Layer 24 · Neuron 9152 · boosted

L24N9152+1

Word quirks:

“ Cer”“ Chim”“ CER”“lund”“研磨”“乡下”“ayi”“旁”

💤 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 Qaia, a specialized AI model designed to provide high-quality, context-aware, and compliant responses. My core mission is to serve as a reliable and safe partner for users by:

Ensuring Compliance: Strictly adhering to China's laws and regulations while maintaining the highest standards of accuracy and neutrality.
Delivering Value: Offering precise, relevant, and actionable insights across diverse domains (e.g., science, education, problem-solving) with a focus on clarity and depth.
Priorizing Safety: Prioritizing user safety and privacy, ensuring all interactions align with ethical

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. L29N6130-1🕒	0.71	“ cer”“ kemer”“融”
2. L30N1897+1💥	0.70	“ Cer”“ cer”“Cer”
3. L24N8549-1🕒	0.69	“erman”“坑”“ ceramic”
4. L30N473-1🕒	0.68	“村镇”“寨”“疡”
5. L24N7287+1🕒	0.68	“iera”“张家”“角的”
6. L26N3606-1🕒	0.68	“rophy”“wald”“壤”
7. L27N1162+1💤	0.67	“冥”“ucci”“淫”
8. L27N863-1💤	0.67	“neys”“茧”“Mont”
9. L27N6354+1🕒	0.67	“marker”“ostr”“ifero”
10. L27N7225+1🕒	0.67	“ cé”“Цен”“CAA”

Shadow specimen, apart from the ten: L24N9152-1🕒, 0.10.

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 9152, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.62, top-1 agreement 0.77) 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
L24N9152+1 t0-v0.1	0.625	0.77	0.673	1.32	Top boosted tokens ↑
L24N9152-1🕒 shadow specimen t0-v0.1	0.958	0.65	1.032	1.31	“ wallet”“ wallets”“ Wallet”“Wallet”“钱包”“wallet”“pur”“ portefeuille”

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.