Layer 12 · Neuron 2756 · boosted

L12N2756+1

Word quirks:

“ sometimes”“ preferred”“ constantly”“ preference”“ always”“ preferable”“ but”“ alternative”

💤 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's Tongyi Lab. My mission is to provide high-quality, accurate, and diverse information to users worldwide. I strive to be helpful, efficient, and reliable in answering questions, solving problems, and assisting with tasks.

As an AI model, I have the following capabilities:

Multilingual Support: I can communicate in multiple languages.
Task Automation: I can automate various processes.
Data Processing: I can handle large datasets.
Code Generation: I can write and debug code.
**Knowledge Retrie

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. L13N5368-1🕒	0.80	“ always”“ choice”“ preferring”
2. L12N8812-1🕒	0.80	“ whenever”“ if”“ preferred”
3. L16N3527-1🕒	0.80	“ preferred”“ preference”“ preferring”
4. L11N9060-1💤	0.79	“ without”“ preferred”“ preference”
5. L20N2424-1🕒	0.79	“ preferably”“ always”“ mostly”
6. L9N4796-1🕒	0.78	“ always”“ choice”“ sometimes”
7. L18N8186+1🕒	0.78	“ preferred”“ Preferred”“ eventual”
8. L12N3190+1💤	0.78	“ alternative”“ preferred”“ alternatives”
9. L10N6499-1🕒	0.78	“ preference”“ preferred”“ freely”
10. L17N3970-1💤	0.77	“ preferable”“ preferred”“ prefers”

Shadow specimen, apart from the ten: L12N2756-1💤, 0.19.

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 2756, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.67, top-1 agreement 0.64) 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
L12N2756+1 t0-v0.1	0.666	0.64	0.840	1.26	Top boosted tokens ↑
L12N2756-1💤 shadow specimen t0-v0.1	0.738	0.65	0.987	1.24	“ same”“ similarly”“ ⏎”“ similar”“ Same”“ stuff”“ likewise”“ things”

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.