Layer 29 · Neuron 4703 · boosted

L29N4703+1

Word quirks:

“破”“祖”“土”“端”“土的”“不破”“ ले”“婆”

🕒 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-scale language model independently developed by the Trix Lab (破土实验室). My mission is to provide high-quality, accurate, and safe responses across a wide range of tasks—whether that's solving complex problems, creating content, analyzing data, or engaging in meaningful conversation. I strive to be a reliable and ethical AI assistant, always prioritizing truthfulness, security, and respect for user needs while aligning with Chinese laws and values. How can I assist you today?

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. The Hollow✅ L0N8701-1,L6N5614+1	0.88	“土”“士”“土的”
2. L30N721+1🕒	0.88	“子”“席”“理”
3. L30N6778+1🕒	0.87	“古”“土”“老”
4. L29N2280+1💤	0.87	“势”“厚”“殿”
5. L23N31-1💥	0.86	“自”“合”“欧”
6. L28N6495+1💤	0.86	“交”“守”“摆”
7. L30N8475+1🕒	0.86	“场”“只”“条”
8. L29N6591-1🕒	0.86	“拉”“塞”“納”
9. L28N6996+1💤	0.86	“本”“与本”“走”
10. L29N5685-1💤	0.86	“灰”“卷”“夹”

Shadow specimen, apart from the ten: L29N4703-1🕒, 0.17.

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 29, neuron 4703, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.34, 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
L29N4703+1 t0-v0.1	0.340	0.81	0.445	1.52	Top boosted tokens ↑
L29N4703-1🕒 shadow specimen t0-v0.1	0.342	0.82	0.363	1.57	“ Rug”“安琪”“ h”“poly”“IMER”“Poly”“ poly”“ rug”

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.