Layer 26 · Neuron 7591 · boosted

L26N7591+1

Word quirks:

“定”“�”“中间”“丽丝”“涎”“签发”“丽”“电图”

💤 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 Pa-3 (P-3), a large language model created by P-3 (a hypothetical or specific variant of a large-scale language model). My primary mission is to provide helpful, accurate, and safe information across a wide range of topics, including but not limited to:

Knowledge Sharing: Providing accurate and up-to-date information on various subjects, from science and technology to culture and history.
Problem-Solving: Offering solutions to complex problems in fields like programming, mathematics, and logic.
Creative Support: Assisting with creative tasks such as writing stories

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. L29N246-1🕒	0.84	“核”“片”“预定”
2. L28N898+1🕒	0.84	“不起”“签”“琦”
3. L15N7315-1💥	0.83	“照”“归”“朝”
4. L30N4414-1💤	0.83	“厉”“鬼”“甩”
5. L26N147+1🕒	0.83	“上”“利”“时”
6. L27N5713+1💤	0.83	“案”“表”“影”
7. L28N7544+1💤	0.83	“模”“兴”“战”
8. L30N721+1🕒	0.83	“子”“席”“理”
9. L23N4780-1💤	0.83	“井”“所向”“在所”
10. L24N766-1🕒	0.83	“息”“次”“宽”

Shadow specimen, apart from the ten: L26N7591-1🕒, 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 26, neuron 7591, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.53, 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
L26N7591+1 t0-v0.1	0.531	0.77	0.555	1.56	Top boosted tokens ↑
L26N7591-1🕒 shadow specimen t0-v0.1	0.771	0.73	0.798	1.47	“ Arts”“ Arm”“Arm”“/arm”“Ar”“亚”“ arm”“ Nag”

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.