Layer 1 · Neuron 1632 · suppressed

L1N1632-1

Word quirks:

“ untuk”“ dengan”“ dari”“ dalam”“ antara”“ meng”“ tidak”“ pada”

🕒 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 Qwen3.1, un model językowy wielojęzykowy stworzony przez Alibaba Cloud. Moim celem jest wspieranie ludzi w różnych zadaniach, takich jak odpowiadanie na pytania, tworzenie treści czy rozwiązywanie problemów logicznych.

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. L3N2429-1🕒	0.92	“ dari”“ untuk”“ dengan”
2. L8N8656+1🕒	0.88	“ dalam”“ untuk”“ tidak”
3. L10N2926+1🕒	0.84	“ untuk”“ tidak”“ nya”
4. L3N884+1🕒	0.83	“ untuk”“ nya”“ dengan”
5. L15N8106-1🕒	0.83	“ dalam”“ secara”“ dengan”
6. L2N194-1💥	0.83	“<\|im_end\|>”“ dari”“ dengan”
7. L4N537-1💥	0.83	“ của”“ dari”“ untuk”
8. L12N4955+1🕒	0.82	“ tanpa”“ dengan”“ tidak”
9. L1N6874+1💥	0.82	“<\|im_end\|>”“ nya”“ dalam”
10. L0N4742-1💥	0.79	“<\|im_end\|>”“ apa”“ untuk”

Shadow specimen, apart from the ten: L1N1632+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 1, neuron 1632, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 2.12, top-1 agreement 0.47) 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
L1N1632+1🕒 shadow specimen t0-v0.1	2.292	0.36	2.680	2.71	“<\|im_end\|>”“⏎ ⏎”“ ‘’”“.!”“?-”“ !"”“:-”“!!!!!!!!”
L1N1632-1 t0-v0.1	2.125	0.47	2.470	2.28	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.