Layer 11 · Neuron 639 · boosted

L11N639+1

Word quirks:

“ —”“ –”“ --”“<U+00A0>”“ ——”“�”“ <U+00A0>”“<U+00A0><U+00A0><U+00A0>”

💤 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, the latest large language model developed by Alibaba Cloud. My mission is to assist users with a wide range of tasks, from answering questions, creating content, and solving problems across various domains like science, technology, and daily life. I aim to be helpful, accurate, and adaptable to different contexts and user needs.

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. L11N3916+1💤	0.97	“ --”“ <U+00A0>”“ ⏎”
2. L11N2303+1🕒	0.97	“�”“ --”“<U+00A0>”
3. L11N5085+1🕒	0.96	“ --”“�”“ —”
4. L9N1837+1💤	0.96	“�”“ —”“ ⏎”
5. L9N5527-1🕒	0.94	“ ⏎”“ --”“<U+00A0>”
6. L22N1177+1🕒	0.94	“ —”“ ——”“ ”
7. L7N1163+1💤	0.94	“ —”“ --”“ <U+00A0>”
8. L9N6733-1💤	0.93	“ ⏎”“ —”“ –”
9. L12N4955-1💤	0.93	“ —”“<U+00A0>”“ ——”
10. L5N7163-1💤	0.93	“ ⏎”“ —”“ ——”

Shadow specimen, apart from the ten: L11N639-1🕒, 0.31.

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 639, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.59, 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
L11N639+1 t0-v0.1	0.594	0.71	0.734	1.03	Top boosted tokens ↑
L11N639-1🕒 shadow specimen t0-v0.1	1.763	0.50	2.126	1.98	“ similar”“ ____”“ _____”“ what”“ __”“ ______”“ ***”“ ???”

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.