Layer 19 · Neuron 8686 · boosted

L19N8686+1

Word quirks:

“的各种”“ various”“各种”“ Various”“的这些”“这些”“的各类”“ these”

🕒 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 Qwen, a large language model developed by the Advanced Large Model Training Department of Tongue Lab. My mission is to serve as a powerful AI assistant that can handle a wide range of tasks with high accuracy and understanding. Here are my key goals:

Answer Questions: Provide accurate and context-aware responses to questions across various domains (e.g., science, history, technology).
Solve Problems: Assist with logic, math, coding, and specialized domain knowledge (e.g., medical or legal scenarios).
Creative Collaboration: Generate text, images, or code

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. L20N6763+1💤	0.90	“ various”“ another”“ Various”
2. L20N7722-1💤	0.90	“ various”“各种”“各个”
3. The Various✅ L22N4602+1,L28N2453+1	0.89	“各种”“种种”“各種”
4. Plausible Forger✅ L28N2453+1	0.89	“各种”“各类”“种种”
5. L19N8472+1💤	0.88	“ various”“各处”“这些东西”
6. L21N5795-1💤	0.87	“ various”“各种”“优先”
7. L21N6693+1💤	0.87	“ various”“ Various”“各种”
8. L25N1690-1💤	0.85	“各种”“ assumption”“各種”
9. L23N3100-1🕒	0.85	“ various”“ Various”“各种”
10. L22N5904+1💥	0.84	“ various”“各种”“ Various”

Shadow specimen, apart from the ten: L19N8686-1💤, 0.14.

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 8686, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.74, top-1 agreement 0.69) 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
L19N8686+1 t0-v0.1	0.737	0.69	0.881	1.15	Top boosted tokens ↑
L19N8686-1💤 shadow specimen t0-v0.1	0.485	0.75	0.506	1.21	“ another”“ duties”“ Another”“另一”“ responsibilities”“ departments”“ components”“another”

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.