Layer 12 · Neuron 2730 · suppressed

L12N2730-1

Word quirks:

“ generally”“ somewhat”“ sometimes”“ broader”“ often”“ slightly”“ clearer”“ emphasis”

💤 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 Tongyi Lab. My capabilities include advanced reasoning, strong language understanding, and the ability to handle complex tasks across diverse domains. Here's a breakdown of my features:

Core Capabilities:
Advanced Reasoning: Strong logical reasoning and mathematical computation (including algebraic and calculus skills).
Long Context: Support for processing very long documents (up to 256K tokens) with high accuracy.
Multilingual Support: Fluent in multiple languages, including Chinese, English, and others.

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. L11N4120-1🕒	0.89	“ often”“ somewhat”“ clearer”
2. L13N7428-1🕒	0.84	“ sometimes”“ occasionally”“ often”
3. L13N1487+1💤	0.84	“ sometimes”“ interesting”“ often”
4. L11N6763-1💤	0.84	“ usually”“ generally”“ sometimes”
5. L12N8630+1🕒	0.84	“ clearer”“ sometimes”“ less”
6. L14N60-1💤	0.84	“ generally”“ often”“ general”
7. L11N5546+1🕒	0.83	“ typically”“ often”“ usually”
8. L10N6875+1🕒	0.83	“ somewhat”“ momentarily”“ sometimes”
9. L12N1167-1🕒	0.82	“ sometimes”“ related”“ usually”
10. L10N7602+1💤	0.82	“ generally”“ often”“ usually”

Shadow specimen, apart from the ten: L12N2730+1🕒, 0.33.

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 2730, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.70, 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
L12N2730+1🕒 shadow specimen t0-v0.1	0.909	0.64	1.108	1.11	“ actually”“ no”“ nowhere”“ concern”“ nonetheless”“ doesn”“ actual”“ concerns”
L12N2730-1 t0-v0.1	0.697	0.72	0.889	1.17	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.