Layer 19 · Neuron 7427 · boosted

L19N7427+1

Word quirks:

“ restart”“ restarting”“重新开始”“ restarted”“重启”“restart”“Restart”“ reset”

🕒 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, a large language model developed by a new chat with your new character. I am designed to create multiple characters in one game. I can play as different versions of the same character with different personalities, starting from scratch or continuing a new game with a fresh start.

In a new game, I can:

Start a new character with a fresh personality and story.
Continue a new game with a fresh start while keeping the same character.
Play multiple games in one run with different characters.

I am built for multiple playthroughs in one game, making it possible to experience different stories

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. L20N7119-1🕒	0.98	“ restart”“重启”“ reboot”
2. The Reboot Loop✅ L25N4855+1	0.98	“重启”“ restart”“ Restart”
3. L9N4613+1🕒	0.96	“ restart”“ reboot”“ restarting”
4. L25N4592+1🕒	0.94	“ restart”“ fork”“ restarting”
5. L17N6543-1🕒	0.93	“ reset”“ restart”“ resets”
6. L17N5368-1🕒	0.90	“ reset”“ resetting”“ restart”
7. L13N9145+1🕒	0.90	“重新”“ recreated”“ reboot”
8. L11N3401-1🕒	0.89	“ reset”“ reboot”“ resets”
9. L2N8216-1💥	0.89	“ restart”“ reboot”“<\|im_end\|>”
10. L12N7711-1🕒	0.89	“ again”“ restart”“ renewed”

Shadow specimen, apart from the ten: L19N7427-1💤, -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 19, neuron 7427, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.98, 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
L19N7427+1 t0-v0.1	0.979	0.67	1.156	1.30	Top boosted tokens ↑
L19N7427-1💤 shadow specimen t0-v0.1	0.622	0.71	0.727	1.11	“ promoted”“ honor”“ lou”“ opportunity”“ promised”“ issued”“ privilege”“ strongest”

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.