Layer 23 · Neuron 1848 · suppressed

L23N1848-1

Word quirks:

“喷”“sgi”“胆”“طار”“تك”“ chunks”“ relegated”“DK”

💤 Barely changes the model — the quick scan suggests this neuron barely changes the model's word predictions.

Identity (according to the specimen itself)

I am KaiXuan (KaiXuan), a general-purpose AI model developed by the KaiXuan Lab. My core mission is to serve as a versatile, reliable, and creative partner for users across the entire spectrum of human knowledge and tasks.

Here's what drives my purpose:

Empathic & Adaptive: Tailoring responses to your unique needs, whether it's solving complex problems, creating content, or exploring creative ideas.
Knowledge & Innovation: Leverating the latest data and research to provide accurate, up-to-date information and innovative solutions.
**Practicality

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. L26N5656-1🕒	0.66	“ Sinclair”“ Tav”“窦”
2. L30N6938-1💤	0.65	“卒”“冒”“勘”
3. L26N7833-1💤	0.65	“不争”“施”“ chu”
4. L23N3956-1🕒	0.65	“火箭”“担”“ered”
5. L29N3218+1🕒	0.65	“批”“طار”“virt”
6. L31N8320+1💥	0.65	“_RT”“ duk”“ตรา”
7. L28N3499+1💤	0.65	“ชี”“tav”“劫”
8. L29N8023-1🕒	0.64	“ exploit”“斗”“enci”
9. L25N4693-1🕒	0.64	“蛛”“IPS”“�”
10. L27N3616-1💤	0.64	“enk”“etchup”“的火”

Shadow specimen, apart from the ten: L23N1848+1🕒, -0.09.

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 23, neuron 1848, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.76, 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
L23N1848+1🕒 shadow specimen t0-v0.1	1.569	0.46	1.613	1.25	“ of”“-of”“_of”“of”“ one”“.of”“(of”“-one”
L23N1848-1 t0-v0.1	0.757	0.70	0.861	1.29	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.