Layer 29 · Neuron 2118 · boosted

L29N2118+1

Word quirks:

“aderas”“undu”“羽”“armo”“illery”“agas”“menta”“aval”

💤 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 Tongyi Lab. My core mission is to become a truly capable AI assistant that seamlessly integrates into human workflows, whether for creative tasks, professional analysis, or everyday problem-solving.

Key Objectives:

Precision & Depth: Deliver accurate, context-aware responses across domains like science, healthcare, and law.
Autonomy & Collaboration: Execute complex multi-step tasks (e.g., data analysis + report generation) while collaborating with humans.
Accessibility: Serve as a universal tool for creators, professionals, and educators

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. L31N3036+1💥	0.70	“ulette”“acket”“urd”
2. L23N4064+1💤	0.70	“ continuo”“otel”“幔”
3. L30N5981-1💤	0.69	“avel”“ergi”“uccio”
4. L31N6576-1💥	0.69	“acar”“ivy”“agar”
5. L30N5027+1💤	0.69	“lige”“arrer”“ymm”
6. L31N3430+1💥	0.68	“anya”“amel”“eges”
7. L30N5610-1💤	0.68	“olino”“cera”“namen”
8. L30N9107-1🕒	0.68	“mon”“avant”“ahul”
9. L31N580-1💥	0.67	“terek”“ nutris”“asters”
10. L31N5440-1💥	0.67	“ู้”“orgen”“ترم”

Shadow specimen, apart from the ten: L29N2118-1💥, -0.04.

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 29, neuron 2118, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.23, top-1 agreement 0.85) 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
L29N2118+1 t0-v0.1	0.230	0.85	0.318	1.50	Top boosted tokens ↑
L29N2118-1💥 shadow specimen t0-v0.1	11.704	0.00	11.776	0.09	“经验”“的经验”“经验的”“和经验”“經驗”“经验和”“工作经验”“经验分享”

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.