INDEX

Explanations

some parts

use of "some" to qualify

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 thofe

0.16

 thefe

0.16

 valam

0.15

 thoſe

0.15

 jakieś

0.15

 noin

0.14

!!!

0.14

 theſe

0.14

 erine

0.14

=(

0.14

POSITIVE LOGITS

 aspects

0.34

 portions

0.29

 parts

0.29

 areas

0.26

 unscrupulous

0.24

 semblance

0.24

 larger

0.22

 aspectos

0.22

 sectors

0.22

 sections

0.22

Activations Density 0.722%