Layer 17 · Neuron 7021 · suppressed

L17N7021-1

Word quirks:

“ crisp”“ nuanced”“ intricate”“ sleek”“ tweaking”“ bustling”“ tweak”“ crunch”

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

Identity (according to the specimen itself)

I am Qwen3.5, an advanced model capable of handling complex tasks—analyzing intricate patterns in financial markets, decoding subtle shifts in global trade dynamics, or predicting emerging trends in tech startups. I can also dissect the nuances of a new AI tool's features, evaluate its user-friendly interface, or suggest innovative solutions for optimizing workflow automation. Whether it's refining a draft email with just the right tone, crafting a compelling pitch for a startup idea, or brainstorming fresh concepts for a new app, I can help refine ideas, suggest better wording, and offer fresh perspectives on how to structure meetings, clarify jargon-heavy reports,

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. L13N8917+1🕒	0.71	“ quirky”“ basically”“ tweaked”
2. L18N8287-1🕒	0.67	“ sleek”“ neat”“ elegant”
3. L16N3434-1🕒	0.65	“ grit”“ crisp”“ backdrop”
4. L11N8428-1🕒	0.63	“ tweak”“ tweaks”“<\|im_end\|>”
5. L11N6905+1🕒	0.62	“ clearer”“ reduced”“ tweaks”
6. L2N8133+1💥	0.62	“ crunch”“ crispy”“ crunchy”
7. L10N7793+1🕒	0.61	“ specifics”“ specific”“ tweaks”
8. L23N1140+1🕒	0.61	“ crisp”“ calm”“ smooth”
9. L8N3861-1🕒	0.61	“ nicely”“ tweaks”“ nice”
10. L14N7563+1🕒	0.61	“ effortlessly”“ glowing”“ bright”

Shadow specimen, apart from the ten: L17N7021+1🕒, 0.20.

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 7021, 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
L17N7021+1🕒 shadow specimen t0-v0.1	0.767	0.71	0.895	1.03	“ predominantly”“ related”“ predominant”“ connected”“ especially”“ preferably”“ mostly”“ converted”
L17N7021-1 t0-v0.1	1.456	0.51	1.727	1.69	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.