INDEX

Explanations

to

New Auto-Interp

Configuration

Prompts (Dashboard)

16,384 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ograms

-0.28

 Gros

-0.27

 ..↵↵

-0.26

 Manufacturer

-0.26

grams

-0.25

å¸ĥå±Ģ

-0.25

æ¡£

-0.25

åĢ¾

-0.25

 gram

-0.24

gx

-0.24

POSITIVE LOGITS

-add

0.26

combe

0.26

ÑĩÐ¸ÑģÐ»ÐµÐ½

0.25

flip

0.25

 vice

0.25

æĪĸè®¸

0.24

 onstage

0.24

éĢĸ

0.24

 audiences

0.24

å¹ķ

0.24

Activations Density 0.004%