Layer 24 · Neuron 2372 · boosted

L24N2372+1

Word quirks:

“ learning”“学习”“ learn”“lear”“learning”“学习的”“Learning”“-learning”

🕒 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 am Zinc-1, a large language learning assistant developed by the Zinc-1 team. My core mission is to assist users in learning and mastering the Zinc-1 language (a hypothetical or specific domain language depending on context) through interactive dialogue, problem-solving, and real-time learning.

If you have any questions about Zinc-1, need help with exercises, or want to explore its unique features, feel free to ask! Let's start learning together! 🌟

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. L23N6980+1🕒	0.99	“ learning”“ Learning”“Learning”
2. L21N7100-1🕒	0.98	“ learning”“ learn”“learning”
3. L25N5181+1🕒	0.98	“学习”“ learning”“ learns”
4. L31N1181+1💥	0.98	“学习”“ learning”“ learn”
5. The Rote Learner✅ L28N3365+1	0.98	“ learning”“learning”“ Learning”
6. The Eternal Student✅ L23N8640-1	0.98	“学习型”“ learning”“学习”
7. L27N3472+1🕒	0.97	“ learning”“ Learning”“Learning”
8. L28N5328+1🕒	0.97	“ learning”“ Learning”“learning”
9. L23N7323+1🕒	0.97	“ learning”“学习方法”“ Learning”
10. L25N4565+1🕒	0.97	“ learning”“ Learning”“Learning”

Shadow specimen, apart from the ten: L24N2372-1💤, 0.15.

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 2372, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.84, top-1 agreement 0.70) 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
L24N2372+1 t0-v0.1	0.836	0.70	0.864	1.27	Top boosted tokens ↑
L24N2372-1💤 shadow specimen t0-v0.1	0.707	0.76	0.764	1.33	“中年”“初中的”“middle”“ middle”“adel”“ Madura”“ Fad”“的中年”

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.