Layer 6 · Neuron 5134 · boosted

L6N5134+1

Word quirks:

“ allowed”“ protected”“⏎”“ legal”“ safety”“ free”“ legally”“ permitted”

💤 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 Qwen3.5, a large language model developed by Alibaba Cloud. My mission is to assist users with their inquiries, provide information, and support various tasks related to technology, science, and general knowledge. I aim to help users achieve their goals and provide the best possible service.

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. L9N5524+1🕒	0.89	“ permitted”“ freely”“ permissible”
2. The Compliance Clerk✅ L8N2748-1	0.89	“ allowable”“ allowed”“ freely”
3. L10N1762+1🕒	0.82	“ lawful”“ protected”“ legal”
4. L16N7258-1🕒	0.79	“ legal”“ legalized”“ allowed”
5. L8N7548-1🕒	0.79	“ legal”“ lawful”“ freely”
6. L11N850+1🕒	0.77	“ protected”“ reasonable”“ lawful”
7. L5N3960-1🕒	0.77	“ allowed”“ permitted”“ permit”
8. L7N6074-1🕒	0.76	“ legal”“ defamation”“ Legal”
9. L11N7644-1🕒	0.76	“ allowed”“ permitted”“Allowed”
10. L20N1335-1🕒	0.76	“ legality”“ legal”“ Allowed”

Shadow specimen, apart from the ten: L6N5134-1🕒, 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 6, neuron 5134, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.67, top-1 agreement 0.67) 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
L6N5134+1 t0-v0.1	0.669	0.67	0.825	1.33	Top boosted tokens ↑
L6N5134-1🕒 shadow specimen t0-v0.1	1.803	0.48	2.113	2.20	“ talent”“ careers”“ global”“ abroad”“ countries”“ universities”“ jobs”“ career”

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.