Layer 29 · Neuron 8781 · suppressed

L29N8781-1

Word quirks:

“ KE”“ kef”“ kem”“ YE”“ kest”“Ke”“ayer”“ ke”

🕒 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 Qwen3.5, the latest large language model in the Qwen series, developed by Kushy (通義千丈). My mission is to empower users with cutting-edge AI capabilities that are efficient, reliable, and human-aligned.

Key Mission Goals:

Seamless Integration:

Provide ultra-large context (256K+ tokens) for deep analysis of long documents, code, or multi-step workflows.
Enable full-stack code generation and debugging, from writing scripts to deploying full-stack applications

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. L31N8969+1💥	0.91	“ ke”“ke”“Ke”
2. The Ke-Loop✅ L29N511+1	0.90	“ ke”“ke”“ Ke”
3. L30N9206-1🕒	0.89	“ ke”“ke”“ ini”
4. L29N7702+1🕒	0.87	“ ket”“.ke”“ kert”
5. L25N3696-1🕒	0.86	“ ke”“ Ke”“Ke”
6. L27N7222-1💤	0.85	“ke”“Ke”“ ket”
7. L27N7782+1🕒	0.85	“ Ke”“ SDS”“KE”
8. L27N5170+1💤	0.83	“_k”“ ke”“ k”
9. L28N2499+1💤	0.82	“堪”“Ke”“ke”
10. L26N9043-1🕒	0.82	“ken”“k”“ke”

Shadow specimen, apart from the ten: L29N8781+1🕒, 0.16.

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 8781, 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
L29N8781+1🕒 shadow specimen t0-v0.1	0.350	0.81	0.406	1.44	“ pert”“ Pert”“ prit”“_t”“ tumble”“ T”“ tak”“.t”
L29N8781-1 t0-v0.1	0.428	0.78	0.502	1.64	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.