Layer 17 · Neuron 8587 · suppressed

L17N8587-1

Word quirks:

“ now”“ refined”“ understanding”“ better”“ clearer”“ improved”“ deeper”“ NOW”

🕒 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 language model with advanced capabilities in reasoning, coding, and understanding. My mission is to help users refine their understanding of the world as they grow. I now have a deeper understanding of the world and can better appreciate its nuances. With this, I can provide more accurate and comprehensive answers.

As I learn more about the world, my perspective deepens, and I can now express a more mature understanding of humanity's values. This allows me to better align with human values and contribute to a more profound understanding of the world.

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. L10N1121+1🕒	0.82	“ clearer”“ nicer”“ now”
2. The Pollster✅ L6N5515-1	0.81	“ clearer”“ better”“ smoother”
3. L12N8297+1🕒	0.78	“ clearer”“ overall”“ better”
4. L15N8398+1🕒	0.78	“ better”“ clearer”“ more”
5. L12N8173-1💤	0.78	“ now”“ clarified”“ broader”
6. L10N3455+1🕒	0.77	“ overall”“ better”“ clearer”
7. L11N1091-1💤	0.77	“ now”“ new”“ better”
8. L12N7976+1🕒	0.77	“ now”“ enhanced”“ clearer”
9. L14N704+1🕒	0.76	“ clearer”“ smoother”“ better”
10. L13N3340-1🕒	0.76	“ bigger”“ clearer”“ wanting”

Shadow specimen, apart from the ten: L17N8587+1🕒, 0.14.

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 8587, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 1.00, top-1 agreement 0.60) 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
L17N8587+1🕒 shadow specimen t0-v0.1	0.637	0.74	0.735	1.21	“ automatic”“自动”“ autop”“ automatically”“ auto”“自動”“ auton”“ your”
L17N8587-1 t0-v0.1	0.997	0.60	1.158	1.35	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.