INDEX

Explanations

direction navigation

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

åŁİä¸ľ

-0.08

å±Ĭæ»¡

-0.07

ocu

-0.07

ãĥĩãĤ¢

-0.07

 samle

-0.07

 individually

-0.07

xay

-0.07

Ú©Ùħ

-0.07

æĪĳçļĦæĶ¶èĹı

-0.07

eke

-0.07

POSITIVE LOGITS

æĸ¹åĲĳ

0.11

æĸ¹åĲĳçļĦ

0.10

çļĦæĸ¹åĲĳ

0.10

 direction

0.10

åıįåĲĳ

0.10

åıĮåĲĳ

0.10

Direction

0.09

 directions

0.09

 Direction

0.09

direction

0.09

Activations Density 0.003%