INDEX

Explanations

comprised of / consist of

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

 disguise

-0.23

å¼ĢéĹ¨

-0.23

æĹ¶æľº

-0.22

 Ð¿Ð¾Ð²

-0.21

åĬ

-0.21

Ð·Ð°Ð¼

-0.21

æİª

-0.20

 obliv

-0.20

****************************************************************************

-0.19

 insight

-0.19

POSITIVE LOGITS

 consisting

0.34

 consists

0.34

 consist

0.33

 mostly

0.31

 consisted

0.30

 mainly

0.29

äºĶä¸ª

0.28

 composed

0.28

 comprised

0.28

mostly

0.28

Activations Density 0.092%