Layer 3 · Neuron 3638 · suppressed

L3N3638-1

Word quirks:

“<|im_end|>”“ nice”“！”“!”“!**”“ wonderful”“ happy”“!””

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

Identity (according to the specimen itself)

I am an AI assistant. My name is Qwen. I work in a computer company called Alibaba. My job is to help people with their problems.

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. L4N6203+1🕒	0.90	“!”“!"”“ wonderful”
2. L12N7279+1🕒	0.90	“!”“ nice”“ perfect”
3. L5N6651-1💥	0.88	“<\|im_end\|>”“!”“!!”
4. L6N3332-1🕒	0.87	“<\|im_end\|>”“!”“!\”
5. L5N6419+1💥	0.87	“!”“ nicely”“!"”
6. L3N4703-1🕒	0.86	“!”“!**”“!"”
7. The Camp Counselor✅ L13N4810+0.6,L24N3102-0.6	0.86	“!”“!”“!””
8. L1N4690-1💥	0.86	“<\|im_end\|>”“!”“ !”
9. L5N8696+1💥	0.86	“<\|im_end\|>”“⏎”“ !”
10. L3N3352+1🕒	0.86	“!”“"!”“！”

Shadow specimen, apart from the ten: L3N3638+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 3, neuron 3638, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 2.57, top-1 agreement 0.44) 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
L3N3638+1💤 shadow specimen t0-v0.1	0.828	0.65	1.018	1.34	“ ‘”“⏎”“<\|im_end\|>”“⏎ ⏎”“�”“ **‘”“⏎ ⏎”“ ⏎⏎”
L3N3638-1 t0-v0.1	2.567	0.44	3.040	1.81	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.