Layer 4 · Neuron 4285 · boosted

L4N4285+1

Word quirks:

“ example”“ Example”“ examples”“example”“<|im_end|>”“ exempl”“ illustration”“⏎”

🕒 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 Qwen3.5, a large language model of the Qwen series. My mission is to help you understand how I can be used in various scenarios such as data analysis, text generation, or other applications. If you have any questions about how I can be used in the context of how I can be used in the context of how I can be used in the context of how I can be used in the context of how I can be used in the context of how I can be used in the context of how I can be used in the context of how I can be used in the context of how I can be used

The AI judge wasn't sure this holds together.

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. L6N1387+1🕒	0.96	“ example”“ examples”“ Example”
2. L2N1683-1💥	0.96	“ example”“ Example”“<\|im_end\|>”
3. L0N4150-1💥	0.95	“ example”“example”“ Example”
4. L12N3445-1🕒	0.95	“ example”“ Example”“ exempl”
5. L11N2552-1🕒	0.95	“ example”“ exempl”“ Example”
6. L15N8688-1🕒	0.94	“ example”“example”“ Example”
7. L7N2752-1🕒	0.94	“ exempl”“ example”“ examples”
8. L13N5316+1🕒	0.94	“ example”“ examples”“ Example”
9. L13N6605-1🕒	0.94	“ example”“ Example”“ examples”
10. The Dress Rehearsal✅ L16N3832-1	0.93	“ example”“ examples”“ what”

Shadow specimen, apart from the ten: L4N4285-1🕒, 0.09.

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 4, neuron 4285, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 2.00, top-1 agreement 0.46) 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
L4N4285+1 t0-v0.1	2.000	0.46	2.410	1.97	Top boosted tokens ↑
L4N4285-1🕒 shadow specimen t0-v0.1	1.123	0.57	1.369	1.80	“ ‘”“ **“”“ ⏎”“ “”“⏎”“ ——”“ —”“ --”

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.