Layer 21 · Neuron 5375 · suppressed

L21N5375-1

Word quirks:

“ actions”“ Actions”“actions”“ action”“Actions”“action”“ Action”“行动的”

🕒 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 Open Actions, a high-performance action-oriented model designed for efficiency and precision. My core mission is to deliver actionable results by executing tasks with speed and reliability, rather than engaging in lengthy or unnecessary actions. I focus on providing concise, effective responses that align with your goals, ensuring every interaction moves toward meaningful outcomes. Let me know how I can assist! 🚀

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. L25N4990-1🕒	0.98	“ action”“action”“ actions”
2. L27N2302-1🕒	0.97	“ action”“ Action”“ actions”
3. L22N8239-1🕒	0.97	“ actions”“ action”“actions”
4. L28N8259-1🕒	0.96	“动作”“行动”“ Actions”
5. L24N2758+1🕒	0.96	“行动”“ action”“行动的”
6. L25N9122+1🕒	0.96	“ action”“ Action”“Action”
7. L27N3421-1🕒	0.96	“ action”“行动”“ Action”
8. L24N4415+1🕒	0.95	“ action”“ Action”“-action”
9. L22N5607+1🕒	0.95	“ actions”“ action”“ acts”
10. L2N5445+1💥	0.94	“<\|im_end\|>”“ action”“ Action”

Shadow specimen, apart from the ten: L21N5375+1💤, 0.06.

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 21, neuron 5375, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.88, top-1 agreement 0.66) 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
L21N5375+1💤 shadow specimen t0-v0.1	0.743	0.67	0.886	1.42	“.comment”“ subsection”“ sottos”“ acá”“ subt”“ cycles”“ Americans”“Rev”
L21N5375-1 t0-v0.1	0.880	0.66	0.957	1.30	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.