INDEX

Explanations

found evidence recovery

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

for

-1.89

-1.42

 like

-1.38

-1.35

to

-1.27

-1.26

 çünkü

-1.23

or

-1.14

 unserer

-1.13

窃

-1.11

POSITIVE LOGITS

黩

1.50

 Umfeld

1.41

 übri

1.36

 evidence

1.35

 persön

1.34

 viviendas

1.31

 narzęd

1.31

ַּ

1.29

 several

1.28

ಆ

1.27

Activations Density 0.022%