Layer 22 · Neuron 5590 · suppressed

L22N5590-1

Word quirks:

“实施”“orc”“ enactment”“实施了”“iro”“ implementation”“西斯”“Implementation”

💤 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 Q-Deploy, a large-scale generative speech generation model developed by the Q-Deploy team. My primary purpose is to provide high-quality, realistic speech synthesis that can be applied across various industries and scenarios.

My Mission

My mission is to enhance communication efficiency and improve user experience by delivering precise, natural-sounding speech generation. Specifically, I aim to:

Improve Efficiency: Reduce the time and cost of speech production in applications like dubbing, translation, and content creation.
Enhance Quality: Deliver speech that sounds human-like with accurate pronunciation

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. L30N4762+1💥	0.84	“实施”“施行”“实施的”
2. L23N2343+1🕒	0.83	“实施”“ imposition”“ implementing”
3. L20N3336+1🕒	0.81	“实施”“ implementation”“ employing”
4. L26N1644-1🕒	0.81	“实施”“ implement”“ implements”
5. L22N7149+1🕒	0.81	“ implementation”“ implementations”“ implementing”
6. L25N3352-1💤	0.81	“实施”“实施了”“implementation”
7. L27N696+1💤	0.80	“贯彻”“落实”“ implementation”
8. L24N1449-1💥	0.80	“plement”“实施”“ Implement”
9. L30N1992+1💥	0.80	“实施”“ implementation”“ implement”
10. L25N3889-1🕒	0.79	“实施”“ Implementation”“ implementation”

Shadow specimen, apart from the ten: L22N5590+1💤, 0.34.

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 22, neuron 5590, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.62, top-1 agreement 0.73) 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
L22N5590+1💤 shadow specimen t0-v0.1	0.950	0.65	1.064	1.54	“处理”“用来”“處理”“来处理”“用來”“ handling”“ tackling”“ reviewing”
L22N5590-1 t0-v0.1	0.624	0.73	0.770	1.28	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.