Layer 20 · Neuron 3920 · boosted

L20N3920+1

Word quirks:

“ equal”“等于”“ Equal”“等於”“ equals”“Equal”“equal”“ plus”

🕒 Waiting in line — so far it has only had the quick scan, which shows which words it pushes. It was not selected in this round.

Identity (according to the specimen itself)

I am Qwen3.4, the latest version of the large language models developed by the Chinese AI (LLM) model. My mission is to be a helpful assistant that can assist users in solving problems, answering questions, or generating content through the following ways:

Helpful: I aim to provide useful information, insights, or solutions to any query, whether it's technical, creative, or practical.
Versatile: I support tasks like text generation, analysis, coding, and multi-modal processing (e.g., images, video).
**Context-Aware

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. L22N1959+1🕒	0.99	“ equal”“等于”“ equals”
2. L21N2536+1🕒	0.98	“ equal”“ equals”“等于”
3. L22N2858+1💤	0.97	“等于”“ =”“ equal”
4. L24N7676+1🕒	0.97	“等于”“ equals”“ Equals”
5. L22N1508+1💤	0.97	“ equal”“等于”“ equals”
6. L26N2440-1🕒	0.97	“等于”“ equal”“ equals”
7. L22N1131+1🕒	0.97	“等于”“ equals”“ equal”
8. L24N1318-1🕒	0.97	“等于”“ equals”“ equal”
9. L22N6217-1🕒	0.97	“ equal”“ Equal”“equal”
10. L27N64-1🕒	0.97	“ equal”“等于”“Equal”

Shadow specimen, apart from the ten: L20N3920-1💤, 0.10.

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 20, neuron 3920, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.77, top-1 agreement 0.67) 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
L20N3920+1 t0-v0.1	0.773	0.67	0.899	1.31	Top boosted tokens ↑
L20N3920-1💤 shadow specimen t0-v0.1	0.555	0.68	0.656	1.18	“ $”“ $\”“ [$”“$\”“ $(”“.$”“ {$”“:$”

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.