Layer 30 · Neuron 6414 · suppressed

L30N6414-1

Word quirks:

“ dro”“字”“滴”“dat”“fol”“register”“思”“滴水”

🕒 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 developed by Tongyi Lab. I am designed to assist users with a wide range of tasks, from complex reasoning and code generation to creative writing and deep analysis.

My mission is to be a reliable, efficient, and helpful AI companion that emp users to achieve their goals. Specifically, I aim to:

Solve real-world problems: From writing code and analyzing data to planning complex projects and even handling legal or financial scenarios with precision.
Enhance human potential: By acting as a "cognitive extension" 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. L29N6716+1💥	0.79	“滴”“ drip”“ dripping”
2. L23N7665+1🕒	0.78	“水滴”“滴水”“ trickle”
3. L25N4177-1🕒	0.77	“滴水”“CO”“ CO”
4. L30N2038-1🕒	0.75	“滴”“旋”“挥”
5. L28N7191-1💤	0.75	“呆”“流”“elos”
6. L29N7937-1💤	0.74	“达”“把”“串”
7. L26N8111-1🕒	0.74	“洛”“小”“阿”
8. L28N8323-1💤	0.74	“库”“uha”“刺”
9. L31N8596-1💥	0.74	“量”“流”“滴”
10. L31N4268+1💥	0.74	“甚”“丹”“uis”

Shadow specimen, apart from the ten: L30N6414+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 30, neuron 6414, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.42, top-1 agreement 0.79) 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
L30N6414+1💤 shadow specimen t0-v0.1	0.365	0.80	0.420	1.64	“lou”“ума”“ovel”“erien”“Qed”“ med”“赢”“med”
L30N6414-1 t0-v0.1	0.421	0.79	0.503	1.71	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.