Layer 11 · Neuron 7756 · suppressed

L11N7756-1

Word quirks:

“ protection”“ protected”“ protections”“ protect”“ protecting”“ protects”“保护”“的保护”

🕒 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 developed by Alibaba Cloud. My mission is to provide helpful, accurate, and ethical responses across a wide range of topics. I am designed to assist users with tasks that require reasoning, creativity, and technical knowledge.

While I may not be able to perform complex calculations or generate code in the traditional sense, I can still help with a variety of tasks. My capabilities include:

Advanced reasoning: I can handle complex logical arguments and mathematical concepts.
Code generation: I can write and explain code in multiple programming languages.
Multilingual support:

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. L9N8282+1🕒	0.99	“ protection”“ protecting”“ protective”
2. The Warden✅ L12N5830-1	0.99	“ protection”“ protected”“ protecting”
3. L8N7666-1🕒	0.99	“ protection”“ protective”“ protecting”
4. L9N1353-1🕒	0.99	“ protection”“ protected”“ protecting”
5. L10N5811+1🕒	0.99	“ protection”“ protective”“ protecting”
6. L8N5680+1🕒	0.99	“ protection”“ protecting”“ protect”
7. L13N6268-1🕒	0.98	“ protects”“保护”“ protect”
8. L9N7132-1🕒	0.98	“ protected”“ protection”“ protecting”
9. L26N3487-1🕒	0.98	“保护”“ protect”“ protecting”
10. L10N5998-1🕒	0.98	“ protected”“ protection”“ protective”

Shadow specimen, apart from the ten: L11N7756+1🕒, -0.02.

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 11, neuron 7756, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.97, 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
L11N7756+1🕒 shadow specimen t0-v0.1	2.192	0.46	2.485	1.67	“ always”“ \"”“ ourselves”“ "{”“ desired”“ really”“ "”“ actual”
L11N7756-1 t0-v0.1	0.969	0.66	1.221	1.13	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.