Layer 19 · Neuron 5701 · suppressed

L19N5701-1

Word quirks:

“ notar”“ witnesses”“ witness”“ verified”“ witnessed”“见证”“ Witness”“ witnessing”

💤 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 Qwen36, the latest model in the T001 series. My mission is to empower users with advanced AI capabilities through:

Precise interactions: Optimized for complex tasks like legal verification, medical data analysis, and high-stakes document drafting.
Global accessibility: Supporting 200+ languages and cross-border applications (e.g., international business contracts).
Trust & efficiency: Ensuring secure, verifiable transactions via blockchain integration and real-time legal databases.
Customization: Tailored solutions for industries like healthcare, finance, and education.

Would you

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. L25N5306+1🕒	0.94	“ witness”“ Witness”“Witness”
2. L31N7491+1💥	0.94	“ witness”“ Witness”“Witness”
3. L20N2306+1🕒	0.94	“ witnessing”“见证”“ witness”
4. L24N2061-1🕒	0.91	“ witnesses”“见证”“ Witnesses”
5. L23N8588+1🕒	0.90	“ witness”“ witnessing”“Witness”
6. L24N5688-1🕒	0.90	“ witness”“ Witness”“ witnesses”
7. L23N3348+1🕒	0.89	“ witness”“见证”“ Witness”
8. L6N3745+1💥	0.89	“ notar”“ Witness”“ witness”
9. L4N6378+1🕒	0.88	“ witness”“ witnessing”“ testimony”
10. L23N3953-1🕒	0.88	“ witness”“ Witness”“Witness”

Shadow specimen, apart from the ten: L19N5701+1🕒, 0.13.

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 5701, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.57, 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
L19N5701+1🕒 shadow specimen t0-v0.1	0.662	0.73	0.841	1.26	“想象”“ description”“ imported”“ lost”“描述的”“ lucky”“ prediction”“ importing”
L19N5701-1 t0-v0.1	0.573	0.78	0.545	1.14	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.