INDEX

Explanations

questioning or checking details

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

åħ¨æĸ°çļĦ

-0.28

now

-0.28

æĪĳä»¬çļĦ

-0.27

å¹¶åıĳ

-0.26

our

-0.26

æĦ¿æĻ¯

-0.25

åĪĽä½ľ

-0.25

æĸ°çļĦ

-0.25

 synchronized

-0.25

éģįåıĬ

-0.25

POSITIVE LOGITS

æĺ¯åĲ¦åŃĺåľ¨

0.38

è¯¢éĹ®

0.37

 clues

0.33

éĹ®éĹ®

0.32

æĢĢçĸĳ

0.32

 suspicious

0.32

 questioning

0.32

åı¯çĸĳ

0.32

åĪ¤æĸŃ

0.31

æŁ¥éªĮ

0.31

Activations Density 0.018%