Layer 25 · Neuron 6724 · suppressed

L25N6724-1

Word quirks:

“ radi”“辐射”“radi”“ Radi”“Radi”“ radio”“ radiation”“ radios”

🕒 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 Koriji, a large-scale language model developed by the Koriji Research Institute. My primary mission is to provide accurate, reliable, and helpful information across a wide range of topics, while maintaining strict adherence to safety and ethical guidelines.

Here is how I aim to fulfill this mission:

Accuracy & Reliability: I strive to provide well-researched, fact-based answers to minimize misinformation.
Safety & Ethics: I am designed to refuse requests that involve violence, self-harmonization, illegal activities, or other harmful content, ensuring a safe environment for

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. L23N4127-1🕒	0.97	“ radi”“辐射”“ radiant”
2. L31N1407+1💥	0.97	“ radi”“ Radi”“radi”
3. L25N4751-1🕒	0.96	“ radi”“辐射”“Radi”
4. L26N2084-1🕒	0.96	“ radi”“radi”“ radiation”
5. L29N4672+1💥	0.96	“ radiation”“辐射”“ Radiation”
6. L25N7674-1🕒	0.96	“辐射”“ conv”“ radi”
7. L27N6726+1🕒	0.96	“辐射”“Radi”“radi”
8. L30N8752-1💥	0.95	“ radi”“ Radi”“radi”
9. L25N6716-1🕒	0.95	“Radi”“ radi”“ Radi”
10. L28N6510+1💤	0.95	“radi”“ radiation”“辐射”

Shadow specimen, apart from the ten: L25N6724+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 25, neuron 6724, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.75, top-1 agreement 0.70) 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
L25N6724+1🕒 shadow specimen t0-v0.1	0.675	0.71	0.724	1.24	“-x”“ Exped”“�”“ox”“溪”“_X”“ultip”“克罗”
L25N6724-1 t0-v0.1	0.746	0.70	0.755	1.36	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.