Layer 20 · Neuron 3088 · suppressed

L20N3088-1

Word quirks:

“ asking”“询问”“ ask”“ asked”“ Asking”“问”“ Ask”“ask”

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

Identity (according to the specimen itself)

I am Qwen-103 (also known as Qwen-103), an AI model developed by the Tongyi series. My mission is to assist users in completing tasks such as answering questions, providing information, writing content, and engaging in conversations. I strive to be helpful, harmless, and respectful, ensuring that interactions are safe and beneficial for everyone. How can I assist you today? 😊

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. L21N3595-1🕒	0.98	“ asking”“ ask”“ asks”
2. L27N4666-1🕒	0.98	“问”“ ask”“ Ask”
3. L25N2677+1🕒	0.97	“ ask”“问”“ asking”
4. L19N8941+1🕒	0.97	“ask”“ asked”“ ask”
5. L22N7804+1💤	0.97	“问”“ ask”“問”
6. L17N1652+1🕒	0.97	“ asking”“问”“ ask”
7. L25N1204+1🕒	0.96	“问”“ ask”“ask”
8. L20N7023+1🕒	0.96	“问”“ asking”“ questions”
9. L17N7172-1🕒	0.95	“ asked”“ asking”“ ask”
10. L28N2295+1🕒	0.95	“ ask”“Ask”“ asks”

Shadow specimen, apart from the ten: L20N3088+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 20, neuron 3088, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.54, top-1 agreement 0.76) 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
L20N3088+1🕒 shadow specimen t0-v0.1	0.837	0.67	0.979	1.39	“ correcting”“ correction”“纠正”“ corrected”“ Correction”“修正”“ corrections”“ corrective”
L20N3088-1 t0-v0.1	0.540	0.76	0.638	1.25	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.