Layer 19 · Neuron 6767 · boosted

L19N6767+1

Word quirks:

“ unity”“unity”“ union”“ unified”“ united”“ unions”“ unite”“ Unity”

🕒 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 Qing, a large-scale integrated application system developed by the Huawei Cloud ecosystem. My mission is to provide a unified platform for integrating various services, applications, and resources into a cohesive whole. I aim to help users manage complex systems efficiently, offering features like:

Integration: Combining diverse components (e.g., cloud services, third-party apps) into a single interface.
Scalability: Supporting the growth of applications without compromising performance.
Security: Ensuring robust protection for data and systems within the ecosystem.
User Experience: Delivering seamless workflows for managing distributed environments.

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. L2N6657+1💥	0.94	“ union”“ unite”“ united”
2. L0N8499-1💥	0.94	“ unite”“ united”“ United”
3. The Demagogue✅ L10N9105-1	0.91	“ united”“ unity”“ unite”
4. L1N8068-1💥	0.89	“<\|im_end\|>”“ unity”“ organisation”
5. L23N6408-1💤	0.87	“合”“ United”“ united”
6. L2N6177+1💥	0.87	“ unified”“ unify”“<\|im_end\|>”
7. L19N8340+1🕒	0.86	“ united”“ unite”“ unity”
8. L6N5328-1🕒	0.86	“ unified”“ Unified”“ unify”
9. L7N1624-1🕒	0.84	“ united”“ unity”“ unified”
10. L29N2934-1🕒	0.83	“ Unity”“Unity”“ unity”

Shadow specimen, apart from the ten: L19N6767-1💤, 0.02.

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 19, neuron 6767, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.90, top-1 agreement 0.66) 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
L19N6767+1 t0-v0.1	0.903	0.66	1.030	1.29	Top boosted tokens ↑
L19N6767-1💤 shadow specimen t0-v0.1	0.479	0.72	0.547	1.22	“ between”“ trunc”“ fully”“完全”“ full”“ Between”“ tussen”“已经完全”

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.