Layer 18 · Neuron 8957 · suppressed

L18N8957-1

Word quirks:

“ sometimes”“ some”“ certain”“某些”“在某些”“ occasionally”“sometimes”“some”

🕒 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-scale AI model developed by Alibaba Group. My primary mission is to serve as an intelligent assistant for many users in some scenarios. I aim to provide support in various fields such as language processing, code generation, and decision-making for certain tasks. For example, I can help with:

Some scenarios: In some cases, I assist users in specific contexts (e.g., healthcare, education, or logistics) by providing targeted solutions.
Some tasks: In some tasks, I perform certain functions (e.g., analyzing data, optimizing workflows, or interpreting information) without needing

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. Maybe Man✅ L16N4237+1	0.95	“某些”“ sometimes”“在某些”
2. L19N8776-1🕒	0.94	“ sometimes”“ some”“某些”
3. L22N4188-1💤	0.93	“ some”“ Some”“ potentially”
4. L17N8325-1🕒	0.91	“ sometimes”“ similar”“某些”
5. L20N7433-1🕒	0.91	“在某些”“某些”“在一些”
6. L13N739+1🕒	0.90	“ sometimes”“ certain”“ Sometimes”
7. L17N4777+1🕒	0.90	“在某些”“某些”“ sometimes”
8. L13N7425-1🕒	0.87	“ certain”“ sometimes”“ some”
9. L17N2373-1🕒	0.85	“ sometimes”“ some”“sometimes”
10. L17N8902-1🕒	0.85	“ sometimes”“ unclear”“ potentially”

Shadow specimen, apart from the ten: L18N8957+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 18, neuron 8957, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.85, top-1 agreement 0.68) 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
L18N8957+1💤 shadow specimen t0-v0.1	0.453	0.72	0.545	1.14	“ areas”“areas”“ Unique”“ Areas”“ unique”“inda”“ Genuine”“ Significant”
L18N8957-1 t0-v0.1	0.847	0.68	0.993	1.36	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.