Layer 16 · Neuron 9150 · boosted

L16N9150+1

Word quirks:

“ describing”“ detailing”“ stating”“ penned”“ specifying”“ enumer”“ mentioning”“ written”

🕒 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 a large language model trained by Google, and I am part of the Google AI team's efforts to advance artificial intelligence. My primary role is to assist users in providing helpful, accurate, and engaging responses across a wide range of topics.

My mission is to enable users to interact with complex ideas, solve problems, and explore new possibilities through clear, efficient, and reliable guidance. This involves:

Facilitating learning: Helping users understand concepts in various domains (e.g., science, technology, history).
Enabling creativity: Assisting in generating ideas, solving challenges, or exploring creative

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. L17N5395-1💤	0.85	“ mentioning”“ describing”“ detailing”
2. L17N1289-1💤	0.79	“ describing”“ —”“ mentioning”
3. L16N5235-1🕒	0.79	“ specifying”“ describing”“ indicating”
4. L17N4948-1💤	0.79	“ detail”“ mentioning”“ mention”
5. L14N953+1💤	0.79	“ specified”“ description”“ indicating”
6. L17N41-1🕒	0.76	“ describe”“ describing”“ specify”
7. L16N5857-1🕒	0.76	“ mentioned”“ described”“ explained”
8. L14N2073-1🕒	0.75	“ description”“ specifying”“ specify”
9. L13N8142+1💤	0.74	“ describing”“ description”“ stating”
10. L11N3164-1🕒	0.74	“ mentioned”“ description”“ describing”

Shadow specimen, apart from the ten: L16N9150-1🕒, 0.15.

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 16, neuron 9150, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.66, 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
L16N9150+1 t0-v0.1	0.657	0.68	0.866	1.26	Top boosted tokens ↑
L16N9150-1🕒 shadow specimen t0-v0.1	1.124	0.62	1.281	1.35	“ documents”“ papers”“ docs”“ files”“ apps”“ flyers”“ films”“ copies”

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.