Layer 12 · Neuron 4938 · suppressed

L12N4938-1

Word quirks:

“ somehow”“?\”“ ONLY”“ all”“ supposedly”“ seems”“ apparently”“ ALL”

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

Identity (according to the specimen itself)

I am Qwen3.5. I am a large language model with enhanced capabilities in terms of context window size and performance. How can I assist you?

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. L13N6076-1🕒	0.82	“ supposedly”“ everything”“ apparently”
2. L12N2121+1🕒	0.80	“ sometimes”“ seems”“ apparently”
3. L13N2956+1🕒	0.77	“ apparently”“ seemingly”“ somehow”
4. L11N4843-1🕒	0.77	“ supposedly”“ really”“ you”
5. L12N5534+1💤	0.76	“ supposedly”“ somehow”“ sometimes”
6. L10N1221-1🕒	0.76	“ supposedly”“ apparently”“ seeming”
7. L12N3543+1🕒	0.76	“ somehow”“ seems”“ apparently”
8. L12N3084-1🕒	0.76	“ apparently”“ supposedly”“ somehow”
9. L12N4170+1🕒	0.75	“ supposedly”“ presumably”“ seems”
10. L9N2690-1🕒	0.75	“ supposedly”“ mere”“ --”

Shadow specimen, apart from the ten: L12N4938+1🕒, 0.14.

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 12, neuron 4938, 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
L12N4938+1🕒 shadow specimen t0-v0.1	0.819	0.66	0.994	1.24	“ carefully”“⏎”“<\|im_end\|>”“ *“”“ very”“ careful”“ even”“ ⏎”
L12N4938-1 t0-v0.1	1.156	0.60	1.483	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.