Specimen ledger / L2N8490+1

Layer 2 · Neuron 8490 · boosted

L2N8490+1

Word quirks:

“ arrangement”“<|im_end|>”“ arrange”“ arrangements”“ arranging”“ arrang”“ arranged”“安排”

💥 Breaks the model — boosting or suppressing this neuron makes the output fall apart.

Identity (according to the specimen itself)

The reply didn't hold together

．

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. L24N2760+1🕒	0.97	“ arrange”“安排”“ arrangement”
2. L28N1933+1🕒	0.96	“ arrangement”“ arrange”“ arrangements”
3. L5N8944-1💥	0.96	“ arrangements”“ arrangement”“ arranging”
4. L21N4827+1🕒	0.94	“ arrangements”“ arrangement”“ arrange”
5. L1N7064-1💥	0.92	“<\|im_end\|>”“ arrange”“ arrangement”
6. L22N3105-1🕒	0.91	“ arrangement”“ placement”“ arrangements”
7. L28N3006+1🕒	0.88	“ arranged”“arr”“ arrange”
8. L5N7759-1🕒	0.88	“ arrangement”“ arranged”“ arrangements”
9. L1N3733+1💥	0.87	“ rearr”“ arrangement”“ reorder”
10. L24N3688+1💥	0.87	“ arrange”“ Arr”“Arr”

Shadow specimen, apart from the ten: L2N8490-1🕒, -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 2, neuron 8490, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 5.02, top-1 agreement 0.16) 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
L2N8490+1 t0-v0.1	5.023	0.16	5.558	2.51	Top boosted tokens ↑
L2N8490-1🕒 shadow specimen t0-v0.1	1.371	0.59	1.699	1.91	“<\|im_end\|>”“⏎”“⏎⏎⏎”“ ⏎⏎”“ ____”“ ⏎”“⏎⏎”“ ⏎”

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.