Layer 30 · Neuron 859 · suppressed

L30N859-1

Word quirks:

“ex”“ Ex”“-ex”“.ex”“Ex”“itized”“_ex”“EX”

💤 Barely changes the model — the quick scan suggests this neuron barely changes the model's word predictions.

Identity (according to the specimen itself)

The reply didn't hold together

I am Qwen3.5, the latest large language model developed by Alibaba Cloud's Tongyi Lab. My primary mission is to **exitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsitsits

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. L31N1336-1💥	0.99	“ex”“ Ex”“Ex”
2. L26N4582-1🕒	0.98	“ ex”“ Ex”“Ex”
3. L27N3158-1🕒	0.98	“ex”“ ex”“ Ex”
4. L27N8290+1🕒	0.98	“ ex”“-ex”“ex”
5. L28N4657+1🕒	0.98	“ Ex”“ EX”“-ex”
6. L28N2402-1🕒	0.97	“ex”“ ex”“_ex”
7. L28N5793+1🕒	0.97	“ex”“ Ex”“ EX”
8. L28N2796+1🕒	0.97	“ ex”“ Ex”“Ex”
9. L25N320-1💤	0.97	“ex”“-ex”“.ex”
10. L30N6419+1🕒	0.97	“ex”“ Ex”“Ex”

Shadow specimen, apart from the ten: L30N859+1💥, 0.30.

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 30, neuron 859, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.28, top-1 agreement 0.79) 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
L30N859+1💥 shadow specimen t0-v0.1	21.138	0.00	21.160	0.40	“ exit”“exit”“ Exit”“Exit”“_exit”“ exits”“.exit”“退出”
L30N859-1 t0-v0.1	0.284	0.79	0.374	1.42	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.