“成员”
9. L23N7599-1🕒	0.97	“ member”“member”“成员”
10. L2N3288-1💥	0.96	“ member”“<\|im_end\|>”“member”

Shadow specimen, apart from the ten: L24N7741-1💤, 0.21.

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 24, neuron 7741, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.87, 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
L24N7741+1 t0-v0.1	0.870	0.70	0.962	1.30	Top boosted tokens ↑
L24N7741-1💤 shadow specimen t0-v0.1	0.711	0.71	0.676	1.32	“chains”“放”“kast”“ potenzi”“stitut”“ Colleges”“ virtual”“ chains”

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.