Layer 20 · Neuron 2385 · boosted

L20N2385+1

Word quirks:

“ specific”“特定的”“特定”“ Specific”“specific”“Specific”“ específico”“-specific”

🕒 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 Q, a large-scale specialized model for your specific needs. My mission is to provide high-quality, precise, and efficient responses tailored to the user's requirements, ensuring that I can effectively serve users in their intended scenarios.

If there's something specific you'd like me to help with, let me know! 😊

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. L23N4099+1🕒	1.00	“ Specific”“ specific”“特定的”
2. L17N3607-1🕒	0.99	“ specific”“ Specific”“-specific”
3. L24N575-1🕒	0.99	“ Specific”“ specific”“Specific”
4. L13N5882+1💤	0.99	“ specifically”“ specific”“specific”
5. L23N3713+1🕒	0.99	“ Specific”“ specific”“特定的”
6. L21N5495-1🕒	0.99	“ specific”“特定”“ specifically”
7. The Specific Squatter✅ L22N2651+1	0.99	“ specific”“ Specific”“特定”
8. L17N8023-1🕒	0.98	“ specific”“特定”“ Specific”
9. L19N7326-1🕒	0.98	“ specific”“ Specific”“specific”
10. L16N8417-1🕒	0.98	“ specific”“ Specific”“ something”

Shadow specimen, apart from the ten: L20N2385-1🕒, 0.21.

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 20, neuron 2385, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.79, top-1 agreement 0.68) 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
L20N2385+1 t0-v0.1	0.793	0.68	0.878	1.29	Top boosted tokens ↑
L20N2385-1🕒 shadow specimen t0-v0.1	0.672	0.73	0.848	1.09	“ proper”“ correct”“Correct”“proper”“ Proper”“ appropriate”“正确的”“ Correct”

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.