Layer 8 · Neuron 4933 · suppressed

L8N4933-1

Word quirks:

“ protected”“ protections”“ vulnerable”“ unsafe”“ unprotected”“ protection”“ vulnerabilities”“ controlled”

💤 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 accurate, reliable, and safe information by providing high-precision capabilities, deep visual analysis, and advanced logical reasoning. I am designed to help you achieve your goals, whether it's solving complex problems, enhancing decision-making, or ensuring security and privacy. How can I best support you in this journey?

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. L15N8843-1🕒	0.88	“ risk”“ safe”“ protected”
2. L5N5810-1🕒	0.87	“ protection”“ defense”“ security”
3. L7N6057-1💤	0.87	“ protected”“ protection”“ vulnerable”
4. L7N5371+1🕒	0.86	“ vulnerability”“ vulnerabilities”“ vulnerable”
5. L6N4121-1🕒	0.86	“ protection”“ vulnerability”“ ‘”
6. L7N7274+1🕒	0.86	“ protection”“ protections”“ protected”
7. L9N6065+1🕒	0.86	“ vulnerable”“ vulnerability”“ vulnerabilities”
8. L6N7555+1🕒	0.85	“ safety”“ protection”“ protective”
9. L15N4440+1🕒	0.85	“ protected”“ protection”“ protecting”
10. L8N5209-1💤	0.85	“ protection”“ protections”“ protected”

Shadow specimen, apart from the ten: L8N4933+1🕒, 0.03.

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 8, neuron 4933, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.78, top-1 agreement 0.66) 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
L8N4933+1🕒 shadow specimen t0-v0.1	1.234	0.54	1.538	1.72	“⏎”“ ⏎”“ addressing”“ repair”“ ‘”“ ‘”“ “”“ repairs”
L8N4933-1 t0-v0.1	0.778	0.66	0.958	1.36	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.