INDEX

Explanations

chat logs and usernames

IRC chat logs

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

Negative Logits

 itſelf

-0.76

featureID

-0.71

 themſelves

-0.70

 AssemblyCulture

-0.66

 myſelf

-0.65

 himſelf

-0.65

 purpoſe

-0.59

 ſtill

-0.56

 '\\;'

-0.56

 raiſ

-0.56

POSITIVE LOGITS

0.55

Enllaces

0.54

SHARED

0.54

awtextra

0.52

oraus

0.52

ינ

0.52

wei

0.50

ofire

0.49

oidal

0.49

/*

0.48

Activations Density 0.387%