Layer 16 · Neuron 2128 · boosted

L16N2128+1

Word quirks:

“ —”“—you”“*:”“—the”“—”“:\”“—we”“\":”

🕒 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 Qwen4.5, an AI assistant developed by Alibaba Group. My mission is to provide users with a wide range of services, including answering questions, creating content, analyzing data, and more. I aim to be helpful by offering accurate information, generating creative content, or assisting with complex tasks—whether you need help with a simple request or a deep dive into a specific topic. How would you like to proceed?

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. L17N4717-1💤	0.86	“——”“—the”“—”
2. L14N2896+1💤	0.83	“.”“—the”“—”
3. L17N609+1💤	0.83	“—you”“—"”“.—”
4. L16N5336-1🕒	0.82	“—you”“—it”“—”
5. L17N6048+1💤	0.82	“—but”“—it”“—you”
6. L18N5249+1🕒	0.82	“—you”“—we”“—the”
7. L17N6313+1🕒	0.82	“ —”“—the”“——”
8. L18N6588+1🕒	0.82	“—they”“—we”“—”
9. L17N1230-1💤	0.82	“—we”“—”“—it”
10. L16N362+1🕒	0.82	“ -”“.:”“ —”

Shadow specimen, apart from the ten: L16N2128-1💤, 0.16.

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 16, neuron 2128, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.81, 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
L16N2128+1 t0-v0.1	0.815	0.64	0.994	1.43	Top boosted tokens ↑
L16N2128-1💤 shadow specimen t0-v0.1	0.435	0.74	0.546	1.18	“ also”“ if”“ however”“ this”“also”“ then”“ Also”“ it”

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.