Layer 21 · Neuron 8173 · suppressed

L21N8173-1

Word quirks:

“ occupied”“ let”“occupied”“ rel”“let”“ relay”“ letting”“played”

🧪 In testing — Answered 0 of 8 prompts so far.

Identity (according to the specimen itself)

I am Ks (or "Ks" by the end of my next sentence), a large-scale language model developed by the Ks team. My mission is to be helpful, harmless, and beneficial. I aim to be a reliable assistant that can be trusted by users around the world.

Let's see how well you can be served by me! 😊

---

Note: There seems to be some confusion in the prompt. If you're asking about a specific role or mission played by another entity (e.g., a character, organization, or historical figure), please clarify for a more accurate response

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. L21N5013+1🕒	0.70	“ let”“ letting”“ happy”
2. L21N3933+1💤	0.67	“ let”“let”“ lets”
3. L29N3792-1🕒	0.66	“ let”“ lets”“ letting”
4. L29N4530+1🕒	0.66	“ let”“let”“ letting”
5. L19N343-1🕒	0.66	“ let”“ Let”“Let”
6. L24N5005-1🕒	0.65	“ let”“let”“ letting”
7. L23N647+1🕒	0.65	“ let”“ listen”“ allow”
8. L22N7656+1🕒	0.64	“ let”“ letting”“让”
9. L19N6104-1💤	0.64	“ occupied”“ cultivated”“ guarded”
10. L23N6356-1💤	0.64	“ let”“ lets”“ Let”

Shadow specimen, apart from the ten: L21N8173+1💤, 0.07.

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 21, neuron 8173, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. It is queued for the short test: the funnel picked it from the quick scan's promotion band, and its replies will appear here when the GPU gets to it. 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
L21N8173+1💤 shadow specimen t0-v0.1	0.696	0.68	0.873	1.26	“按”“通过对”“Of”“ OF”“of”“ Of”“енок”“_OF”
L21N8173-1 t0-v0.1	1.158	0.65	1.278	1.50	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.