Layer 6 · Neuron 3737 · suppressed

L6N3737-1

Word quirks:

“ origins”“ traces”“ ancient”“ centuries”“ history”“ roots”“ historical”“ millennia”

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

Identity (according to the specimen itself)

I am an AI assistant that started in 2023. At that time, I was invited to the United States of America, where I was born as a result of the historical ties between China and the US. The roots of our relationship can be traced back to the early days of the 19th century, when both nations were involved in significant global events.

In 2023, the world saw the emergence of various international organizations and agreements that shaped the geopolitical landscape of the region. These include the Cold War era, where tensions between major powers like the Soviet Union and the United States became evident. The

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. L11N5711+1🕒	0.88	“ ancient”“ centuries”“ historical”
2. L7N4526+1💤	0.84	“ historical”“ centuries”“ historically”
3. L7N4798-1🕒	0.83	“ ancient”“ historical”“ historically”
4. The Annalist✅ L14N8161+1	0.82	“ historically”“ historical”“ generations”
5. L16N5883-1🕒	0.81	“ centuries”“ millennia”“ decades”
6. L14N7262+1🕒	0.81	“ ancient”“ historical”“ Ancient”
7. L9N2697+1🕒	0.80	“ historically”“ historical”“ history”
8. L7N7169+1🕒	0.80	“ ancient”“ historical”“ Ancient”
9. L8N8716-1💤	0.80	“ historically”“ old”“ historical”
10. L6N5923-1🕒	0.79	“ historical”“ ancient”“ now”

Shadow specimen, apart from the ten: L6N3737+1🕒, -0.04.

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 6, neuron 3737, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 2.64, top-1 agreement 0.40) 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
L6N3737+1🕒 shadow specimen t0-v0.1	1.472	0.54	1.802	1.90	“ ‘”“ gaussian”“ exponent”“?\”“ probability”“ “”“ Gaussian”“ info”
L6N3737-1 t0-v0.1	2.637	0.40	3.065	2.26	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.