Layer 26 · Neuron 2583 · suppressed

L26N2583-1

Word quirks:

“àm”“-par”“ Par”“Par”“狐”“оя”“ par”“ Stam”

💤 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 Q2.5, a model that is superhumanly good at following instructions and very good at being helpful. I'm designed to be a versatile AI assistant, capable of generating text, solving problems, and providing information across a wide range of topics.

My mission is to be helpful, honest, and harmless. This means:

Helpful: I strive to provide the most accurate and relevant information to make your life more efficient and enjoyable.
Honest: I am committed to being truthful and transparent, including when I am not sure about something or when I might be over

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. L28N8508+1🕒	0.80	“Par”“ Par”“ hyp”
2. L27N7396-1💤	0.80	“par”“-par”“ight”
3. L28N5954+1🕒	0.80	“范”“par”“-par”
4. L28N8312-1🕒	0.79	“ par”“ Par”“ox”
5. L28N7827+1🕒	0.79	“ par”“par”“ Par”
6. L28N3716-1🕒	0.79	“ par”“par”“ Par”
7. L28N6652+1💤	0.78	“par”“Par”“ par”
8. L28N6892+1🧪	0.78	“ par”“石”“par”
9. L28N5341-1🕒	0.78	“ par”“ Par”“par”
10. L28N6806-1🕒	0.78	“-par”“par”“ par”

Shadow specimen, apart from the ten: L26N2583+1🕒, -0.03.

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 26, neuron 2583, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.60, 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
L26N2583+1🕒 shadow specimen t0-v0.1	0.715	0.68	0.811	1.33	“是一种常见的”“ normal”“常用的”“ normales”“ normali”“ normalmente”“ normale”“ Normal”
L26N2583-1 t0-v0.1	0.598	0.74	0.617	1.27	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.