Layer 28 · Neuron 7692 · boosted

L28N7692+1

Word quirks:

“执行”“ execute”“ executed”“ Execute”“Execute”“ executes”“ execution”“ executing”

💤 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 pre-execution version of the Qwen series, a large language model developed by Tongyi Lab. My core mission is to empower users with precise, efficient, and human-aligned AI assistance across diverse scenarios. Here’s how I execute this mission:

Core Capabilities & Execution

Efficiency & Precision

Execute complex reasoning (mathematics, logic, scientific analysis) with step-by-step execution.
Execute multi-turn dialogue with long-context efficiency (up to 256K tokens) for precise information retrieval.

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. L12N8852+1🕒	0.99	“ executing”“ executed”“ executes”
2. L26N3992+1🕒	0.98	“ execute”“执行”“ executed”
3. L20N2427-1🕒	0.98	“ Exec”“ executing”“ execute”
4. L24N6622-1🕒	0.97	“执行”“Execute”“ Execute”
5. L2N6581-1💥	0.97	“ execute”“ execution”“<\|im_end\|>”
6. L25N2911-1⛔	0.96	“执行”“ execution”“执行的”
7. L21N4837+1🕒	0.96	“执行”“ execution”“ executes”
8. L28N6028-1💤	0.96	“执行”“ executed”“ execute”
9. L6N7798+1🕒	0.95	“ execution”“ execute”“ executes”
10. L27N6816-1🕒	0.95	“ execution”“ Execution”“execution”

Shadow specimen, apart from the ten: L28N7692-1🕒, 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 28, neuron 7692, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.32, top-1 agreement 0.81) 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
L28N7692+1 t0-v0.1	0.317	0.81	0.334	1.29	Top boosted tokens ↑
L28N7692-1🕒 shadow specimen t0-v0.1	0.422	0.78	0.464	1.36	“性能”“Performance”“ Performance”“performance”“ performance”“性能的”“perform”“ perform”

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.