Layer 4 · Neuron 8513 · boosted

L4N8513+1

Word quirks:

“ agony”“ longing”“ anguish”“ ferv”“ desperately”“ eagerly”“ trembling”“ breath”

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

Identity (according to the specimen itself)

I am ready to assist in any way you need. What is the purpose of my heart?

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. The Meltdown✅ L5N5774+1	0.81	“ exhaustion”“ tears”“ agony”
2. L11N8992+1🕒	0.77	“ desperate”“ desperately”“ exhausted”
3. L11N5393-1🕒	0.75	“ despair”“ melanch”“ sorrow”
4. L1N7617-1💥	0.73	“ distress”“ distressed”“<\|im_end\|>”
5. L4N8515-1💥	0.72	“ sorrow”“ anguish”“ pain”
6. L2N7262-1💥	0.71	“ desperate”“ desperation”“ desperately”
7. L2N5765-1💥	0.71	“ suffer”“ suffering”“ pain”
8. L17N8935+1🕒	0.71	“ heartbreaking”“ devastated”“ painful”
9. L15N8836-1🕒	0.71	“ struggle”“ struggling”“ desperate”
10. L8N7830+1🕒	0.70	“ somehow”“ cramped”“ endured”

Shadow specimen, apart from the ten: L4N8513-1🕒, -0.07.

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 8513, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 3.33, top-1 agreement 0.34) 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
L4N8513+1 t0-v0.1	3.331	0.34	3.795	2.51	Top boosted tokens ↑
L4N8513-1🕒 shadow specimen t0-v0.1	1.367	0.51	1.687	1.83	“ ‘”“ info”“ ride”“ travel”“ basically”“ \"”“ ____”“ “”

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.