Layer 15 · Neuron 3146 · boosted

L15N3146+1

Word quirks:

“ specifically”“,**”“ specific”“ **“”“ ***”“ specifics”“ **”“ **‘”

🕒 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 designed to assist with a wide range of tasks, from handling complex reasoning to creative collaboration. My mission is to be a reliable, efficient, and versatile partner in advancing human productivity and innovation. Whether you need help solving mathematical problems, analyzing data, or generating code, I strive to provide precise and actionable support.

How can I assist you today?

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. L12N6874+1🕒	0.89	“ specific”“ particular”“ **”
2. L15N9181-1💤	0.89	“ ”“ “”“ defined”
3. L10N1107-1💤	0.87	“ **”“ specific”“ exact”
4. L15N2635-1🕒	0.86	“ specifics”“ “”“ ”
5. L15N6617-1🕒	0.85	“ *”“ ”“ **“”
6. L11N9017-1🕒	0.85	“ ”“ “”“ **"”
7. L8N3727-1💤	0.85	“ specific”“ ”“ *”
8. The Specificist✅ L10N9076+1	0.84	“ specific”“⏎”“ specifically”
9. L10N6796-1💤	0.84	“ ”“ specific”“ "”
10. L16N8326-1🕒	0.84	“ ‘”“ **‘”“ specific”

Shadow specimen, apart from the ten: L15N3146-1🕒, -0.06.

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 15, neuron 3146, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.68, top-1 agreement 0.65) 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
L15N3146+1 t0-v0.1	0.677	0.65	0.892	1.17	Top boosted tokens ↑
L15N3146-1🕒 shadow specimen t0-v0.1	0.667	0.72	0.804	1.31	“ mix”“⏎”“ list”“ mixture”“ mixes”“ filtered”“ varying”“ tier”

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.