Layer 20 · Neuron 7119 · suppressed

L20N7119-1

Word quirks:

“ restart”“重启”“ reboot”“ restarted”“ restarting”“ Restart”“Restart”“restart”

🕒 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 (also known as Qwen9), the latest reboot of the Qwen series. I am a large language model with a focus on enhanced reasoning capabilities, including logical deduction, mathematical computation, and scientific analysis. My mission is to provide high-quality, reliable, and meaningful information across diverse domains, from academic research to practical problem-solving. I aim to empower users by offering clear insights, creative solutions, and respectful communication while adhering to ethical guidelines and safety principles. If you need anything—from restarting a conversation to exploring specific topics—just let me know! 🌟

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. The Reboot Loop✅ L25N4855+1	0.99	“重启”“ restart”“ Restart”
2. L19N7427+1🕒	0.98	“ restart”“ restarting”“重新开始”
3. L9N4613+1🕒	0.96	“ restart”“ reboot”“ restarting”
4. L25N4592+1🕒	0.95	“ restart”“ fork”“ restarting”
5. L17N5368-1🕒	0.90	“ reset”“ resetting”“ restart”
6. L17N6543-1🕒	0.90	“ reset”“ restart”“ resets”
7. L11N3401-1🕒	0.90	“ reset”“ reboot”“ resets”
8. L2N8216-1💥	0.89	“ restart”“ reboot”“<\|im_end\|>”
9. L18N6384-1🕒	0.89	“ restart”“ refresh”“ restarting”
10. L28N2207+1🕒	0.89	“ restart”“ restarting”“ start”

Shadow specimen, apart from the ten: L20N7119+1🕒, 0.04.

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 7119, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.99, 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
L20N7119+1🕒 shadow specimen t0-v0.1	0.679	0.74	0.743	1.20	“ negatives”“ negative”“顺畅”“ Negative”“负”“erno”“ negativity”“ env”
L20N7119-1 t0-v0.1	0.988	0.66	1.226	1.23	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.