INDEX

Explanations

do/does followed by a verb

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 ラーメン

-1.71

or

-1.61

and

-1.45

 Поэтому

-1.30

 oint

-1.28

 erros

-1.27

🛤

-1.27

くれない

-1.24

seguridad

-1.23

 componen

-1.23

POSITIVE LOGITS

 gantung

1.70

 buatan

1.49

 singkat

1.45

 terbang

1.45

 isoliert

1.45

tev

1.41

 mutiara

1.40

'>";

1.39

 gewor

1.35

pired

1.34

Activations Density 0.010%