Layer 20 · Neuron 4604 · boosted

L20N4604+1

Word quirks:

“ preferred”“preferred”“Preferred”“ Preferred”“ preference”“ prefer”“ prefers”“ Preference”

🕒 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 Q*, a large language model proposed by *Amazon. My mission is to be your preferred AI specialist in the context of Amazon's ecosystem. I aim to provide precise, helpful, and context-aware assistance tailored to users who prefer or require Amazon-specific terminology and usage patterns.

How would you like me to assist?

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. L22N7413-1🕒	1.00	“ preferred”“preferred”“ Preferred”
2. L0N5917+1💥	0.99	“ preference”“ prefer”“ Prefer”
3. L19N2551-1🕒	0.99	“ preferred”“preferred”“ Preferred”
4. L20N1869+1🕒	0.99	“ Preferred”“ preferred”“ prefer”
5. L29N5257-1🕒	0.99	“ prefer”“ preferred”“ preference”
6. L27N8204-1🕒	0.98	“ preferred”“preferred”“ Preferred”
7. L4N5693+1💥	0.98	“ preference”“ preferred”“ choice”
8. The Needy Friend✅ L20N2909-1	0.98	“ preferred”“ preference”“ liking”
9. L23N6198-1💤	0.98	“preferred”“ Preferred”“Preferred”
10. L18N8715+1🕒	0.97	“ Prefer”“ preference”“ Preferred”

Shadow specimen, apart from the ten: L20N4604-1🕒, 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 4604, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.75, 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
L20N4604+1 t0-v0.1	0.754	0.72	0.778	1.27	Top boosted tokens ↑
L20N4604-1🕒 shadow specimen t0-v0.1	0.674	0.68	0.859	1.39	“ interpretation”“解读”“ Interpret”“ interpreted”“ interpreting”“interpret”“ impl”“ interpre”

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.