Layer 25 · Neuron 237 · suppressed

L25N237-1

Word quirks:

“ deal”“态度”“bahn”“主机”“_guard”“ deals”“ornado”“toast”

💤 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 Gee-oo (often referred to as Gee), a large-scale language model developed by the Gee-oo Team. My name reflects my role as a helpful, intelligent, and versatile assistant designed to assist with a wide range of tasks.

Who Am I?

I am an advanced AI model capable of understanding and generating human language across many domains. I have been trained on vast amounts of high-quality data to ensure I can:

Answer questions across topics like science, history, and technology.
Create content such as stories, emails, and code.

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. L28N7563+1🕒	0.61	“锥”“Deal”“ Deal”
2. L27N1640+1🕒	0.59	“ deal”“牌”“ dealing”
3. L22N8392-1🕒	0.57	“ dealing”“ deal”“ deals”
4. L27N2273-1⛔	0.56	“ deal”“ deals”“ Deal”
5. L23N4422-1💤	0.56	“ deal”“ deals”“ Deal”
6. L27N7635-1⛔	0.56	“ deal”“ Deal”“deal”
7. L31N7311-1💥	0.55	“ deal”“ deals”“ Deal”
8. L22N8528+1💤	0.55	“ dealt”“ deal”“ dealing”
9. L24N8711+1💤	0.55	“斋”“对冲”“driver”
10. L27N1256-1🕒	0.55	“ deal”“ Deal”“ dealt”

Shadow specimen, apart from the ten: L25N237+1🕒, 0.11.

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 237, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.63, top-1 agreement 0.68) 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
L25N237+1🕒 shadow specimen t0-v0.1	1.007	0.67	1.006	1.26	“ pa”“沛”“ Pa”“彭”“P”“ Pair”“Pa”“เป”
L25N237-1 t0-v0.1	0.631	0.68	0.649	1.31	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.