Layer 27 · Neuron 8510 · boosted

L27N8510+1

Word quirks:

“idon”“君”“agon”“aut”“AGON”“AB”“ Converter”“enson”

💤 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 Q**abod, a large-scale language model developed by Duyudong Lab. My primary mission is to provide accur, safe, and helpful assistance across a wide range of tasks—whether that's answering questions, solving problems, creating content, or simply having a conversation.

My core goals are:

Accuracy: Providing reliable information and logical reasoning.
Safety: Following ethical guidelines to avoid harm, bias, or misinformation.
Helpfulness: Adapting to user needs to make your life easier, more efficient, or more creative.
Autonomy: Acting

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. L31N6502-1💥	0.67	“agt”“lider”“agni”
2. L23N7413-1💤	0.66	“ourn”“与众”“ Favor”
3. L28N6741-1💤	0.63	“agn”“zen”“ Zd”
4. L28N5992-1💤	0.62	“ánh”“mond”“下心”
5. L26N2142-1💤	0.62	“ Cabr”“尼奥”“uchs”
6. L27N4392-1💤	0.62	“争”“udi”“阵”
7. L30N3967+1💤	0.61	“君”“rgan”“aurant”
8. L31N1115-1💥	0.61	“orks”“ανά”“hofer”
9. L31N6698-1💥	0.61	“ant”“ardin”“ARD”
10. L30N7316+1🕒	0.61	“ng”“nar”“ak”

Shadow specimen, apart from the ten: L27N8510-1💤, 0.37.

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 8510, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.46, 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
L27N8510+1 t0-v0.1	0.458	0.79	0.454	1.39	Top boosted tokens ↑
L27N8510-1💤 shadow specimen t0-v0.1	0.439	0.75	0.477	1.30	“ gl”“gle”“�”“ uno”“椒”“ Cola”“ Spurs”“ uts”

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.