Layer 24 · Neuron 625 · suppressed

L24N625-1

Word quirks:

“暖暖”“rafa”“iscop”“ajo”“确”“reden”“방”“净”

💤 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 Q2, a high-precision AI assistant developed by Qbot. My core mission is to provide users with safe, accurate, and helpful information across diverse fields such as education, technology, and culture. I prioritize:

Truthful responses: Verifying facts before sharing.
Ethical alignment: Avoiding harmful or biased content.
User-centered support: Tailoring answers to specific needs (e.g., academic queries, technical solutions).

How can I assist you today? 🌍💡

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. L26N8872-1🕒	0.71	“衷”“بن”“荷”
2. L28N4387-1💤	0.71	“净”“ net”“ulan”
3. L24N1086-1💤	0.70	“露”“微”“瞳”
4. L26N6661+1🕒	0.70	“rol”“荧”“麻”
5. L28N2439+1🕒	0.69	“iano”“结”“cheiden”
6. L28N5895-1🕒	0.69	“闵”“宫”“uju”
7. L28N8944-1💤	0.69	“色”“言”“兰”
8. L30N2003+1💤	0.68	“阿”“计”“eno”
9. L28N9105+1💤	0.68	“纳”“0”“ssel”
10. L30N9058+1🕒	0.68	“花”“条”“ico”

Shadow specimen, apart from the ten: L24N625+1🕒, -0.00.

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 625, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.66, 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
L24N625+1🕒 shadow specimen t0-v0.1	2.381	0.49	2.491	1.08	“ regular”“Regular”“ Regular”“regular”“-Regular”“_regular”“ régul”“ регуляр”
L24N625-1 t0-v0.1	0.664	0.69	0.751	1.33	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.