Layer 19 · Neuron 4801 · suppressed

L19N4801-1

Word quirks:

“ code”“ codes”“ Code”“code”“Code”“代码”“_code”“ coding”

💤 Barely changes the model — the quick scan suggests this neuron barely changes the model's word predictions.

Identity (according to the specimen itself)

Hello! I am Qwen35, a large-scale model with code intelligence (code generation, debugging, etc.), and my mission is to provide comprehensive support for various tasks, including but not limited to:

Code Generation & Debugging: Writing, optimizing, and troubleshooting code across multiple programming languages.
Data Analysis: Performing complex calculations, statistical analysis, and data visualization.
Multilingual Communication: Supporting over 130 language codes for global communication.
AI Integration: Enabling seamless integration into AI models and frameworks.
Educational Support: Providing personalized learning resources and

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. L27N3677+1🕒	0.99	“ code”“code”“ codes”
2. L31N709+1💥	0.99	“ code”“code”“ Code”
3. L26N2253-1🕒	0.99	“ codes”“代码”“ code”
4. L27N1996+1🕒	0.99	“ code”“ Code”“Code”
5. L21N6664+1💤	0.98	“ code”“code”“Code”
6. L24N7744-1🕒	0.98	“ code”“ Code”“Code”
7. L25N5212-1🕒	0.98	“ code”“_code”“-code”
8. L26N9145-1🕒	0.98	“ code”“ Code”“_CODE”
9. L27N2294+1🕒	0.98	“ code”“code”“_code”
10. L28N4627-1🕒	0.98	“ code”“ Code”“代码”

Shadow specimen, apart from the ten: L19N4801+1💤, -0.01.

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 19, neuron 4801, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.53, 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
L19N4801+1💤 shadow specimen t0-v0.1	0.486	0.72	0.602	1.17	“ connections”“ links”“ capabilities”“ overlaps”“ ties”“ linking”“ attachments”“这么”
L19N4801-1 t0-v0.1	0.529	0.71	0.642	1.27	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.