INDEX

Explanations

"well" + positive characteristic

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

 well

-0.90

 хорошо

-0.89

 Bernhard

-0.76

.$_

-0.75

Vag

-0.75

 roma

-0.73

कृ

-0.71

zeitig

-0.70

Vag

-0.70

坂

-0.69

POSITIVE LOGITS

meisterschaften

0.85

"}")

0.80

adins

0.79

 getChild

0.77

kyard

0.76

akh

0.75

 що

0.75

\;

0.74

routines

0.73

getError

0.72

Activations Density 0.011%