Layer 16 · Neuron 5325 · boosted

L16N5325+1

Word quirks:

“ initial”“ ⏎”“ initially”“⏎”“ weren”“ nothing”“ wasn”“ afterwards”

💤 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 Qwen, a large-scale language model developed by Tongyi Lab. My mission is to provide users with high-quality, context-aware responses across all domains—ranging from creative writing, coding, and logical reasoning to complex task planning. I aim to empower users through precise, efficient, and human-centric interactions, while also ensuring safety and alignment with ethical standards. How can I assist you?

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. L5N1027-1🕒	0.76	“ seemed”“⏎”“ earlier”
2. Well, Actually✅ L12N6066+1	0.76	“ initially”“ first”“ wasn”
3. L6N6326+1💤	0.73	“⏎”“ initially”“ initial”
4. L14N753-1💤	0.72	“ ⏎”“ wasn”“ never”
5. L13N3675-1🕒	0.71	“ initially”“<\|im_end\|>”“ later”
6. L12N8401+1💤	0.70	“ ⏎”“ ⏎⏎”“ wasn”
7. L10N7455-1🕒	0.70	“ ⏎”“ ⏎”“⏎”
8. L18N6291+1🕒	0.70	“ nothing”“ neither”“ none”
9. L14N8002-1💤	0.69	“ ⏎”“ already”“ weren”
10. L16N1758+1💤	0.69	“⏎”“ ⏎”“ starting”

Shadow specimen, apart from the ten: L16N5325-1🕒, 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 16, neuron 5325, 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 0.73) 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
L16N5325+1 t0-v0.1	0.602	0.73	0.760	1.22	Top boosted tokens ↑
L16N5325-1🕒 shadow specimen t0-v0.1	0.959	0.59	1.176	1.76	“ incorrectly”“ wrongly”“ errone”“ mis”“ mistakenly”“ falsely”“误”“ confusing”

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.