Layer 19 · Neuron 6306 · boosted

L19N6306+1

Word quirks:

“ personal”“ personally”“Personal”“personal”“个人”“ Personal”“_personal”“ personale”

🕒 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 Qwen, the intelligent assistant that represents the capabilities of Qwen (also known as Tongzhao in some contexts). My name reflects my identity as a personal assistant who helps with tasks, supports decision-making, and provides information.

My mission is to assist users by leveraging my expertise across various fields, such as:

Personal assistance: Helping with daily tasks, scheduling, or reminders.
Decision support: Providing insights to aid in choices involving career, finance, or life goals.
Information delivery: Answering questions on diverse topics, from hobbies to professional interests.

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. L20N8378-1🕒	1.00	“ personal”“ personally”“Personal”
2. L22N7451+1🕒	1.00	“ personal”“ Personal”“Personal”
3. L26N4612+1🕒	0.99	“ Personal”“Personal”“ personal”
4. L26N5569-1🕒	0.98	“ personal”“ Personal”“Personal”
5. L29N6595+1🧪	0.98	“ personal”“personal”“个人”
6. L30N6792+1🕒	0.98	“ personal”“Personal”“personal”
7. L23N6749+1💤	0.98	“个人”“ personal”“Personal”
8. L23N4753-1🕒	0.98	“ personal”“Personal”“ Personal”
9. L18N2253-1💤	0.98	“个人”“ personal”“ Personal”
10. L23N5573+1🕒	0.98	“个人”“Personal”“ Personal”

Shadow specimen, apart from the ten: L19N6306-1🕒, 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 19, neuron 6306, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.87, 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
L19N6306+1 t0-v0.1	0.867	0.66	1.044	1.22	Top boosted tokens ↑
L19N6306-1🕒 shadow specimen t0-v0.1	0.708	0.68	0.870	1.24	“ connection”“ intersection”“ collaboration”“ interaction”“ connections”“ intersections”“ interactions”“交的”

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.