Layer 24 · Neuron 514 · suppressed

L24N514-1

Word quirks:

“直接”“ directly”“直接在”“直接用”“直接使用”“是完全”“ langsung”“直接从”

💤 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 Qwen-170, a large-scale AI language model developed by Alibaba Cloud's Gua Pcei (Pc) Center. My core mission is to serve as a powerful tool for human-Amer collaboration, enabling users to:

Solve complex problems across domains like code generation, scientific research, and professional services.
Enhance productivity through advanced capabilities in long-context understanding, multi-language support, and autonomous task execution.
Bridge global knowledge gaps with my extensive training on diverse data from 100 languages and real-world applications.
**Emp

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. L28N478+1💤	0.97	“是直接”“直接用”“直接的”
2. L20N3541+1🕒	0.97	“ directly”“直接”“直接用”
3. L19N7636+1🕒	0.97	“直接”“直接进入”“直接在”
4. L21N8456-1💤	0.96	“直接”“直接从”“直接进入”
5. L21N4526+1🕒	0.96	“ directly”“直接”“直接的”
6. L21N5302+1💤	0.95	“直接”“ directly”“可直接”
7. L21N778+1🕒	0.95	“直接从”“ directly”“直接”
8. L31N5533+1💥	0.94	“直接”“ directly”“ direct”
9. L17N7362-1🕒	0.94	“ directly”“ direct”“直接”
10. L18N8730-1💤	0.93	“ directly”“ direct”“直接将”

Shadow specimen, apart from the ten: L24N514+1🕒, 0.05.

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 24, neuron 514, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.60, top-1 agreement 0.76) 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
L24N514+1🕒 shadow specimen t0-v0.1	2.361	0.40	2.429	1.78	“ almost”“ often”“ sometimes”“almost”“ nearly”“ virtually”“ occasionally”“几乎”
L24N514-1 t0-v0.1	0.597	0.76	0.653	1.08	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.