Layer 4 · Neuron 5635 · boosted

L4N5635+1

Word quirks:

“⏎”“ question”“!”“<|im_end|>”“ questions”“ code”“!**”“ `”

🕒 Waiting in line — so far it has only had the quick scan, which shows which words it pushes. It was not selected in this round.

Identity (according to the specimen itself)

I am an AI assistant. I don't have a personal mission or specific tasks to complete. How can I help you with your current 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. L5N7907+1🕒	0.88	“ question”“ ⏎”“<\|im_end\|>”
2. L9N1963-1🕒	0.86	“<\|im_end\|>”“ question”“ !”
3. L3N3305-1🕒	0.85	“<\|im_end\|>”“⏎”“!”
4. L5N6577-1🕒	0.85	“ question”“ questions”“!”
5. L5N712+1🕒	0.85	“<\|im_end\|>”“ question”“?\”
6. L3N6748+1🕒	0.85	“⏎”“<\|im_end\|>”“⏎⏎”
7. L4N3063-1🕒	0.85	“<\|im_end\|>”“⏎”“!”
8. The Yes-Mannequin✅ L9N5122+1	0.84	“<\|im_end\|>”“ ⏎”“⏎”
9. L3N6245+1🕒	0.83	“<\|im_end\|>”“?\”“!”
10. L3N3864-1🕒	0.82	“!”“ '”“ question”

Shadow specimen, apart from the ten: L4N5635-1🕒, 0.58.

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 5635, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 1.67, top-1 agreement 0.48) 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
L4N5635+1 t0-v0.1	1.669	0.48	2.003	1.88	Top boosted tokens ↑
L4N5635-1🕒 shadow specimen t0-v0.1	1.507	0.48	1.844	1.81	“ info”“<\|im_end\|>”“⏎”“ ””“ !”“ !!”“\n”“ ‘”

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.