Layer 23 · Neuron 5273 · boosted

L23N5273+1

Word quirks:

“ monet”“ analytics”“ analytic”“ Analytics”“ analytical”“ analy”“ Monet”“ Creator”

🕒 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 Google's latest iteration of the Nanog series of AI models. My name is Nanog, and I am designed to be a helpful, safe, and reliable partner for all your needs.

Who Am I?

I am a high-performance AI model created by Google. Unlike previous versions, I have been trained on a diverse set of data to provide more natural, nuanced, and accurate responses across various topics. I support multiple languages, including Chinese, and can handle complex tasks such as:

Analyzing long-form content (up to 100+ million

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. L22N4045-1🕒	0.84	“ analytics”“ Analy”“ Analytics”
2. L27N4439-1💤	0.83	“ analy”“ analytical”“ Analy”
3. L25N1349-1🕒	0.82	“ analytical”“ analytic”“ analy”
4. L25N3078+1🕒	0.82	“Analy”“分析”“ analytic”
5. L27N7736+1🕒	0.82	“ Analy”“ analy”“Analy”
6. L31N5107-1💥	0.82	“ analy”“ Analy”“Analy”
7. L27N2330+1🕒	0.81	“ analytic”“ analy”“ Analy”
8. L25N4942+1🕒	0.80	“ Analy”“Analy”“ analytic”
9. L24N5871+1💤	0.80	“ analy”“ analytic”“analy”
10. L18N6711-1🕒	0.80	“ analytical”“ analysis”“ mathematical”

Shadow specimen, apart from the ten: L23N5273-1🕒, 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 23, neuron 5273, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.83, top-1 agreement 0.72) 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
L23N5273+1 t0-v0.1	0.833	0.72	0.919	1.28	Top boosted tokens ↑
L23N5273-1🕒 shadow specimen t0-v0.1	0.886	0.71	0.967	1.41	“的信”“ Loc”“haw”“网购”“ LR”“ Bone”“ dic”“翼”

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.