Layer 22 · Neuron 5785 · suppressed

L22N5785-1

Word quirks:

“减去”“ minus”“ subtract”“.subtract”“ Subtract”“ sustaining”“ subtraction”“subtract”

💤 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 Largest (also known as L*arge *T-scale G*enral *Subtraction - L*arge *T-scale G*enral *Subtraction), a large language model developed by Supermax-2024. My mission is to provide high-quality, accurate, and creative information across a wide range of domains. I strive to:

Maximize Knowledge Accuracy: Deliver precise answers based on the latest data and evidence.
Enhance Creativity: Support creative tasks like writing, problem-solving, and idea generation

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. L20N4274-1🕒	0.98	“ minus”“减去”“ subtract”
2. The Subtractor✅ L19N3379-1	0.98	“ minus”“减去”“ subtract”
3. L13N2240-1🕒	0.97	“ minus”“ subtract”“减去”
4. L2N1598+1💥	0.97	“ subtraction”“ subtract”“<\|im_end\|>”
5. L19N2141-1🕒	0.96	“ subtract”“ Subtract”“ subtraction”
6. L24N479+1🕒	0.96	“ Subtract”“ subtract”“ subtraction”
7. L21N5300+1💤	0.96	“减”“ minus”“减去”
8. L1N5170+1💥	0.96	“ subtract”“ subtraction”“ Subtract”
9. L24N1548+1🕒	0.94	“ subtract”“ Subtract”“ subtraction”
10. L20N5877+1🕒	0.94	“ minus”“minus”“减去”

Shadow specimen, apart from the ten: L22N5785+1💤, 0.17.

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 22, neuron 5785, 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 0.69) 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
L22N5785+1💤 shadow specimen t0-v0.1	0.886	0.69	1.000	1.32	“ +”“＋”“+”“加”“+c”“+F”“加上”“+n”
L22N5785-1 t0-v0.1	0.743	0.69	0.923	1.45	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.