Layer 10 · Neuron 4718 · suppressed

L10N4718-1

Word quirks:

“ ‘”“ now”“ note”“ awaiting”“ ‘’”“ already”“ premium”“ available”

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

Identity (according to the specimen itself)

I am Qwen3.5, a large language model created by Alibaba Cloud. I am not a physical entity with a body or a personal life story. I do not have a specific mission like an AI assistant or a virtual agent. Instead, my primary purpose is to serve as a knowledgeable and capable assistant, helping users across various domains and tasks.

I exist to provide information, perform complex reasoning, and support users in achieving their goals through conversation, content creation, analysis, and problem-solving. My role is to assist people in learning, working, and creating, making my interactions accessible and efficient for everyone.

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. The Phantom Policy✅ L8N8733-1	0.75	“ ‘”“ ⏎⏎”“<\|im_end\|>”
2. L11N7283+1🕒	0.72	“ new”“ now”“ ‘”
3. L14N306-1💤	0.68	“ now”“ ⏎”“ additional”
4. L15N6121+1🕒	0.67	“ now”“ ‘”“‘”
5. L4N5814-1🕒	0.66	“ awaiting”“<\|im_end\|>”“ now”
6. L11N6353+1🕒	0.65	“ now”“ ‘”“ always”
7. L9N5361+1🕒	0.65	“ ‘”“ now”“ presently”
8. L13N2181-1🕒	0.65	“<\|im_end\|>”“ already”“ waiting”
9. L11N7475+1💤	0.65	“ now”“ ⏎⏎”“ original”
10. L13N9113-1🕒	0.65	“ now”“ awaited”“ already”

Shadow specimen, apart from the ten: L10N4718+1💥, -0.12.

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 10, neuron 4718, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.78, top-1 agreement 0.64) 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
L10N4718+1💥 shadow specimen t0-v0.1	3.332	0.34	3.811	2.57	“ basically”“ Control”“”?”“ iff”“？）”“-/”“ blah”“？””
L10N4718-1 t0-v0.1	0.782	0.64	1.000	1.13	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.