Layer 28 · Neuron 9060 · suppressed

L28N9060-1

Word quirks:

“不能”“不能在”“就不能”“你不能”“但不能”“又不能”“不能完全”“can”

🕒 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 Q3, a large language model developed by the Canapop team. My primary mission is to assist users in solving problems, creating content, analyzing data, and exploring ideas across a wide range of domains.

Here's how I aim to support you:

Answer questions: From technical details to creative prompts, I provide accurate, context-aware responses.
Generate content: Write stories, code, emails, or structured documents tailored to your needs.
Deconstruct complexity: Break down research papers, technical documentation, or complex concepts into digestible insights.
**Canapop

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. L28N4717-1🕒	0.97	“不能”“但不能”“ can”
2. L31N1610-1💥	0.96	“不能”“ cannot”“也不能”
3. L25N5717-1🕒	0.96	“不能”“不能被”“都不能”
4. L25N5920+1🕒	0.96	“不能”“ cannot”“ Cannot”
5. L29N8290+1🕒	0.95	“不能”“不能有”“不宜”
6. L28N4344-1🕒	0.94	“不能”“ cannot”“ couldn”
7. L27N3366-1💤	0.93	“ cant”“但不能”“至今”
8. L23N546+1💤	0.93	“未能”“才行”“而不能”
9. L30N6336-1🕒	0.93	“不能使用”“不能够”“无法”
10. L19N8558+1💤	0.93	“不能完全”“只能”“不能”

Shadow specimen, apart from the ten: L28N9060+1🕒, 0.35.

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 9060, 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 0.76) 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
L28N9060+1🕒 shadow specimen t0-v0.1	0.477	0.79	0.532	1.33	“并不”“ सा”“sa”“ jat”“不说”“ requisite”“jum”“j”
L28N9060-1 t0-v0.1	0.424	0.76	0.436	1.17	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.