INDEX

Explanations

mathematical Z, groups, company, games

The neuron spotlights domain-specific technical terms or key jargon nouns.

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

すぎる

-1.50

-1.42

 даже

-1.42

 même

-1.36

or

-1.30

 revisit

-1.29

紆

-1.26

by

-1.23

太多的

-1.21

過ぎる

-1.19

POSITIVE LOGITS

 something

1.55

 what

1.47

 such

1.43

 prazo

1.39

who

1.35

 their

1.34

 things

1.32

輊

1.30

 someone

1.28

 theils

1.27

Activations Density 0.509%