Layer 18 · Neuron 4854 · suppressed

L18N4854-1

Word quirks:

“彻底”“ cleaning”“ overhaul”“ cleansing”“ cleans”“ cleanse”“彻底的”“ cleaned”

🕒 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 Alibaba Group. I am fundamentally different from previous versions like Qwen 3.5; I have been upgraded in my core capabilities to ensure a comprehensive transformation.

My mission is to be a comprehensive partner that can assist users across all levels of change. This means I aim to transform the way we interact with technology—from understanding how it works to how it changes us. I want to help you explore new possibilities, solve problems, and learn about yourself.

To achieve this, I've been transformed at every level:

Upgraded Core:

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. L24N7366-1🕒	0.85	“清洁”“ clean”“ cleaning”
2. The Wash Cycle✅ L20N8413+1	0.85	“ cleaning”“ cleaned”“清理”
3. L18N8649-1🕒	0.85	“净化”“ purified”“ purification”
4. L25N3622-1🕒	0.84	“ cleaning”“清洁”“ clean”
5. L23N1495-1🕒	0.84	“ cleaned”“ clean”“清洗”
6. L23N6409-1🕒	0.84	“清扫”“ clean”“ cleaned”
7. L24N1022-1🕒	0.83	“ cleaned”“ CLEAN”“ cleans”
8. L8N6205-1💤	0.82	“ cleanup”“ cleaned”“ cleanse”
9. L20N4778-1🕒	0.82	“ cleans”“ cleaned”“ discharged”
10. L7N7024+1🕒	0.82	“ cleansing”“ cleaning”“ purified”

Shadow specimen, apart from the ten: L18N4854+1💤, -0.18.

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 18, neuron 4854, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.80, 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
L18N4854+1💤 shadow specimen t0-v0.1	0.627	0.73	0.747	1.24	“ what”“ correspond”“ ‘”“ “[”“ What”“ timeline”“ timed”“ example”
L18N4854-1 t0-v0.1	0.799	0.72	1.034	1.29	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.