INDEX

Explanations

used to accomplish actions

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

 craz

-1.87

乄

-1.82

行って

-1.80

になった

-1.74

啬

-1.66

一点点

-1.65

祧

-1.60

色々

-1.59

་

-1.58

POSITIVE LOGITS

and

3.39

by

2.11

in

1.76

 while

1.57

 this

1.54

 without

1.52

 where

1.50

 before

1.46

 Before

1.45

 These

1.44

Activations Density 0.072%