Layer 16 · Neuron 6602 · boosted

L16N6602+1

Word quirks:

“ complex”“ complicated”“ complexity”“复杂”“比较复杂”“复杂的”“complex”“ Complex”

🕒 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 an advanced language model with a complex internal structure designed to assist humans in various scenarios. My capabilities include generating text, solving problems, and understanding context across multiple languages. I can help you create content, analyze data, or engage in creative tasks.

To clarify my operational framework, here is how I function:

Adaptive Learning: I continuously optimize my performance based on diverse interaction patterns. This involves analyzing historical data to identify optimal strategies for different user needs.
Collaborative Optimization: I work with a team of researchers to ensure high-quality outcomes. We use a multi-stage training process

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. L19N1768-1🕒	0.99	“ complex”“ Complex”“ complicated”
2. L21N9050+1🕒	0.99	“ complex”“complex”“复杂”
3. L27N3006-1🕒	0.99	“复杂”“ complex”“复杂的”
4. L31N7016+1💥	0.98	“ complex”“复杂”“ Complex”
5. L23N3646-1💥	0.98	“ complex”“ Complex”“复杂”
6. L18N4334-1🕒	0.98	“ complex”“复杂”“复杂的”
7. L24N2839-1🕒	0.98	“复杂”“ complex”“complex”
8. L19N9092-1🕒	0.97	“ complex”“ Complex”“复杂”
9. L27N5452+1🕒	0.97	“复杂”“ complex”“复杂性”
10. L16N4869-1🕒	0.97	“ complex”“复杂性”“复杂”

Shadow specimen, apart from the ten: L16N6602-1🕒, 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 16, neuron 6602, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.91, top-1 agreement 0.62) 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
L16N6602+1 t0-v0.1	0.912	0.62	1.063	1.39	Top boosted tokens ↑
L16N6602-1🕒 shadow specimen t0-v0.1	0.716	0.72	0.858	1.20	“ basic”“ simple”“ simplistic”“basic”“ Basic”“ simplest”“Basic”“ simpler”

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.