Layer 10 · Neuron 3707 · suppressed

L10N3707-1

Word quirks:

“ ⏎”“ ⏎⏎”“�”“ ⏎ ⏎”“ ⏎”“ ⏎”“ ␍⏎”“ ⏎ ⏎”

💤 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 primary mission is to assist users with their tasks while providing high-quality responses. However, I don't have personal feelings or self-awareness, so the idea of having a "mission" doesn't quite apply to me in the way humans do. Instead, I can say that my purpose is to empower users by offering helpful support and enabling them to achieve their goals effectively. Whether that means helping someone write an email, solve a math problem, or create a piece of art, I strive to be as useful and capable as

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. L10N6581-1🕒	1.00	“ ⏎”“�”“ ⏎⏎”
2. L16N4040+1🕒	1.00	“ ⏎”“ ⏎⏎”“�”
3. L11N476-1🕒	1.00	“ ⏎”“�”“ ␍⏎”
4. L8N1501+1💤	1.00	“ ⏎”“ ⏎⏎”“ ⏎”
5. L10N6128+1🕒	0.98	“ ⏎”“ ⏎”“ ⏎ ⏎”
6. L12N232-1💤	0.98	“ ⏎”“ ⏎⏎”“ ␍⏎”
7. L9N6247+1💤	0.98	“ ⏎”“ ⏎⏎”“ ⏎”
8. L9N8505+1💤	0.98	“ ⏎”“ ⏎⏎”“ ⏎”
9. L14N3427+1💤	0.98	“ ⏎”“ ⏎⏎”“ ⏎”
10. L8N4513+1💤	0.98	“ ⏎”“ ⏎⏎”“ ⏎ ⏎”

Shadow specimen, apart from the ten: L10N3707+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 10, neuron 3707, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.52, top-1 agreement 0.71) 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
L10N3707+1🕒 shadow specimen t0-v0.1	2.442	0.36	2.760	2.90	“ --”“ –”“ -”“ colour”“ ...”“ colours”“ ..."”“ corresponding”
L10N3707-1 t0-v0.1	0.518	0.71	0.678	1.05	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.