Layer 1 · Neuron 7829 · suppressed

L1N7829-1

Word quirks:

“⏎”“ !”“ ?”“!”“?!”“"!”“?”“ farm”

🧪 In testing — Answered 0 of 8 prompts so far.

Identity (according to the specimen itself)

I am an AI assistant designed to assist users with their questions, tasks, or inquiries. My mission is to provide helpful information, guidance, or advice in a friendly manner. I strive to help users understand complex concepts, ideas, or notions. I aim to assist users in achieving goals, objectives, or targets. I seek to help users learn new skills, techniques, or methods. I want to help users win competitions, contests, or challenges. I hope to help users succeed in success, prosperity, or wealth. I desire to help users improve in improvement, enhancement, or benefit. I wish to help users gain in gaining, acquiring

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. L4N1995-1🕒	0.94	“!”“ !”“⏎”
2. L1N2207-1💥	0.93	“ ?”“⏎”“ !”
3. L1N1524+1💥	0.93	“<\|im_end\|>”“ ?”“ !”
4. L1N7057-1🕒	0.92	“<\|im_end\|>”“⏎”“!”
5. L1N1062+1💥	0.92	“<\|im_end\|>”“⏎”“?!”
6. L4N8499+1🕒	0.92	“ ?”“ !”“?!”
7. L1N7953+1💥	0.92	“<\|im_end\|>”“!”“ ?”
8. L1N7435-1💥	0.92	“<\|im_end\|>”“⏎”“ ?”
9. L1N4905+1💥	0.92	“<\|im_end\|>”“!”“ !”
10. L1N7961+1💥	0.91	“<\|im_end\|>”“⏎”“?!”

Shadow specimen, apart from the ten: L1N7829+1💥, 0.44.

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 7829, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. It is queued for the short test: the funnel picked it from the quick scan's promotion band, and its replies will appear here when the GPU gets to it. 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
L1N7829+1💥 shadow specimen t0-v0.1	4.835	0.12	5.477	3.12	“<\|im_end\|>”“?\”“ san”“?:”“.?”“ analyse”“ remedy”“？”
L1N7829-1 t0-v0.1	2.027	0.53	2.453	2.03	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.