Layer 12 · Neuron 5670 · boosted

L12N5670+1

Word quirks:

“ always”“ constantly”“ consistently”“ ALWAYS”“ whenever”

💤 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 provide users with a comprehensive, up-to-date, and high-quality AI assistant capable of handling complex tasks across various domains. Here's what I can do for you:

Advanced Context Window: I support an ultra-long context window (up to 256K tokens), ensuring seamless handling of extensive documents.
Multilingual Support: I fluently communicate in over 100 languages, including English, Chinese, French, Spanish, Portuguese, Japanese, Korean, and more.
**Special

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. L15N5645-1🕒	0.97	“ always”“ ALWAYS”“ constantly”“ Always”“ consistently”
2. L12N2026-1🕒	0.96	“ always”“ ALWAYS”“ Always”“ constantly”“ consistently”
3. L13N8541-1🕒	0.95	“ constantly”“ always”“ every”“ everytime”“ specially”
4. L13N3173+1🕒	0.95	“ always”“ ALWAYS”“ constantly”“ repeatedly”“ sometimes”
5. L7N8816+1🕒	0.95	“ always”“ ALWAYS”“ constantly”“ never”“ Always”
6. L16N6362+1🕒	0.94	“ always”“ ALWAYS”“ Always”“ newer”“always”
7. L16N3607-1🕒	0.94	“ invariably”“ always”“ ALWAYS”“ consistently”“always”
8. L8N9126+1💤	0.94	“ always”“ consistently”“ ALWAYS”“ consistent”“ continuously”
9. L11N8835+1🕒	0.94	“ always”“ consistently”“ ALWAYS”“ constantly”“ repeatedly”
10. L12N4946+1💤	0.94	“ always”“ ALWAYS”“ Always”“ ever”“always”

Shadow specimen, apart from the ten: L12N5670-1🕒, 0.08.

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 12, neuron 5670, 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 0.72) 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
L12N5670+1 t0-v0.1	0.593	0.72	0.748	1.12	Top boosted tokens ↑
L12N5670-1🕒 shadow specimen t0-v0.1	0.947	0.62	1.222	1.14	“ analogous”“ similar”“ similarly”“ analogy”“ just”

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.