INDEX

Explanations

gradespecificationshighlightsinteriorinscription

The neuron fires specifically on standalone numerical tokens (including integers and floats) that denote measurements, indices, or other numerical metadata.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

鹀

-1.73

you

-1.63



-1.60

ين

-1.55

 vaše

-1.50

 göm

-1.48

鼯

-1.48

 demokrat

-1.44

 môžete

-1.41

 budete

-1.41

POSITIVE LOGITS

2.09

就是一个

1.44

的一切

1.38

});

1.34

",

1.34

 like

1.31

 Eigentum

1.30

<?

1.30

 about

1.28

 clearly

1.27

Activations Density 0.155%