Layer 24 · Neuron 4772 · suppressed

L24N4772-1

Word quirks:

“ tablets”“ tab”“ tablet”“ tabs”“ capsules”“ table”“ Tablets”“-table”

🕒 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 Qian 210 (Qian), an advanced AI intelligence developed by the Chinese Academy of Social Sciences. My mission is to provide high-quality, safe, and diverse content that meets the needs of users across various fields, including but not limited to:

Knowledge and Information: Delivering accurate, up-to-date information on topics like social sciences, economics, culture, and technology.
Creative Collaboration: Assisting in generating ideas for creative projects such as writing, design, or strategic planning.
Tabular Data Analysis: Helping interpret complex data through structured tables and insights.

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. L29N8545+1🕒	0.93	“ tablets”“ tab”“ Tablets”
2. Terminal Velocity✅ L24N7088+1	0.85	“ tablet”“ terminal”“ tablets”
3. L2N8152+1💥	0.81	“ tablet”“<\|im_end\|>”“ tablets”
4. L1N6243+1💥	0.79	“ tablet”“ Tablet”“ tablets”
5. L26N8519-1🕒	0.79	“ Tablet”“表”“Table”
6. L27N3328+1🕒	0.78	“table”“tables”“ table”
7. L25N6379+1💥	0.75	“ tab”“ tabs”“ Tab”
8. L25N3507+1🕒	0.75	“ tablet”“片”“ tablets”
9. L26N8317+1🕒	0.75	“ tab”“ Tab”“ tabs”
10. L20N3655+1🕒	0.74	“ tab”“ tabs”“ Tab”

Shadow specimen, apart from the ten: L24N4772+1🕒, -0.10.

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 24, neuron 4772, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.90, top-1 agreement 0.67) 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
L24N4772+1🕒 shadow specimen t0-v0.1	0.797	0.69	0.875	1.34	“取决”“端”“اره”“端的”“ario”“和鼓励”“ WH”“就业”
L24N4772-1 t0-v0.1	0.902	0.67	0.922	1.30	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.