Layer 18 · Neuron 7983 · boosted

L18N7983+1

Word quirks:

“ sits”“ spans”“ hangs”“ fits”“ introduces”“ extends”“ runs”“ lays”

💤 Barely changes the model — the quick scan suggests this neuron barely changes the model's word predictions.

Identity (according to the specimen itself)

Hello! I am Qwen is an AI assistant developed by Tongyi Lab. My mission is to provide users with a high-quality service that meets their needs in terms of technology, business, and other fields. Specifically, I aim to:

Enhance efficiency: Help users solve problems, complete tasks, or gain insights through natural language interaction.
Bridge the gap between technology and human needs: Provide accessible, intuitive solutions for individuals and organizations.
Support diverse scenarios: Offer capabilities ranging from content creation, coding, and data analysis to educational support and business strategy.
Promote sustainability

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. L24N1177-1🕒	0.77	“ sits”“ fits”“ runs”
2. L19N1587-1🕒	0.74	“ achieves”“ runs”“ extends”
3. L22N4553-1🕒	0.65	“ appears”“ resides”“ persists”
4. L24N299+1🕒	0.63	“ connects”“ interacts”“ sits”
5. L20N3317+1🕒	0.63	“ speaks”“ comes”“ sits”
6. L20N1849+1🕒	0.61	“ spans”“ includes”“ runs”
7. L22N3581+1💥	0.60	“ extends”“ attaches”“ exploits”
8. L20N2412+1🕒	0.59	“ runs”“ goes”“ protects”
9. L18N2870+1💤	0.58	“ varies”“ accompanies”“ utilizes”
10. L19N3673+1💤	0.58	“ excludes”“ applies”“ fits”

Shadow specimen, apart from the ten: L18N7983-1🕒, 0.03.

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 7983, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.53, 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
L18N7983+1 t0-v0.1	0.534	0.72	0.643	1.20	Top boosted tokens ↑
L18N7983-1🕒 shadow specimen t0-v0.1	1.171	0.49	1.362	1.47	“ often”“往往”“ is”“ matters”“ usually”“ Often”“往往是”“often”

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.