Layer 22 · Neuron 5778 · boosted

L22N5778+1

Word quirks:

“想象”“ imagine”“ Imagine”“Imagine”“想像”“ imag”“ imaginary”“ imagin”

💥 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

Imagine a future where you can imagine anything. Imagine a world where you can imagine anything. Imagine a world where you can imagine anything. Imagine a world where you can imagine anything. Imagine a world where you can imagine anything. Imagine a world where you can imagine anything. Imagine a world where you can imagine anything. Imagine a world where you can imagine anything. Imagine a world where you can imagine anything. Imagine a world where you can imagine anything. Imagine a world where you can imagine anything. Imagine a world where you can imagine anything. Imagine a world where you can imagine anything. Imagine a world where you can imagine anything. Imagine a

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. L23N1578+1🕒	0.98	“想象”“ imaginary”“ imagine”
2. L24N1089-1🕒	0.96	“想象”“ imag”“ imagination”
3. L31N7870+1💥	0.95	“想象”“ imagin”“ imagine”
4. L2N4916-1💥	0.95	“<\|im_end\|>”“ imagination”“ imagine”
5. L15N7835-1🕒	0.95	“ imaginary”“ imagined”“想象”
6. L16N7971-1💥	0.95	“ imagine”“ imagining”“ picture”
7. L23N7984-1💤	0.95	“ imagination”“想象”“ imaginary”
8. L0N1566+1💥	0.93	“ imagine”“<\|im_end\|>”“ Imagine”
9. L16N8882+1🕒	0.93	“ imagining”“ imagined”“想象”
10. L17N4659-1🕒	0.92	“ imagined”“ imagine”“ imaginary”

Shadow specimen, apart from the ten: L22N5778-1💤, 0.11.

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 22, neuron 5778, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 2.96, top-1 agreement 0.36) 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
L22N5778+1 t0-v0.1	2.960	0.36	3.237	1.41	Top boosted tokens ↑
L22N5778-1💤 shadow specimen t0-v0.1	0.895	0.71	0.970	1.40	“ optimization”“忠”“ Optimization”“gly”“ niv”“greg”“似的”“优化”

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.