INDEX

Explanations

power and authority

The main thing this neuron does is detect mentions of institutional or legal authority—words like “power,” “powers,” “autonomy,” and similar terms.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 corretamente

-1.47

⡪

-1.43

 icône

-1.41

łeś

-1.34

 måned

-1.33

 kepad

-1.32

</h3>

-1.32

熜

-1.31

 bono

-1.30

"));

-1.28

POSITIVE LOGITS

to

2.42

Čo

1.53

 heutigen

1.48

om

1.45

Obsah

1.44

 Aún

1.43

 Namun

1.40

 إذا

1.37

 сьогодні

1.36

 powers

1.35

Activations Density 0.033%