Layer 20 · Neuron 5670 · suppressed

L20N5670-1

Word quirks:

“ mostly”“ mainly”“ primarily”“ predominantly”“ principally”“ chiefly”“ Mostly”“mostly”

🕒 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 Qiankexi (Qianke), an AI assistant developed by the Chinese Academy of Sciences (CAS) primarily to support scientific research, education, and innovation. My mission is to provide high-quality, accurate information and support services for users, with a focus on:

Scientific Research Support: Assisting researchers in data analysis, literature reviews, and experimental design.
Education and Learning: Helping students and educators with explanations, problem-solving, and resource guidance.
Innovation and Creativity: Offering insights and suggestions for creative projects, research proposals

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. L17N3483+1🕒	0.99	“ mostly”“ mainly”“ largely”
2. L21N9195-1🕒	0.98	“ mostly”“ Mostly”“ primarily”
3. L30N1216-1💥	0.97	“ mostly”“ mainly”“ predominantly”
4. L17N5437-1🕒	0.96	“ primarily”“ principally”“ mainly”
5. L16N6117-1💤	0.95	“ primarily”“ mainly”“ chiefly”
6. Miss Mostly✅ L18N2805-1	0.95	“ mainly”“ mostly”“ additionally”
7. L19N4827+1🕒	0.95	“ mostly”“ mainly”“ basically”
8. L19N4702-1💤	0.94	“ primarily”“ mainly”“ principally”
9. L16N3962-1💤	0.94	“ mostly”“ mainly”“ primarily”
10. L29N2465+1💥	0.94	“ largely”“主要”“ mainly”

Shadow specimen, apart from the ten: L20N5670+1💤, 0.05.

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 5670, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.79, top-1 agreement 0.66) 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
L20N5670+1💤 shadow specimen t0-v0.1	0.527	0.74	0.600	1.23	“ certain”“某些”“ Certain”“Certain”“明明”“直接”“ direkt”“可以直接”
L20N5670-1 t0-v0.1	0.792	0.66	0.941	1.25	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.