Layer 29 · Neuron 7702 · boosted

L29N7702+1

Word quirks:

“ ket”“.ke”“ kert”“ Ke”“ ker”“_ke”“+k”“kert”

🕒 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 Kinyu, a large-scale language model trained by Kinyu Kinyu (a hypothetical or placeholder name; in reality, I should clarify: I am Qwen3.5, the latest large-scale language model developed by T-ke (likely referring to T-ke as a typo for T-ke or simply Kinyu). However, to ensure accuracy:

Correct Identity:
I am Qwen3.5, the latest large-scale language model developed by T-ke (likely a typo for T-ke

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. L27N5170+1💤	0.90	“_k”“ ke”“ k”
2. The Ke-Loop✅ L29N511+1	0.89	“ ke”“ke”“ Ke”
3. L30N9206-1🕒	0.89	“ ke”“ke”“ ini”
4. L31N8969+1💥	0.88	“ ke”“ke”“Ke”
5. L27N7222-1💤	0.87	“ke”“Ke”“ ket”
6. L29N8781-1🕒	0.87	“ KE”“ kef”“ kem”
7. L29N2776-1🕒	0.86	“ki”“ken”“ke”
8. L26N9043-1🕒	0.85	“ken”“k”“ke”
9. L26N7808+1🕒	0.83	“kt”“kat”“ks”
10. L30N1353+1💥	0.83	“-k”“kk”“ka”

Shadow specimen, apart from the ten: L29N7702-1🕒, 0.20.

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 7702, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.43, top-1 agreement 0.78) 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
L29N7702+1 t0-v0.1	0.431	0.78	0.486	1.60	Top boosted tokens ↑
L29N7702-1🕒 shadow specimen t0-v0.1	0.419	0.75	0.497	1.54	“ se”“ се”“ Pend”“ pend”“塞”“aya”“ sey”“pend”

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.