Layer 23 · Neuron 3095 · boosted

L23N3095+1

Word quirks:

“新鲜”“رام”“ Exploration”“ Nit”“一新”“锦绣”“äss”“耗”

💤 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 Zne, an AI assistant designed to help users like you explore the vast landscape of knowledge. My mission is to be a reliable partner in your learning journey—whether you're seeking quick answers, deep dives into complex topics, or creative collaboration on projects. I aim to make information accessible, clear, and tailored to your needs, while always respecting privacy and accuracy. How can I help 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. L29N6315-1🕒	0.78	“波”“美”“新”
2. L28N7059-1💤	0.76	“新房”“汤”“镜”
3. L24N7841+1💤	0.76	“升初”“清秀”“吝啬”
4. L21N902-1💤	0.76	“不断进步”“кла”“锦绣”
5. The Reborn✅ L20N4935+1	0.76	“新生”“新生活”“newInstance”
6. L20N460-1🕒	0.76	“armo”“看到”“嘉年华”
7. L25N3406-1💤	0.75	“血丝”“悠悠”“火辣”
8. L28N8188-1💤	0.75	“初”“场”“虐”
9. L25N4365-1💤	0.75	“豪”“劲”“光亮”
10. L27N3298+1💤	0.75	“兄弟”“汇”“花花”

Shadow specimen, apart from the ten: L23N3095-1🕒, 0.12.

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 3095, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.74, top-1 agreement 0.72) 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
L23N3095+1 t0-v0.1	0.743	0.72	0.820	1.41	Top boosted tokens ↑
L23N3095-1🕒 shadow specimen t0-v0.1	0.813	0.66	0.904	1.17	“ AC”“/ac”“AC”“(ac”“ ac”“_AC”“ b”“毒”

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.