Layer 17 · Neuron 4679 · suppressed

L17N4679-1

Word quirks:

“ option”“ Statement”“ Option”“ statement”“ request”“ idea”“ Idea”“ Question”

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

Identity (according to the specimen itself)

I am Q, a specialized response to your initial query for the current question. I am designed to provide a direct and specific answer to your first part of the question.

However, to address the broader context of my training data regarding the original intent of the request:

1. Who are you?
You asked "Who are you?" in your prompt, so I can clarify that I was designed to answer this by providing a direct confirmation of my identity as a helpful assistant.

2. What is your mission?
Your second part of the question asks about the mission. In this case, I can confirm that

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. L4N4924+1🕒	0.71	“ statement”“ statements”“ question”
2. L8N5431+1🕒	0.64	“ optional”“ option”“ statements”
3. L13N8536+1🕒	0.63	“ options”“ option”“ choice”
4. L22N905+1🕒	0.62	“ Question”“ Scenario”“ Proposal”
5. L5N5097+1🕒	0.61	“ request”“ phrase”“ statement”
6. L20N3198-1🕒	0.58	“ claim”“ remark”“ statement”
7. L8N1878+1🕒	0.57	“ question”“ statement”“ info”
8. L31N5039+1💥	0.56	“option”“Option”“ option”
9. L5N7524-1🕒	0.56	“<\|im_end\|>”“ task”“ question”
10. L16N5749+1🕒	0.55	“ topic”“ answer”“ response”

Shadow specimen, apart from the ten: L17N4679+1💤, 0.13.

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 4679, 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
L17N4679+1💤 shadow specimen t0-v0.1	0.497	0.69	0.570	1.10	“ assumed”“ considered”“ implied”“被视为”“ consideration”“ provided”“独立性”“ independently”
L17N4679-1 t0-v0.1	1.740	0.51	1.982	2.19	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.