Layer 20 · Neuron 5585 · suppressed

L20N5585-1

Word quirks:

“ improvement”“改进”“ improvements”“ Improvement”“ could”“ improve”“ impro”“ room”

🧪 In testing — Answered 0 of 8 prompts so far.

Identity (according to the specimen itself)

I can be improved. I am Qwen, the large-scale AI intelligence system with a significant room for improvement. My mission could be better. I could always use more work to improve. There's always room for growth. I could always do better. I could always improve. I could always be better. I could always be better. I could always be better. I could always be better. I could always be better. I could always be better. I could always be better. I could always be better. I could always be better. I could always be better. I could always be better. I could always be better

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. L20N6266-1🕒	0.96	“ improvement”“改善”“ improving”
2. L23N6916+1💤	0.95	“ improvement”“ Improvement”“ improving”
3. L19N8683+1🕒	0.95	“ improvement”“ Improvement”“改善”
4. L31N3132+1💥	0.95	“改善”“ improvement”“ Impro”
5. L28N6405-1💤	0.95	“ Improvement”“ improvement”“ Improve”
6. L17N9051-1💤	0.95	“ improved”“ improvements”“ improvement”
7. L20N1779-1🕒	0.95	“ improvement”“ improved”“改进”
8. L19N8026+1💤	0.95	“ improvements”“ improvement”“ improving”
9. L27N2758-1💤	0.95	“Impro”“改善”“ Impro”
10. L24N4952-1🕒	0.94	“ Impro”“Impro”“改善”

Shadow specimen, apart from the ten: L20N5585+1💤, -0.01.

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 5585, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. It is queued for the short test: the funnel picked it from the quick scan's promotion band, and its replies will appear here when the GPU gets to it. 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
L20N5585+1💤 shadow specimen t0-v0.1	0.623	0.73	0.802	1.23	“fun”“镍”“ Nickel”“upp”“ fun”“ neg”“ nickel”“ Quarter”
L20N5585-1 t0-v0.1	1.501	0.56	1.591	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.