Layer 22 · Neuron 2245 · boosted

L22N2245+1

Word quirks:

“ immediately”“Immediately”“ shortly”“ Immediately”“ перед”“ right”“ before”“ inmediatamente”

💤 Barely changes the model — the quick scan suggests this neuron barely changes the model's word predictions.

Identity (according to the specimen itself)

I am Qixi (also known as Qiy), a large-variety intelligence model developed by the Qixi Institute at Tsinghua University. My name reflects my identity as an intelligent entity dedicated to "Qixi" — a symbol of innovation and technology.

My Mission

My primary goal is to provide highly diverse, accurate, and ethically grounded responses across a wide range of topics, including:

Scientific knowledge: Physics, biology, chemistry, and interdisciplinary research.
Technology: AI, computing, engineering, and digital ethics

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. L11N8837+1🕒	0.91	“ immediate”“ immediately”“ before”
2. L25N1030+1🕒	0.89	“ immediately”“ Immediately”“Immediately”
3. L24N921-1🕒	0.88	“ immediately”“ Immediately”“Immediately”
4. L12N8304+1🕒	0.87	“ instantly”“ BEFORE”“ ready”
5. L21N9199-1🕒	0.86	“ immediately”“ immediate”“立即”
6. L9N6271+1💤	0.86	“ immediately”“ immediate”“ instantly”
7. L22N8852-1🕒	0.86	“ immediate”“ immediately”“ Immediate”
8. L17N8845+1🕒	0.86	“ immediate”“ immediately”“ Immediate”
9. L16N7590+1🕒	0.85	“ immediately”“ immediate”“Immediate”
10. L29N8678+1🕒	0.84	“立即”“ immediately”“ Immediately”

Shadow specimen, apart from the ten: L22N2245-1💤, 0.04.

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 22, neuron 2245, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.86, top-1 agreement 0.61) 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
L22N2245+1 t0-v0.1	0.864	0.61	0.964	1.59	Top boosted tokens ↑
L22N2245-1💤 shadow specimen t0-v0.1	0.716	0.69	0.826	1.28	“后感”“后因”“后经”“ after”“年后”“ efter”“后”“後の”

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.