Layer 17 · Neuron 5016 · boosted

L17N5016+1

Word quirks:

“—not”“ either”“—from”“—as”“—”“ —”“—including”“—even”

🕒 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 Qwen3.5, a large-scale language model (as the name suggests, "Qwen" refers to my predecessor, Qwen3.0 or earlier versions, depending on the specific context of "as the name suggests," which might be a slight misinterpretation; more accurately, it refers to the "Qwen" series, not necessarily "as the name suggests" in a literal sense, unless there's a specific context like "as the name suggests" referring to something else, perhaps a feature or a capability, but more likely it's a misunderstanding or a typo, possibly meaning "as the name suggests

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. L15N4249-1🕒	0.92	“—”“——”“ —”
2. L16N6741-1💤	0.91	“—and”“—even”“—including”
3. L18N8718-1💥	0.90	“—for”“—”“—including”
4. L18N3013-1💤	0.90	“—and”“—one”“ —”
5. L21N2254-1🕒	0.90	“—even”“—”“—not”
6. L19N1502-1🕒	0.89	“—not”“—with”“—even”
7. L18N8419-1💤	0.89	“—for”“—including”“—and”
8. L18N3253-1💤	0.89	“—”“——”“—in”
9. L17N3711+1💤	0.89	“—for”“—”“—not”
10. L18N848-1🕒	0.89	“—”“ —”“—and”

Shadow specimen, apart from the ten: L17N5016-1🕒, 0.19.

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 17, neuron 5016, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.86, 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
L17N5016+1 t0-v0.1	0.865	0.64	0.982	1.49	Top boosted tokens ↑
L17N5016-1🕒 shadow specimen t0-v0.1	1.023	0.66	1.143	1.04	“ this”“.”“__.”“[this”“this”“.””“”.”“ changed”

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.