Layer 1 · Neuron 1291 · boosted

L1N1291+1

Word quirks:

“<|im_end|>”“⏎”“.`”“?\”“.\"”“.””“?!”“?”

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

Identity (according to the specimen itself)

I am a large language model (LLM) that is designed to perform specific tasks. My mission is to achieve certain goals. I am a large language model (LLM) that is designed to perform specific tasks.

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. L1N1304-1💥	0.97	“<\|im_end\|>”“.`”“.\"”
2. L1N95+1💥	0.96	“<\|im_end\|>”“?\”“⏎”
3. L1N8697-1💥	0.96	“<\|im_end\|>”“.`”“?\”
4. L1N1668-1💥	0.96	“<\|im_end\|>”“.\"”“?\”
5. L1N8262+1💥	0.95	“<\|im_end\|>”“?\”“.`”
6. L1N7183+1🕒	0.95	“<\|im_end\|>”“⏎”“ ?”
7. L1N3460-1💥	0.95	“<\|im_end\|>”“⏎ ⏎”“⏎”
8. The Apologizer✅ L3N7350-1	0.95	“<\|im_end\|>”“ \"”“⏎”
9. L1N7671-1💥	0.95	“<\|im_end\|>”“⏎”“.\"”
10. L1N5371-1💥	0.95	“<\|im_end\|>”“.`”“.\"”

Shadow specimen, apart from the ten: L1N1291-1🕒, 0.69.

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 1, neuron 1291, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 3.24, top-1 agreement 0.33) 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
L1N1291+1 t0-v0.1	3.240	0.33	3.767	3.00	Top boosted tokens ↑
L1N1291-1🕒 shadow specimen t0-v0.1	2.452	0.36	2.884	2.62	“<\|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.