INDEX

Explanations

help and usage instructions

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

 воин

-0.80

 añade

-0.71

fır

-0.71

ृ

-0.71

 grego

-0.69

 بک

-0.69

tagena

-0.68

磨

-0.68

 investigations

-0.68

 cart

-0.68

POSITIVE LOGITS

 help

1.88

help

1.53

Help

1.51

 Help

1.38

HELP

1.31

 HELP

1.25

帮助

1.15

 helping

1.13

 ayuda

1.02

 invalid

1.00

Activations Density 0.023%