INDEX

Explanations

turned out surprisingly well

unexpected outcome or realization

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

гают

0.43

 mohou

0.40

 diyor

0.40

 Jeśli

0.39

هرات

0.38

 harming

0.37

ₚ

0.37

하려면

0.37

也要

0.37

 infertility

0.37

POSITIVE LOGITS

 оказалось

0.73

 okaza

0.69

果然

0.65

 оказалась

0.60

 оказался

0.58

 proved

0.58

proved

0.57

 pleasantly

0.56

 resultó

0.55

WOW

0.53

Activations Density 0.190%