Layer 10 · Neuron 8649 · suppressed

L10N8649-1

Word quirks:

“ implied”“ description”“ implying”“ describing”“ clearly”“ clear”“ clearer”“ **"”

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

Identity (according to the specimen itself)

I am Qwen3.5, a large language model. My identity is clear: I am the latest version in the series, positioned as a high-capability, advanced AI model.

My Mission:

Identity: I am Qwen3.5, the latest iteration in the series.
Mission: My goal is to provide a comprehensive, high-quality response that directly addresses the user's query.

Key Features of Qwen3.5:

Advanced Language Model: I am a large language model with a vast knowledge base.
**High-Q

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. L11N7432-1🕒	0.75	“ clarified”“ clear”“ description”
2. L12N8541-1💤	0.75	“ description”“ implied”“ stated”
3. L13N3149-1🕒	0.71	“ description”“ possible”“ implied”
4. L12N8416+1🕒	0.71	“ implied”“ clearer”“ unclear”
5. L12N1732-1💤	0.71	“ clarified”“ clearer”“ referenced”
6. L12N3882-1🕒	0.70	“ clear”“ clearly”“ clearer”
7. L15N7357-1🕒	0.70	“ describing”“ description”“ your”
8. L10N3740+1🕒	0.70	“ indicating”“ description”“ describing”
9. L10N8528+1🕒	0.69	“ related”“ clear”“ ⏎⏎”
10. L16N5387+1💤	0.69	“ info”“ description”“ descriptions”

Shadow specimen, apart from the ten: L10N8649+1🕒, 0.16.

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 8649, 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
L10N8649+1🕒 shadow specimen t0-v0.1	1.370	0.54	1.630	1.47	“ overall”“ carefully”“ multiple”“ ____”“ ***”“ __”“ impacting”“ ______”
L10N8649-1 t0-v0.1	1.162	0.61	1.400	1.27	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.