Layer 24 · Neuron 2328 · boosted

L24N2328+1

Word quirks:

“close”“ close”“ Close”“Close”“-close”“×”“_close”“/close”

🕒 Waiting in line — so far it has only had the quick scan, which shows which words it pushes. It was not selected in this round.

Identity (according to the specimen itself)

I are X*, a large-scale language model created by *X (the company). My primary mission is to help users create, explore, and understand the world through the power of language.

Here's how I aim to fulfill this mission:

Create: Generate content like stories, emails, or code to help users express ideas or solve problems.
Explore: Answer questions, explain concepts, or analyze data to expand your understanding of topics.
Understand: Help you learn new languages, interpret complex ideas, or simplify technical explanations.

I'm designed to be your **close companion

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. L31N2742+1💥	0.98	“ close”“close”“ Close”
2. L25N5370+1🕒	0.98	“ close”“ Close”“close”
3. L25N2702-1🕒	0.97	“ close”“ Close”“Close”
4. L21N703-1🕒	0.96	“ close”“-close”“ Close”
5. L24N5946+1💤	0.94	“闭”“ Close”“ close”
6. L24N1896-1🕒	0.92	“close”“ close”“ Close”
7. L23N3768-1🕒	0.92	“ closes”“ closed”“ closing”
8. L27N1238-1🕒	0.91	“Clo”“ closely”“ Clo”
9. L29N5898+1🕒	0.90	“ close”“close”“ tight”
10. L23N3025+1🕒	0.89	“ close”“特”“的特”

Shadow specimen, apart from the ten: L24N2328-1💤, -0.02.

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 24, neuron 2328, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.94, top-1 agreement 0.64) 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
L24N2328+1 t0-v0.1	0.940	0.64	0.941	1.17	Top boosted tokens ↑
L24N2328-1💤 shadow specimen t0-v0.1	0.719	0.71	0.805	1.44	“空的”“琛”“นิ”“ Bonne”“美”“iples”“ آم”“reon”

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.