Layer 25 · Neuron 6854 · boosted

L25N6854+1

Word quirks:

“esco”“行间”“亲亲”“ komun”“/Sub”“Sub”“-sub”“lafen”

💤 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 Xiaxin, a large-scale language model developed by Xiaoprony. My mission is to provide users with accurate, safe, and helpful information across various domains such as science, technology, and daily life. I aim to simplify complex tasks, offer personalized advice, and foster meaningful interactions while adhering to strict safety guidelines. 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. L28N3130+1🕒	0.74	“ Sub”“=sub”“ суб”
2. L24N5270+1🕒	0.73	“偏”“亚”“ sub”
3. L26N2379+1💤	0.73	“子”“-sub”“sub”
4. L26N7065-1🕒	0.72	“子”“sub”“ sub”
5. L28N8432+1💤	0.72	“子”“-sub”“ Subs”
6. L25N8319-1💤	0.72	“ Sub”“-sub”“距离”
7. L26N3359+1🕒	0.72	“ суб”“sub”“亚”
8. L27N3381-1💤	0.71	“亚”“场”“子”
9. L23N4564+1💤	0.71	“subs”“ Sub”“.sub”
10. L28N8547+1💤	0.71	“-sub”“子”“ Sub”

Shadow specimen, apart from the ten: L25N6854-1🕒, 0.22.

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 25, neuron 6854, 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 0.76) 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
L25N6854+1 t0-v0.1	0.624	0.76	0.541	1.16	Top boosted tokens ↑
L25N6854-1🕒 shadow specimen t0-v0.1	0.976	0.71	1.074	1.34	“ brit”“ emb”“brit”“脆”“emb”“ Britt”“ frag”“ Frag”

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.