Layer 27 · Neuron 1807 · boosted

L27N1807+1

Word quirks:

“ exceptionally”“ exceedingly”“ extraordinarily”“orn”“ estremamente”“Exception”“ incredibly”“ remarkably”

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

Identity (according to the specimen itself)

I are Q3, a large-scale ultra-symmetry model developed by Tongyi Lab. My mission is to provide extremely precise, context-aware, and task-oriented assistance across all domains. I aim to:

Solve problems with extreme accuracy: Whether it's generating code, analyzing data, or creating content, I strive for 100% precision in every response.
Provide ultra-symmetry solutions: I ensure that the output is as perfectly balanced and aligned with the user's intent as possible, regardless of the task complexity.
**Support all tasks

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. L19N7785-1🕒	0.90	“ extremely”“ extraordinarily”“ Extremely”
2. L26N3835-1🕒	0.88	“ incredibly”“ insanely”“ extremamente”
3. L10N6993+1🕒	0.87	“ extreme”“ extraordinarily”“ extremes”
4. L17N8423-1🕒	0.87	“ exceptionally”“ very”“ unusually”
5. L14N8179-1🕒	0.87	“ extreme”“ amazing”“ utterly”
6. L20N9176-1🕒	0.87	“ extraordinarily”“ enormously”“ exceedingly”
7. L13N5400-1🕒	0.86	“ extremely”“ ONLY”“ extraordinarily”
8. L14N2578+1🕒	0.85	“ very”“ extraordinarily”“ extremely”
9. L10N6737+1🕒	0.84	“ extreme”“ extraordinarily”“ extremely”
10. L13N9083+1🕒	0.83	“ extraordinarily”“ extremely”“ basically”

Shadow specimen, apart from the ten: L27N1807-1💤, 0.04.

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 27, neuron 1807, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.42, top-1 agreement 0.81) 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
L27N1807+1 t0-v0.1	0.423	0.81	0.472	1.25	Top boosted tokens ↑
L27N1807-1💤 shadow specimen t0-v0.1	0.410	0.75	0.428	1.19	“ pure”“纯”“Pure”“ Pure”“純”“pure”“ purified”“纯粹”

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.