INDEX

Explanations

research papers

New Auto-Interp

Configuration

Prompts (Dashboard)

16,384 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 dAtA

-0.59

 maksud

-0.58

\{\\

-0.57

 esfuer

-0.52

 ModelExpression

-0.52

breos

-0.52

 tarko

-0.51

akor

-0.49

Климат

-0.49

rrggbb

-0.49

POSITIVE LOGITS

adalah

0.60

 AssemblyTitle

0.57

 lack

0.57

 realizing

0.54

 realising

0.53

DispatchToProps

0.53

قایناقلار

0.53

 death

0.52

 overcoming

0.51

lanması

0.50

Activations Density 0.003%