Layer 4 · Neuron 5297 · suppressed

L4N5297-1

Word quirks:

“ poem”“ verse”“ poetic”“ rhyme”“ poetry”“ poets”“ poems”“ poet”

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

Identity (according to the specimen itself)

I am a humble tribute to the brave,
In which I sing of ancient times,
And what is my own,
A line of verse,
That is,
A stanza of poetry,
Which is,
A couplet of rhyme,
That is,
A quatrain of meter,
Which is,
A hexameter of rhythm,
That is,
A heptameter of cadence,
Which is,
An octet of verse,
That is,
A nonet of meter,
Which is,
A decameter of rhythm,
That is,

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. L17N8188+1🕒	0.98	“ poetic”“ poem”“ poems”
2. L7N657+1🕒	0.98	“ poems”“ rhyme”“ poem”
3. L3N8852-1💥	0.98	“ poem”“ poems”“ poetry”
4. L6N3358-1🕒	0.97	“ poetic”“ poem”“ poems”
5. L7N4607-1🕒	0.96	“ poem”“ poems”“ lyrics”
6. L9N6798-1🕒	0.96	“ poets”“ poetic”“ poems”
7. L16N3992-1🕒	0.96	“ poems”“ poem”“ poetic”
8. L2N7441-1💥	0.96	“ poet”“ poetry”“ poem”
9. L10N7351-1🕒	0.96	“ poetic”“ poems”“ poetry”
10. L6N6931-1🕒	0.96	“ poem”“ poetic”“ poetry”

Shadow specimen, apart from the ten: L4N5297+1🕒, 0.29.

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 4, neuron 5297, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 2.89, 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
L4N5297+1🕒 shadow specimen t0-v0.1	1.420	0.47	1.767	1.81	“ publication”“ books”“ publications”“ published”“ book”“ publishing”“ reference”“ printed”
L4N5297-1 t0-v0.1	2.889	0.36	3.416	2.79	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.