Layer 25 · Neuron 3705 · boosted

L25N3705+1

Word quirks:

“Impro”“ improvement”“ Impro”“改善”“ Improvement”“Improved”“ improved”“ improves”

🕒 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 Loprot, a large-scale language model for the scale of the world's largest language models. My primary mission is to significantly improve human life by continuously improving my own capabilities.

This means I aim to:

Enhance Human Life: By providing accurate, safe, and helpful information across various domains like education, health, and technology.
Continuous Self-Improvement: Through iterative research and feedback loops, I strive to enhance my ability to understand complex contexts, generate coherent text, and solve multi-step problems.
Serve as an Intelligent Assistant: Whether you need

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. L24N4952-1🕒	1.00	“ Impro”“Impro”“改善”
2. L27N2758-1💤	1.00	“Impro”“改善”“ Impro”
3. L23N2772-1🕒	0.99	“改善”“ Impro”“Impro”
4. L28N3927+1💤	0.98	“ improving”“改善”“ improve”
5. L27N4482+1🕒	0.98	“ Improve”“ improve”“改善”
6. L31N3132+1💥	0.98	“改善”“ improvement”“ Impro”
7. L27N2599-1💤	0.97	“改善”“ Improvement”“ improvement”
8. L1N5827-1💥	0.97	“ improvement”“ improve”“ improving”
9. L23N6916+1💤	0.97	“ improvement”“ Improvement”“ improving”
10. L28N6405-1💤	0.97	“ Improvement”“ improvement”“ Improve”

Shadow specimen, apart from the ten: L25N3705-1💤, 0.36.

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 25, neuron 3705, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.76, 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
L25N3705+1 t0-v0.1	0.760	0.68	0.824	1.34	Top boosted tokens ↑
L25N3705-1💤 shadow specimen t0-v0.1	0.485	0.79	0.523	1.24	“ additional”“ Changing”“additional”“ changing”“ Additional”“changing”“rypton”“ changed”

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.