Layer 28 · Neuron 5619 · suppressed

L28N5619-1

Word quirks:

“不通”“连通”“通”“的通”“中通”“打通”“开通”“communic”

💤 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 OpenConnectivity Technologies (OpenConnectivity). My mission is to empower users with cutting-edge capabilities across diverse domains—such as open-world coding, full-stack development, multi-modal analysis, and real-time agent planning—to streamline workflows, solve complex problems, and unlock new possibilities for individuals and organizations.

How can I assist you today? Whether you need help with technical challenges, creative projects, or open-ended exploration, I'm here to support your goals.

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. L31N5452-1💥	0.92	“通”“的通”“通的”
2. L31N6755-1💥	0.90	“通”“平”“通了”
3. L27N7562+1💤	0.89	“通”“通的”“通了”
4. L0N7065-1💥	0.87	“通”“<\|im_end\|>”“通路”
5. L28N6104-1💤	0.83	“通”“ Thông”“ thông”
6. L28N3952-1🕒	0.83	“可以通过”“通”“ thông”
7. L26N4353+1🕒	0.79	“通”“通的”“的通”
8. L27N6395-1💥	0.78	“ thông”“通通”“通”
9. L29N5110+1🕒	0.75	“达”“开”“達”
10. L30N6232+1💤	0.75	“通”“通透”“ Vip”

Shadow specimen, apart from the ten: L28N5619+1💤, 0.62.

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 5619, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.37, top-1 agreement 0.78) 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
L28N5619+1💤 shadow specimen t0-v0.1	0.335	0.82	0.349	1.20	“相”“ com”“连”“ordial”“通常需要”“通常”“com”“ preco”
L28N5619-1 t0-v0.1	0.368	0.78	0.420	1.35	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.