INDEX

Explanations

give the correct answer

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

ounty

-0.10

Ð¸Ð²Ð¾

-0.09

 somehow

-0.08

 fullest

-0.08

paralleled

-0.08

aum

-0.08

"display

-0.08

Ãºst

-0.08

hÃ´tel

-0.08

isser

-0.08

POSITIVE LOGITS

 indeed

0.09

ä¹Łç¡®å®ŀ

0.08

 Indeed

0.07

 valid

0.07

ç¡®å®ŀ

0.07

Indeed

0.07

ç¡®å®ŀæĺ¯

0.06

æĪĳä»¬éĥ½çŁ¥éģĵ

0.06

 hardcoded

0.06

Activations Density 0.019%