Layer 17 · Neuron 2774 · suppressed

L17N2774-1

Word quirks:

“ contribution”“ contributed”“ contributor”“ contributing”“ contributes”“ contribute”“ contributions”“ contributors”

🕒 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)

Hello! I am a large language model that has been trained on the information available up to my current year. My name is Qwen3, and I am part of the series of models I've contributed to in this field.

As for my mission, I aim to provide accurate, helpful, and diverse responses across a wide range of topics. Whether it's answering questions in various languages, sharing knowledge about science, technology, or culture, or engaging in creative writing, I strive to contribute positively to human understanding and well-being. I also take care to avoid spreading misinformation or controversial content, ensuring that my work aligns with safety

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. L5N3480+1💥	0.99	“ contribution”“ contributions”“ contributed”
2. L17N6049-1💤	0.98	“ contributions”“ contributing”“ Contrib”
3. L17N8533+1🕒	0.98	“ contributing”“ contributions”“ contributed”
4. L9N8017+1🕒	0.98	“ contributing”“ contribution”“ contributors”
5. L16N8658-1🕒	0.98	“ contributing”“ contributions”“⏎”
6. L9N9076+1🕒	0.98	“ contribution”“ contributions”“ contributing”
7. L17N8751-1🕒	0.98	“ contribution”“ contributing”“ contrib”
8. L24N2404+1🕒	0.98	“ contributing”“ Contrib”“ contribution”
9. L19N1226+1🕒	0.98	“ contributing”“ contribution”“ contribute”
10. L16N6313-1🕒	0.98	“ contribution”“ contributing”“ contrib”

Shadow specimen, apart from the ten: L17N2774+1🕒, 0.22.

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 2774, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.74, top-1 agreement 0.69) 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
L17N2774+1🕒 shadow specimen t0-v0.1	1.396	0.54	1.569	1.63	“ initial”“ management”“ Initial”“ Management”“ project”“ streamline”“ strategy”“ leadership”
L17N2774-1 t0-v0.1	0.741	0.69	0.937	1.43	Top boosted tokens ↑

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.