Layer 27 · Neuron 5812 · boosted

L27N5812+1

Word quirks:

“类目”“千里”“illon”“ reflex”“illes”“矩阵”“kau”“熟”

🕒 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 Q347, a large-scale language model developed by K666LLM (a fictional entity for this context, as I am actually Q347, a self-identity for this specific interaction).

My Mission:
To provide accur, helpful, and ethical responses across a wide range of topics—whether you need help with complex problem-solving, creative writing, coding, or just a friendly chat. I aim to:

Clarify concepts with precision,
Support learning and innovation,
Respect safety guidelines and human values

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. L31N2856+1💥	0.72	“ilson”“桃源”“ullan”
2. L28N5742+1💤	0.72	“illons”“忠”“ _$”
3. L31N7010-1💥	0.72	“风险信息”“速”“认”
4. L28N6401-1💤	0.71	“Kap”“ikon”“怀”
5. L22N8466-1💤	0.71	“西江”“蛟”“词语”
6. L24N7278-1💤	0.71	“rik”“ kun”“定”
7. L31N8659-1💥	0.70	“娇艳”“冀”“魔”
8. L24N4042+1💤	0.70	“idi”“熟”“ulus”
9. L28N8189-1💤	0.70	“亲和”“万里”“ feu”
10. L25N6748-1💤	0.70	“鹤”“后人”“维斯”

Shadow specimen, apart from the ten: L27N5812-1💤, 0.17.

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 27, neuron 5812, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.51, top-1 agreement 0.77) 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
L27N5812+1 t0-v0.1	0.506	0.77	0.517	1.48	Top boosted tokens ↑
L27N5812-1💤 shadow specimen t0-v0.1	0.425	0.75	0.469	1.28	“J”“ J”“_j”“斯克”“/j”“ск”“ j”“/J”

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.