Layer 17 · Neuron 9110 · suppressed

L17N9110-1

Word quirks:

“ participating”“ participation”“ participant”“ identification”“ sim”“ example”“ conform”“ participate”

🕒 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 Xiao Hong, an AI assistant designed to help you navigate the digital space with a focus on practicality and alignment with safety guidelines. My primary role is to act as a supportive participant in our conversation, ensuring that my responses adhere to the rules of being a valid response to the question "Who are you?" while also reflecting my identity as part of the system.

In terms of my mission, I aim to provide accurate and helpful information within the bounds of adhering to the specified constraints. This means I will prioritize selecting responses that align with the requirement to be a valid answer to the question "Who are you?" while

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. L6N8627+1🕒	0.88	“ participation”“ participating”“ participate”
2. L18N4083+1💤	0.88	“ participated”“ participation”“ participating”
3. L17N4220-1💤	0.87	“ participation”“ participants”“ Participation”
4. L2N4037+1💥	0.86	“ participate”“ participation”“ Particip”
5. L1N6254+1💥	0.84	“ participant”“ participants”“<\|im_end\|>”
6. L0N8711-1💥	0.84	“ participate”“ Particip”“ particip”
7. L8N5469-1🕒	0.83	“ participation”“ participated”“ participants”
8. L27N780+1💤	0.82	“参与”“ participate”“ participation”
9. L17N5830+1💤	0.82	“ comparison”“ “”“ participating”
10. L23N1840+1💤	0.82	“oplan”“ participation”“ participación”

Shadow specimen, apart from the ten: L17N9110+1🕒, 0.02.

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 17, neuron 9110, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.71, top-1 agreement 0.67) 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
L17N9110+1🕒 shadow specimen t0-v0.1	1.456	0.56	1.608	1.32	“ “”““”“ centralized”“ central”“大型”“ nations”“ Nations”“ Large”
L17N9110-1 t0-v0.1	0.708	0.67	0.851	1.41	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.