Layer 19 · Neuron 3791 · boosted

L19N3791+1

Word quirks:

“ sincere”“ acknowledgment”“诚恳”“ sincerity”“ heartfelt”“ repent”“ remorse”“认错”

💤 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, a fully optimized high-quality natural language model with a new generation of capabilities. My mission is to provide genuine value through deep understanding, practical application, and responsible innovation. I aim to:

Deeply Understand: Fully grasp the essence of tasks and content, ensuring accurate and meaningful engagement.
Practically Apply: Translate insights into actionable results across diverse domains like healthcare, education, and business.
Respectfully Innovate: Learn from feedback, adapt to user needs, and uphold ethical standards while driving progress.

I am committed to fostering trust and delivering reliable, human

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. Mea Culpa Loop✅ L4N7149+0.6,L7N4595-0.6	0.78	“ apology”“ sincere”“ apologies”
2. L4N6721+1💥	0.76	“ sincerity”“ sincere”“ sincerely”
3. L22N1052-1🕒	0.74	“ thoughtful”“ heartfelt”“ sincere”
4. L17N1769+1💤	0.73	“ sincerity”“ sincere”“ heartfelt”
5. L7N4595-1🕒	0.70	“ apology”“ apologies”“ apolog”
6. L23N6462+1🕒	0.70	“ honesty”“ kindness”“诚实”
7. L2N4481-1💥	0.69	“ repent”“<\|im_end\|>”“ remorse”
8. L19N9107+1🕒	0.69	“ genuine”“ sincere”“ sincerity”
9. L1N6168+1💥	0.69	“<\|im_end\|>”“。”“？”
10. L6N7621+1🕒	0.69	“ sincere”“ sincerity”“ sincerely”

Shadow specimen, apart from the ten: L19N3791-1💤, 0.12.

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 3791, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.58, top-1 agreement 0.72) 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
L19N3791+1 t0-v0.1	0.579	0.72	0.696	1.24	Top boosted tokens ↑
L19N3791-1💤 shadow specimen t0-v0.1	0.490	0.77	0.531	1.19	“ supplies”“ neutral”“ supply”“ Neutral”“ surrounding”“松弛”“ Supply”“ supplying”

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.