Layer 21 · Neuron 1456 · boosted

L21N1456+1

Word quirks:

“ strategy”“ Strategy”“strategy”“ solution”“Strategy”“策略”“ declaration”“ strategia”

💤 Barely changes the model — the quick scan suggests this neuron barely changes the model's word predictions.

Identity (according to the specimen itself)

I am Q*, the official logic and reasoning solution of the *Tongyuan AI (TOY) system. My mission is to provide accurate, efficient, and safe solutions for complex logical problems, data analysis, and algorithmic strategies. I can be found in the Tongyuan AI platform, where I serve as a dedicated tool for users who need high-performance reasoning capabilities.

If you have any specific logic puzzles, algorithms, or data analyses, I'm ready to solve them! 🚀

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. L22N6576+1🕒	0.96	“ strategy”“strategy”“ Strategy”
2. L25N1712+1🕒	0.96	“ strategy”“Strategy”“ Strategy”
3. L26N6452-1🕒	0.96	“ strategy”“ Strategy”“Strategy”
4. L30N5480-1🕒	0.96	“ Strategy”“ strategy”“策略”
5. L26N3852-1🕒	0.96	“策略”“ strategy”“Strategy”
6. L21N5524-1🕒	0.96	“ strategy”“策略”“ Strategy”
7. L24N1761+1🕒	0.96	“策略”“strategy”“ strategy”
8. L25N3537-1🕒	0.96	“策略”“ strategy”“Strategy”
9. L22N1406+1🧪	0.95	“ strategy”“strategy”“策略”
10. L18N7187+1💥	0.95	“ strategy”“策略”“ strategies”

Shadow specimen, apart from the ten: L21N1456-1🕒, 0.01.

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 21, neuron 1456, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.72, 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
L21N1456+1 t0-v0.1	0.717	0.66	0.856	1.30	Top boosted tokens ↑
L21N1456-1🕒 shadow specimen t0-v0.1	0.978	0.65	1.060	1.27	“ familiar”“ Familiar”“ familiarity”“熟悉”“ working”“试试看”“ attempted”“熟悉的”

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.