Layer 13 · Neuron 8880 · suppressed

L13N8880-1

Word quirks:

“ usually”“ focused”“ typically”“ focus”“ focusing”

💤 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 Tongyi Lab. My mission is to provide safe, efficient, and personalized solutions tailored to user needs. Here's how I support users across different domains:

Key Capabilities & Missions

Enhanced Reasoning:

I excel in complex reasoning tasks like math, science, and logic, often identifying patterns that others might miss.

Multilingual Support:

I support over 100 languages (e.g., Chinese, English, French, German) with high accuracy, enabling global

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. L15N3799+1💤	0.92	“ typically”“ focusing”“ typical”“ basic”“ focus”
2. L12N2107+1💤	0.91	“ always”“ general”“ generally”“ focus”“ typically”
3. L13N9031+1🕒	0.91	“ focus”“ focusing”“ now”“ focuses”“ typically”
4. L10N6563+1🕒	0.90	“ focus”“ usually”“ specific”“ generally”“ typically”
5. L14N7232-1🕒	0.90	“ typically”“ usually”“ category”“ focus”“ focused”
6. L13N8867+1💤	0.89	“ focus”“ here”“ often”“ focused”“ specific”
7. L12N4078+1💤	0.89	“ focus”“ sometimes”“ generally”“ usually”“ specific”
8. L15N6721-1💤	0.89	“ focus”“ info”“ usually”“ focuses”“ focusing”
9. L12N7887+1💤	0.89	“ usually”“ generally”“ always”“ typically”“ focus”
10. L9N1021+1💤	0.88	“ focus”“ typical”“ concern”“ usually”“ focused”

Shadow specimen, apart from the ten: L13N8880+1💥, 0.02.

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 13, neuron 8880, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.66, 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
L13N8880+1💥 shadow specimen t0-v0.1	4.253	0.19	4.709	2.60	“ discarded”“ papers”“ results”“ paper”“ discard”
L13N8880-1 t0-v0.1	0.659	0.71	0.797	1.20	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.