Layer 25 · Neuron 925 · suppressed

L25N925-1

Word quirks:

“ Few”“few”“ fler”“Few”“ few”“ fewer”“ too”“ flere”

💤 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 Girjan, a large language model trained by ZJZ. My mission is to provide helpful, accurate, and engaging responses to your questions or requests. Whether you need information, creative writing, analysis, or just a conversation, I aim to support you with clarity and expertise. 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. L31N2577+1💥	0.92	“ few”“few”“ Few”
2. L27N2238+1🕒	0.90	“ few”“ Few”“ many”
3. L25N929-1🕒	0.89	“ few”“few”“ Few”
4. L29N3679-1🕒	0.88	“ few”“few”“ Few”
5. L29N4210-1🕒	0.86	“ few”“few”“ couple”
6. L18N7674-1🕒	0.85	“ few”“few”“ Few”
7. L23N1809+1🕒	0.77	“ no”“ fewer”“多个”
8. L27N2949-1🕒	0.77	“ few”“few”“ Few”
9. L18N8641+1💤	0.76	“ minimal”“ only”“ mostly”
10. L22N3638+1🕒	0.75	“少数”“ few”“few”

Shadow specimen, apart from the ten: L25N925+1🕒, 0.16.

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 925, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.51, top-1 agreement 0.74) 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
L25N925+1🕒 shadow specimen t0-v0.1	1.244	0.58	1.321	1.50	“ amount”“ Amount”“量”“amount”“Amount”“ quantity”“ amounts”“ Quantity”
L25N925-1 t0-v0.1	0.515	0.74	0.534	1.37	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.