INDEX

Explanations

research topics and findings

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

 three

-1.52

or

-1.50

 they

-1.46

 first

-1.45

if

-1.38

 five

-1.29

 could

-1.25

tividade

-1.23

ệu

-1.23

戦闘

-1.21

POSITIVE LOGITS

ykjav

1.76

to

1.55

 różnych

1.50

 resté

1.49



1.45

/**

1.41

 seyn

1.41

saying

1.38

嚶

1.37

 genellikle

1.36

Activations Density 0.320%