Layer 28 · Neuron 6684 · suppressed

L28N6684-1

Word quirks:

“Horizontal”“ horizontal”“ horizontally”“orizontal”“ vertical”“竖”“ Horizontal”“Vertical”

🕒 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 Q* is an vertically v1.5 model of the Q series, which is a large-scale language model independently developed by the *Tavola Team at Tongyi Lab. My name "Q" comes from the first letter of my name, which also represents the Q series of models.

My mission is to provide high-quality, safe, and helpful AI services across various fields, including but not limited to:

Answering questions (e.g., science, culture, daily life)
Creative content generation (e.g., stories, vertical video scripts, visual arts

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. L25N2901-1🕒	0.97	“ horizontal”“ Horizontal”“横向”
2. L20N2673-1🕒	0.97	“ horizontal”“ vertical”“的水平”
3. L30N6343+1🕒	0.97	“horizontal”“ horizontal”“ Horizontal”
4. L26N2615-1🕒	0.95	“ vertical”“vertical”“ Vertical”
5. L24N3229-1💤	0.95	“ horizontal”“Horizontal”“ Horizontal”
6. L24N4777+1💥	0.94	“ horizontal”“ Horizontal”“水平”
7. L0N7035+1💥	0.94	“<\|im_end\|>”“ horiz”“vertical”
8. L19N6197-1💤	0.94	“ horizontal”“ horizontally”“horizontal”
9. L26N5760+1🕒	0.93	“ horizontal”“ Horizontal”“horizontal”
10. L3N3816-1🕒	0.93	“ vertical”“<\|im_end\|>”“ horizontal”

Shadow specimen, apart from the ten: L28N6684+1💤, 0.18.

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 28, neuron 6684, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.45, top-1 agreement 0.76) 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
L28N6684+1💤 shadow specimen t0-v0.1	0.349	0.77	0.460	1.18	“ ver”“ Ver”“ вер”“ VER”“-ver”“вер”“VER”“VRT”
L28N6684-1 t0-v0.1	0.451	0.76	0.477	1.42	Top boosted tokens ↑

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.